Downloaded from [Link]

com/book/58964 by guest on 29 August 2025

 Downloaded from [Link] by guest on 29 August 2025
Science of Science
Downloaded from [Link] by guest on 29 August 2025
 Science of Science
Understanding the Foundations and Limits of
Science from an Interdisciplinary Perspective

Downloaded from [Link] by guest on 29 August 2025

A LE X A N D ER K R AUS S
 Great Clarendon Street, Oxford, OX2 6DP,
United Kingdom
Oxford University Press is a department of the University of Oxford.
It furthers the University’s objective of excellence in research, scholarship,
and education by publishing worldwide. Oxford is a registered trade mark of
Oxford University Press in the UK and in certain other countries
© Alexander Krauss 2024
The moral rights of the author have been asserted

Downloaded from [Link] by guest on 29 August 2025

This is an open access publication, available online and distributed under the
terms of a Creative Commons Attribution-Non Commercial 4.0
International licence (CC BY-NC 4.0), a copy of which is available at
[Link]
Subject to this license, all rights are reserved.

Enquiries concerning reproduction outside the scope of this licence should be sent
to the Rights Department, Oxford University Press, at the address above
Published in the United States of America by Oxford University Press
198 Madison Avenue, New York, NY 10016, United States of America
British Library Cataloguing in Publication Data
Data available
Library of Congress Control Number is on file with the LOC.
ISBN 9780198937371
DOI: 10.1093/9780198937401.001.0001
Printed and bound by
CPI Group (UK) Ltd, Croydon, CR0 4YY
Links to third party websites are provided by Oxford in good faith and
for information only. Oxford disclaims any responsibility for the materials
contained in any third party website referenced in this work.
 Contents

Introduction 1
1. Describing Science of Science 16
2. Biology of Science 23

Downloaded from [Link] by guest on 29 August 2025

3. Archaeology of Science 27
4. Cognitive Science of Science 33
5. Psychology of Science 38
6. Sociology of Science 41
7. Economics of Science 48
8. History of Science 53
9. Anthropology of Science 60
10. Methodology of Science 66
11. Scientometrics and Network Science 74
12. Computer Science of Science 79
13. Statistics and Mathematics of Science 83
14. Philosophy of Science 91
15. Linguistics of Science 101
16. Science of Science: An Integrated and Methods-Driven
Understanding of Science 104
17. Science of Science: An Integrated Field Grounded in
the New-Methods-Drive-Science Theory 114
18. The Limits of Science: An Overview 118
19. The Limits of Science: Grounded in the Boundaries of
Our Methods and Mind 128
20. The Limits of Science: Expanding the Limits by
Expanding Our Methodological Toolbox 144
vi CONTENTS

Conclusion and Implications 161

Appendix 169

References 170
Acknowledgements 184
Index 185

Downloaded from [Link] by guest on 29 August 2025

 Introduction

Science has driven our remarkable advances in modern society. But we do

not yet understand well some of the most fundamental questions about sci-
ence: What are the origins, foundations and boundaries of science? How have
we learned what we know about the world around us? Why is it that how we

Downloaded from [Link] by guest on 29 August 2025

advance science is poorly understood, even though it has an enormous influ-
ence on our lives through the medicine we take and the technology we use?
We go through our daily lives understanding little about the answers to these
fundamental questions about science and the present limits of what we know
about the world. Yet these are among the most important questions about
science and understanding the world around us. No consensus yet exists on
these fundamental questions, as researchers who do study them mainly do so
from their own disciplinary perspective. Answering these questions is essential
because the answers help us understand science better, reduce the constraints
and biases we face when doing science and make new scientific breakthroughs.
The answers also enable us to provide a comprehensive foundation for the
field of science of science: a field that holds the potential to understand the
foundations of science, its present boundaries and how to push those limits.
Improving science is very important for the simple reason that science
shapes nearly all aspects of our lives. Scientific advances are behind all the
technologies that dominate our lives, such as computers, smartphones, elec-
tricity and high-speed transportation that connect us to people and resources
around the world. Science has provided us with theories that explain how
our species evolved and the nature of matter and the universe. Science drives
our unprecedented prosperity and has enabled us to reduce disease, malnu-
trition and poverty more than we ever have in human history. Yet science has
also enabled us to endanger our own existence, through challenges like cli-
mate change and overpopulation. And we turn to science again to make new
breakthroughs that can help tackle these challenges. During the coronavirus
pandemic, scientists directly steered society and the decisions of governments,
with much public debate about the role of science in our lives.
So it is key for our success as a species and for improving the conditions
of our lives that we understand fundamental aspects of science: What is sci-
ence? How did we start science? Why did science evolve the way it did? What
forces drive science and new advances? What are the present limits of science

Science of Science. Alexander Krauss, Oxford University Press. © Alexander Krauss (2024).
DOI: 10.1093/9780198937401.003.0001
2 SCIENCE OF SCIENCE

and how can we push those limits? It is key that we understand the what, how,
when and why of science. But most scientists are generally busy doing science
and do not have the time to study what drives science, the constraints facing
the methods and instruments we use or how we can do science better. Scien-
tists focus on producing results, evidence and theories. Rarely do they have the
time to take a step back and study deeper questions about the foundations and
present boundaries of science and what we know—about what enables and
constrains the knowledge we acquire and the methods and instruments we
use to acquire that knowledge. This book addresses these foundational ques-

Downloaded from [Link] by guest on 29 August 2025

tions about what makes science possible and how we can better conduct and
advance science.
When attempting to answer these questions, most researchers take one
individual disciplinary perspective—studying only the history of science,(1,2,3)
scientometrics,(4,5,6,7,8,9) sociology of science,(10,11,12,13) philosophy of
science(14,15,16,17,18,19) or psychology of science.(20,21,22,23,24) But science
is a highly complex and multidimensional phenomenon and we cannot
understand science from just one disciplinary perspective. To understand
the environment and climate change for example, environmental sciences
combine methods and evidence from ecology, physics, chemistry, physical
geography, natural resource management, economics and atmospheric
science. To understand the human body, medical sciences integrate meth-
ods and evidence from biomedicine, genetics, physiology, epidemiology,
neuroscience, nutrition science and biostatistics. Taking an interdisciplinary
approach is also the only way we can comprehensively understand a system
as complex as science and its foundations and limits. There is no way around
it. As we have not yet taken an integrated approach to studying science, we do
not yet understand well what the foundations of science are and what drives
science—at least not nearly as well as we understand the foundations and
driving forces of the environment, the human body and other complex phys-
ical, biological and social phenomena. The best way we gain comprehensive
knowledge is by using diverse methods from different fields that grounds our
knowledge in those separate strands of evidence, as each method captures the
phenomenon differently and strengthens our understanding when evidence
is coherent across the different methods.
From an interdisciplinary perspective, when we look into any field study-
ing science it can appear astonishing that the approach to studying science
and the most studied and thus most important feature of science can vary
so widely across isolated fields—and still be the ‘most important’ feature of
science. For economists, the central feature has commonly been funding and
 INTRODUCTION 3

incentive structures for rewarding and advancing science.(25,26,27,28) For statis-

ticians, it has been the methodological constraints and biases in designing
and improving our experimental studies in science.(29,30) For scientometri-
cians and network scientists, it has been scientists’ number of citations and
publication records.(4,5) For sociologists of science, it has been the role of
social influences on science.(10,11) For historians and philosophers of science,
it has been the role of scientific theories including models,(1,14,15,18,31) and
so on. These are all important aspects of science, but they have not yet been
integrated together into a coherent account.

Downloaded from [Link] by guest on 29 August 2025

Consider the most well-known historian of science Thomas Kuhn and
the most well-known philosopher of science Karl Popper. Kuhn dedicated
his research to investigating the evolution of science, focusing on how sci-
entific theories evolve. For him, science is not cumulative but undergoes
paradigm shifts—that is, foundational changes in theories. Kuhn’s account
of science is arguably the most well-known account of science yet proposed.
Despite widespread debate since Kuhn, the nature and growth of science
remains elusive. Kuhn studied particular theories mainly in physics up to
the early 20th century to develop his hypothesis of paradigm shifts.(1,32) Yet
here we shift the focus across fields and to scientific methods and instru-
ments used to do science, which illustrates the vast accumulation of scientific
knowledge. Major scientific methods and tools used across fields, such as
mathematics, microscopes and X-ray methods, and major scientific fields,
such as physics, geography and genetics, have not been entirely discarded.
And it is difficult for us to imagine that they could be. We instead constantly
expand and refine them, which demonstrates the cumulative advancement of
science.
Popper in turn adopted a different approach to the study of science and
attacked Kuhn’s account of non-cumulative science and constant paradigm
changes, describing it as a ‘lunatic fringe.’ For Popper, the only scientific
methodology needed and the defining criterion of what science is is whether
we can falsify (refute) a scientific theory or not.(15, 33) This principle of falsi-
fiability states: if we are not able to falsify a theory (such as string theory or
a theory of astrology), then the theory is not scientific. This normative prin-
ciple, for Popper, should define scientific investigation and the evaluation of
theories. But it has not had much effect on the scientific community outside of
philosophy of science. Falsifiability also tells us little about questions such as
why science evolved the way it did and how we can best advance science in the
future. Popper also argued that there is no logical structure of how scientific
breakthroughs emerge.(34) Here we will provide integrated answers to such
4 SCIENCE OF SCIENCE

questions by adopting a holistic approach that goes beyond the perspective of

any one discipline.
Consider also the work of network scientists who study science by lever-
aging the methods of scientometrics and big data. Scientometrics, a scientific
discipline using large-scale data, studies science by analysing features of scien-
tific publications and citation counts (how many times a publication is cited
by other publications as a measure of its impact). Scientometricians including
network scientists study topics such as scientists’ collaborations, careers, pro-
ductivity and networks.(4,5,8,9,35,36,37,38) This big data perspective to science

Downloaded from [Link] by guest on 29 August 2025

has generated important insights about the dynamics and outputs of science.
But it has not yet been able to explain well the general origins and limits of
science and what drives science and the methods of science. These researchers
are aware that the success of the field ‘depends on us overcoming traditional
disciplinary barriers’ and are aware of the limitations of using scientometric
methods to study how science works and advances: ‘this bias toward citations
is reflective of the current landscape of the field, [and] it highlights the
need to go beyond citations as the only “currency” of science.’(4) cf. (35,37,39)
Critics have also highlighted the dangers for the scientific community of
overrelying on scientometrics that has generated a publish-or-perish sys-
tem of adverse incentives and an increase in lower quality research across
science.(37)
There is much research on science and these are just three examples among
the range of different subfields studying science. These leading researchers,
like those in other fields such as psychology of science and economics of sci-
ence, provide important insights into our understanding of science. Yet each
adopts their own perspective and focuses on their own area of research. This
is observed looking at who they cite and who cites them—largely researchers
within their particular subfield. No integrated account of how science evolves
and advances has yet been developed that is coherent across fields. We cannot
address most questions about the nature of science by adopting the com-
mon unidisciplinary approach—for example taking a scientometric approach
that studies citation patterns. We depend on cognitive science to be able to
explain how our cognitive abilities enable us to perceive, reason, do science
and acquire knowledge about the world, and how our mind and senses present
constraints on how we perceive and understand the world. We rely on method-
ology and statistics to be able to explain how we advance science by reducing
such constraints and developing new methods and instruments, such as new
experimental techniques and radio telescopes, that expand our scope to the
world, and so on.
 INTRODUCTION 5

Because researchers studying science commonly address one aspect of sci-

ence from one perspective,(37) this has led to blind spots in existing accounts of
science. This book aims to shift this debate that, to date, has taken place within
individual fields—not across fields. The interdisciplinary approach we take
here enables us to understand the interconnected factors (methodological,
social, cognitive, financial etc.) that make science possible and constrain sci-
ence. As we will see, different factors influencing science, highlighted as highly
important within a field, turn out to be much less important when compared
against the range of factors across all fields. This has important implications

Downloaded from [Link] by guest on 29 August 2025

on understanding what actually drives science most and on shifting the focus
of our research in a number of fields.
Disciplinary specialisation is important and has provided much of what we
know about the world. Yet we also need a meta-approach that pulls the dis-
parate pieces together and provides us with an overall picture of science. Only
then can we establish which factors are most and least important across fields
in understanding and advancing science, and identify the common mechanism
underlying the different fields. The central shortcoming in existing studies is
that despite generally viewing the field from their own disciplinary viewpoint,
researchers still attempt to explain, often entirely, what drives science. Here
we broaden the scope of science of science by integrating the range of sub-
fields outlined above and others that have been largely, or almost entirely,
disconnected from the other fields studying science to date—from method-
ology of science,(30,40,41,42) economics of science(25,26,27,28,43) and computer
science of science,(44,45,46) to cognitive science of science,(47,48,49,50) biology
of science,(51,52,53,54) anthropology of science(55,56,57,58,59) and archaeology of
science.(60,61,62) cf. (4,5,35) The central challenge of science of science is account-
ing for and integrating the existing empirical and theoretical knowledge and
different perspectives from across disciplines into a holistic field. Only this
way can we then build on that integrated knowledge. This has also tradition-
ally been the central challenge towards greater understanding in related fields
like philosophy of science. This book aims to address this challenge.
We will see here how our evolved mind and sensory abilities (to observe,
experiment and process information) make doing science possible but also
shape what and how we observe. Our scientific methods and instruments
(such as statistics and electron microscopes) enable us to study a much
broader set of phenomena, but also have constraints to how we measure them.
Institutions, funding and societal challenges help influence what knowledge
and research methods we produce, distribute and use. Scientific norms and
methodological assumptions shape the way we evaluate our evidence, among
6 SCIENCE OF SCIENCE

other influences. A wide range of fields have insights that are very relevant in
explaining aspects of science. This book is about fitting the pieces of this large
puzzle together. Researchers studying science largely did not know that many
pieces or even the puzzle existed. As we will see, some of the pieces of the puz-
zle are much bigger than others—that is, some fields are much better able to
explain aspects of science than others.
This book aims to provide a unified framework for the field of science of sci-
ence by combining methods and evidence from across the natural, behavioural
and social sciences and thus offers a comprehensive understanding of how

Downloaded from [Link] by guest on 29 August 2025

we drive knowledge and science, what shapes their limits and how we can
improve them. The hope here is that the science of science will emerge as an
integrated field that incorporates the range of subfields that study science, just
as the environmental sciences and medical sciences are integrated fields that
help overcome previously fragmented and disconnected areas of knowledge—
which defines, to date, how we have studied and understood science. Within
this integrated field, we can better comprehend the origins, foundations and
limits of science. Understanding its limits is inseparably linked to understand-
ing its origins and foundations, and vice versa. Identifying the most important
features driving science and the limitations facing science is also important
because it enables any scientist or scientific institution to direct greater atten-
tion to those features that are best at advancing science and to addressing
those limitations. A better understanding of science can help scientists bet-
ter identify bottlenecks and become aware of their own constraints and blind
spots in doing science and how to address them. A number of insights pro-
vided throughout the book can even be directly applied by researchers in their
own career—insights on increasing one’s research productivity, questioning
the assumptions one adopts and how one can develop new methodological
techniques to do science better.
A central argument throughout the book is that to understand science com-
prehensively we need to integrate evidence of the abilities and conditions that
have enabled us to develop science (biological, cognitive, social and method-
ological), the abilities and conditions shaping the scope of science (including,
in addition, historical, economic etc.) and, most importantly, the abilities and
conditions allowing us to expand the present limits of science (mainly method-
ological and instrumental but also cognitive, sensory and social). The account
of science presented here explains how the scientific methods and instruments
we develop play a central role in shaping the foundations and present limits of
science by setting the boundaries of how we are able to observe, measure and
experiment—that is, how we do science.
 INTRODUCTION 7

This methodological toolbox of ours sets the scope and present limits of
what we can know and what is possible in science—while economic, social
and historical influences help shape what we study within that scope and those
limits. Methods and instruments we develop are the main mechanism through
which we develop new knowledge and make new advances—while the role
of other factors that can foster science (such as funding, incentive structures
and the scientific community) vary widely; we observe in ground-breaking
scientific publications that teams can be small or large, low or high funded,
young or old, at low and top ranked universities, or interdisciplinary or not.

Downloaded from [Link] by guest on 29 August 2025

On the one hand, we thus have our mind’s internal methodological abilities
to observe, experiment and process information (our universal methodolog-
ical toolbox). On the other, we have the sophisticated external methods we
have developed using our mind, such as statistics, radar telescopes and X-ray
devices (our adaptive methodological toolbox). Taken together, our method-
ological toolbox has enabled us to develop and advance science, medicine and
technology. We will show how scientific methods underlie the different factors
explored across all subfields studying science. Our methods are the best way
to bring together the different subfields and many are already interested in the
important role of methods and tools. Given that the foundations and present
limits of science are largely determined by our available scientific methods
used to study the world, a central focus on scientific methods is essential to the
integration of the science of science—as we will see.

The (methodological) tower of science: a holistic framework

for science

We can think of science as a massive (methodological) tower of science that

consists of three elements. The first element of the tower is its foundation
at the bottom that is made up of human abilities—our senses and cogni-
tive abilities—which enable us to observe the world around us and solve
problems, test hypotheses, imagine and reason causally. As we are all born
with these methodological abilities, we call this here our universal (built-in)
methodological toolbox.
The second element of our tower of science are the different floors of the
structure, and each major scientific method and instrument we have invented
to improve our universal methodological toolbox represents a different floor.
There is a floor on which the microscope is built, another on which statistics
are built, another on which X-ray methods are built, and so on. The collection
8 SCIENCE OF SCIENCE

of all our methods and tools of science makes up the structure of the tower of
science. This includes all scientific tools that vastly extend our visual scope of
the world, such as microscopes, telescopes, X-ray methods and spectrometers.
It consists of all quantitative methods that greatly enhance our cognitive abili-
ties to process complex phenomena and study vast amounts of information,
such as statistical and mathematical methods and controlled experimental
methods. It encompasses all instruments that enable us to separate previously
inaccessible substances like proteins, DNA and viruses, such as the centrifuge,
electrophoresis and chromatography, and so on. Making such methodological

Downloaded from [Link] by guest on 29 August 2025

innovations requires us to build on the foundation of the tower (our univer-
sal methodological toolbox) and on its existing floors—that is, on previous
methodological innovations (our adaptive methodological toolbox).
This means that electron microscopes are useful as they enhance exist-
ing light microscopes, and light microscopes are useful as they vastly extend
our evolved ability for vision. The same applies with our basic sense of
quantity we have used to develop arithmetic, then statistics and eventually
big-data methods, and so on. Each of these upgrades to our methodological
toolbox immensely expands our ability and scope to study and understand
phenomena—opening up the world of atoms, cells, molecules, ecosystems and
galaxies. The first two elements of the tower make up the foundation and struc-
ture of science, without which we would not be able to conduct, develop and
advance science. Without them, science would not exist.
The third element of the tower of science are the rooms of different scien-
tific fields that are commonly made up and defined by the particular scientific
methods and instruments that are applied. Each room is a body of knowledge
we acquire using particular tools of science. That is, bodies of knowledge are
(methodologically) organised in rooms. Cell biology for example was devel-
oped as a scientific field after inventing the electron microscope, and molecular
biology after developing X-ray diffraction methods. A number of biology’s sub-
fields are on the floors of the tower that rest on microscopes, centrifuges and
X-ray methods. Yet the rooms on a given floor also rest on other foundational
floors below it, with methods like statistics and controlled experimentation
near the bottom of the tower since most experimental research across science
heavily relies on them. Complex knowledge and fields commonly build on
multiple methods and instruments.
The shape and form of our tower of science has been continually extended
over time. A few hundred years ago, the tower was much smaller. With the
invention of foundational methods and instruments of the 17th century the
central floors of the tower were built and enabled researchers to continue
 INTRODUCTION 9

expanding the edifice of science. Each new major scientific method and instru-
ment we design has amplified our scope and understanding of the world.
Six new revolutionary methods and instruments developed around the 17th
century brought about most major discoveries at the time: the first micro-
scope was developed in 1590, telescope in 1608, barometer in 1643, air pump
in 1659, statistics in 1663 and calculus in 1675 (Figure 0.1). The pioneering
scholars Galileo, Hooke, Boyle and Newton and their contemporaries each
applied one or more of these new tools of science to expand our understanding
in astronomy, biology, physiology, pneumatics, mechanics and optics. These

Downloaded from [Link] by guest on 29 August 2025

methods, including instruments, in their extended form, remain central and
foundational to science today.
The tower houses the entire scientific community—about nine million
researchers presently globally.(63) Funding also helps expand the tower and the
shape it takes. It can foster the construction of new floors and rooms with new
sophisticated methods and instruments and fill the shelves with new books of
knowledge.
With each new major method and instrument we develop, a new floor is
constantly being built onto the tower and existing floors are restructured. The
shape and height of our tower of science is determined by the range of scientific
methods and instruments we have invented thus far that enable us to study and
understand the world. The strength of the tower is determined by the reliabil-
ity of our best tools of science. We redraw the borders of the tower and thus the
present borders of science as we expand our methodological toolbox, and they
are also influenced by basic factors such as the size of the scientific community
and available funding, especially in some research fields. Like filling the miss-
ing chemical elements of the periodic table of elements that make up the world,
we continue to fill in more and more of the floors and rooms of the tower. But

dp
F (t) =
dt

microscope telescope barometer air pump statistics calculus

Figure 0.1 The (methodological) tower of science in the 17th century.

10 SCIENCE OF SCIENCE

its future shape and form are not predetermined and we will later discuss the
possibilities for expanding and stretching science in new directions.
The present shape of the tower of science is depicted here with central
methods and instruments of science (Figure 0.2). These powerful tools have
revolutionised science, making up the foundational and most important floors
and pillars of the tower.
Our tower of science is today’s version of Egypt’s Great Library of Alexan-
dria in the 2nd century BCE which aimed to gather all existing knowledge in
the world. As we go through the chapters of the book, we will go on a tour of

Downloaded from [Link] by guest on 29 August 2025

the tower of science, learning about how it is built, what conditions shape it,
how it evolves and how we can improve it. As we will see, different subfields of
the science of science provide different evidence and insights into the three ele-
ments of the methodological tower of science: the foundation of the tower (our
universal methodological toolbox), the floors (our adaptive methodological
toolbox) and the rooms (our scientific fields grounded in our methodological
toolbox). To date, researchers who study science work on different floors and
largely within isolated rooms—with most only publishing in journals within
their own field. Yet studying science by making observations out of one win-
dow, as we will see, provides an understanding of science from one perspective.
We will illustrate how to restructure the tower, so all researchers studying the

Floors of the tower:

central methods and instruments of contemporary science

Rooms of the tower:

all bodies of knowledge
and scientific fields
grounded in our methods
telescope chromatography
centrifuge

electrophoresis
X-ray diffraction/
methods thermometer
laser

Foundation of the
tower:
the methodological
abilities we are born statistics
with: microscope/
observation, problem solving, imagination etc. spectrometer
electron microscope

Figure 0.2 The (methodological) tower of science in contemporary science.

 INTRODUCTION 11

same topic—science—work on joint floors to gain a much more holistic view

of science that complements and does not compete with each other (Chapters
16 and 17). Such an integrated approach will help us tackle our constraints
and blind spots in the way we presently understand science. We will also later
outline the different pathways for how we can further develop the tower in the
future—that is, the pathways for the future of science and how we can push
the present limits of science.

Downloaded from [Link] by guest on 29 August 2025

How the book is organised and the central argument

Let us provide an overview of the book’s structure. We will first describe

the integrated field of science of science by combining methods and evi-
dence from 14 relevant fields that are grouped here into four areas:
internal factors (biology, cognitive science, psychology and linguistics), exter-
nal factors (economics and sociology), historical and cross-cultural factors
(history, anthropology and archaeology) and meta-level and methodologi-
cal factors (methodology, scientometrics/network science, computer science,
statistics/mathematics and philosophy) (Figure 1.2 and Chapter 1). We identi-
fied these subfields of the science of science using the criterion of whether they
help explain the origins, foundations or limits of science. Fields like physics
and chemistry are not included as they are not as directly relevant; for science
is a complex system in which we humans and the methods we apply are at
the centre, and thus human sciences and methodological fields provide most
insight, as we will see. By pulling these fields together, we provide here an
overview of the literature across science that enables readers to gain a compre-
hensive understanding of science, on the one hand. On the other, it enables us
to identify the foundational role of scientific methods at the centre of the field,
explaining how methods are connected to factors across all 14 disciplinary
perspectives (Chapters 2–15).
The ultimate test of a successful account of science is how useful it is in
understanding science—in providing a comprehensive understanding of the
foundations, boundaries and advancement of science. On this test, the holistic
and methods-driven account presented here can take us further than individ-
ual fields of study alone (such as psychology of science, scientometrics and
history of science) (Chapter 16). What emerges is an integrated and coherent
theory, the new-methods-drive-science theory, that explains how we develop
new scientific ideas and discoveries through the new methods and instru-
ments we develop and apply collectively. The book aims to offer a unifying
12 SCIENCE OF SCIENCE

theory and foundation for the field of science of science (Chapter 17). With this
integrated understanding of science, we then outline the present boundaries
of science and how to push those boundaries. We also look forward and lay out
the possible pathways that the future of science holds (Chapters 18–20). In the
Conclusion we then draw implications for the field of science of science. We
outline how we can help establish the field, how we can better measure science
and discoveries with a broader set of methods and, most importantly, how we
can foster ways to advance science.
In a nutshell, we can summarise the central argument of the book as follows:

Downloaded from [Link] by guest on 29 August 2025

the different subfields that study science each capture one aspect of science, its
evolution or its boundaries, but not the whole story. We need to integrate the
different subfields to reveal the bigger picture of science. Each subfield has
something specific to contribute to our understanding of science, and of sci-
entific methods and instruments as the foundation of how we conduct and
advance science (Chapters 1–15). To integrate the subfields, we need to bet-
ter focus on scientific methods and instruments because they are the common
thread where the subfields all overlap. When we then combine these different
disciplinary perspectives, we learn that our methodological toolbox—which
consists of our best microscopes, mathematical methods and X-ray methods—
is the main driver of science by enabling us to perceive, measure and explain
the world in new ways (Chapters 16 and 17). Within this methods-driven
framework, we can see how the present limits of our methodological toolbox
largely account for the present limits of science (Chapters 18 and 19). We need
to thus expand our methodological toolbox to expand the present limits of
science and the research frontier (Chapter 20).
In terms of the research methodology, explaining the foundations, limits
and advancement of science requires adopting a cross-disciplinary frame-
work that integrates methods from across the natural and social sciences.
Evidence is derived here from studies using large-scale statistical analysis,
experiments, surveys of scientists, historical analysis, big data analysis and
other data sources. We take advantage here of studies using different quan-
titative and qualitative methods to examine the complexities of science.
Each method provides evidence for different aspects of science. Importantly,
the independent strands of evidence derived from the different methodolo-
gies are consistent with each other and, when pulled together, enable us
to understand science more coherently and uncover which factors are most
important.
While there are many advantages to quantitative studies, there are also cases
in which a broader, synthesising study that adopts a qualitative approach is
 INTRODUCTION 13

needed. To be able to provide a comprehensive overview of a multifaceted

field like science of science requires pulling together insights from the diverse
subfields studying science across the sciences and humanities. In doing so,
qualitative studies enable exploring complex phenomena and patterns across
those diverse disciplines. They can contribute to a holistic and nuanced under-
standing of the different factors influencing scientific processes and outcomes
by integrating them together—from the evolution of ideas and methodologies
to scientists’ motivations and social norms. Qualitative studies can provide a
bridge between disciplines and foster interdisciplinary understanding. Over-

Downloaded from [Link] by guest on 29 August 2025

all, they can enable us to better understand the broader context influencing
the foundations of science. A companion book, The Motor of Scientific Discov-
ery, analyses over 750 major scientific discoveries including all Nobel-Prize-
winning discoveries and the methods and instruments used to make them
and provides statistical evidence of how we drive science and discovery—and
complements this book with such quantitative data.(64) Together, this qualita-
tive and quantitative work enables developing this theoretical and conceptual
framework that integrates diverse insights into the study of science.

Defining science and outlining the personal motivation

for the book

A book on science needs to clarify what is meant by science, so we briefly

do that here. The scope and complexity of science will become clearer as we
progress through the chapters. Different views exist about what is involved in
the seven-letter word science—with its Latin origin scientia, meaning knowl-
edge. Commonly, science is defined as the study of the ‘world through obser-
vation, experimentation, and the testing of theories,’ according to the Oxford
English Dictionary.(65) Yet scientists engage in many activities: they identify
new research problems, design and conduct experiments, collect and analyse
data, reveal new patterns in datasets, use imagination and analogies, develop
explanations and theories, run simulations and make predictions, develop
methods and instruments, generate models, visually represent phenomena,
and derive implications for policy. Science is driven by many methodological
approaches and redefined here as follows: science is the study of the natu-
ral and social world by using our cognitive abilities (including observation,
experimentation and problem solving) and the methods and instruments we
develop (including statistics and microscopes) with the aim of describing,
explaining, predicting and controlling phenomena. Scientific methods include
14 SCIENCE OF SCIENCE

statistical techniques, controlled experimentation and algebra, and scientific

instruments (tools) include particle accelerators, electron microscopes and
electrophoresis (not other features of science such as concepts, theories and
language).
A further clarification: since this book offers a different approach and views
that are at odds with the disciplinary approach to studying science in exist-
ing studies, we briefly sketch out here the evolution of this book and its
interdisciplinary and methods-driven approach. With my PhD completed, I
worked with governments and the World Bank for five years doing research in

Downloaded from [Link] by guest on 29 August 2025

applied statistics, economics and behavioural sciences. This included quanti-
tative evaluations of the causal effects of public policies and reforms on topics
like health, education and energy. Over time, I became more aware of the lim-
itations, assumptions and biases of leading scientific methods and aware of the
problems related to the causal results we derive from experiments and simu-
lations. On the one hand, I increasingly began to question the foundations of
the methods and evidence we use. I was constantly confronted with questions
about what we can know with which level of certainty, what robust evidence is
and how we can improve the way we do research. That is, I was confronted with
questions about the foundations and present limits of our knowledge and the
methods we use to develop that knowledge. On the other hand, I became aware
that the common approach to scientific questions of using one method and dis-
ciplinary perspective will not take us very far in understanding and addressing
complex challenges—whether malnutrition or climate change. These are com-
monly tackled from an interdisciplinary perspective in the applied research
environment of governments and the World Bank.
I began to question why scientists in academia, in contrast, commonly
approach a problem in a more narrow way rather than adopting multiple
methods and an interdisciplinary approach that allows for more reliable and
coherent evidence grounded in the different independent methodologies. It
appeared that researchers’ training in a given method together with constraints
on time and resources drove the status quo. With the aim of addressing these
questions about science I then transitioned back into academia. This applied
research and concrete experiences have been central to adopting this book’s
big-picture approach that integrates methods and evidence across multiple
fields.
Finally, this book is written with a broad audience in mind, including any
academic and reader interested in understanding the origins, foundations and
present limits of science. In education systems we are largely taught the outputs
of science: facts, theories and laws about the world. We do not generally learn
 INTRODUCTION 15

about the foundations of science, how science works (the process of science)
and how to improve science. Yet with such an understanding, we are better
equipped to broaden the scope of science and advance science more broadly.
Within academia, this book could also be used as a textbook for a course on
Science of Science, Metascience, Methodology of Science, Science and Tech-
nology Studies, History of Science and also Philosophy of Science. More than
a general background and interest in science is not required and terms that
readers may not be familiar with are defined as they are introduced.

Downloaded from [Link] by guest on 29 August 2025

 1
Describing Science of Science

We first describe the existing landscape of the disparate fields studying science
and then outline what an integrated and unified science of science looks like by

Downloaded from [Link] by guest on 29 August 2025

providing a framework for the field. Science of science is an unexpected combi-
nation of terms at first glance. It involves scientists doing science to understand
science. When studying science itself, we thus practice the activity being stud-
ied. We use scientific evidence to explain what science is and how it works.
For methodologists and statisticians, it means studying the constraints and
assumptions of scientific methods that they themselves adopt. For sociolo-
gists and psychologists, it involves studying the biases and norms in science
that they themselves can be influenced by. For scientometricians and network
scientists, it means studying publications and citation patterns in science by
producing publications that they expect will be cited. For evolutionary biol-
ogists, it entails studying the evolution of our mind and its abilities to reason
and perceive the world that they themselves have also inherited.
Given the paramount importance of science in society and about nine mil-
lion scientists currently worldwide(63) it is surprising that there are only few
dedicated full-time to studying science itself and how to improve science and
its methods. It is surprising that science of science, as a discipline, did not
develop as science developed. There are not yet interdisciplinary journals or
university departments specifically dedicated to science of science. Yet across
the various subfields that study science, the vast majority of existing publi-
cations cluster in five areas of knowledge, with the largest concentrations in
philosophy of science (31%), history of science (25%), scientometrics/network
science (16%), cognitive science of science (9%) and sociology of science
(5%) (Figure 1.1). That these five subfields account for about 86% of all
publications is a historical contingency. As we will see, they do not intrinsi-
cally provide a more comprehensive understanding of the foundations and
limits of science than other subfields, like methodology of science and cog-
nitive science of science. Each subfield contributes in part to a richer and
more integrated understanding of science by studying a different aspect and
providing one perspective to the larger picture. Overall, only about 3% of pub-
lications studying science use the term ‘science of science’ (or ‘metascience’

Science of Science. Alexander Krauss, Oxford University Press. © Alexander Krauss (2024).
DOI: 10.1093/9780198937401.003.0002
 DESCRIBING SCIENCE OF SCIENCE 17

40% 11000

10000
35%
9000
30%
Share of publications (out of 100%)

8000

Number of publications
25% 7000
Philosophy, 31%
6000
20%
History, 25% 5000

15% 4000

Downloaded from [Link] by guest on 29 August 2025

Cognitive science, 9% Scientometrics/network 3000
10% science, 16%
Sociology, 5%
2000
5% Science of science, 3%
Linguistics, 2% Statistics/math, 4%
1000
Methodology, 1% Anthropology, 0.4%
Economics, 1%
0% 0
Biology, 2% Psychology, 1% Computer science, 0.2% Archaeology, 0.1%

Subfields studying:
internal external metalevel and historical and
factors factors methological factors cross-cultural factors

Figure 1.1 Share and number of publications across the subfields of science of
science.
Note: data are derived from Scopus (the largest citation database of scientific journals), and reflect
estimates for all existing publications—up to early 2024—within each subfield.(66) The estimated
shares add up to 100% of publications across all 14 subfields including the 3% of publications using
the term ‘science of science,’ ‘metascience’ or ‘metaresearch’ (on the far right). The shares provide a
rough estimation of the distribution of research across fields provided in Scopus while they do not
capture all publications in each research area. For interested readers, a note in the Appendix outlines
how the shares are calculated.

or ‘metaresearch’) (Figure 1.1). As an indication of limited integration in the

field, very few publications across all subfields use the common terms interdis-
ciplinary, cross-disciplinary or multidisciplinary, at less than 3% in total. For
example, within the over 8000 publications in ‘history of science,’ only 3.5%
mention the term ‘inter-/cross-/multidisciplinary’ and 0.1% the term ‘science
of science.’ Within the over 5000 publications in ‘scientometrics,’ only 12% and
0.6% mentioned these terms respectively.(66)
Different researchers studying science use a different method and unit of
analysis and thus study different features of science. Leading scientometri-
cians and network scientists like Fortunato, Wang and Barabási have focused
on and stressed the key role of publications and citations;(4,5) leading histo-
rian of science Kuhn, the paradigm shifts in scientific theories;(1,32) leading
philosopher of science Popper, the evaluation principle of falsification of sci-
entific theories;(14,15,31) leading sociologists of science Latour, Woolgar and
Bourdieu, the social practices of scientists;(10,11) and so on. Such disciplinary
18 SCIENCE OF SCIENCE

isolation has led to simplified and at times contradictory views. Leading

researchers studying science have not studied how important, relatively speak-
ing, the particular ‘key’ factor is that they study and how it relates to the
other ‘key’ influencing factors in different fields which they do not study.
We will later assess the role of each factor using the criteria of its scope
in explaining the foundations, limits and advancement of science, and the
direct influence we have on that factor in shaping science. Using these two
criteria, we observe that these central factors proposed by the most cited
researcher studying science within a particular field—namely Kuhn in his-

Downloaded from [Link] by guest on 29 August 2025

tory of science, Popper in philosophy of science, Latour and Woolgar in
sociology of science, etc.—are not able to explain as much or have direct
influence compared to other factors (Chapter 16). These leading researchers
have overinterpreted the particular role of the factor they study compared to
other factors, especially the foundational role of our scientific methods and
our mind in enabling and constraining science that are not as commonly
studied.
Classic work in the early origins of science of science goes back at least to
Znaniecki in 1923,(67) Ossowska and Ossowski in 1935(68) and more generally
to Galton’s English Men of Science: Their Nature and Nurture in 1874(69) and
later de Solla Price’s Little Science, Big Science in 1963(70) and Zuckerman’s
Scientific Elite: Nobel Laureates in the United States in 1977.(71)
In Figure 1.1 we outline the landscape of existing research studying sci-
ence and its concentration in particular subfields. In Figure 1.2 in turn we
describe the unified approach to the field presented here that is needed to com-
prehensively understand science, combining the different bodies of research
and methods which, to date, have been fragmented and isolated from each
other. It provides an overview of what we will be covering in the chapters
to come. The set of factors, and thus fields, are grouped here into four
areas: internal factors (biology, cognitive science, psychology and linguistics),
external factors (sociology and economy), historical and cross-cultural factors
(history, anthropology and archaeology) and meta-level and methodologi-
cal factors (methodology, scientometrics/network science, computer science,
statistics/mathematics and philosophy).
Within the scientometric community that includes network scientists, the
field has however been viewed narrowly as the ‘field that relies on big
data to unveil the reproducible patterns that govern individual scientific
careers and the workings of science’ by studying primarily publications and
citations.(4) cf. (5,35,39) This common view among scientometricians illustrates
 DESCRIBING SCIENCE OF SCIENCE 19

Cognitive science

Biology Psychology
subfields studying
internal factors
Archaeology Linguistics

Anthropology Sociology
subfields studying

Downloaded from [Link] by guest on 29 August 2025

historical and cross subfields studying
-cultural factors external
factors
Our universal and adaptive
History methodological toolbox is at the
centre of science and underlies Economics
factors across all disciplinary
perspectives

Philosophy subfields studying Methodology

meta-level and methodological
factors

Scientometrics & Computer science

Network science
Statistics &
Mathematics

Figure 1.2 Science of science: explaining the foundations, limits and

advancement of science from 14 subfields—an integrated field that combines
multiple methods.
Note: this is the first study to demarcate the field of science of science by identifying the 14 relevant
fields that contribute to our understanding of science and it groups them into four areas. Internal
factors refer largely to the human body and mind; external factors refer to our broader environment;
historical and cross-cultural factors refer to the past and different contexts; metalevel and
methodological factors refer to metascientific aspects of science and scientific methodology. Other
research domains that can provide insight into science of science can be categorised in one of these
14 subfields—for example public policy is included in economics of science, communication sciences
in linguistics of science, data science in statistics, and the arts in terms of imagination and creativity
in cognitive science of science. Finally, cognitive science of science broadly covers our evolved
cognitive abilities and constraints (related to observation, memory and abstraction), while
psychology of science narrowly covers psychological biases and personality traits. For interested
readers, a note in the Appendix outlines which methods are commonly used in each subfield.

how they view the field in terms of one method. Though they recognise that
the success of the field ‘depends on us overcoming traditional disciplinary
barriers,’(4) there are a range of other disciplinary methods, evidence and per-
spectives from across the other fields studying science that are not taken into
account. Scientometrics has however dominated the research studying science
20 SCIENCE OF SCIENCE

in leading multidisciplinary science journals to date.(4,5,35,37) Figure 1.2 out-

lines what an integrated science of science, without disciplinary divisions,
looks like. The integrated field presented here can be defined as follows:

The field of science of science is the study of science, and especially the
foundations, limits and advancement of science and scientific methods, that
integrates methods and evidence from across the natural, behavioural and
social sciences.

Downloaded from [Link] by guest on 29 August 2025

The field addresses foundational, methodological and meta-scientific aspects
of science. It studies fundamental questions that span the scope of the field:
What drives science? How do we develop science? What constrains science?
How can we improve and advance science? What are the structures, pro-
cesses and dynamics that underlie the production and evolution of scientific
knowledge? More specifically, what particular factors drive scientific progress
and innovation? Many factors like methodological and technological advance-
ments, collaborations, diversity of perspectives, funding and incentives are still
not fully understood. How can we more accurately measure and evaluate the
impact of scientific research? Current metrics, such as citation counts, have
many limitations and do not capture the immediate impact of new ideas in sci-
ence or the broader impact of research on policy or society. How can we better
ensure that science has a positive impact on policy, society and people’s lives?
How can we improve and incentivise the replicability of scientific research?
How do cognitive biases and social factors influence scientific research, such as
the interpretation of data and the publication of results? How can we minimise
these biases and promote more objective and rigorous research? What are the
ethical implications of scientific research, especially in fields like gene edit-
ing and artificial intelligence? How can we better integrate interdisciplinary
approaches in scientific research? How can we promote more effective com-
munication and dissemination of scientific research, including on social media
and digital platforms? These are some of the important questions in the science
of science.
The field assesses the methods and instruments of science, the process
of science, how we design, implement and evaluate scientific studies, and
domain-specific topics from its subfields. This is done by applying a range of
methodologies. These include empirical studies, experiments, surveys of sci-
entists, historical analysis, big data analysis and conceptual analysis, which are
integrated across domains and subfields. The field studies science from across
disciplinary borders—and from both high altitude and the bottom up. The
 DESCRIBING SCIENCE OF SCIENCE 21

objective of the field is straightforward: by better understanding the founda-

tions and present limits of science and scientific methods, we can do science
better and drive new knowledge and discoveries. This is the field of science of
science. And given that scientific methods are at the foundation of science—
developed and used to conduct, advance and improve science—we need to
place scientific methods at the centre of focus to integrate the science of science.
No factor other than our scientific methods and instruments is implicit in all
14 factors and fields and can we more directly influence and improve to do
science and make new breakthroughs.

Downloaded from [Link] by guest on 29 August 2025

Interdisciplinary fields like environmental sciences are excellent examples
of why we can only understand a complex, multifaceted subject by adopting a
range of methods and evidence from across fields—just as in the case of science
of science. Environmental sciences are the study of environmental problems.
And defining, analysing and addressing such problems are only possible by
taking natural, social, economic, political, historical, psychological and other
dimensions into account simultaneously.(72) The natural dimension of envi-
ronmental sciences involves studying environmental problems that arise in
our natural environment through physical, chemical and biological interac-
tions and that are linked to our awareness of those problems and how we
measure them. Similarly, science of science studies how we measure nature
using our methods and instruments and how we have evolved in our natu-
ral environment that has shaped our abilities for observing, solving problems,
causal reasoning and acquiring knowledge. The social dimension of environ-
mental sciences studies how population growth and cultural norms can lead
to environmental problems. Similarly, science of science studies how popula-
tion size, cooperation, specialisation and societal influences shape science and
how we conduct science.
The economic dimension evaluates the role of natural resources, their
depletion and their interrelationship with economic growth. Similarly, science
of science evaluates the incentive structure of scientists and how economies
of scale foster cumulative knowledge. The policy dimension assesses how
we can mitigate environmental challenges and how international institutions
can coordinate initiatives. Similarly, science of science assesses the public
institutions that help plan, finance and manage how knowledge is produced,
distributed and used. The historical dimension analyses the environment on
a large timescale that includes geological and human history. Similarly, sci-
ence of science analyses the evolution of science, discoveries, methods and
their complexities throughout history. The psychological dimension exam-
ines how citizens perceive environmental challenges such as climate change
22 SCIENCE OF SCIENCE

and the level of willingness to change their behaviour. Similarly, science of sci-
ence examines our psychological biases and the scientific methods developed
to reduce them. Assessing the evolution of global warming and simulating its
future course requires using sophisticated statistical programmes and com-
puter technology. Similarly, science of science uses complex statistical and
computational methods to assess large datasets of discoveries, scientists and
publications. Importantly, environmental sciences, like science of science,
requires applying multiple methods to be able to capture the multiple factors
at different levels and thus multiple dimensions of the field.

Downloaded from [Link] by guest on 29 August 2025

The emergence of the environmental sciences enabled a much more com-
prehensive understanding of the environment by developing integrative meth-
ods like the dynamic integrated climate and economy (DICE) model. It has
enabled us to evaluate climate change and its multiple environmental, social,
economic and historical dimensions more realistically than with previous iso-
lated assessment models. For developing this integrative approach William
Nordhaus received the Nobel Prize. No such integrated methods and analyses
yet exist in science of science. Yet, commonly, attempting to understand the
multifaceted phenomenon of science from one perspective is like attempting
to understand the environment by just studying oceans or changes in tempera-
ture. We learn a lot, but the approach is deficient in providing a comprehensive
understanding.
Just as this interdisciplinary approach has already been adopted in such
fields, it is just as feasible and necessary to adopt it in the science of science.
We outline here a vision of science of science in which many, if not most,
researchers studying science can continue to pursue disciplinary specialisa-
tion, but all would begin to spend a share of their time to be broadly informed
about existing research on the same topic across the other subfields of sci-
ence of science. This way we can ensure our understanding of science is more
coherent—that is, coherent across the subfields of science of science. Just as
environmental sciences arose as an integrated field in the 1960s and 1970s to
be able to understand complex environmental problems,(72) the aim of this
book is that science of science arises as an integrated field driven by the same
need for a cross-disciplinary approach to be able to understand the complex
nature of science and discovery. In the next chapter we begin by assessing the
evidence derived from biology as we work through the different topics that
help us understand aspects of the foundations and limits of science.
 2
Biology of Science

What are the evolutionary origins of science and how can they help us under-
stand how we do science today? Like other animals, our ancestors evolved

Downloaded from [Link] by guest on 29 August 2025

abilities for vision and other senses that enable us to perceive the world, and
evolved other related physiological functions. These provide insights into
science’s evolutionary origins. Species like ours require making observations
and acquiring knowledge about the world to be able to survive and meet basic
needs. This requires knowing about what foods they can and cannot eat, and
about their ecological environment and other animals. They have to identify
regularities in nature.(60,73) Chimpanzees for instance forage using different
types of tools and techniques. They crack nuts using stone and wood ham-
mers and extend their reach and extract termites, ants and honey using sticks
(Figure 2.1). They gather water using leaf sponges, throw stones as weapons
and use levers for different tasks.(21,74,75,76) Using tools to solve problems,
chimpanzees thus have a toolkit that they acquire through social learning and
experimenting. To use such tools requires chimpanzees to have a clear objec-
tive of the tool in mind, predict how the tool can enable them to achieve that
objective and understand how the tool must be applied. They thus must com-
prehend the interactions needed between the physical tool, their hands and
the desired outcome, which illustrates their ability to acquire knowledge and
manipulate the world around them. Tool use is widespread among many ani-
mals, from crows to sea otters and octopuses, that manipulate objects for their
purposes.(74,77)
Chimpanzees are also able to think of reality abstractly and create and test
hypotheses about the behaviour of others, when they for instance deceive
others.(21) Non-human primates also reason about objects, space, quanti-
ties and the mental states of others, use classification systems and recognise
causal relationships.(21,74,78) Omnivorous mammals—from rats to humans—
for example, reason causally by perceiving similarities and patterns in the
world and basing their behaviour on them. They make for instance the connec-
tion between food with a peculiar taste and gastrointestinal sickness occurring
afterwards.(79)

Science of Science. Alexander Krauss, Oxford University Press. © Alexander Krauss (2024).
DOI: 10.1093/9780198937401.003.0003
24 SCIENCE OF SCIENCE

Downloaded from [Link] by guest on 29 August 2025

Figure 2.1 Chimpanzees reason causally when using tools like sticks to ‘fish’ for
termites.
Source: Mike R., Wikimedia Commons.

We can also better understand our constraints to studying the world by

comparing them with the evolved abilities of non-human animals. In some
ways, these abilities surpass our human abilities—and, in other ways, they
have abilities that we humans do not.(21,52) Different birds for example use
magnetic fields to orient themselves, perceive their altitude and create mental
geographic maps.(74) Certain chimpanzees have extraordinary working mem-
ory, particularly flash memory. This allows them to quickly memorise things
within less than a second (far exceeding humans), as experiments with chim-
panzees and humans illustrate.(21,74) Ultraviolet-seeing insects such as bees are
directed towards the centre of many flowers that have developed ultraviolet
streaks, allowing bees to find nectar and foster pollination.(80) Echolocat-
ing bats perceive their surroundings, including moving prey, using sound
waves.(50,74) We humans did not evolve these and other abilities to the same
extent. The sensory abilities of any species (including ours) thus only pick up
parts and fragments of information and energy in the world compared to all
the sensory abilities across millions of species.(81,82) We have a limited scope to
the world. To support our limited perception, we have developed methods and
 BIOLOGY OF SCIENCE 25

tools. These enhance different aspects of how we can perceive the world—as
we discuss in Chapter 10.
The parts of our surroundings that we, with our evolutionary adaptations,
are able to observe and sense do not thus reflect the only way to view real-
ity. Different species, with their own set of adaptations, survive using their
particular means of perception. Bees, gorillas, condors and humans each dis-
tinguish between types of plants and animals in different ways, and classify
geographic areas and their species’ behaviour in different ways.(81) There are
different ways to perceive and categorise phenomena in the world given that

Downloaded from [Link] by guest on 29 August 2025

there are different species in the world.
In fact, while doing science or even just reading this book, our brain cells
require sufficient water and food to function properly and retain informa-
tion, and we require sufficient sleep and warmth to be able to concentrate.
We depend on our environment and physical resources to be able to survive
and acquire knowledge about the world around us.
There is another biological dimension of science: our human perspective
to the world—as the biological animals we are—shapes what knowledge and
objectives we pursue as we use our mind and the methods we develop to do
science. Just by being human and members of our species, we direct more
attention to some phenomena than to others, namely phenomena that fall
within the environmental niche of the world we have evolved in and live in.
Nearly all scientists study aspects of reality relevant to our needs and wants—
human biology, human technology, human society, human diseases, human
behaviour and other problems and objectives we humans face. Science is
human-driven.
Large science funding agencies (like the European Commission, National
Science Foundation and other public funding bodies) generally require
researchers to outline the human impact of their research to be able to receive
funding. Most science funding worldwide is spent on studying human beings,
with for example 52% of total public research funding in the US allocated
to medicine/health, life sciences and psychology. The remaining 48% is allo-
cated to all other disciplines that also generally aim to benefit human beings
and human progress, including engineering, physical and social sciences, envi-
ronmental science and computer science.(83) So why do tens of thousands of
scientists worldwide work on explaining and predicting illnesses, pandemics,
population dynamics, financial markets, our behaviour and weather but only
few scientists on explaining and predicting phenomena such as dark energy
of the universe, the flora and fauna of the earth’s deep oceans and the mind
of insects? For one more directly impacts our lives while the other does not.
26 SCIENCE OF SCIENCE

And scientists study the latter topics often to the extent that it may benefit
and help better understand ourselves. We humans are our own point of ref-
erence. We want to enhance our conditions and lifespan but not generally
those of all living organisms, including all animals and plants, unless they ben-
efit us—as evidenced by the vast majority of research across the biological,
medical, behavioural and social sciences. Overall, we study the world from
our human perspective, our anthropocentric context, that shapes the present
scope of science.
In sum, smart animals are able to meet their needs and create knowledge

Downloaded from [Link] by guest on 29 August 2025

using their biological abilities for vision and their other senses that lay the basis
for our methodological abilities that we use to be able to do science (Chapters 3
and 4). Understanding these common abilities that our species shares to differ-
ent degrees with other species illustrates how we are able to develop knowledge
and thus what abilities have enabled us to start science.(82) These abilities make
up the foundation at the bottom of the tower of science that we will be recon-
structing as we go through the chapters. These abilities thus provide insight
into the foundations of science and help us understand aspects of our cur-
rent scientific landscape, including our research focus. But biology of science
alone is (just like scientometrics, psychology of science or any science of sci-
ence subfield) incomplete. This is because science is shaped by our mind and
the methods and instruments we use to do science, while also influenced by
our broader social, economic and historical context.
 3
Archaeology of Science

Archaeological artefacts that include increasingly sophisticated tools devel-

oped by early humans provide evidence for the origin of science. They offer

Downloaded from [Link] by guest on 29 August 2025

historical evidence of the evolution of our methodological abilities to reason
and acquire knowledge that are needed to construct those artefacts and to do
science today. We humans have evolved abilities to observe, solve problems,
experiment, categorise, reason causally and test ideas or hypotheses.(22,74)
These, together, account for our methodological abilities of the mind we use
to be able to develop knowledge and make sense of the world around us. Using
these abilities, early humans such as Homo erectus and Neanderthals created
complex tools such as hand axes at least about 1.5–2 million years ago.(84,85)
Making tools like hand axes requires the ability to imagine and plan what
they will look like before creating them. Early stone toolmakers needed to
make mental representations, inferences and predictions. Making stone tools
requires the ability to systematically observe, experiment, reason causally,
test a hypothesis, imagine and plan in order to produce a clear preconceived
object. cf. (86,61) Early humans demonstrated an incredible advance in control-
ling fire about 600,000 to 1 million years ago and developing sophisticated
fire-hardened spears about 400,000 years ago.(60,87) This requires reason-
ing complexly, interconnected inferences and evaluating hypotheses.(88,89)
These are also the evolved human abilities commonly used in contemporary
scientific practice.
To understand the foundations of science, we have to understand the
evolution and abilities of early hominines. This can present a challenge in
drawing conclusions from fossils and material artefacts about our early species’
abilities. cf. (90) Because the past is beyond the scope of experimentation and
because our ancestors’ cognitive abilities and methods do not fossilise, our
understanding of the past and their specific abilities can be constrained. Yet
we can extend our reach into the past and partly explain it by combining his-
torical evidence we do have with current evidence.(91) Given that we know
early human species developed such complex tools(60) and given that for us to
develop them requires a systematic approach and refining the tools, we know
they also used these methodological abilities (Chapter 4). No other way exists

Science of Science. Alexander Krauss, Oxford University Press. © Alexander Krauss (2024).
DOI: 10.1093/9780198937401.003.0004
28 SCIENCE OF SCIENCE

to explain the foundations of how early humans reasoned and gained knowl-
edge other than our methodological abilities of the mind they all required
using.
Turning to our species, we Homo sapiens are the evolutionary product of
millions of years and slowly emerged as a distinct species in Africa an estimated
250,000–300,000 years ago.(21,90) The way we perceive and view the world
is shaped by our evolution. The ability to abstract, imagine and use analo-
gies developed in our early ancestors over time.(49) This essential ability is
used by contemporary scientists when reasoning, using methods to represent

Downloaded from [Link] by guest on 29 August 2025

information about the world and developing theories. Cave paintings, sculp-
tures and other symbolic art created to represent phenomena abstractly during
the Palaeolithic can be seen in some ways analogously to models created to
represent phenomena abstractly in science.(60) Niels Bohr for instance used
our solar system as an analogy to develop a quantum model of the hydrogen
atom, imagining electrons in the hydrogen atom behaving like planets orbit-
ing around the nucleus. Charles Darwin modelled and illustrated his theory
of evolution using the branches of a tree, with all species being related and
humans just on one branch of the same tree.(51)
Abstraction and modelling reality are fundamental to how we reason and
simplify the complexity of the world around us. We use representational mod-
els of the world in most scientific fields—from theories and analogies, to
experiments, statistical simulations and mathematical equations. Our cog-
nitive and social abilities for greater cooperation, language, abstraction and
imagination have developed in a symbiotic way. They have built on each other
through continual cumulative feedbacks that have made these abilities increas-
ingly complex over time,(52,21) together with our abilities for developing more
complex methods and tools (Chapters 4 and 10).
Combining our abilities, we have become better able to develop a remark-
able set of increasingly complex methods and tools. We created for example
bone harpoons and a range of flake tools to hunt and cut things more eas-
ily. We built dwellings to provide protection from the elements of nature.(60)
We created what are viewed as early systems of notation, by using an ordered
set of spatially distributed marks engraved on stones and bones to record,
process and pass along information.(62) We produced symbolic drawings and
depictions to help document observations and provide an external source
of memory.(60) Developing such methods and tools requires (as in the case
of constructing a dwelling) the cumulative use of systematic observation (to
identify the best wood, mud and stones for construction). It requires causal
reasoning (to understand that a well-constructed dwelling will protect us
 ARCHAEOLOGY OF SCIENCE 29

from rain and predators) and experimentation (to identify better construc-
tion methods). It requires imagination (to conceive how the dwelling will look
before constructing it) and collectively planning within groups (to collect the
materials needed for its construction). These are the human abilities we use
across contemporary science, though more explicitly. Our early methods and
tools can be seen analogously to those we develop today—such as microscopes
and telescopes to extend our visual abilities (Chapter 10).
We eventually, using our methodological abilities, learned how to domesti-
cate animals, which requires us to have knowledge of biological reproduction,

Downloaded from [Link] by guest on 29 August 2025

the nutritional needs of animals and selective breeding techniques to foster
particular traits. We learned how to cultivate crops, which requires us to under-
stand the causal interactions between seeds, rain, soil fertility and erosion and
often knowledge of annual cycles. It requires us to continually experiment with
seeds to produce more productive yields and involves collective knowledge
about selective planting, storing seeds and often about irrigation methods.(92)
Early farmers had comprehensive botanical, zoological and ecological knowl-
edge and were able to control parts of their environment through agriculture
and livestock—as hunter-gatherer groups commonly still do today.(21,56,87,88)
As societies became more complex, we increasingly specialised in differ-
ent professions and crafts. An estimated 9000 years ago in the Middle East,
architectural improvements led to the development of rectangular buildings
constructed out of stone and mud bricks that were often two storeys. Some
of these well-designed buildings exist today. Such construction requires us
to understand geometric dimensions, the weight that different materials can
support and the need to heat limestone to temperatures of 750–850 ∘ C to be
able to produce plaster floors.(93) Constructing them is not possible without
extensive collective planning, systematic measurement and experimentation.
Such specialised professions enabled us to develop new bodies of knowl-
edge that cumulatively built on existing knowledge. In the same way, without
large communities of specialised professionals and cumulative knowledge,
contemporary science would not exist today.
With the emergence of the first civilisations, we made large leaps towards
science.(82) We developed the earliest known systems of written language
and mathematics around 6000 years ago.(92) The shift from oral to writ-
ten systems marked an important transformation that enabled us to use our
cognitive and methodological abilities increasingly systematically and explic-
itly. These systems make recording what we observe easier. They reduce our
cognitive constraints in processing and remembering information, making
mathematical calculations and building on existing knowledge.(21)
30 SCIENCE OF SCIENCE

The development of systems of writing reflects a pivotal historical shift, not

only in the cumulative development of science but also in understanding sci-
ence. Written language has enabled us to better comprehend the scientific and
technological advances of our ancestors, as they were able to describe them in
written form. Without a written record, we are limited to only studying other
archaeological artefacts.
In the civilisations of Mesopotamia, Egypt, China, Central America and
India, astronomy was for example a common area of knowledge developed
using systematic observation and written records. It involved sophisticated

Downloaded from [Link] by guest on 29 August 2025

astronomical models for making reliable predictions.(94) We learned that we
can use the sky as a clock for time keeping, a compass for orientating ourselves
and a calendar for planning agriculture, weather and temperature. Despite
acquiring this understanding, our early ancestors could not yet explain stars,
comets, the sun and other astronomical phenomena that largely remained
intellectual puzzles.
Early written use of geometry can be traced back to around 5000 years ago
in Mesopotamia and Egypt. It involved principles of areas, lengths, angles
and volumes that were identified empirically and used for surveying, agri-
culture, construction and astronomy.(95) Ancient Egyptian civilisation and
Norte Chico civilisation (modern Peru) constructed vast pyramids at least
4500 years ago (Figure 3.1). Building a pyramid requires—both then and
today—applying principles in engineering, architecture and geometry that
are grounded in systematic measurement, planning and experimentation.
Farming in such civilisations was improved by experimenting using con-
trols, namely testing different crops and comparing the outcomes, and thus
acquiring knowledge by experimenting comparatively. An Egyptian medical

Figure 3.1 Pyramids of the Norte Chico civilisation constructed using

systematic measurement, planning and experimentation, circa 2600 BCE.
Source: Kyle Thayer via Wikimedia Commons.
 ARCHAEOLOGY OF SCIENCE 31

textbook from about 1600 BCE provides detailed cumulative experimental

knowledge of dealing with injuries, fractures, tumours and various surg-
eries and it applies the methods of examining, diagnosing, treating and
prognosis.(96)
A controlled experimentation is for instance described in the Old
Testament.(97) The book of Daniel (1: 12–13) outlines an experimental trial
with control groups that tests the influence of a vegetarian diet: ‘Test your
servants for ten days. Give us nothing but vegetables to eat and water to
drink. Then compare our appearance with that of the young men who eat the

Downloaded from [Link] by guest on 29 August 2025

royal food [and drink wine], and treat your servants in accordance with what
you see.’ We require combining our cognitive abilities to test such a hypoth-
esis and conceive the design for such a controlled experimentation. Causal
reasoning is required to test whether a potential cause (in this case a vege-
tarian, water-based diet as opposed to a non-vegetarian, alcohol-based diet)
has an observable effect on people’s physical appearance. Carrying out a trial
is required, including systematically recording and comparing the outcomes
of the physical appearance between the two groups after 10 days, and then
deriving inferences from the trial outcomes to modify people’s diets in the
future. We then combined controlled experimentation with further method-
ological features such as randomisation and blinding in the 19th century (so
that participants do not know which group in an experiment they are assigned
to and are randomly allocated to groups) which further reduces human
bias.(98)
Overall, developing such technological, agricultural and architectural inno-
vations requires us to manipulate the world by applying our methodological
abilities of the mind that we have used collectively throughout human history,
including in contemporary science. Developing such innovations throughout
history—and even just surviving—requires us to be able to understand and
predict phenomena in the world. This involves knowledge of the character-
istics of physical objects and forces (intuitive physics), the characteristics of
animals and plants (intuitive biology), the behaviours and views of others and
our own (intuitive psychology),(99) the characteristics of places, directions and
shapes (intuitive geometry)(100) and distinctions between quantities (intuitive
arithmetic).
In sum, contemporary science builds on these intuitive conceptions of the
world and on our mind’s methodological abilities that are evidenced in vast
archaeological records. These have enabled us to reason and develop knowl-
edge and tools, and they provide evidence of what abilities have enabled us to
start science. The foundation at the bottom of the tower of science is grounded
32 SCIENCE OF SCIENCE

on these methodological abilities, which we will see in the next chapter also
shape the scope of how we do science today. Archaeology of science accounts
for one piece of the foundation of science and thus of science of science.
Understanding these methodological origins of science, as we will later see,
is important to understanding the current dynamics of science and how we
can best advance science.

Downloaded from [Link] by guest on 29 August 2025

 4
Cognitive Science of Science

Our mind makes doing science and creating knowledge about the world
possible. It allows for vision needed to make observations, memory to recall

Downloaded from [Link] by guest on 29 August 2025

what we observe, language to express what we observe, and reason to solve
problems and develop scientific methods. We have evolved these method-
ological abilities of the mind over time and we use them to be able to study,
experiment and acquire knowledge of the world. Our mind thus shapes how
we get by and make sense of our natural and social environment, on the one
hand. On the other, we face cognitive and sensory constraints imposed by
nature and evolution, and also constraints on the methods and instruments
we develop using our mind. They set the scope within which we are able to
do science and create knowledge about the world. Cognition thereby refers to
our ability to perceive, process, retain and act on information from the world
around us.(101) Because our mind makes reasoning and creating knowledge
possible, we can best understand our mind’s abilities by studying how they
have evolved to enable us to reason and create knowledge the way we do.
Our cognition, like all features of human anatomy and other natural phe-
nomena, is the outcome of our evolution. It is the outcome of a largely
unpredictable path of trial and error over the past few million years. This basic
fact helps understand our cognitive abilities and their limitations: our mind
has evolved largely reacting to problems we have faced up to now, within our
environmental and cultural niche.(21,48,52,87,102,103) Our mind has developed
over much of our history within the savannahs and natural landscapes around
us. It has evolved by making sense of those parts of reality (especially people
and our surroundings including plants and animals) that are relevant for our
species’ needs and survival and that we can access with our senses.(81,104)
Our mind has thus evolved over our species’ history in large part for the
pragmatic purpose to help us meet our basic needs. Our evolutionary his-
tory is one that did not directly involve observing and mentally modelling
phenomena such as the size and nature of the universe, the historical origin
of life, global financial markets and the emergence of conscious experience.
Only in more recent human history and only by creating methods and instru-
ments have we been able to develop such complex theories about phenomena

Science of Science. Alexander Krauss, Oxford University Press. © Alexander Krauss (2024).
DOI: 10.1093/9780198937401.003.0005
34 SCIENCE OF SCIENCE

that we do not have direct sensory experience of and that generally do not
directly affect our biological fitness. This has been one of the great mysteries
of our mind and science:(105) how have we, given our evolutionary history,
evolved the ability to do science and develop highly elaborate knowledge
about the world? We have illustrated how we have evolved cognitive abili-
ties to systematically observe, solve problems and experiment that enabled our
species to develop stone tools, shelters and eventually agriculture and helped
us meet our basic needs. Creating sophisticated scientific knowledge today
is made possible by using these same evolved abilities but for purposes that

Downloaded from [Link] by guest on 29 August 2025

do not directly influence our survival, as when they first developed. These
abilities also include abstraction, imagination and creativity which we have
developed over the last few hundred thousand years within social groups
(Figure 4.1).(21,60,78)
We experience the world through our senses that send signals to our brain
which creates an image of the world around us.(50,103) Through our senses,
we perceive the world directly (in an unaided way), as three-dimensional and
generally as well ordered by expecting future phenomena to resemble past
phenomena.(21) We collectively develop methods such as statistics to study
phenomena in the world using observational data that requires us to create
variables (using values such as 0 or 1) to be able to calculate statistical rela-
tionships for those observations—which are often viewed as one-dimensional
(Chapter 13).(42) Our mind and the methods we create using our mind deter-
mine what we are able to observe in the world and the way we are able to
process what we observe.
Many phenomena in the world—given our cognitive and perceptual
limitations—fall below or lay beyond the directly observable conditions in
which our mind and senses have developed. The further we move away from
these conditions—from the surface of the earth, from our ecological niche,

observing
solving problems experimenting
reasoning causally
abstracting categorising
testing hypotheses...

Figure 4.1 Our mind has evolved a

range of abilities that we have used …and these evolved abilities lay the
more systematically over time. foundation for how we do science today.
 COGNITIVE SCIENCE OF SCIENCE 35

from our particular context—we generally require greater abstraction of the

phenomena we study. Our mind is generally not able, without the aid of
methods and tools, to access most phenomena studied in science beyond our
senses—when we move to minuscule phenomena such as atoms and photons
at the quantum level, vast phenomena such as magnetic fields and gravita-
tion, extremely fast phenomena such as the speed of light, extremely distant
phenomena such as the earth’s core and planetary systems, or unobservable
social phenomena such as global economic markets and political systems that
lie beyond the observable actions of its actors. The same applies to many phe-

Downloaded from [Link] by guest on 29 August 2025

nomena accessible to our senses, such as biological functions and chemical
reactions that we have observed throughout our history. But to understand
them systematically we have had to first develop new methods and instruments
to access them. Just as our evolved visual abilities do not allow us to observe
very small or large phenomena in the world, our evolved cognitive abilities do
not allow us to process large sets of observations or understand highly com-
plex phenomena well. Methods and instruments we have created collectively
using our flexible mind explain most of the expansion of science by enabling
us to study phenomena that would otherwise lie beyond our basic cognitive
and sensory reach (Chapter 10).
We create knowledge either using only our methodological abilities of the
mind (to observe, solve problems and experiment) or applying scientific meth-
ods and instruments that we can only develop using these abilities. Our
methodological abilities thus always drive the knowledge we develop—while
we do not always apply the tools of science that we create using these abilities.
Our tools of science are products of our methodological abilities that we use to
better study the world or solve problems more effectively. Our methodological
abilities, scientific methods and knowledge also interact through a cumula-
tive bootstrapping process in which they extend each other, in increasingly
complex ways.
What is observable and non-observable also importantly shapes the knowl-
edge we develop. Our vision, by allowing us to observe phenomena around
us, provides our main and most used sensory input to make sense of the
world (including for other primates).(50,106) Vision is a fundamental biological
adaptation that we—and most animals—benefit from. It enables us to extract
information about the position, movement and features of objects around us. It
is through a narrow range of radiation in the form of light that we thus acquire
most knowledge about reality.(50,106) Our eyesight, together with instruments
we create to enhance our vision, set the scope and bounds within which we are
able to observe phenomena. Our visual experience is the most common form
36 SCIENCE OF SCIENCE

of evidence we use to explain things. Human experience and understanding of

reality is thus strongly shaped by the objects that fall into our visual range.(107)
Theorising and visual perception are also generally inseparable in science. To
theorise about something, we generally fall back on observations of the phe-
nomenon we are studying or observations of related phenomena and then infer
or extrapolate from them. Even theories developed in fields like theoretical
physics and theoretical economics thus rely on observations of the world.
We cannot directly observe with our own eyes or sense most phenomena we
study—from physical forces, fields and subatomic particles, to proteins, neu-

Downloaded from [Link] by guest on 29 August 2025

ral signals and complex economic systems. It is the methods and instruments
we create that enable us to access and study them. To improve our ability to
observe more easily accessible phenomena, we have created microscopes and
telescopes (Chapter 10). Even for phenomena not directly observable we gen-
erally develop visual representations to better understand them, ranging from
models of DNA and diagrams of chemical compounds, to figures of statistical
distributions and economic graphs.
We generally understand phenomena more abstractly as we move from what
is observable to what is non-observable. In the behavioural sciences, we can
see this in the difference between studying the observable actions of people
from a behavioural perspective and studying their unobservable mental states
from a neurological perspective which are more difficult to understand. In
physics, we can see this in the difference between Newton’s focus on the more
easily observable aspects of the world and the less observable world of quan-
tum mechanics which is more difficult to grasp. When we study phenomena
not observable with our sensory organs and not directly testable, we require
greater mental abstraction, imagination and interpretation.
Our perception enables us to conceptualise not only observable phenomena
but also non-observable phenomena. Light for example is viewed as acting
as both a particle and a wave, depending on the measurement methods and
experiments we apply.(81,108) And as we humans have not evolved to be able
to perceive or conceive phenomena with both these properties at the same
time, our cognitive ability to conceptualise phenomena at the quantum level
is constrained. The peculiarities of wave-particle duality are the product of a
complex interaction between how the world is and how we are capable of per-
ceiving and thinking about it.(81,108) It is because we can imagine phenomena
in the world as being a particle or a wave that we can create theories about
non-observable phenomena at the quantum level possibly possessing both of
these properties. We thus describe non-observable phenomena based on what
 COGNITIVE SCIENCE OF SCIENCE 37

we know about observable phenomena or what is conceivable using informa-

tion about observable phenomena. We are not generally able to conceptualise
non-observable phenomena in ways that are fundamentally different from
our observation-based concepts (such as particles or waves). And when we
do conceptualise phenomena differently—for example abstractly as statistical
variables for medical treatments, mental states or inflation—we generally alter
the dynamic phenomena to some degree. This is because we capture them in
quantitative variables that are amenable to statistical analysis. To conceptualise
phenomena like dark matter and global economic systems, because they are

Downloaded from [Link] by guest on 29 August 2025

not directly accessible with our retinas or senses, requires greater abstraction
than studying phenomena like rocks, plants and people.
Because many topics that scientists study—from theoretical physicists to
theoretical economists—are not visible and we cannot sense them, the chal-
lenge of making sense of parts of the world is often related to our limited
human senses and the ways we can observe phenomena using methods and
instruments we develop to enhance our senses. This helps explain the greater
limitations that confront theories in theoretical fields given the constraints we
often face in collecting data to test and verify them—such as string theory and
theories of multiple universes (Chapters 18 and 19).
In sum, we are only able to do science and create knowledge about the
world by using our methodological abilities of the mind for observing, solv-
ing problems, experimenting and abstracting. Our mind, sensory abilities and
evolutionary niche shape the knowledge we develop today. Cognitive science
of science lays a fundamental basis of science of science. The central founda-
tion of the tower of science rests on these methodological abilities that shape
the scope of science and the scientific methods and tools we create to extend
these abilities and thus perceive and measure the world in new ways—the topic
of Chapter 10.
 5
Psychology of Science

When we do science and acquire knowledge about the world, our mind also
faces psychological biases. We face limited mental resources, time constraints

Downloaded from [Link] by guest on 29 August 2025

and incomplete information, so we use simplified heuristics like rules of thumb
or shortcuts when reasoning.(102,109) We for example often rely on existing
assumptions and evidence when formulating a hypothesis or applying a given
scientific method rather than questioning and testing them every time. In
general, our mind has largely evolved to be able to absorb and process a lim-
ited amount of information and then make quick assumptions, decisions and
conclusions based on that (incomplete) information. We think fast, are habit-
based and use heuristics most of the time. This can result in unconscious
biases.(109,110)
Experiments illustrate that scientists often have limited knowledge of their
own unconscious biases in their work. A study of ecology scientists showed
that they have low awareness and understanding of the importance of biases
and how to mitigate them.(111) A study of forensic specialists illustrated that
they generally viewed their own judgements as almost infallible and they
demonstrated limited understanding of cognitive biases.(112) A study of medi-
cal doctors illustrated that they regularly made errors in clinical practice due to
cognitive biases, and such biases were made throughout the diagnostic process
when collecting, processing and confirming information.(113)
In general, our human cognitive constraints can present biases at all
steps of the scientific process. This includes when designing and conducting
experiments, such as confirmation bias when searching for evidence con-
sistent with the hypothesis being tested. It includes when analysing data,
such as omission of some results and poor understanding of statistical meth-
ods used. It also includes when writing up results for publication, such as
HARKing bias (hypothesising after results are known) and confirmation bias
when only reviewing existing literature consistent with the tested hypothesis
(Chapter 13).(114) We are thus more likely to accept new evidence if it supports
our already held views and theories (confirmation bias).(114,115) Researchers
at times develop a new hypothesis or modify an existing hypothesis after
analysing the results of a study—that is, they present an ex-post hypothesis as

Science of Science. Alexander Krauss, Oxford University Press. © Alexander Krauss (2024).
DOI: 10.1093/9780198937401.003.0006
 PSYCHOLOGY OF SCIENCE 39

an ex-ante hypothesis (HARKing bias).(116) Experiments show that we often

accept existing evidence and fall back on the same decisions and assumptions
we made in the past (status quo bias). Our expectations can help shape the out-
come of our observations by influencing interpretations we make and infor-
mation we use (expectancy bias).(50,109) Our expectations and hypotheses can
at times shape our results by making us more inclined towards observations
and outcomes that fit them and more easily rejecting those that do not.(117)
Such biases can appear to be evolutionary design flaws—given their empiri-
cal inaccuracy in different contexts. Yet we can better think of them as cognitive

Downloaded from [Link] by guest on 29 August 2025

features adapted to quickly solve problems under resource and time con-
straints that have often been important despite trade-offs for less accuracy.(102)
The way we reason and do science, when viewed as evolutionarily adaptive, is
not always unbiased and accurate. In science, ‘we all find it difficult to see the
flaws in our own work—it’s a normal part of human cognition.’(115) Yet we can
reduce some of our individual psychological biases when reasoning by using
methods, such as statistical techniques, randomisation and blinding, designed
to mitigate such biases (Chapters 10 and 13). We can also reduce such biases
through greater awareness, peer review, pre-registration of study designs and
independent replication of studies by other researchers (Chapters 6 and 13).
Our reasoning is influenced not only by psychological biases but also by
personality traits.(118) Drive and discipline foster systematic reasoning, just as
curiosity, creativity and analogical reasoning foster how we do science.(119)
As the historian of science Thomas Kuhn stated: ‘A man may be attracted
to science for all sorts of reasons. Among them are the desire to be useful,
the excitement of exploring new territory, the hope of finding order, and the
drive to test established knowledge … Many of the greatest scientific minds
have devoted all of their professional attention to demanding puzzles,’ that is,
to ‘solving a puzzle that no one before has solved.’(1) And Einstein famously
said: ‘I have no special talent, I am only passionately curious.’ Intellectual
stimulation and recognition can also provide motivation to solve a problem
or develop a new theory. Goals and needs influence and motivate us. And
personal interest and social contribution can coincide for mutual gain.(118,120)
Newton’s seminal book Philosophiae Naturalis Principia Mathematica
in 1687 had the greatest impact on physics for several centuries.(121) And
Newton made continual efforts to receive priority for his work, including
in his manuscripts at least 12 different defences for his precedence in
developing calculus before Leibniz whom he charged with plagiarism.(120)
James Watson, who helped develop the theory of the structure of DNA,
portrays himself as aggressive and arrogant in his book The Double Helix.(122)
40 SCIENCE OF SCIENCE

He depicts competition as a defining trait of his thinking and research, and

is largely motivated by the desire for fame.(120) Watson even later decided
to auction off his Nobel Prize medallion for $4.1 million in 2014.(123) In his
autobiography, the Nobel laureate in physics Max Planck also expressed his
sense of competition and wrote that he had ‘the desire to win, somehow, a
reputation in the field of science.’(124) Darwin wrote that ‘It seems hard on me
that I should lose my priority of many years’ standing’ in reference to Alfred
Russel Wallace developing a similar evolutionary theory at the time.(120) Since
scientists build on the work of others and attempt to go beyond that work,

Downloaded from [Link] by guest on 29 August 2025

scientific advancement is often grounded in a symbiotic relationship between
cooperation and competition. Competition can help ensure quality control
and independent testing of others’ work. While a desire for recognition is
a basic human trait, the sociologist Robert Merton highlights that most
renowned academics are known to be driven by a desire for fame—from
Galileo and Descartes, to Faraday and Freud.(120)
Scientists are not a special breed of Homo sapiens but are driven by goals,
interests and personality traits like everyone else. A belief that scientists may be
more clever or quicker at solving problems (or more objective or ethical) than
people pursuing other professions is largely that, a belief—as a comparative
study found among engineers, neurosurgeons and the general population that
put into question the common expressions that ‘it is not rocket science’ and
‘it is not brain surgery.’(125) This is connected to the widely held 10,000-hour
hypothesis in psychology that states that about 3.5 years of full-time training
(8 hours times 365 days) are needed to excel in most human domains including
a scientific field.(126) Overall, scientists are not at birth entirely different from
other people (who do not wear white coats or do not have the label PhD after
their names).
In sum, we face psychological biases that influence our reasoning and our
scientific results. Though, we can reduce them through research methods
designed for that purpose. Personality traits also influence our motivation for
doing research. Reasoning and acquiring knowledge are however not mind-
bound but occur in our social and physical world. To better understand them,
we need to study not only our individual biological, cognitive and psycho-
logical constraints and biases within us—that is, internal factors that largely
revolve around our methodological abilities of the mind (Chapters 2–5)—
but also the collective methods we develop using these abilities and the range
of external factors that include social, economic and historical influences
(Chapters 6–15; Figure 1.2). We have to thus study both the foundation of the
tower of science and the factors that influence the shape and form of the tower.
 6
Sociology of Science

If the cognitive abilities of children at birth a few hundred or even thou-

sand years ago and today are likely not very different, what can explain

Downloaded from [Link] by guest on 29 August 2025

the large differences in their theories of the world as adults? While our
mind’s methodological abilities to observe, solve problems and experiment
are a precondition of science and we have used them more systematically
over time, changes in broader demographic, social(21) and economic factors
have fostered developing vast knowledge and methods. There were hun-
dreds of scientists a few centuries ago.(127) Today, there are about nine
million full-time scientists worldwide.(63) Demographic growth and com-
plex social organisation have been crucial for the scientific community to
grow and for greater collaboration, cumulative knowledge and methodolog-
ical development (Chapters 10 and 11). Doing science is thus not just a
cognitive activity conducted by individuals (Chapters 2–4). It has become
an increasingly complex social activity conducted among a community of
researchers.(21,78,87,88)
After birth we are socialised into a system of language and mathematics and
we acquire much of our knowledge of the world through institutions designed
for that purpose—schools and universities. We are raised in a cognitive and
social environment full of information. We do not just use cognition but cul-
turally embedded cognition when we reason and do science.(128) Without
formal education and methodological training we would not be able to develop
highly complex knowledge (Chapters 10, 12 and 13).
We are not disinterested and detached observers of reality but guided by
institutions, shaped by scientific norms and motivated by values that can influ-
ence our research. The scientific community we are embedded in influences
which problems, questions and objectives we find relevant to pursue and how
we frame them. It shapes which methods we consider credible for analysing
them. It influences which experimental designs we choose, the way results are
assessed and what assumptions are allowed (Chapter 14). It shapes how we
classify phenomena and how we define variables for them that influence how
we measure and understand phenomena. It influences how we use results in
medicine, technology and society. Our scientific community also sets norms

Science of Science. Alexander Krauss, Oxford University Press. © Alexander Krauss (2024).
DOI: 10.1093/9780198937401.003.0007
42 SCIENCE OF SCIENCE

such as the kinds of evidence accepted (and not accepted) to support a claim,
and the types of hypotheses and theories that are suitable (and not suitable).
It sets norms such as the kinds of information that needs to be included
(and not included) for publishing articles, and the forms of peer review that
are appropriate (and inappropriate) for publication in journals. Such norms
are ingrained in our scientific communities and help direct our scientific
activities. They are defined differently across fields and change over time
(Chapter 8). Together, they account for the rules of the game for doing
science—rules that each player needs to abide by if they are to stay within their

Downloaded from [Link] by guest on 29 August 2025

field. Acquiring knowledge through systematic observation, robust method-
ological analysis, significant results and reliable conclusions thus first requires
agreeing on common criteria for evaluating what counts as systematic, robust,
significant and reliable.
Researchers across and within fields also adopt different methodologi-
cal approaches to studying the same phenomena. US-American scientists
for example have adopted a more narrowly focused approach and under-
standing to genetics and natural selection than German scientists, who in
contrast view evolution as also operating at the macro level.(129) In sci-
ence of science, the study of topics like collaborations among scientists or
scientific impact is approached using different methodologies. Scientometri-
cians including network scientists commonly adopt a descriptive empirical
approach, economists a causal empirical approach, sociologists largely a qual-
itative empirical approach, philosophers a conceptual approach, and so forth.
This often leads to different answers to the same question as approaches are
not yet integrated.
The sociologist of science Harriet Zuckerman published the seminal book
Scientific Elite: Nobel Laureates in the United States in 1977. Zuckerman
embarked on a journey travelling across the US to interview Nobel laure-
ates about their personal background, family and research.(71) Her pioneering
research investigated the demographic and social traits of eminent scientists,
including attributes like age, sex, religion and ethnicity. She describes the pro-
cess of how being ‘a Nobel laureate is, for better or for worse, to be firmly
placed in the scientific elite.’(71) A central argument of hers is that science
is a collective effort that is embedded in a social and cultural context, with
the scientific elite made up of talented individuals influenced by social and
institutional factors like education, social stratification and mobility. cf. (119)
She emphasises the role of power, authority and influence related to science’s
award system. Studying these ultra-elite of science, Zuckerman suggests that
the process of discovery is highly competitive and hierarchical, in which only
 SOCIOLOGY OF SCIENCE 43

a privileged few can ascend to a high status. Her research also underscores
that Nobel laureates are dedicated to their work, often at the detriment of
other facets of their lives, with many laureates working tirelessly, at times
over decades, in search of a single breakthrough over the course of their
careers.(71)
Robert Merton—a central inspiration and later collaborator of
Zuckerman(71) —was another eminent sociologist of science. In his influential
book Science and Technology in a Democratic Order in 1942, he argues that
science and scientific advances take place within a scientific community with

Downloaded from [Link] by guest on 29 August 2025

shared scientific norms, values and institutions.(130) Merton identified the
influential ‘Matthew Effect’ in science, which states that renowned scientists
receive much more credit for their research than less renowned scientists
with equally important research contributions.(131) The social structure of
science thus generates a self-perpetuating cycle of scientific success among
established scientists. He also highlights the broader relationship between
science and society, exploring how scientific advances can have important
implications for technology, social change and policy. In 1993, at the age of
83, Merton married Zuckerman, with the duo leaving behind a large mark in
the sociology of science.
Two other leading sociologists of science, Latour and Woolgar, investigated
the role of social influences within the laboratory.(10) They observed that sci-
entists within a leading biological laboratory are exposed to peer and social
pressures and seek influence. And influence is not just achieved by the theo-
ries they develop but by the scope of their social networks and their ability
to mobilise support for their work.(10,11) In attempting to explain science,
they argued that social influences are the most important factor in creating
knowledge—though they only focus on one factor in studying science, namely
social influences. In the book Science of Science and Reflexivity, another lead-
ing sociologist of science, Bourdieu, argued that science may be ‘in danger of
becoming a handmaiden to biotechnology, medicine, genetic engineering, and
military research,’ given the possible risks and interests of corporations.(11)
Science can in turn also bring negative effects for society and the environ-
ment, by contributing to global challenges like overpopulation and climate
change. Bourdieu’s aim—similar to Latour and Woolgar—is ‘to identify the
social conditions in which science develops.’(11) But by focusing almost exclu-
sively on social conditions, they strongly overvalue the role of social influences
in science—while neglecting the role of methods and our mind, and the broad
range of other interrelated factors (biological, historical, economic) that help
drive science (Chapters 2–15).
44 SCIENCE OF SCIENCE

In studying science, the highly influential sociologist of science Latour had

very limited knowledge of science—in fact, ‘Latour’s knowledge of science was
non-existent’(10) and he thus adopted a largely naive, anti-science position that
lacked knowledge of most other features of science. Since then, much related
work in the sociology of science has taken such a constructivist stance that has
been parodied(132) and rejected(133) as it is not supported by rigorous empir-
ical evidence in other fields of science. Such work has done more harm to the
perception of studies of science than contributing to our understanding of sci-
ence. And it is a case in point for why we must take the range of cognitive,

Downloaded from [Link] by guest on 29 August 2025

methodological, demographic and other factors into account when analysing
science to avoid misattributing the role of a single factor—social influences
like power and peer pressure. This strong position also further constrained
disciplinary integration that can be traced to the so-called science wars that
further divided sociology of science from history and philosophy of science,
and embodied vast disagreement about what the study of science should entail.
The scientific process is not only influenced by social factors but is itself a
social process in which we generate knowledge by building on each other’s
work and methods. We read the articles of others. We apply methods and
instruments developed by others. We discuss and verify our results with col-
leagues, peer reviewers and editors. In science, we observe organised curiosity
(in co-authored publications), coordinated critique (in peer review) and
public funding for research teams (through institutional grants).
No single person can develop a complex method, field or theory on their
own. A complex method (from statistics to randomised controlled experimen-
tation) has not been created by an individual mind but by many individuals
building on the work of others. An academic field (from molecular biology to
nuclear physics) has always been developed by collectively working together
in a cumulative way. A complex theory (from quantum theory to a theory of
the origin of life) has not been created by the mind of a single person without
relying on much previous knowledge. To create such knowledge we need to
acquire and share information cumulatively—we need cumulative knowledge.
And the larger a scientific community, the more researchers can cumulatively
build on each other’s research.
One of the greatest discoveries of the 20th century was uncovering DNA’s
double-helix structure in 1953. Commonly, the names of two scientists are
mentioned when referring to the discovery: Francis Crick and James Wat-
son. Yet the discovery built on initial work by Miescher, based on the pivotal
X-ray work produced by Franklin and Gosling, without which producing
the image of the double helix would not have been possible. That in turn
 SOCIOLOGY OF SCIENCE 45

required applying X-ray diffraction methods developed by von Laue, who used
X-radiation identified by Röntgen. The work was also supported by parallel
research on DNA structure by Wilkins and his group of colleagues, among
many others.(122) Rosalind Franklin arguably contributed the key missing
piece by applying the method of X-ray diffraction using DNA fibre to be able to
identify the structure of DNA (Figure 6.1). Crick, Watson and Wilkins however
won the Nobel Prize (directly after Franklin’s death) for the work that builds
on her research. The discovery of the molecular structure of DNA revealed
how information is transferred in living material and opened vast new areas

Downloaded from [Link] by guest on 29 August 2025

of research.
Another social dimension of science are the gender disparities we observe
across scientific fields, with science remaining heavily biased towards males.
Assessing all over 500 Nobel-Prize-winning discoveries shows that only 3%
of all Nobel laureates in science are women.(64) Physics is particularly male-
dominated, with only 2% of Nobel-Prize discoveries in the field made by
women. The share is 6% in astronomy and 7% in medicine. While shares
remain very low, we can observe a positive trend, as more than half of all female
Nobelists who were ever awarded the Prize received it since 2000.(64) A num-
ber of major discoveries have however been in large part made by women who
did not receive credit or a Nobel Prize for their work. Rosalind Franklin is a
classic example. Another example is the microbiologist Esther Lederberg. She
made the discovery of a virus which infects bacteria, and with her husband

Figure 6.1 Discovering DNA’s double-helix structure was the outcome of

multiple scientists working together, with Rosalind Franklin (left) and her X-ray
‘photo 51’ (middle) playing the key role in revealing the double-helix structure
(right).
Source: Elliot & Fry/National Portrait Gallery (left); Raymond Gosling/King’s College London via
Wikipedia (middle); Qimono via Den Store Danske (right).
46 SCIENCE OF SCIENCE

created a technique to transfer bacteria between petri dishes. But in 1958 her
husband Joshua Lederberg received the Nobel Prize for the research carried
out with his wife, and he only mentioned her once in his Nobel lecture.(123)
Another example is the astrophysicist Jocelyn Bell Burnell who discovered pul-
sars in 1967, and she published an article together with Antony Hewish. But
in 1974 Hewish was awarded the Nobel Prize for the work.(134)
An explanation for the very low levels of female Nobel laureates is that
women have been systematically discriminated in accessing education and sci-
ence throughout history. The unfavourable norms about the role of females in

Downloaded from [Link] by guest on 29 August 2025

science have begun to improve since the second half of the 20th century and
especially in the 21st century. But women still remain especially underrepre-
sented in STEM (science, technology, engineering and mathematics) fields.
We also still often hear sexist statements in science. The Nobel laureate and
biochemist Sir Tim Hunt for example said in 2015: ‘Let me tell you about my
trouble with girls. Three things happen when they are in the lab: you fall in
love with them, they fall in love with you, and when you criticize them they
cry.’ Hunt was heavily criticised by social media for such sexist statements.(123)
Moreover, knowledge across science is not just created or discovered, but
must also be explained, justified and generally replicable. When for example
Darwin had an initial idea about natural selection or Einstein about a pos-
sible relationship between mass and energy, they did not yet necessarily
produce reliable knowledge. Knowledge gains reliability and validity when
others independently arrive at the same results using the same and other meth-
ods, others can replicate the results and we can justify the methods used to
others (Chapters 10 and 14). cf. (135) Theories require gaining acceptance and
consensus among members of a scientific community. And that is not always
straightforward for some major new discoveries when they are first made.
Quantum theory was at first a particularly controversial discovery in physics,
which motivated Max Planck to polemically claim that: ‘A new scientific truth
does not triumph by convincing its opponents and making them see the light,
but rather because its opponents eventually die, and a new generation grows
up that is familiar with it.’(124)
In sum, we develop complex knowledge, methods and science within a sci-
entific community and social context. Science cannot be entirely independent
of our scientific practices, norms and institutions. These help coordinate how
we produce and distribute new knowledge. Newton in 1675 said that ‘If I have
seen further, it is by standing on the shoulders of giants.’ cf. (136) And his claim
holds for contemporary science. Returning to the tower of science analogy
we see that sociologists of science are among the most isolated from other
 SOCIOLOGY OF SCIENCE 47

researchers studying science on different floors of the tower and they take
their own perspective and publish research in journals within their own field.
Later we will illustrate how to restructure the tower so all researchers studying
the same topic, science, work on the same floor to gain a much more holis-
tic view of science that complements and does not compete with each other.
Social and demographic features play a crucial role in understanding and fos-
tering science, and thus in science of science. And they are closely connected
to economic features.

Downloaded from [Link] by guest on 29 August 2025

 7
Economics of Science

We can foster science through economies of scale, a reward system, sci-

ence policy and targeted research funding.(25,26,27,28,43,137) As society becomes

Downloaded from [Link] by guest on 29 August 2025

more productive, diversified and efficient in providing goods and services,
more individuals can dedicate themselves to scientific activities. Larger com-
munities of scientists have a comparative advantage over individual isolated
scientists in cumulatively building on research. Science can function like
an economy: just as a growing and more specialised labour force generally
develops more diversified goods and technologies, a growing and more spe-
cialised scientific community generally develops more diversified knowledge
and methods. Economies of scale and agglomeration facilitate greater division
of labour across and within scientific fields—and thus greater methodological
diversity and knowledge.
Science runs on a priority-based reward system that motivates research
and innovation.(71) It works by giving priority to the first person to publish
a new idea or develop a new method. It also requires us to make research
and methods publicly available.(25,28,138) As a form of intellectual property
right, priority is rewarded through social recognition from the scientific com-
munity and through the potential of contributing to society (Chapter 5).
This winner-takes-all system thus incentivises scientists to produce and share
knowledge,(71,120) and generally more so than monetary incentives.(25)
Public institutions also help plan, finance and manage how we produce, dis-
tribute and use knowledge. They intervene in how we produce different areas
of knowledge by setting government priorities, science policy and resource
allocation—with funding also influenced by overall economic development.
They shape how we distribute knowledge through schools, universities and
research institutions and how efficient that knowledge is mediated.(28,139)
To practise science we generally require some level of funding. During the
centuries between Newton and Einstein, research was commonly conducted
on low budgets and a small scale. As Einstein put it, ‘Science is a wonderful
thing if one does not have to earn one’s living at it.’ Scientists funded them-
selves or received small funds from universities or private foundations. Then
companies like Bell Telephone and General Electric began to conduct basic

Science of Science. Alexander Krauss, Oxford University Press. © Alexander Krauss (2024).
DOI: 10.1093/9780198937401.003.0008
 ECONOMICS OF SCIENCE 49

research in their laboratories. It took governments until around the time of

World War II to realise that science had driven the enormous advances in
technology, medicine and human well-being up to that point. This is when
public funding for science began to be strategically targeted for the first time
in a coordinated way.(140) The Chinese Academy of Sciences was grounded
in 1949, the National Science Foundation in the US in 1950 and the German
Research Foundation in 1951. In the US for example, total annual expendi-
ture on science was negligible before the war, accounting for 0.1 billion US$
in 1930. After the war, the US became the most important player in science

Downloaded from [Link] by guest on 29 August 2025

funding and output. There was a nine-fold increase in total annual expendi-
ture in research and development in the US, increasing from 1.5 billion US$
to 13.6 billion between 1945 and 1965.(141)
Since most science spending today is public and funded by taxpayers, we
can view much scientific research as a public good that needs to be made pub-
licly and freely available.(25,26) Scientists do not generally have at their disposal
large amounts of funds to conduct research in certain areas of science, like
basic research, cost-intensive research or research only yielding returns after
many years.(5,25) The costs of laboratories, equipment and running certain
experiments in some fields within chemistry, biology, medicine and espe-
cially physics and astronomy are at times high for individual researchers to
bear on their own.(5,25) Without government funding, some areas of research
would not be possible or would be underinvested. Developing some of the
most expensive scientific instruments in the world would not be feasible, such
as CERN’s (Conseil Européen pour la Recherche Nucléaire) large hadron
collider—the world’s largest particle accelerator—which cost billions of dol-
lars (Figure 7.1). The collider led to the discovery of a new particle, the Higgs
boson, in 2012 which is a central part of the standard model of physics.
There are however large shares of discoveries that have been made with only
limited funding. Assessing over 750 major discoveries, including all Nobel-
Prize-winning discoveries, we observe that hundreds of discoveries have been
made using low-cost methods and instruments, such as statistical and math-
ematical methods, light microscopes, electrophoresis, assay techniques, ther-
mometers, chromatography methods and centrifuges. These are among the ten
most commonly used methods and instruments in science and today we can
acquire them new for less than a thousand or even a few hundred dollars.(64)
Large-scale funding is today not always needed to make novel breakthroughs.
The most expensive scientific instruments are in contrast concentrated espe-
cially in subfields of physics and astronomy. They include the massive Hubble
space telescope that enabled discovering the accelerating expansion of the
50 SCIENCE OF SCIENCE

Downloaded from [Link] by guest on 29 August 2025

Figure 7.1 Large-scale public funding is needed for some expensive scientific
instruments, such as CERN’s large hadron collider (particle accelerator).
Source: CERN.

universe in 1998. The massive laser interferometer gravitational-wave observa-

tory (LIGO) led to the discovery of the existence of gravitational waves in 2015.
The Manhattan Project required developing new isotope separation tech-
niques and reactors which led to the invention of the first atomic bomb in 1945.
Exceptionally expensive projects in genetics and medicine have also invested
vast amounts of funding targeted to cancer research and the human genome
project. These well-known examples of the most costly scientific instruments
and initiatives required vast government funding. Such large-scale funding is
fortunately not the norm but the exception, and hundreds of discoveries are
made using low-cost methods and instruments.
Funding agencies have also increasingly required research to have social
impact—for example to influence outputs like diagnostic tools, technologies,
medications and public policies.(27) This can shape the supply of research by
determining which scientific areas receive more funding. For some areas of
research, funding can shift in response to societal challenges that we face, such
as climate change, energy efficiency and health pandemics. Funding can also
shift as scientific challenges emerge, such as the replication crisis in science
and big data. Research agendas are often set to meet the needs and objec-
tives of those funding the research—mostly wealthy societies. For example,
 ECONOMICS OF SCIENCE 51

diseases more common in wealthier countries such as diabetes and cancer

receive much more research funding than tropical diseases that are poten-
tially much easier to cure but mainly affect those in poorer countries with
less financial resources to fund science. Private funding in science, through
pharmaceutical companies for instance, is also biased towards the demand of
wealthier countries. Governments and funding agencies have become, in most
fields, increasingly focused on science that produces quicker results through an
incentive system based on shorter-term outputs, especially articles. This can
lead to a shorter research horizon that neglects long-term research projects

Downloaded from [Link] by guest on 29 August 2025

(Chapter 11).(25,27) How resources in biomedicine within the US are allocated
for example is more strongly associated with existing allocations and research
than with the present burden of diseases, illustrating a mismatch between
allocated resources and actual needs.(142)
Importantly, some studies, especially by economists, have begun assessing
the causal effects of interventions on questions in science of science. A study
carrying out a field experiment evaluates how search costs affect new scientific
collaborations by randomly assigning 402 scientists to different sessions (shar-
ing information on grant funding opportunities) at a research symposium. It
finds that scientists assigned to the same session (the treatment) were more
likely to build a new collaboration for a grant at 28% than scientists assigned
to different sessions (the control) at 16%.(143) A study using a regression dis-
continuity research design evaluates the Matthew effect (‘the rich get richer’)
in science funding among 3660 grant applicants. It finds that researchers do
not win subsequent grants because of greater achievements but rather because
they already won a previous grant as a younger researcher, which suggests
that receiving funding early on provides a strong advantage for receiving later
funding.(144)
A study using a difference-in-differences research design assesses whether
the premature death of 452 prominent scientists changes the evolution (mea-
sured by publication rates and funding flows) of subfields that they published
in before passing away, compared to control subfields. It finds that when emi-
nent scientists pass away, fields more easily evolve in new directions.(145) A
study conducting a randomised controlled trial (RCT) assesses the effects of
open peer review by assigning reviewers to either a group who signed their
reviewer report (revealing their identity to authors) or a group who did not
sign the report (remaining anonymous) for 408 manuscripts. It finds that
signed reviews had higher quality, required longer time to complete and were
more likely to recommend publication than unsigned reviews.(146) A greater
shift towards conducting such studies with research designs that estimate the
52 SCIENCE OF SCIENCE

causal effects of interventions driving science is a critical step to move the field
of science of science forward and bring greater rigour—which we will discuss
in the Conclusion.
In sum, economic factors and public institutions shape funding priorities,
research objectives and science policy. A reward system also helps provide
some incentives for scientists to develop new scientific advances, discoveries
and the methods needed to do so. Economic factors influence the research
and scientific tools that we can fund and thus the size and shape of the tower
of science. But economics of science alone is—like any science of science

Downloaded from [Link] by guest on 29 August 2025

subfield—incomplete and cannot explain the foundations, limits and advance-
ment of science. For science is largely shaped by our methods and mind, while
also influenced by our broader social and historical context.
 8
History of Science

Science has a history, and when we study the history of science including the
discoveries, theories and the methods used to create them, we can trace their

Downloaded from [Link] by guest on 29 August 2025

origin to the individuals who developed them. Thomas Kuhn, the most cited
and well-known historian of science in the 20th century, offered an expla-
nation of the history of science that rejected the view of scientific change
as being cumulative.(1,32) The history of science can be viewed as a cycle in
which established ideas and facts are doubted, new problems and evidence
then lead to new revolutionary ideas and facts (and replace the established
ones), which eventually over time are also doubted once problems and anoma-
lies associated with them become apparent, and the cycle begins again. For
Kuhn, this process of science is not cumulative. It represents revolutionary
paradigm shifts, in which a scientific community rejects existing assumptions,
concepts and theories and adopts entirely new ones. Kuhn thus describes sci-
ence as going through paradigm shifts. For him, ‘revolutionary’ scientists help
redefine which research questions and assumptions are important and what
methods are best. In short, they help rewrite the rules of the game. This notion
of science may seem to apply to shifts in physics in the past, namely shifts in
our theories of physical reality from Aristotle to Newton and then to Einstein.
Kuhn focused on such cases largely in physics up to the early 20th century.(1)
The shift from the Ptolemaic earth-centred theory of the universe to the
Copernican sun-centred theory characterised the classic paradigm change,
which Kuhn focused much research on. In geology for example the shift from
a stabilist to a plate-tectonic mobilist view of the earth can also, according
to Kuhn’s thesis, be viewed as a paradigm shift. In economics, it can reflect
the shift from a neoclassical view of individuals acting as perfectly rational
actors towards more behavioural approaches that view individuals as acting
irrationally at times. In physics, it can reflect the shift from a classical Euclidean
view of time and space as flat to an Einsteinian view reflecting the curvature
of time and space. And it can reflect the shift from a classical mechanics’ view
of physical reality in which particles act in a deterministic way (Newton) to
a quantum mechanics’ view of physical reality in which phenomena act as
both a wave and a particle at the same time and in a probabilistic way (Bohr,

Science of Science. Alexander Krauss, Oxford University Press. © Alexander Krauss (2024).
DOI: 10.1093/9780198937401.003.0009
54 SCIENCE OF SCIENCE

Heisenberg and others). Each of these changes in theory has been made possi-
ble by changes in the methods we use to study the phenomenon. Such changes
in theory in the past can appear as large shifts in science. But they have become
very rare and are in fact based on much cumulative knowledge, as we will
see. Quantum theory, the mechanisms of evolution, computer science, micro-
scopes and the like have all been continually and extensively refined over time.
No major scientific methods used across fields (such as statistics, X-ray
methods or controlled experimentation) and no major scientific fields (such as
biology, chemistry, nuclear physics and computer science) have been entirely

Downloaded from [Link] by guest on 29 August 2025

discarded. And it is difficult for us to imagine that they could be. Rather,
we cumulatively extend them over time. Our methods and fields of science
encompass our extensive bodies of empirical and theoretical knowledge con-
solidated over time. In contrast, a hypothesis or theory put forth by a scientist
can be tested by others and we abandon many of our ideas and hypotheses
and a number of our theories, as depicted in Figure 8.1. In analysing over 750
major discoveries, including all Nobel-Prize-winning discoveries, only 1% of
discoveries (8 in total) have been abandoned and they were mostly theoretical
in nature—not empirical or methodological.(64) They were not grounded in
rigorous empirical evidence.
It is in particular our central methods of science that are cumulative.
Mathematics, RCTs, chromatography, statistics, centrifuges, particle detectors,
telescopes, electrophoresis, spectrometers and the like have all been constantly
extended and fine-tuned, often over centuries. In contrast to hypotheses and
theories, we have not abandoned any of these established scientific methods
or instruments. We use them in multiple scientific fields and we continue to
improve all of them, which have enabled making dozens of new discoveries.
Our methods and instruments make science highly cumulative. In contrast to
Kuhn’s hypothesis, the history of science tells a cumulative and unified story,
and the story is driven by the methods and instruments we develop. Shifting
our attention from individual hypotheses and select theoretical discoveries to
all major scientific discoveries, methods and fields is the best way to measure
and assess the cumulative nature of science. For they make up the foundation
of science and how we conduct science and they encompass our established
bodies of knowledge. Cumulative knowledge is thus commonly on a spectrum:
from unestablished ideas and hypotheses, and then experimental findings and
theories (experimental and theoretical discoveries), and finally to established
methods and fields (Figure 8.1).
Our best light microscopes today do not compete as a different paradigm
to the first light microscope developed in 1590. Our methods of arithmetic
 HISTORY OF SCIENCE 55

Spectrum of increasingly cumulative knowledge

higher

likelihood
of being
abandoned

lower

scientific laying scientific

ideas formulated hypo- tested experiment- general- theories made by
in theses with al findings ised in applying methods foundat- fields
ion of

Downloaded from [Link] by guest on 29 August 2025

can be developed by an individual researcher developed by a community
of researchers building on existing and
established methods and instruments

Figure 8.1 The cumulative nature of science: decreasing likelihood of

abandoning ideas, hypotheses, experimental findings, theories, methods and
fields.

today do not compete as a different paradigm with those developed by the

Sumerians. As a vast cumulative project, we have extended our major tools of
science over time—from statistics to microscopy. No major scientific method
or instrument that is used across fields goes through competing methodologi-
cal paradigms. They are the foundation of science and being able to do science
across fields (so that a possible paradigm shift that would abandon them would
so fundamentally change how we conceive science that we could likely no
longer call what we do science).
Within the new-methods-drive-science framework presented here, scientific
progress is viewed as primarily driven by the new methods we develop that
enable us to do science in new ways and make new discoveries—not just by
largely theoretical developments as Kuhn(1) argues. The Kuhnian focus on
a theoretical shift reflects a final output but not the methodological process
we take to develop, refine and replicate the output. This new-methods-drive-
science framework better reflects actual scientific practice and enables us
to better analyse and explain how scientific progress and new discoveries
arise. It provides a more coherent alternative to the notion of broad Kuhnian
paradigm shifts. While Kuhn provides insight into scientific change in theo-
ries, such change in theories is not just historical and intrinsic to science, as he
outlines.(1) It is also influenced by a range of other factors, such as cognitive
and methodological factors (Chapters 2–15).
The title of Kuhn’s best-selling book The Structure of Scientific Revolutions
portrays the notion of vast scientific changes resembling vast political changes
56 SCIENCE OF SCIENCE

or revolutions (Figure 8.2).(1) A more accurate title, given his specific focus,
could have been An Account of Theory Change Mainly in Early Physics, as
he did not identify an underlying structure across such shifts and they are
generally slow, collective processes rather than revolutions. A more accurate
account of science could be described as An Account of Cumulative Scientific
Progress Embodied in Scientific Methods and Fields.
To better understand scientific theories and how we develop them
cumulatively—such as Darwin’s theory of evolution and Einstein’s theory

Downloaded from [Link] by guest on 29 August 2025

Figure 8.2 Kuhn’s seminal book is a foundational text in the science of

science, 1962.
 HISTORY OF SCIENCE 57

of special relativity—we cannot view them independent of their historical

context and the existing knowledge on which most of their work builds.
Our theories are, when contextualised, small contributions within an exist-
ing web of cumulative knowledge. Arguments in favour of evolution, that an
organism can descend from another organism, have been made at least since
antiquity. They were made by ancient Chinese (like Zhuang Zhou) and pre-
Socratic Greeks (like Empedocles and Anaximander),(147) and Romans (like
Lucretius) and medieval Arabs (like Al-Jāḥiẓ and Nasīr al-Dīn Tˉusī).(148) In
1748, De Maillet also argued that different species emerge by life forms adapt-

Downloaded from [Link] by guest on 29 August 2025

ing to different circumstances.(149) Erasmus Darwin—Charles’s grandfather—
viewed living beings as self-improving, arguing in 1794 against a mecha-
nistic view of human life and theorising about evolution that could occur
through natural selection.(150) Lamarck developed a similar evolutionary the-
ory in which animals adapt to their local environment and thus use and
stop using certain traits. He proposed this as the reason why moles became
blind.
In the first few decades of the 1800s, different scholars such as Wells,
Matthew, Blyth and Chambers also already developed a theory of natural
selection in one form or another. Around the mid-19th century, theories of
transmutation were widely accepted among scholars.(149) Then, by drawing
on the work of Erasmus Darwin, Lamarck, Chambers and others, Wallace also
developed a theory of natural selection in two papers published in 1855 and
1858. He then published work together with Charles Darwin in 1858—before
Charles published his expanded ideas in On the Origin of Species in 1859.(51)
Other scholars had thus already developed the notion of evolution and natu-
ral selection before Charles Darwin observed on the Galápagos Islands how
mockingbirds varied from island to island, which provided him further evi-
dence of the variation of species.(151) His critical contribution was collecting a
range of detailed observations and specimens to support the theory, realising
that theories are best grounded with detailed documentation.(149) If Charles
Darwin had not written On the Origin of Species and remained a hobby geol-
ogist, we would still have a well-developed theory of evolution by natural
selection.
Einstein is similarly widely credited for developing the famous equation
E = mc2 , associated with his theory of special relativity in 1905. The simple
algebraic equation states that energy E within a system (an atom, the human
body or a planetary system) can be calculated as its mass m multiplied by the
speed of light c squared. Special relativity explains how speed affects mass,
time and space and it transformed our understanding of the relationship
58 SCIENCE OF SCIENCE

between space and time. Yet before Einstein, others already developed vari-
ants of this mass-energy equation, and even the very same equation.(152,153)
Umov in 1873 and Preston in 1875 developed similar mass-energy formulas,
such as Е = kmc2 , as did Thomson in 1881, Heaviside in 1889 and Poincaré
in 1900.(154) In 1904, Hasenöhrl formulated the equation E = 3/8mc2 that
he arrived at by studying the properties of blackbody radiation within a
moving cavity. And his work was consistent with special relativity.(152) In a
paper published in 1903 within the Proceedings of the Royal Veneto Insti-
tute of Science, De Pretto developed the same formula E = mc2 that later

Downloaded from [Link] by guest on 29 August 2025

brought Einstein international fame.(153) This work was all published before
Einstein’s paper, containing the same formula, was published in 1905 in the
German-based Annalen der Physik—in which Hasenöhrl also published his
work the previous year (which Einstein did not cite).(152) Einstein did later give
credit in 1906 to Poincaré’s mass-energy equivalence and acknowledged that
their formulas were mathematically equivalent regarding the centre of gravity
problem.(155)
In 1907, Planck formulated, by building on the existing work of Hasenöhrl
and Poincaré, the final derivation of E = mc2 . In that same year, Einstein stated
that ‘It seems to me to be in the nature of things that other authors might
have already elucidated part of what I am going to say. However, bearing in
mind that the problems under consideration are being treated here from a new
standpoint, I felt that I should be permitted to forgo a survey of the literature
(which would have been very troublesome for me).’(156) Einstein connected the
formula with the theory of relativity, but without Einstein we would still have a
theory of relativity which other scientists, like Hendrik Lorentz, were working
on at the time. And special relativity was originally called the Lorentz–Einstein
theory. Overall, other researchers generally also deserve credit for their contri-
butions to developing nearly all leading scientific theories, from the equation
E = mc2 by Hasenöhrl, De Pretto, Einstein and others, to the theory of evolu-
tion by Lamarck, Wallace, Darwin and others. It is however easier for science
textbooks, teaching science and awarding prizes to associate a discovery with
a single name rather than with the community of scholars who developed it
(Chapter 6). Imagine if we would have to remember all these names instead of
just Darwin and Einstein. But simplicity comes at a cost: it distorts the image
of science, how the scientific process actually works and how we foster new
advances.
What any one individual can contribute, when we historicise their work, is a
piece or connection between already existing pieces, compared to the extensive
knowledge and methods developed before them on which they build and make
 HISTORY OF SCIENCE 59

their work possible. Science is, on the whole, a cumulative and iterative pro-
cess of continual refinement. Darwin, Einstein and the like are not scientific
geniuses but people endowed with the same senses and motor skills as others,
while generally very good at synthesising and building on existing knowledge
and methods. Often they are just smart normal people in the right place at the
right time.
In sum, science is cumulative and scientific theories and methods are not
independent of their historical context. They are provisional and have been
expanded by new evidence, experiments and methodological advances over

Downloaded from [Link] by guest on 29 August 2025

time. There is a history of science—not just one science that is constant over
time. Yet people in the future are not likely to think the same way about the
level of validity and rigour of our current theories as we do about those a few
centuries ago. This is because today’s best methods and instruments (including
electron microscopes, X-ray methods and radar telescopes) and the theories
developed using them are much better able to explain and predict phenomena
with much greater accuracy. Historians of science do not just develop grand
accounts of scientific progress, like Kuhn, but also investigate historical docu-
ments and correspondences, research science in particular historical contexts
and conduct case studies and field work. Interestingly, Kuhn stated that ‘in the
early stages of the development of any science different men confronting the
same range of phenomena … describe and interpret them in different ways,’
and this is evident in the field of science of science.(1) History of science helps
us understand the highly cumulative nature and evolution of our knowledge
and methodological toolbox, and thus the highly cumulative nature of how we
build our tower of science. History of science forms an essential piece of the
greater picture of science of science—which is linked to the anthropology of
science.
 9
Anthropology of Science

Anthropology of science is the cross-cultural study of humanity which retraces

how we have developed science, from the past to the present. So how did we get

Downloaded from [Link] by guest on 29 August 2025

from being hunters and gatherers using axes, basic observation and numerical
reasoning, to being scientists hunting explanations and gathering data using
systematic controls and mathematical methods? How did we get from being
nomads running in the savannahs and conducting trial and error, to being sci-
entists running sophisticated experiments and conducting statistical analysis?
We do not have a definitive answer as we face constraints to reconstructing
past events. This holistic and methods-driven account here however provides
an integrated explanation.
Our species began to observe, solve problems, experiment and acquire
knowledge more and more systematically, on the one hand, as our biological,
cognitive and social abilities evolved over time by interacting with our envi-
ronment (Chapters 2–5). On the other hand, we did so as broader social, eco-
nomic and historical factors enabled us to use these abilities more effectively
and build on what we observe in larger groups (Chapters 6–8). We developed
more and more complex language an estimated 50,000–100,000 years ago.(92)
Language is essential to be able to express our observations and ideas verbally
and explain them to others—both then and today (Chapter 15). Expanding our
language abilities allowed, in an interconnected way, for greater cooperation
in larger groups and greater tool-making. This made it easier to pass along the
methods and tools we developed and the knowledge we acquired about flora
and fauna. We became increasingly better botanists, zoologists, geographers
and engineers within our particular environment. Together, these expanding
cognitive and social abilities gave us an increasing advantage, likely for the first
time in history, over many other smart animals—animals that also have such
abilities, though to a lesser extent (Chapters 2–4).
We then created basic measurement tools such as simple tally mark systems
at least 35,000 years ago. Eventually, by developing agriculture (an estimated
12,000–13,000 years ago)(92) we increased our availability of food and our
labour productivity. This freed up our time and cognitive resources as we
gradually abandoned our hunter-gatherer lifestyle to become sedentary village

Science of Science. Alexander Krauss, Oxford University Press. © Alexander Krauss (2024).
DOI: 10.1093/9780198937401.003.0010
 ANTHROPOLOGY OF SCIENCE 61

dwellers. We could increasingly dedicate our time to other cognitive activ-

ities, beyond meeting our basic needs such as food, shelter and clothing.
We created a wider range of goods and technologies and experimented with
more productive crops. We eventually developed written language and com-
plex numerical systems—two essential features of contemporary science—an
estimated 5000–6000 years ago.(92) These incredible new tools enabled us to
document, calculate and plan food production at a mass scale for growing
populations. With written language, we could then record what we observed
and experienced and better share it across generations, providing an exter-

Downloaded from [Link] by guest on 29 August 2025

nal source of memory (without forgetting) (Chapter 15). At the time, to be
able to develop technologies and tools, we used trial and error, other forms
of basic experimentation and tested hypotheses—though often implicitly. As
villages expanded into cities and some cities into empires over the past several
thousand years, growing populations were able to build on cumulative knowl-
edge and increasingly specialise.(92) We could then develop more sophisticated
methods, tools and knowledge (Chapter 7).
Some cultures maintained a degree of stability over centuries, and even
millennia, such as the ancient Chinese, Greeks and others.(57) Such stabil-
ity allowed us to build extensively on what we know and to develop a wider
range of scientific and technological tools. With more complex numerical
systems, we became able to record and measure what we observed more
systematically—from agricultural plots to the movements of the moon. Popu-
lation density, specialisation (division of labour) and methodological diversity
increase together, with changes in one generally affecting the others. Together,
they have allowed us to grow up in a cultural context with much cumulative
knowledge about the world (Chapter 7). Scholars especially in ancient China
and Greece studied a wider range of phenomena than in earlier civilisations,
from astronomical events and the properties of living animals, to magnetism
and sound. They did so with a logical view of how the world is broadly con-
strued and they viewed certain phenomena as operating according to general
principles.(57,157)
Adopting a pragmatic experimental approach, ancient Chinese developed,
as the first or independently, many more advancements than their Mediter-
ranean counterparts, the ancient Greeks.(157) These include effective immuni-
sation techniques, magnetic compasses, negative numbers and the ‘Pascal’ tri-
angle, astronomical observations of novae, seismographs, paddlewheel boats,
irrigation systems and quantitative cartography, as well as papermaking and
printing that fostered the spread of knowledge.(57,157,158) Ancient Chinese
created smallpox vaccines, which required a complex understanding of the
62 SCIENCE OF SCIENCE

causes and effects of infectious disease, their interactions and how to con-
trol them.(157) It is because the Chinese created a more complex system of
astronomical records than any other culture—including star catalogues and
observations of eclipses and novae—that our records today are able to go back
millennia.(94,157) Such advancements allowed ancient Chinese to control and
predict nature through technology.(57,159) Not only medicine but also geol-
ogy, alchemy/chemistry, geography, technology and engineering were bodies
of knowledge supported by the Chinese state.(94)
In the centuries leading to the 1500s and 1600s, we began to exchange

Downloaded from [Link] by guest on 29 August 2025

technologies more rapidly and eventually for the first time globally.(92) We
had already widely used the methodological approaches of more system-
atic observation, measurement and experimentation to create increasingly
sophisticated technologies. These include eyeglasses, windmills and mechan-
ical clocks in 13th-century Europe (though mechanical clocks were already
created in 8th-century China) and the microscope in the 16th century.(160)
Eventually, these methods and instruments were applied to questions whose
practical relevance was not only directly observable (technological knowl-
edge) but also increasingly not always directly observable (purely scientific
knowledge). At the time, we increased diversification and productivity at a
scale not yet experienced in history.
What made the work of 17th-century scholars possible is a cumulative pro-
cess of greater technological advances and greater awareness of these more
systematic methods already widely used for such advancements. These meth-
ods were then adopted to also study more theoretical questions about the
world, commonly using the newly developed instruments including micro-
scopes, barometers and telescopes that made many of the discoveries possible.
At the time, scholars expanded science by combining our evolved methodolog-
ical abilities and adopting written language, mathematical systems including
geometry and algebra, and diverse technological and scientific knowledge
developed by our ancestors over thousands of years. They built on the existing
work of scholars like Aristotle, Archimedes and Ibn Al-Haytham. The politi-
cal environment in parts of Europe then became increasingly open to question
the status quo and authority.
The printing press, first developed in China and later brought to Europe,
supported the spread of ideas and freedom of thought.(157) Scholars like
Copernicus, Galileo, Boyle and Newton were thus able to grow up in a
context with much cumulative knowledge and favourable demographic,
economic, technological and environmental conditions. They grew up in a
context rich in natural resources and with productive agricultural practices
 ANTHROPOLOGY OF SCIENCE 63

that supported the necessary surplus of labour that enabled them to focus
on studying the world.(92) Using methods in more systematic ways became
institutionalised through newly established education systems as centralised
governments spread in the 19th and 20th centuries. These became our main
means to systematically transfer knowledge and complex methods across
generations. Since the second half of the 20th century, developing digital
technology and computers has exponentially accelerated our ability to pro-
duce and distribute knowledge. Today, we continue making extensions to our
methodological toolbox that allows us to acquire increasingly sophisticated

Downloaded from [Link] by guest on 29 August 2025

bodies of knowledge (Chapter 10).
These are, taken together, the historical hallmarks in the development of
our human mind, social organisation and methods and instruments that have
enabled us, in increasingly complex ways, to develop science. This account
outlines how we over time developed our mind’s methodological abilities and
more complex abilities for language and a numerical and measurement system,
and eventually we became more systematic in observing, experimenting and
creating theories about the world, and ultimately began gathering large sets of
data observations and creating complex methods such as statistics to analyse
them. The combination of these multiple abilities and factors, and especially
the methods we developed, gave rise to growing communities in which we
more and more systematically experimented with and developed better hunt-
ing techniques, then better agricultural crops, eventually better plant-based
medicines and more complex technology, and ultimately sophisticated theo-
ries about the origins of life and the universe. It is an account of the history of
human reasoning, knowledge and science, summarised in this chapter. The
evolution of our knowledge is marked by continual progress when viewed
over long historical periods, though it is not entirely linear. Religious dogma,
famine and war have led to knowledge stagnating at different times throughout
history, such as in parts of the Middle Ages.
Also, differences arise across cultures in the methodological approaches we
develop using our universal methodological abilities, and thus in the ways we
think and our world views. These are shaped by the particular environment
we are exposed to.(57) For ancient Chinese, thinking can be characterised as
being more holistic, continuous and observation-based, and ancient Greek
thinking as more reductionist, principle-searching and abstract.(57) Ancient
Chinese, by adopting a holistic and experimental approach to reasoning,
developed a vast range of systematic methods and remarkable technologies
as outlined above.(157,158) The common view is that scholars began to sys-
tematically study and understand the world only around the 17th century in
64 SCIENCE OF SCIENCE

Europe.(121) But that view neglects the technological and scientific advances
and knowledge made using systematic observation and experimentation by
the ancient Chinese, Arabs and others that Europeans built on.(82)
Ethnographic research of indigenous groups worldwide describes how they
have also acquired comprehensive knowledge, over generations, about our
natural environment including plants and animals, their characteristics, prop-
erties and lifecycles and developed sophisticated classification systems for
them.(55,56) An estimated 70% of medical drugs are plant based and much
of what we know in science about the healing potential of plants has been

Downloaded from [Link] by guest on 29 August 2025

acquired from indigenous cultures.(161) From these cultures, contemporary
science has adopted much experimental knowledge about nutrition, botany,
agriculture, natural pharmaceutics, ecology and agroforestry (Chapter 3).
Now consider a group of people who do not commonly develop abstract
theories about reality and who are viewed, by contemporary scientists, as
being constrained in acquiring reliable knowledge. They use sensory and
reasoning abilities including observation and experimentation, such as trial
and error, as the main means to solve problems and acquire knowledge
about the world. And they are thus not familiar for example with the theory
of relativity or quantum mechanics. Consider also another group of peo-
ple who, in addition, create and use an astronomical map for mathematical
and observational purposes, and apply a 0 to 60 scale as the basis of their
numeral system. Both of these groups have rich knowledge of plant-based
medicines, the characteristics and behaviour of animals and many other fea-
tures of the world. They understand buoyancy and construct boats. They can
think in mathematical terms. But for contemporary scientists, their meth-
ods may—despite developing much practical knowledge—be viewed as rather
primitive. The first group refers to hunter-gatherers, and the second group to
Sumerians.
In sum, science has developed as a gradual process over our species’ his-
tory, in which we have used our evolved methodological abilities of the mind
increasingly systematically and collectively. Over time we have strengthened
the foundation and structure of our tower of science. We have gained, used
and further developed extensive knowledge and methods from our earlier
ancestors. These have been absorbed into contemporary science—from the
plant-based medicines of indigenous populations to the technological, sci-
entific and mathematical advances of ancient cultures. Science, commonly
defined as the study of the ‘world through observation, experimentation,
and the testing of theories,’(65) is not just a product of 17th-century Europe.
Only when we integrate evidence about human evolution and the mind
 ANTHROPOLOGY OF SCIENCE 65

(Chapters 2–5) with social organisation (Chapter 6) can we gain a more coher-
ent understanding about the foundations of how we acquire knowledge and
do science—and how we expand science, when we also integrate evidence
from the broader range of factors (Chapters 7–15). Anthropology of science
is just one integral part of understanding the foundations of science. It helps
us explain how we have gotten to this point in history and provides much
evidence of how the methodological roots of science have evolved.

Downloaded from [Link] by guest on 29 August 2025

 10
Methodology of Science

Over our species’ history, we evolved methodological abilities of the mind

(observation, problem solving and experimentation) that we use together

Downloaded from [Link] by guest on 29 August 2025

with increasingly complex methods developed using these abilities (controlled
experimentation, statistics and X-ray methods). Science has always been
grounded in these evolved methodological abilities (our universal method-
ological toolbox) that have enabled us to develop vast bodies of knowledge
by creating sophisticated methods and tools (our adaptive methodologi-
cal toolbox). As we face constraints when using our evolved abilities to
do science, we have developed methods and instruments to reduce these
constraints. Such constraints are cognitive (such as limited sensory abil-
ities, cognitive bandwidth and memory), social (such as cultural values,
norms and interests), geographic (such as differences across contexts that
require conducting studies in multiple contexts) and so forth (Chapters 2–8
and 13).
A central argument throughout the book is that we develop new methods
and instruments that enable us to better access and understand the world and
make new scientific advances by addressing our human and methodological
constraints (Chapter 4). Mathematical methods for example are used across
fields, from theoretical physics to economics, to help us systematically calcu-
late and measure phenomena and represent them using algebraic equations
(Chapter 13). Controlled experimentation and randomisation (such as ran-
domly allocating participants to groups in an experiment) are used across
fields, from biomedicine to psychology, to reduce human biases in designing,
implementing and analysing studies. Datasets allow us to store informa-
tion and sets of observations externally. Magnetic resonance imaging (MRI)
enables us to detect phenomena like magnetic fields and radio waves that we
do not have sensory receptors for and is used widely for medical purposes.
Computers aid us in efficiently processing, analysing and sharing large sets
of observations in ways that would not be possible without them (Chapter
12). Electron microscopes vastly enhance our visual capability and enable us
to perceive miniscule objects using the wavelength of an electron, far exceed-
ing the magnification of light microscopes. Nanoparticles, microorganisms,

Science of Science. Alexander Krauss, Oxford University Press. © Alexander Krauss (2024).
DOI: 10.1093/9780198937401.003.0011
 METHODOLOGY OF SCIENCE 67

crystals and molecules come into our visual scope with an electron microscope
(Figure 10.1).
Such methods and instruments we create using our mind greatly extend
our evolved mind. Our methods and instruments, once created, are mainly
external resources that can function as an efficient external extension of
our mind (Chapter 4). In general, when we design any new method or
instrument, we are asking how we can answer a question by improving our
existing cognitive, sensory or methodological abilities. We are asking how
we can address a problem by reducing a human constraint or bias to doing

Downloaded from [Link] by guest on 29 August 2025

science.
A companion study analysing the methods and instruments used to
make over 750 major scientific discoveries, including all Nobel-Prize-
winning discoveries, provides direct evidence of how we drive science.
It illustrates that 94% of discoveries used observation, 75% experimenta-
tion and 100% developed a new method or instrument (X-ray methods,
radar telescope etc.) that was applied to make the discoveries.(64) The main
conclusion is that scientific methods and instruments we develop drive
new discoveries—that is, they are the main mechanism through which

Figure 10.1 An electron microscope, one of the most important scientific

instruments used across fields (left), with a photograph of coronavirus taken
using it (right).
Source: David Morgan; NIAID.
68 SCIENCE OF SCIENCE

we have directly made scientific discoveries. The study has five main
findings:

• Directly after developing a method like modern statistics and X-ray

diffraction, or an instrument like the electron microscope and chro-
matography, we have been able to make dozens of scientific discoveries
with each of these methods and instruments.
• Discoveries can only ever be made after we develop the needed methods
or instruments and are commonly made soon after developing them: 10%

Downloaded from [Link] by guest on 29 August 2025

of discoveries have been made simultaneously or in the same year of cre-
ating the new method or instrument used to make the discovery, 20%
within two years, 32% within four years and 52% within ten years.
• The ten big methods and instruments that have been most commonly
used to make discoveries and have revolutionised science are statisti-
cal/mathematical methods, spectrometers, (electron) microscopes, X-ray
methods, chromatography, centrifuges, electrophoresis, telescopes, ther-
mometers and lasers.
• There are three ways new methods and instruments drive new break-
throughs. One way is that a researcher develops a method or instrument
themselves that they apply to make a scientific discovery (25% of all dis-
coveries). The second way is that a researcher develops a new method or
instrument that is applied by another researcher to make a scientific dis-
covery (50%). The third way is that a researcher develops a new method
or instrument that is the major discovery itself (25%), such as the electron
microscope and particle detector.
• Many new scientific tools enable making multiple discoveries, at present
and in the future, and in different fields beyond which the tools were
designed. They thus causally enable discoveries unforeseeable by the
inventors of the tools. The discoveries of cells, bacteria and mitochondria,
and Uranus, galaxies and pulsars were only possible after developing the
microscope and telescope and they were not being searched for. Many
breakthroughs are exploratory (not hypothesis-driven) that we make by
applying a new scientific tool.(64)

Scientific advances are thus made possible through new methodological inno-
vations. While conventional research applies existing methods, cutting-edge
research that produces new discoveries uses newly developed methods or
instruments that enable studying the world in a new way. This direct link
between new methodological innovations and new discoveries is observed
 METHODOLOGY OF SCIENCE 69

in Figures 10.2 and 10.3. And this direct link has been largely overlooked in
understanding what drives science and its limits.(64)
New scientific methods and instruments drive not only new scientific
discoveries but also new scientific fields. Developing particle detectors and
accelerators gave birth to high-energy physics/particle physics in the early
1900s, and the field continually expanded by developing more advanced detec-
tors and accelerators. Creating the electron microscope in 1933 helped open
the field of molecular genetics. Developing X-ray diffraction methods in 1912
gave rise to protein crystallography in 1946. Femtosecond spectroscopy cre-

Downloaded from [Link] by guest on 29 August 2025

ated in 1985 led to the emergence of femtochemistry in 1988 (Figure 10.4).
Newly developed methods and instruments are often the defining feature of
new fields.(64)
Another dimension of scientific methods is that each method uses a com-
mon measurement, and a common measurement is closely linked to being
able to develop more replicable and reliable knowledge among independent
scientists. Rigorous findings can then be viewed independent of individual
scientists, as they can be tested for accuracy by a group of scientists using
Number of new methods/instruments and discoveries
0 20 40 60

1800 1820 1840 1860 1880 1900 1920 1940 1960 1980 2000 2020

Methods/instruments developed to make the discoveries

Discoveries

Figure 10.2 Trends in the development of new methods and instruments

closely follow trends in subsequent discoveries.
Data reflect the number of central methods/instruments and discoveries made in a given period
among 653 major discoveries, from 1800 to 2022, and include all Nobel-Prize discoveries.(64)
70 SCIENCE OF SCIENCE

Legend:

2000
Methods and instruments
Year that main method/instrument was developed
developed (|), and the
subsequent discoveries
made using them (•)
1800 1900

Downloaded from [Link] by guest on 29 August 2025

1700 1600

1600 1700 1800 1900 2000

Year that discovery was made using it

Figure 10.3 New discoveries are sparked by developing methods and

instruments needed to make the discoveries.
Data reflect 734 major discoveries since 1575, including all Nobel-Prize discoveries, and the
central methods/instruments used to make them. Data illustrate that the year the main method or
instrument is developed is strongly correlated with the year the discovery is made.(64)

the same measurement method. This reduces the possibility of individual

scientists influencing results (Chapter 13).
However, methods and instruments, while reducing human influences and
enhancing our limited abilities, can bring constraints and possible biases in
developing knowledge. Each mathematical technique, X-ray method and sta-
tistical method generally has limits as to which questions we are able to study
and what results and conclusions we are able to derive using it. Each has a
set scope within which we can capture or model phenomena in the world,
design, implement and evaluate experiments, and interpret results.(42) Each
faces constraints and requires making assumptions.
One of the best ways for us to reduce individual methodological constraints
in science is by applying multiple methods.(40,42) For each method can pro-
vide different evidence and perspectives into a phenomenon. Each method
can confirm whether results consistently point in the same direction. To better
understand for example how a medical treatment may affect patients or how a
government policy may influence citizens, it is best to apply multiple methods.
These range from quantitative methods including RCTs and observational
 Downloaded from [Link] by guest on 29 August 2025
Econometrics-driven fields Legend
econometrics (1933)
Scientific fields
corporate finance (1954) (those that are
financial econometrics (1982)
economic forecasting (1941) simultaneously a method/
instrument in italics)
Particle detector/accelerator–driven fields
particle detector–Wilson cloud chamber (1932) Method/instrument that
particle detector–cloud chamber (1911) particle detector–bubble chamber (1952) opened new field (but is
particle accelerator/cyclotron (1929) particle detector–hydrogen bubble chamber (1959) not itself a field)
high–energy physics/particle physics
(~1900s/1910s)

X-ray–driven fields
radiation genetics (1927)
X-ray analysis (1895) X-ray crystallography (1912) protein crystallography (1946)

diagnostic radiography (1895) molecular biology (1938) X-ray astronomy (1962)

Microscopy-driven fields
microscopy, improved–with silver staining technique … electron microscopy (1933) cryo-electron microscopy (1981) STED microscopy/super–resolution fluorescence microscopy (1994)

neuroscience (1873) phase–contrast microscopy (1930) electron crystallography (1962) single–molecule microscopy (2006)
modern cell biology (1945) scanning tunnelling microscopy (1982)

Spectroscopy-driven fields spectroscopy (femtosecond) (1985)

spectrography (1859)* mass spectrometry (1919) molecular spectroscopy (1957) photoelectron spectroscopy (1958) frequency comb (spectroscopy)/frequency combs (2000)

neutron diffraction/spectroscopy (1946) NMR spectroscopy (1946) MRI (1973) femtochemistry (1988)
protein NMR spectroscopy (1985)

1850 1900 1925 1950 1975

Figure 10.4 Developing a new central method or instrument opens new scientific fields (illustrations from Nobel Prizes for
developing methods/instruments and subsequent fields).
∗
The first spectrograph, developed in 1859, is the only instrument included that did not receive a Nobel Prize, which was used to open the field of mass
spectrometry.(64) NMR, nuclear magnetic resonance; STED, stimulated emission depletion microscopy.
72 SCIENCE OF SCIENCE

studies, to qualitative methods including a set of rich single cases, consen-

sus among a group of experts and historical evidence.(ibid.) Methodological
constraints can at times also be reduced through statistical simulations, boot-
strapping (a resampling method in statistics) and calibration. Yet assumptions,
limitations and biases that can influence our results and knowledge cannot
always be entirely avoided. Some form the basis of our very methods (dis-
cussed in Chapter 13).(ibid.)(162) Others enter the scientific process before and
after we apply methods and thus cannot be reduced using methods. These
range from data collection techniques and selecting study design features, to

Downloaded from [Link] by guest on 29 August 2025

interpreting outcomes and using the results beyond the study context.(42)
At a deeper epistemological level, with the range of methods and instru-
ments we have developed thus far in history, we do not have an inherent
inclination to view the world using one particular method or classification sys-
tem instead of another. We are not naturally inclined to use the metric instead
of imperial system, Fahrenheit instead of Celsius, probabilistic instead of non-
probabilistic methods, a numerical system with 10 as its base instead of 60 and
quantitative instead of qualitative methods for all questions (Chapters 8 and
9). We do not have an innate predisposition to acquire knowledge and concep-
tualise phenomena in the world using mathematical equations and universal
laws instead of using no mathematics and no laws. Which method we develop
and use depends on our particular question and scientific field. We do not
inevitably explain phenomena using one disciplinary perspective or method
instead of multiple. But it is how we are methodologically trained and which
methods we have developed thus far that leads us to do so (Chapters 2–6). Our
scientific methods such as computational and statistical techniques and X-ray
methods that we recently developed, and thus our theories we create using
them, are in a process of continual refinement over time.
In sum, because our evolved mind faces constraints in perceiving, process-
ing and explaining our complex world (Chapters 4 and 5), we have developed
a powerful methodological toolbox that significantly expands our cognitive
abilities and bodies of knowledge. Our tools of science in turn can face
constraints in how we represent and model the dynamic character of phenom-
ena using them. Our mind’s methodological abilities and complex methods
and instruments we create using these abilities (our universal and adaptive
methodological toolbox) are at the centre of understanding science. Methods
and instruments we develop are a factor we can directly influence and are very
important in shaping the foundations, limits and advancement of science, as
we illustrate comparatively to other factors in Chapter 16. Our scientific meth-
ods and instruments are also the only factor that underpins all other factors
 METHODOLOGY OF SCIENCE 73

(Chapters 2–15). A central focus on scientific methods is thus essential to the

integration of science of science. Our tower of science is made possible by our
evolved methodological abilities (that account for its foundation) and our sci-
entific methods and instruments (that account for the different floors of its
structure), which together determine how far we can observe, measure and
understand the world. Ultimately, the discovery of the methods and instru-
ments we apply are rarely cited and referenced in our studies even though
they make our studies and advances possible. So citations do not capture their
enormous impact on science and this is a shortcoming in our scientometric

Downloaded from [Link] by guest on 29 August 2025

system.
 11
Scientometrics and Network Science

Science describes and explains the world through research articles and books
that are organised into scientific fields. Scientometricians including network

Downloaded from [Link] by guest on 29 August 2025

scientists analyse this scientific literature. To do so, they rely on the indi-
cators of citations and publication counts to study issues such as research
productivity, team collaborations, career dynamics, networks of scientists and
institutions, and novelty in science.(4,5,8,9,35,163,164,38) They use large-scale data
(big data) and network analysis and search for patterns in such data. That
is scientometrics in a nutshell. Scientometrics is of interest to all researchers
as citations and publication counts largely determine whether researchers get
research grants, academic jobs and promotions. Most measures of scientific
impact and success use citations (the number of times a publication is cited by
other researchers). Our current reward system in science is deeply embedded
in this metric.
The British bibliometrician and historian of science Derek John de Solla
Price is commonly considered the father of scientometrics. He pioneered work
in the quantitative analysis of scientific literature and citation analysis to assess
scientific productivity. In his book Little Science, Big Science published in
1963, de Solla Price examined the structure and growth dynamics of scientific
knowledge.(70)
Since then, a central topic in scientometrics, and in science of science in
general, has been innovation. Studies illustrate that researchers are gener-
ally risk-averse, choosing to study phenomena in which they already have
expertise and with which they are familiar. This limits what is studied in the
future and making potential new discoveries. Researchers willing to explore
new areas and undertake a riskier career, moving from traditional topics to
riskier innovation, are more likely to expand a field and make new discoveries.
Research strategies that are conservative foster individual careers rather than
science as a whole. What characterises high-impact science are conventional
combinations of existing work that integrates novel combinations of not-yet-
connected topics(165,166,167) or research methods. To increase the impact of
innovative and cross-disciplinary research, scientists need to show how it con-
tributes to established research. Otherwise research is less likely to achieve

Science of Science. Alexander Krauss, Oxford University Press. © Alexander Krauss (2024).
DOI: 10.1093/9780198937401.003.0012
 SCIENTOMETRICS AND NET WORK SCIENCE 75

its maximum impact.(166) When it comes to incentivising innovation, a study

finds that medical researchers receiving greater freedom in experimentation
and failure and greater rewards for long-term success are more likely to pro-
duce higher impact research.(168) Funding bodies can thus strategically foster
higher risk research projects that test unexplored ideas, areas and diseases(5)
and unexplored research methods. Funding bodies and academic journals also
evaluate submitted manuscripts and project proposals partly by predicting
their potential future impact.(35)
In terms of researcher productivity and impact, major discoveries are gener-

Downloaded from [Link] by guest on 29 August 2025

ally made by younger researchers and explained by their higher productivity
and not yet securing permanent positions.(4,7) Studies generally illustrate a
median age of discoverers between their mid-30s and mid-40s.(4,7,169,170) Sci-
entists of all ages can make major discoveries as long as they are productive,
though the likelihood decreases over scientists’ careers.(7) Assessing all Nobel-
Prize-winning discoveries shows that only 7% of discoveries were made by
scientists after the age of 50 and only 1% after the age of 60.(64) So if researchers
do not make their great contribution to science earlier in their career, they are
unlikely to do so and receive a Nobel Prize in their lifetime.
As science expands, each generation has a larger body of knowledge and
methods at its disposal. This generally leads to more specialised (and at
times longer) education and greater teamwork in science (Chapters 6 and
7).(38,171,119) So reaching the research frontier takes longer and longer.(8)
A study of about 20 million research articles reflects a related trend of a
general shift towards teams across all scientific fields: from slightly over
one to about five individuals per team in science and engineering between
1900 and the early 2000s.(172,173) Larger teams can often be better able to
develop new combinations of ideas and apply different methods (Chapters
6 and 7). Greater collaboration also generally increases research quality
and impact, with prominent European researchers collaborating with other
prominent European researchers more often than prominent North American
researchers collaborate with each other.(174)
Another central question in scientometrics, and in science of science in
general, has been what drives the world’s top researchers.(4,5,7,35,164) A study
of early-career factors driving success in science identifies four key factors
among the world’s top 100 researchers across fields (measured by the high-
est h-index): co-authoring with other top 100 researchers, working at a top 25
ranked university, publishing a paper in a top 5 ranked journal and publish-
ing most papers in first quartile (high-impact) journals. Over 95% of the 100
researchers with the highest h-index, across multiple fields, had at least one of
76 SCIENCE OF SCIENCE

these four features in the first five years of their career.(175) Co-authoring with
leading researchers is one of the best ways to become visible early on in a sci-
entist’s career, through more citations and mentorship.(176) Being at one of the
top 25 ranked universities in one’s early career is related to greater research
impact later, as researchers can enjoy a high-quality research environment
and access to greater resources. Researchers’ early career is thus important,
with the choices and effort that young researchers make shaping their later
success (Figure 11.1). Yet because these early-career attributes of successful
scientists are predictable of later success, it suggests that the scientific sys-

Downloaded from [Link] by guest on 29 August 2025

tem is already relatively closed. It also points to shortcomings in using the
common and highly influential indicators of success, namely citation metrics.
This is because early career advantages—measured using these metrics—
are so strong that they help predefine ‘highly successful scientists’ without
further information about the content or social and policy impact of their
research.(175)
Measuring scientific success using citation counts can however constrain us
in developing new ideas since it gives an advantage for highly-cited researchers
to become even more cited—that is, the rich just get richer. In turn, it pro-
vides a bias for researchers, with limited time, to use more cited and thus at
times older research. It also disadvantages younger researchers and innova-
tive researchers working between disciplines and paradigms.(35,131) Scientific
institutions need to place greater focus on other metrics of success beyond cita-
tions, such as levels of innovation and societal relevance of research(5) and the
development of new methods and instruments.

Endline
Baseline Prominent researchers
with top h-index

Early-career factors:
most co-
a top 5 papers in being at a authoring
journal first top 25 with
paper quartile university prominent
journals researchers
1st Citations: at 10 at 15 at 20
at first 5 years of scientists’ career
publication h-index: years years years

Figure 11.1 Early-career choices and factors can shape later success in
science.(175)
 SCIENTOMETRICS AND NET WORK SCIENCE 77

Tracing science and the scientific process through published literature and
citations (the scientometric approach) does not uncover many relevant factors
to understanding science. Citation counts are a metric that does not capture
the impact of most major past discoveries that account for our vast bodies
of knowledge that make up science, because most major discoveries through-
out history were made before the widespread use of references and citations
that began in the second half of the 20th century.(177) Citation counts do not
capture the immediate impact of new ideas or breakthroughs in science (as
citations take time to accrue) or the impact on policy or society (as they cannot

Downloaded from [Link] by guest on 29 August 2025

be cited).(39) Citation counts do not capture well the powerful role of our sci-
entific methods and instruments in conducting and advancing science. Studies
at most mention techniques and procedures used in an experiment but rarely
cite and reference the discovery of the method or instrument they use. Our
most important scientific methods and tools, such as Martin and Synge’s parti-
tion chromatography method,(178) Townes’ maser/laser,(179) Ruska’s electron
microscope(180) or Svedberg’s centrifuge,(181) would each have received mil-
lions of citations as they are mentioned in millions of studies, according to
Google Scholar. This far exceeds any scientific studies in any field. But these
method-making studies only have received between a few hundred and a few
thousand citations each.
We require a new norm in science of always systematically citing the meth-
ods and instruments we use, so their vast importance becomes powerfully
evident. This book reflects a fundamental rethinking needed in how we study
and understand science, moving away from just using ex-post indicators of
output, mainly citations and publications,(4,5,6,7,8,35,36) in order to under-
stand all other features of science that they do not capture. It reflects a shift
towards also studying the equally important ex-ante indicators of the process
and inputs of science and discovery. This requires placing our methods and
instruments at the centre of study but also greater focus on the broader demo-
graphic, social and economic contexts of science—which are not captured with
citations (Chapters 2–15). Also, most existing publications in scientometrics
study one factor using descriptive data. This common approach in the field
resembles the early emergence of the social sciences before the advent of more
rigorous experimental designs, controls, regression analysis and causal identi-
fication strategies. Using such methods, we can identify the importance of one
factor relative to other factors and can provide causal explanations. Big-data
has not been able to push the field forward with robust causal knowledge. For
that, we have to turn to more rigorous methods used in fields like public health
and economics, and we discuss these in the conclusion.
78 SCIENCE OF SCIENCE

In sum, studying scientific literature and citations contributes to our under-

standing of science by providing insight into the conditions shaping scientific
impact, collaboration and productivity. Scientometrics alone, like any sci-
ence of science subfield, cannot however provide us with a comprehensive
understanding of science. Scientometric research relies heavily on using com-
puters and statistical methods, which are both used in all scientific fields and
are important pillars of the tower of science that we discuss in the next two
chapters.

Downloaded from [Link] by guest on 29 August 2025

 12
Computer Science of Science

We are constrained by our limited cognitive and computing capacity when

studying the world. We are flooded with vast amounts of new data and pub-

Downloaded from [Link] by guest on 29 August 2025

lications each year, at a pace far exceeding our human abilities to process the
expanding influx of information and data. Computers play a central role in
science and studying science by expanding our limited cognitive resources,
memory and capacity for data processing, statistical analysis and simulations
(Chapter 4).(44) Computers are used across all research fields, completely
transforming the way we do science. Computational methods are also the
foundation of a number of scientific fields—from computational physics, biol-
ogy and cognitive science to data science and bioinformatics. Computers are
so central to science that they have become an integral piece of the overall
picture of science of science.
New discoveries and methods can now spread across the globe and reach
scientists with computers and internet basically instantaneously. We now have
almost immediate access to the vast range of existing methods and bod-
ies of knowledge in science, for the first time in history. The World Wide
Web is today’s version of China’s Yongle Encyclopaedia in the 15th cen-
tury that aimed to bring together all available knowledge in the world in an
encyclopaedia.(182) Computers are crucial in conducting science much more
efficiently by automating scientific processes and making it much easier to
organise, store and retrieve enormous amounts of data through databases.
Computers and the internet also connect scientists in ways and in num-
bers unparalleled before, allowing for much greater collaboration around the
world.(44) In short, computers do such a good job at maximising efficiency that
we have become highly dependent on them—and it is difficult to imagine our
lives without them.
A critical bottleneck in making computers and the internet possible was
overcome in a landmark article titled A mathematical theory of communica-
tion, in 1948. The American mathematician and electrical engineer Claude
Shannon published this seminal article in which he addressed the question
if there was a unified theory for communication—a general theory for how
we produce and transfer information.(183) His answer was to conceive the

Science of Science. Alexander Krauss, Oxford University Press. © Alexander Krauss (2024).
DOI: 10.1093/9780198937401.003.0013
80 SCIENCE OF SCIENCE

digital nature of information as binary digits (0 or 1). This completely shaped

how we began to use computational data. The general model of communi-
cation he developed is simple: sent information (data) becomes the received
information (data) (that can possibly be distorted by noise)—that is, a trans-
mitter converts information into a signal (that can be distorted by noise) and
is then decoded by the receiver (Figure 12.1). Central to the idea is that Shan-
non modelled the transfer of information probabilistically. Today, data are
widely used as binary digits across science, for example in statistical anal-
ysis to capture phenomena in the form of variables.(183) By demonstrating

Downloaded from [Link] by guest on 29 August 2025

how we can quantify digital information, Shannon’s work has been called the
Magna Carta of the digital age.(184) It helped pave the way for scientific compu-
tation, complex quantitative analysis and mathematical modelling. Scientific
computation enables simulating phenomena—from physical and biological
systems to economic actors and societies. Algorithms embedded in statistical
computer programmes (like Python, R and Stata) enable computing complex
regression analyses and simulations using thousands or even millions of data
observations.
A rapid increase in computing power and available data has also acceler-
ated growth in artificial intelligence, providing new opportunities to apply
computer technology in science.(44,185) We most directly advance science by
developing new methods and instruments that address our cognitive and
methodological constraints (Chapters 4 and 10) and there is great potential for
methods of artificial intelligence to expand and automate aspects of the way we
do and study science beyond our mind and conventional methods. Machine
learning applies computer algorithms that improve automatically through
an iterative process of using a given dataset. It allows us to delegate some
aspects of data collection and analysis to automated computer programmes.

Information
source Transmitter Receiver Destination

Signal Received
signal
Message Message

Noise
source

Figure 12.1 Diagram of Claude Shannon’s model of communication that made

our digital age possible.(183)
Source: photograph from Tekniska Museet via Wikimedia Commons.
 COMPUTER SCIENCE OF SCIENCE 81

This is especially relevant when we study phenomena for which we have large
amounts of data or when we require making quick decisions.
In the biomedical sciences for example, we have methods for drug design
that automate many mechanical tasks performed by biomedical researchers.
Researchers provide the collected data that are coded and inputted into
robotic platforms that automatically conduct a series of experiments and
generate results.(46) Deep learning methods can help medical practitioners
decide whether brain scans and images of patients indicate signs of a stroke
or tumours, with algorithms reaching similar accuracy as medical specialists

Downloaded from [Link] by guest on 29 August 2025

but much more quickly. These new methods can complement (not replace)
human expertise.(45,186) Machine learning has also been applied to help pre-
dict protein structures using large genomic datasets, to estimate the effects
of climate change on cities using large climate datasets and to detect signals
in large astronomical datasets.(187) Artificial intelligence can help us solve
problems and make decisions when we deal with greater complexity and it
can help us reveal new structures and patterns in large amounts of data—
data mining.(187) Artificial intelligence programmes are also being designed
to rerun data analyses with an existing or new computational method and
produce new hypotheses that can verify or diverge from existing research.
Artificial intelligence systems may reduce some biases and errors that arise
due to human influence in the scientific process (Chapters 4 and 5),(4,44)
although they also give rise to other biases and errors that only manual data
collection and analyses can detect and avoid. This is due to the diversity in
methods, evidence and language across studies and fields that constrains gen-
eralised automatisation using computers. Such programmes can help conduct
part of the routine work in science, so researchers could dedicate more atten-
tion to other (including creative) aspects of science. Yet we face important
constraints in understanding what factors actually lead such programmes to
particular results (that is, understanding how they arrive at results) which are
thus not always replicable.(4,44,185)
Computer scientists have also offered a computational account of dis-
covery by attempting to model and simulate scientific discovery processes.
(188,189,190,191)
Computational accounts have been ambitious, attempting to
develop computer programmes and algorithms that could drive new discover-
ies. But they have mostly only focused on the path from data to scientific laws,
and do not analyse the role of methods and broader background factors in the
discovery process that are taken as given. To date, they have had only limited
success in reproducing past scientific discoveries and do so at a high level of
abstraction.(187,188)
82 SCIENCE OF SCIENCE

Greater computational power has led to big data and the analysis of vast
complex datasets. This can help us overcome small sample sizes and low sta-
tistical power, but it generally brings more noise to data and faces the same
challenges that arise with smaller datasets.(192) In general, artificial intelli-
gence and machine learning are becoming increasingly important for some
tasks of collecting, cleaning and analysing data in different fields. Large lan-
guage models (like ChatGPT) can assist certain researchers with various tasks,
functioning as a kind of research assistant. These include quickly summaris-
ing key points and trends in a field and suggesting potential hypotheses and

Downloaded from [Link] by guest on 29 August 2025

questions to explore based on existing knowledge. They can help with coding,
identifying patterns in data and offering preliminary interpretations of results,
but can also assist with editing and improving the clarity of text. All outputs
must be validated and updated by researchers. This enhances productivity and
enables researchers to focus on higher-level tasks. At earlier conceptual stages
of research, they seem to be more helpful than at later stages. Overall, these
tools are expanding rapidly, with new developments continually being made.
Though, the early hype around new methods like initially with big data
analysis and network analysis can fade as we become aware over time
of the limitations and biases facing the methods across different fields.
Science, despite computational advances and our new dependency on com-
puters, remains human centred. Experimentation and deep methodological,
empirical and theoretical understanding are central aspects of science and they
remain largely driven by humans, not computational machines. While we del-
egate more tasks in science to computers and while some fields become more
computational, science continues to be led by humans—by our mind, meth-
ods and instruments that we are presently better able to understand than many
artificial intelligence programmes.
In sum, computers have become an important tool for us to conduct and
study science much more efficiently, particularly how we share and use data,
methods and knowledge—quickly and globally. By transforming how science
is conducted, computers have become an essential part of the foundation of
science, and thus our tower of science. We have used artificial intelligence
methods in tasks such as processing and analysing data, and classifying and
detecting statistical patterns—though they have not yet been able to make new
major discoveries [Link]. (187) Using computer technology in science is also
closely linked to data science, with scientists using statistical computer pro-
grammes and binary digital data (0 or 1) across all fields for data analysis,
modelling and predictions.
 13
Statistics and Mathematics of Science

Revolutions are easy to spot when they happen as a single extraordinary event.
The impact of some methodological discoveries, such as X-ray analysis in 1895

Downloaded from [Link] by guest on 29 August 2025

and the gene editing method CRISPR in 2012, was immediately known around
the world. Other methodological discoveries are difficult to spot as they can be
driven by multiple advances over centuries. The development of mathematical
and statistical methods reflects such slow and quiet methodological revolu-
tions that did not make international headlines. But they have fundamentally
transformed how we do science and how we understand the world.
Statistics and mathematics are arguably the two most widely used methods
across science, with statistics involving the collection, analysis and interpre-
tation of data. Assessing over 750 major scientific discoveries, including all
Nobel-Prize discoveries, illustrates that, since 1900, 99% of discoveries have
used mathematics and 77% used statistics.(64) In physics, the field’s two central
theories are quantum theory (which incorporates probabilities and exhibits
indeterministic behaviour) and relativity theory (which is described by math-
ematical formulas and is deterministic). In economics, the field is divided
into applied economics (using statistical analysis) and theoretical economics
(using algebraic equations). Most scientific theories are formalised using alge-
bra and calculus, such as the equation E = mc2 . In many fields of science,
inferential statistics (the analysis of data to make inferences about phenom-
ena) has however become synonymous with the scientific method. Inferential
statistics has revolutionised empirical sciences like biomedicine, psychology
and social science by enabling us to generate causal evidence. Not only in sci-
ence but also in our everyday life we often use statistical reasoning: we ask
what is the likelihood of a medication we take to work, or our planet’s cli-
mate to increase by 3 ∘ C within a century or getting lung cancer from smoking.
Statistics and mathematics are at the foundation of most of science. And like
computers, it is difficult to imagine what contemporary science would look
like without them.
We humans, with our limited cognitive resources and processing capacity
(Chapter 4), have developed statistical methods that we apply together with
computers to analyse hundreds and even millions of observations in a single

Science of Science. Alexander Krauss, Oxford University Press. © Alexander Krauss (2024).
DOI: 10.1093/9780198937401.003.0014
84 SCIENCE OF SCIENCE

dataset. Without these methodological tools, we would not be able to study

the world in such ways, and the advent of computers was key in quantifying
and digitalising nearly all scientific fields (Chapter 12). Modern statistics trans-
formed empirical science by allowing us to study the world with vast amounts
of data in more complex ways and conduct and analyse larger-scale exper-
iments. We apply statistical methods to study basically any phenomenon in
science, from cells and viruses in populations, to planets, economic markets
and science itself. In general, experimental science involves collecting statisti-
cal data about a given phenomenon from a sample, generating results and often

Downloaded from [Link] by guest on 29 August 2025

estimating causal relationships, and then developing an explanation or theory
for the evidence. Science commonly aims to establish cause–effect relation-
ships in the world. And our understanding of causation is inherently statistical
in experimental science: the more data we collect to observe a given phe-
nomenon in different contexts and time periods, the more reliable our findings
become.
Modern statistics developed in stages largely around the early 20th cen-
tury and culminated in the landmark book Statistical Methods for Research
Workers published by the British statistician and biologist Ronald Fisher in
1925.(193) It marked the first full-length book on statistical methods and was
critical in establishing and spreading modern statistics. In this seminal book,
Fisher developed some of today’s most important and widely used statisti-
cal methods by scientists across all fields. He created the analysis of variance
(ANOVA) technique that enables us to analyse the differences among sub-
groups or subexperiments within an experiment. It namely enables assessing
the statistical differences between the means of the groups of a study sample—
for instance of individuals or agricultural plots receiving different treatments.
Fisher also popularised the use of the p-value, or probability value, in statis-
tics, and he proposed using a 1 in 20 probability of a result arising by chance
as a threshold for testing the statistical significance of experimental results
(null-hypothesis testing).(193) His proposed threshold has become a universal
constant. Today, most academic publications report the statistical significance
of their results using his threshold, although debates are still ongoing about
whether 0.05 is the best threshold (Figure 13.1). In light of the essential role of
statistically rigorous research designs in shaping later study outcomes, Fisher
stated that ‘To consult the statistician after an experiment is finished is often
merely to ask him to conduct a post mortem examination.’ Fisher also pio-
neered the use of randomisation in agricultural experiments to reduce bias
by randomly (rather than arbitrarily) selecting a sample from the general
population.
 STATISTICS AND MATHEMATICS OF SCIENCE 85

p = 0.05

–3 –2 –1 0 1 3
z = 1.96

Downloaded from [Link] by guest on 29 August 2025

Figure 13.1 Most experimental results in science are tested with a threshold of
statistical significance of 0.05 (1 in 20 odds of a result arising by chance), as
proposed by Ronald Fisher.
Source: photograph from Wikimedia Commons, 1913.

A central limitation of statistical and mathematical methods is however that

many aspects of the world cannot be easily studied using them since they can-
not be captured well numerically. These range from ecosystems and many
long-term health issues to psychological states and political institutions. Quan-
titative measurement is generally viewed as a central feature of scientific rigour
and replicability. And robustness is a feature of measurement tools like statis-
tics that enable us to test and reproduce results. Scientists generally direct their
attention to those phenomena in the world that are most quantifiable—and
thus measurable and amenable to statistical analysis. The level of quantifi-
ability can range widely. On the one hand, there are highly stable physical
phenomena (such as measuring gravity and the volume of certain objects) and
phenomena that only exist as numerical constructs (such as age and income).
On the other, there are not directly quantifiable phenomena (such as com-
plex ecological systems and biological evolution) and historical events (such
as the given year of the extinction of dinosaurs or the political causes of the fall
of the Berlin Wall). Where a phenomenon falls on this spectrum determines
whether we can more easily measure it (and thus whether it often appears more
scientific) than other phenomena.
We make our observations of phenomena quantifiable and amenable to sta-
tistical analysis to be able to study them. We must fit observations of the world
into variables (using for example binary values like 0 or 1) to be able to cal-
culate statistical relationships between phenomena. Such binary variables we
create capture an aspect of the world in quantitative terms. Creating such vari-
ables influences the phenomena we study by altering how we measure and
86 SCIENCE OF SCIENCE

observe the world. For phenomena in the human sciences are not naturally
binary. No traits or behaviour fall squarely into a black or white description
of the world. A common measure in medical studies is for example mortality,
with which the effectiveness of a treatment is assessed by the duration that a
study participant survives while taking the treatment. Yet such a binary mea-
sure does not capture participants’ quality of life or level of pain, including the
extent to which they suffer adverse events or are hospitalised before passing
away.(42) This is a typical example of how scientists’ value judgements enter
into scientific evidence depending on how we define a variable used in a study

Downloaded from [Link] by guest on 29 August 2025

(Chapter 6).
There are a number of problems that commonly face statistical studies
and affect the quality of our evidence. Common problems in producing
reliable and replicable scientific results are small sample size and low sta-
tistical power,(194) p-hacking and selective reporting, small effect sizes and
HARKing.(29,40,162,195,196,197) Small sample size and low statistical power can
affect studies negatively by increasing the chance of false positive results. P-
hacking occurs when researchers for example collect additional data after
assessing the statistical significance of results or exclude some outliers to
improve the statistical significance of results. It arises given pressure to report
only statistically significant or positive results, since journals are less likely
to publish studies with statistically insignificant or negative results (publica-
tion bias). HARKing arises when researchers present unexpected results as if
they were the initial intended hypothesis of the study, or present exploratory
research as if it was confirmatory research (i.e. testing a pre-established
hypothesis) (Chapter 5). Such research practices often lead to statistical biases
in studies. This includes sampling bias (when individuals in a study sample are
not representative of the broader population) and measurement bias (when
data in a study are inaccurately measured or classified).(29,42,162,194,195)
While such biases and constraints face some studies, a broader set of com-
plex biases and constraints generally face to some extent all statistical studies
across the biomedical, behavioural and social sciences. A degree of variation
between outcomes of different studies is expected given different conditions
between people, contexts and time periods in which statistical studies are
conducted. This is because researchers infer causal evidence from estimated
statistical results that are the outcome of multiple complex processes. In the
human sciences, researchers require selecting a sample, randomising, blind-
ing and controlling the sample of participants, and carrying out treatments.
Studies involve many actors, who include study designers, all participants, data
collectors, implementing practitioners and study statisticians. These actors
 STATISTICS AND MATHEMATICS OF SCIENCE 87

make multiple decisions at different steps when designing and implementing

studies, and when analysing and extrapolating their results within a particular
context and time period. Some difficulties in producing replicable statistical
studies thus generally arise across the human sciences, since they are con-
ducted by collecting data in a particular context (such as a clinic, laboratory or
field) with particular people and conditions, and the phenomena being stud-
ied often change across contexts and can evolve over time. Replication studies
are thus generally not able to obtain the same results (with the same effect sizes
and significance levels) as in the original studies.(198)

Downloaded from [Link] by guest on 29 August 2025

The replication crisis across the human sciences—a widespread inability
to replicate existing studies using new data—is thus partly driven by differ-
ences in contextual factors across studies, together with problems in using
and reporting statistical results.(198) This constrains 1 to 1 replication from
one study context to another.(42,162,196) All phenomena, when measured sta-
tistically, face some degree of variability, and researchers generally report in
studies the degree of precision and error bars (that estimate the degree of error)
for their data.
An important related dimension of science is the geography or context of
science that deals with the generalisability of evidence beyond the context in
which we collect data. It deals with the external validity of our evidence—the
conditions under which, and extent to which, data we gather in one context
apply to another context. We live in a limited pocket of the universe (the
planet Earth) and much of the universe is inaccessible to us and our best
telescopes and spectrographs. We also live at one point in time, within our
current century, and cannot directly collect data or run experiments on the
past. For us to do science and create knowledge depends on our geographic
and temporal context in which we can gather data. We conduct our studies
in a medical clinic, a geological site, an ecosystem and our universe—that is,
within a particular geographic location and time period.
Our most reliable knowledge is commonly viewed as knowledge that tran-
scends context, that is universally valid. Examples include that living organ-
isms require energy in the form of food and that electrons are negatively
charged. For most phenomena we are however not generally able to collect
data and run experiments universally but rather only locally, within partic-
ular laboratories or fields, especially across the human sciences.(199,200) The
data generated in an experimental context are the basis for the scope of results
and causal claims we can derive from the experiment. Our evidence becomes
more reliable the more data we collect across studies conducted in different
contexts and at different times. Our evidence also becomes more reliable the
88 SCIENCE OF SCIENCE

more diverse methods we apply that produce comparable results supporting

existing evidence.
These issues are related to the external validity of evidence that we illustrate
here with the method of RCTs. RCTs are commonly viewed as the leading
method across the biomedical, behavioural and social sciences for assessing
how effective a medical treatment or policy intervention is. RCTs estimate the
causal effect of an intervention and we often face some limitations in using
those estimated causal effects outside the study context. If for example a medi-
cal treatment for diabetes reduces blood sugar levels by 5% on average among

Downloaded from [Link] by guest on 29 August 2025

treated participants in a trial study, it can be difficult to attain the same effects
among the general population after the trial. This is because of demographic
and clinical differences between study participants and later the broader pop-
ulation for which the intervention is actually intended. For study participants
are in a controlled trial context while the general population does not undergo
experimental controls.(201) Limitations also arise due to differences in the
extent to which institutional, economic, educational and behavioural condi-
tions are met. In no two contexts are all conditions met to the same extent for
a treatment to produce the same causal effect. When we use results derived
from a trial population in a different population, we always face some con-
straints given variation between populations. A trial’s estimated causal results
are relative to particular factors under particular conditions within a particular
sample at a particular time.(42)
How we explain evidence thus depends on how and where we collect the
data to do so. How we express our evidence and knowledge is influenced by
the context in which we collect data as well as by our human mind and senses,
methods and practices, categories and definitions (Chapters 2–4, 10–15). We
do not have a perspective from nowhere through which to gather data that
transcends time and space, and that is independent of our methods. Yet the
more data we collect in different contexts and time periods, and the more
methods we use, the more reliable and robust our knowledge becomes.
In general, methods are typically passed along to the next generation who
were not exposed to the debates that had to first take place before those
methods became accepted. Methods however face assumptions, limitations
and potential biases that are not generally outlined in studies that use them
(Chapter 10). At many universities, graduate students in fields like biology and
medicine are generally only required to take at most one course in research
methods like statistics. But later they require using those methods to publish
their results and they may not be aware of the range of assumptions, limitations
and potential biases. This can constrain research quality.
 STATISTICS AND MATHEMATICS OF SCIENCE 89

Let us turn now to steps we can take to improve research and its poten-
tial impact. Because science relies heavily on statistical methods, the quality
and replicability of our evidence also depends on how we design our statis-
tical studies, refine our statistical methods and report our statistical results.
Improving research practice by conducting studies in multiple contexts, pre-
registering studies, better blinding and randomising participants in studies,
and updating reporting standards for studies are among the important ways
we can reduce bias. Pre-registering study designs can for instance be an effec-
tive way to blind participants and reduce bias, as the data and outcomes do not

Downloaded from [Link] by guest on 29 August 2025

yet exist and cannot be later altered. Pre-registration, while common in clinical
medicine, is not always feasible. And importantly, most scientific research will
remain exploratory, not confirmatory, and so cannot be pre-registered.(195)
We can also make experimental designs more robust by building greater rigour
and replicability into the research design. In the human sciences we can
achieve this by integrating comparability of groups, treatments and contexts
within a single study—that is, building replicability into our study designs. (198)
Research needs to be not only statistically robust and reliable but also, with
most research publicly funded, useful for society or policy (Chapter 7).
Taken together, such statistical issues have contributed to a replication cri-
sis in science. Methodological causes of the crisis include small samples and
effect sizes in studies, p-hacking and statistical standards including the statis-
tical significance level that researchers apply in their studies.(162,195) Structural
causes contribute to the crisis that are related to scientific institutions and
journals, including the current reporting guidelines for studies, and the prior-
ities of funding agencies and reward systems (Chapter 7).(162) Psychological
causes of the crisis arise due to confirmation bias, heuristics-based reason-
ing and our bounded rationality (Chapter 5).(109) Just as the replication crisis
within science is driven by a complex of methodological, structural and psy-
chological causes,(202) science in general is driven by such a broad range of
factors, outlined throughout the book. But each science of science subfield has
focused on one dimension. The scientometric community and history of sci-
ence community for example have largely not taken into account the range
of methodological and psychological factors shaping science. The best way
to understand and address the diverse causes of the replication crisis is also
the best way to understand and advance science: by adopting an integrated
approach.
In sum, methods like statistics and mathematics are fundamental to how
we conduct science. Yet we face constraints when using such methods and
making phenomena in the world amenable to them. Challenges arise in
90 SCIENCE OF SCIENCE

replicating statistical studies as we face difficulties in designing, implementing

and analysing studies in different contexts, especially in the human sciences.
The geographic location and time period in which we gather data also shapes
the parameters, content and scope of our evidence and knowledge, particularly
when dealing with human beings. We need to thus conduct studies in multiple
contexts to make our evidence more reliable. Much ongoing debate on reduc-
ing biases and enhancing quality and replicability of our scientific evidence
revolves around improving statistical methods and reporting of studies. Our
statistical and mathematical methods thus help shape the foundations, present

Downloaded from [Link] by guest on 29 August 2025

limits and advancement of science, and account for central floors in our tower
of science. Since its birth, statistics has been the practical solution across sci-
ence to the problem of identifying causal relationships in the world thanks to
statisticians like Fisher and the practical solution to the problem of induction
that philosophers like Hume aimed to address, which we turn to next.
 14
Philosophy of Science

What science is and its foundation have been explored by philosophers for
centuries, including Francis Bacon, David Hume and Karl Popper, and what

Downloaded from [Link] by guest on 29 August 2025

knowledge is and its foundation for over two and a half millennia, includ-
ing Plato, Aristotle and Ludwig Wittgenstein. Philosophers have addressed
central questions of science of science longer than researchers in any other
subfield. Major debates that have dominated philosophy of science include
paradigm shifts, justification, induction, demarcation and realism, as evi-
denced in leading philosophy of science textbooks.(18,19,203,204,205) Paradigm
shifts refer to fundamental changes in the theories of a scientific field.(1) Jus-
tification deals with principles such as falsification and verification to justify
our theories of the world (and in statistics it refers to falsifying ‘by chance’
hypotheses).(15,33,107) Induction addresses the question of whether observa-
tions we make can or cannot justify generalising about the observations in
other contexts or in the future, which is called the problem of induction (and in
statistics it refers to the problem of external validity of results).(206,207) Demar-
cation involves defining criteria for what is and is not science.(15,16,130,208)
Realism concerns whether scientific theories provide a reliable approxima-
tion and true description of reality, for observable and not directly observable
phenomena.(209,210)
In general, what we can know, as human beings with cognitive limitations,
has arguably been the central question of epistemology since Hume and Kant
in the 1700s.(206,211) These philosophers stressed the foundational role that
our mind plays in how we express our ideas about the world. Yet there is
still no consensus among philosophers on which explanations of knowledge
and science are most accurate. The reason is because philosophers have theo-
rised about what knowledge and science are mainly using conceptual analysis,
individual case studies or abstract normative methods rather than more sys-
tematic scientific methods that allow for generalising about knowledge and
science.(15,16,107,206,207,210,211,212,213)
In the 4th century BCE, Plato famously described knowledge as ‘justified
true belief.’(212) Most philosophers (and nearly all humans) view observa-
tion as the main source of our knowledge of the world. It is the common

Science of Science. Alexander Krauss, Oxford University Press. © Alexander Krauss (2024).
DOI: 10.1093/9780198937401.003.0015
92 SCIENCE OF SCIENCE

sense view throughout human history going back to early Homo sapiens
(Chapters 3 and 4), stressed by the ancient Chinese,(157) outlined in detail
by Al-Haytham in 1021, and reiterated again by Robert Grosseteste, Roger
Bacon and Thomas Aquinas in the 1200s, and later Francis Bacon in his
book Novum Organum in 1620(214) and then by positivist philosophers like
David Hume and John Stuart Mill, among many others. The empiricist view
is thus that our sensory experiences in the world are the basis of our ideas
and what we can know. That is, nothing goes through our intellect that did
not first go through our senses (Chapters 2–4). Francis Bacon stressed the

Downloaded from [Link] by guest on 29 August 2025

need for studying and intervening in reality by conducting experiments and
empirically testing reality (or, as he said, by ‘twisting the lion’s tail’ and seeing
what happens),(214) although this did not become the standard approach in
philosophy, as it did in science. In the 1920s, the philosophers of the Vienna
Circle again reemphasised the importance of observation but went beyond
the positivist focus on the strictly observable world. They also stressed the
need for using methods of verification, mathematics and logic when creating
and justifying knowledge (Chapter 13).(18,107)
Explaining how we create knowledge and how science operates are in
philosophy at times reduced for simplicity to two theories of scientific method-
ology: induction and falsification. Bacon’s theory of induction and scientific
methods is commonly viewed in philosophy as the conceptual origin of mod-
ern science, namely using the methods of observation, experimentation and
deriving conclusions.(214) These are the same methodological abilities of the
mind that our species have always used and will continue to always use to
acquire knowledge of the world, and we have done so in increasingly system-
atic ways over time (Chapters 2–4). In general, inductive reasoning is when
we for example repeatedly observe the sun rising and then infer that it will
rise tomorrow or even that it will always rise every morning. It allows us to go
beyond our current set of observations and draw conclusions about the future
and is directly linked to how we make statistical inferences across science
(Chapter 13).(206,207,214)
For falsificationists, in contrast, scientists need to do science by constructing
hypotheses and theories, testing them and attempting to falsify them.(14,15,215)
Popper argues that the defining trait of scientific investigation, evaluation and
justification of theories is the principle of falsification—which is the most
influential account in the philosophical literature and most well known out-
side of philosophy.(215) As Popper stated, ‘The growth of knowledge depends
entirely on disagreement.’ Scientists have not however adopted falsification as
a guiding philosophical principle for evaluating and justifying theories. While
 PHILOSOPHY OF SCIENCE 93

falsificationists make theories the foundation of their inquiry, inductionists

focus on the process of observing and experimenting while trying to reduce
underlying assumptions and theory.(18,19) For both inductionists and falsifica-
tionists however, theories like Freud’s psychoanalytic theory are not scientific
as we cannot easily empirically test or falsify them.
These two central theories of scientific methodology in philosophy (fal-
sification and induction) have a long tradition in statistics. For statisticians
there is not much new to Popper’s principle of falsification. The notion of
falsifying hypotheses was first developed in statistics and has a long practice

Downloaded from [Link] by guest on 29 August 2025

rooted in Fisher’s null-hypothesis testing and falsifying ‘by chance’ hypothe-
ses (i.e. not proving but disproving hypotheses). This is one of the most widely
used statistical techniques across all scientific fields today, and was outlined
in Fisher’s seminal book in 1925 (Chapter 13).(193) For statisticians, what they
call the problem of external validity is, for philosophers, also known as the
problem of induction. For both, it refers to the constraints we face in gen-
eralising from observations we make to observations in other contexts—for
example, extrapolating results from a particular experimental sample of data
to a broader or future population. Statistical methods have been developed
and are widely applied across scientific fields to address both falsification and
induction—such as null-hypothesis testing, random and stratified sampling,
and replication of studies (Chapter 13). As philosophers of science—like other
researchers in science of science subfields—work largely in isolation, they have
not yet resolved these two major debates by adopting the practical solutions
long developed and used in other fields, although thousands of articles in
statistics exist on how to address these two issues. Another related central
debate in the philosophy of science is on demarcation (identifying criteria for
what is and is not science). We will later discuss the topic and provide a new
criterion for defining the boundaries of science (Chapter 19). Philosophy of
science moreover covers domains such as philosophy of physics, biology and
economics that provide insights into the complexities of these fields’ concepts,
definitions and assumptions.
We can also better understand science by comprehending the metaphysi-
cal aspects underlying science—that is, digging under the surface of science
to identify its ontological aspects. Metaphysics of science is a branch of phi-
losophy of science that can be classified into two broad categories: on the one
hand, it studies causal and mechanistic explanations, measurement of causes
and the nature of scientific regularities/laws; on the other hand, it studies the
metaphysical assumptions of science inherent in research, namely in methods,
definitions, concepts and theories.(216,217,218,219)
94 SCIENCE OF SCIENCE

We observe causal relationships for example in the different degrees of

stability found in phenomena in the world—protons are positively charged;
a balanced nutrition improves health; and so on. For each proton or per-
son’s health we assess, we do not need to start from scratch by conducting
experiments to identify whether protons are positively charged or a bal-
anced nutrition improves health. Identifying regularities and causal rela-
tionships enables us to explain and predict them.(216,217,218) Across sci-
entific fields, we measure these using different methods. Each method
generally provides us with a different perspective to better understand

Downloaded from [Link] by guest on 29 August 2025

the given phenomenon. In studying consumer behaviour or depression
for example, psychologists and applied economists generally apply statis-
tical and experimental methods, sociologists conduct qualitative studies
and interviews, and theoretical economists use deterministic mathematical
equations. Each group studies the same phenomenon but captures it in dif-
ferent ways (Chapter 13). How we measure cause–effect relationships in
the world is embedded in and are properties of the particular methods we
apply.
When we describe regularities in the world there is also often a trade-off
between simplicity and strength of an explanation: the greater the simplicity
used to describe phenomena, the greater the loss in power to explain them.(220)
The best explanation, theory or ‘law’ would ideally account for strength and
simplicity. Newtonian physics for example has in general less strength but
is simpler than quantum physics. For the philosopher Nancy Cartwright,
the fundamental principles and theories in physics provide an idealised rep-
resentation between abstract theory and empirical reality. Also, scientific
theories and laws are not universally applicable, as Cartwright argues, but
are true only ceteris paribus (all other things being equal) within a given
context.(220)
In general, different scientific communities develop their own ontologies
(such as definitions and concepts), systems of measurement (such as descrip-
tions of quantity and mathematical representations), causal classifications
(such as deterministic, probabilistic and ceteris paribus causal relationships),
levels of evaluation (such as the property or system, or process or outcome)
and thus epistemologies (explanations, theories and worldviews). These dif-
ferent features are embedded in each scientific method we develop and use.
Our methods and instruments (experimental, statistical, chromatographic)
shape the metaphysical aspects of science through these different domains. In
this way, our tools of science standardise and automate features of conducting
science (Chapters 10, 12 and 13).
 PHILOSOPHY OF SCIENCE 95

We illustrate this point here with RCTs. When we use the RCT method,
we automatically focus on studying an isolated single causal effect (the tar-
geted intervention) instead of multiple causal effects within the larger context,
which we cannot study using RCTs. We focus on quantitative instead of quali-
tative data and on the outcome instead of the process of the phenomenon. We
also focus on the average causal effect instead of the distribution of effects, on
simple instead of complex interventions and on an isolated instead of holis-
tic explanation of phenomena.(42) It is the features of a method and research
design that thus shape our ontological views, measurement practices, causal

Downloaded from [Link] by guest on 29 August 2025

understanding and evaluation criteria. In turn, these shape our epistemology
of the world. The RCT method embodies different features within these dif-
ferent domains, compared to other methods such as observational studies or
individual case studies used to study similar questions. While the RCT method
has transformed how we understand interventions targeting disease, human
behaviour and public policy, it neglects other aspects of the same phenomena
that are better captured by those other methods. Overall, how we understand
phenomena cannot be viewed separately from the way we methodologically
measure and represent them. Our scientific methods and instruments are our
lenses through which we perceive the world. An important implication for
science is that we need to apply multiple methods so we can better under-
stand different aspects of the same phenomenon using different evidence
and improve our aggregate understanding of the phenomenon—the approach
adopted in this book.
Turning now to the metaphysical assumptions of science, scientists rarely
outline the assumptions implicit in their research and the methods they adopt.
These can be classified here into seven main traditional assumptions underly-
ing science that guide how most of science needs to be done and are inher-
ent in our common conceptions and definitions of science: mathematical,
material, reductionist, universal, causal, unidisciplinary, and one-scientific-
method. These traditional assumptions have been important in bringing
about medical, technological and scientific achievements. They guide scien-
tific research and specify what counts as scientific. These assumptions did not
however always exist. They have been adopted in a piecemeal fashion increas-
ingly since the 17th century and are observed for example in the works of
Kepler, (221) Galileo,(222) Newton(223) and Einstein.(224) Scientists are not
always aware that they are assumptions. They often view them as established
scientific facts needed to do science, as observed in scientific publications.
Physicists commonly adopt most, if not all, of them—as for example captured
in E = mc2 associated with Einstein’s theory of special relativity or in F = ma
96 SCIENCE OF SCIENCE

associated with Newton’s law of motion.(223,224) These assumptions provide

methodological rules for doing science, but they all face exceptions:

• The mathematical assumption: Scientific method is the ‘mathemat-

ical and experimental technique employed in the sciences [and]
used in the construction and testing of a scientific hypothesis,’
as defined by Encyclopaedia Britannica.(225) Nature thereby speaks
the language of mathematics.(221,222,223) However, many phenomena
cannot be measured well mathematically—from evolutionary pro-

Downloaded from [Link] by guest on 29 August 2025

cesses and historical events, to molecules, viruses, anatomy and
botany.
• The material assumption: Science studies ‘the nature and behaviour
of the material and physical universe,’ as defined by Collins English
Dictionary.(226)cf.(65,227,228) However, many phenomena are not entirely
material or physical—from consciousness and space and time, to scien-
tific reasoning, institutions and social norms.
• The reductionist assumption: Science aims to reduce the complexity of
the world in the simplest way.(229,230,231,232) However, many phenomena
cannot be reduced for example to mathematical equations and statisti-
cal models that can neglect the complexity of such phenomena—from
ecological, biological and social systems, to economic reforms.
• The universal assumption: Science studies the world with the aim of estab-
lishing universal facts and formulating ‘laws to describe these facts in
general terms,’ as defined by Collins English Dictionary.(226)cf.(228,233)
However, many phenomena do not have a universal structure but are
context-specific—from natural catastrophes and evolutionary processes
in biological organisms, to financial markets.
• The causal assumption: Science studies cause–effect relationships and
all phenomena underlie them.(233,234) However, many phenomena do
not exhibit a discernible cause and effect—from gravity and human
evolution, to weather patterns and psychological states.
• The unidisciplinary assumption: Science is conducted with each
academic field studying a different domain (as discussed in the
Introduction).(235,236) However, we best understand most complex
phenomena from across disciplinary perspectives that provide dif-
ferent evidence by applying different methods—from climate change
studied by physicists, geographers, political scientists and others, to
human behaviour studied by evolutionary biologists, cognitive scientists,
economists and others.
 PHILOSOPHY OF SCIENCE 97

• The one-scientific-method assumption: Science requires applying the sci-

entific method based on ‘the collection of data through observation and
experiment, and the formulation and testing of hypotheses,’ as defined
by Merriam-Webster Dictionary.(237)cf.(214,226) However, some phenom-
ena cannot be studied using these methodological approaches—from not
being able to conduct experiments on fossils, planets or the develop-
ment of cities, to not being able to test hypotheses when surveying rock
formations or describing historical events.(238)

Downloaded from [Link] by guest on 29 August 2025

Given the great success of physics in explaining vast parts of the world—
in terms of relativity theory and quantum theory—these assumptions have
become an ideal for researchers. Researchers in other fields generally adopt
several or more of them. The lens through which we study the world—the
lens of mathematics, reductionism, unidisciplinarity etc.—portrays atoms, the
universe, economic actors and markets as mathematical, mechanical, unidis-
ciplinarily etc. These foundational assumptions help frame the questions we
ask, the methods we use, the way we interpret data, the theories and mod-
els we develop. These common assumptions of science, when viewed as how
science should be done, can however constrain science. With an awareness of
these assumptions, we can open up science and adopt alternative methodolog-
ical approaches (such as non-mathematical and cross-disciplinary) to address
many questions and challenges we face. Despite the various philosophical
insights into the nature of science, philosophy of science faces a number of
shortcomings.
Overall, philosophy of science has, to date, not yet provided a comprehensive
understanding of its subject matter—science—for four reasons. Firstly, most lit-
erature in philosophy of science has focused on studying theories (one output
of science) and has done so focusing on the field’s major debates of paradigm
shifts (in theories), justification (of theories), induction (for developing the-
ories), demarcation (for testing whether theories are scientific) and realism
(whether theories provide a reliable approximation of reality), as observed
in leading textbooks in the field.(18,19,203,204,205) That philosophers of science
focus on one output of science—theories—is understandable because their
knowledge of science is centred mainly around this scientific output and they
do not commonly work in labs and conduct experiments using scientific meth-
ods (the process of science). The field has not given sufficient attention to
the fact that its main subject of study, science, is developed and advanced by
creating new scientific instruments and methods, such as novel X-ray diffrac-
tion methods, computational methods, electrophoresis, statistical methods,
98 SCIENCE OF SCIENCE

particle accelerators and chromatography. And these particular methods and

instruments of science have not received much, if any, attention in terms of
studying their assumptions, biases and limits and how to push those limits
to drive new discoveries (Chapters 10, 16–20). Related research has instead
focused on general scientific methodology, in abstract terms. Leading philoso-
phers of science who have led the major debates in the field outlined above
have thus not focused on the major methods and instruments of science as
the main mechanism driving new scientific advances, theories, discoveries and
fields.(1,15,33,206,210)cf.(18,19,203,204,205)

Downloaded from [Link] by guest on 29 August 2025

Secondly, most leading publications in philosophy of science have studied
what science is using theoretical, conceptual and normative approaches, rather
than also using systematic (scientific) methods from across scientific fields—
experimental, statistical, computational etc.(1,15,33,206,210)cf.(18,19,203,204,205)
Thirdly, most philosophers have used their mind as the main source of acquir-
ing knowledge. They have largely not taken into account the powerful role
of systematically gathering empirical data and carrying out representative
studies to make robust and generalisable inferences—as used in all fields
of science.(1,15,33,206,210)cf.(18,19,203,204,205) Fourthly, what most publications
in philosophy of science have studied and the proposed explanations are
commonly an internal response to philosophical questions. They are not com-
monly a response to problems facing contemporary science, including those
related to methods, instruments, biases and institutions that contributes to
existing scientific literature.(ibid.)
The Nobel laureate in physics Richard Feynman claimed that ‘philosophy
of science is as useful to scientists as ornithology is to birds.’ The philosopher
Hasok Chang also argued that ‘most scientists today would regard most dis-
cussions currently taking place in professional philosophy as utterly irrelevant
to science’.(239) Such claims appear to also include the philosophy of science ‘in
practice’ that generally does not build on the vast existing scientific literature.
This can be seen by looking at a sample of philosophy of science publications—
via Scopus or Google Scholar—and exploring who they cite and who cites
them, namely other philosophers. Yet it can seem surprising for an outsider
that philosophers of science study science, without aiming to influence or
improve science and thus to also publish their studies in science journals. Can
internal dialogue with other philosophers be the purpose of philosophy of sci-
ence rather than engaging with and informing scientists and helping improve
science?
The old epistemic questions about what knowledge, science and their foun-
dations are and how we develop them, which philosophers have studied for
 PHILOSOPHY OF SCIENCE 99

millennia, can be more rigorously answered scientifically using empirical evi-

dence (Chapters 2–15). And only then can we develop a coherent concept
and understanding of knowledge and science that is consistent with existing
empirical evidence about knowledge and science across fields and grounded
in scientific practice (Chapters 16 and 17). This requires integrating a range
of methods and evidence to be able to arrive at conclusions that are coher-
ent with (do not contradict) what is already known across scientific fields. It
is for these reasons why the field of philosophy of science, which raises some
of the best questions about the foundations and limits of science, has to date

Downloaded from [Link] by guest on 29 August 2025

largely not been able to address those questions. This traditional philosoph-
ical project, by adopting a unidisciplinary approach to studying science, has
not sufficiently taken the empirical world, scientific methods and actual sci-
entific practice into account—at least not in a way comparable to the rigorous
methods applied across science. Philosophy of science is another case in point
illustrating why we can only understand science comprehensively by taking
an integrated approach. The integrated account adopted here can provide a
more empirically founded and scientific understanding to these foundational
questions about science (Chapters 2–20; Figure 1.2).
In studying the nature of science, the key difference between the old field
of philosophy of science and the new field of science of science is that the for-
mer relies mainly on theoretical and conceptual analysis, while the latter relies
on empirical methods and evidence. Providing scientific responses using sci-
entific methods to old philosophical questions is how many scientific fields,
throughout history, emerged out of natural philosophy. This is also the case
for the field of science of science outlined here. The general trend has however
changed. Throughout history, science has often begun where philosophical
questions ended. Philosophical questions now often begin where science ends.
While scientists push forward and continually seek to expand our under-
standing of the world, philosophers of science commonly seek to demonstrate
problems and limitations inherent to science and our limited understanding of
the world. It is often a difference between academic optimists and pessimists.
Finally, often ‘a scientist formulates problems in a way which requires
for their solution just those techniques in which he himself is especially
skilled.’(240) Researchers may often express the questions they address in such
a way that the answers to those questions can be best provided with the meth-
ods that they already have at their disposal, as opposed to methods in other
fields that they may not be familiar with. If scientometricians’ main method is
large-scale data analysis, if philosophers of science’ main method is conceptual
and theoretical analysis, and so on, then questions look like those that only
100 SCIENCE OF SCIENCE

large-scale data or only conceptual and theoretical clarity can address. This
provides another perspective to the disciplinary isolation among the science
of science community to date—a community that collectively studies science
(when viewed from outside) but is largely not aware that it is part of this much
larger community across many fields aiming to address the same questions
(when viewed from within each subfield). It also provides a further expla-
nation for why the science of science community has not yet jointly adopted
multiple methods and a truly interdisciplinary approach that stretches across
all fields, as outlined here.

Downloaded from [Link] by guest on 29 August 2025

In sum, over millennia, philosophy has contributed to understanding the
theoretical and conceptual foundations and boundaries of what we know.
Philosophers of science have widely debated how to theoretically explain and
justify knowledge and science, especially using methodologies like induction
and falsification—which statisticians have long provided practical solutions
to. Such philosophical debates often revolve around theoretical and normative
principles of what knowledge and science should look like and how to demar-
cate them—that is, principles of what grounds the tower of science, what its
borders are and what shape it should take. But such theoretical principles have
not generally been directly used when doing science and applying methods in
practice. Science also involves studying causal relationships, mechanisms and
regularities, and adopting metaphysical assumptions. Understanding these
provides insights into the foundations of science and can improve how we do
science by expanding our methodological approaches and how we view and
describe the world. And as the philosopher Wittgenstein stated, ‘the limits of
my language are the limits of my world.’
 15
Linguistics of Science

We turn now to the role of language in science and understanding science.

Without a system of language we would not be able to reason complexly,

Downloaded from [Link] by guest on 29 August 2025

express our knowledge and do science (or even read books on science like
this one). It enables us to describe and explain to others what we observe,
how we solve problems and the knowledge we acquire about the world.
(21,90,92)
With language we can quickly obtain and pass along methods and
bodies of knowledge. How we use language determines how scientific stud-
ies are expressed and disseminated and how accessible they are to researchers
in the same and other fields—or whether research can only be accessed by
few. (21,90,92,52,87)
Written and especially digital documentation allows us to more efficiently
share and cumulatively build on vast bodies of knowledge and methods
across generations (Chapter 6).(21) It frees up our limited working memory by
making it much easier to process information such as multiple ideas simultane-
ously and quickly expand on them. It helps us address complex problems and
questions more effectively. It also extends our long-term memory to system-
atically record what we observe and to store knowledge. A system of written
language is a precondition for cumulative knowledge and creating scientific
methods including a system of mathematics. (241) Only by using language can
we express our methods of science, including statistical coefficients and alge-
braic equations. Our methods and knowledge cannot thus be independent of
the language we have developed.
Technical language divides the scientific community in general, including
the science of science community. A specialised language connects researchers
in one subfield with a common language but often presents a barrier for
researchers in other subfields to understanding the content of studies. Some
fields are laden with technical jargon. Think of common terms used among
economists of science, such as elasticity or hyperbolic discounting (which
likely cause a blank face when outsiders read them). Elasticity refers to how one
factor (such as demand for knowledge) responds to a change in another fac-
tor (such as supply of government funding). Hyperbolic discounting refers to
delayed discounting or decreasing returns over time for choices we make (such

Science of Science. Alexander Krauss, Oxford University Press. © Alexander Krauss (2024).
DOI: 10.1093/9780198937401.003.0016
102 SCIENCE OF SCIENCE

as related to financial rewards for making future breakthroughs). Think of

common terms among philosophers of science, such as the metaphysics of sci-
ence or logical positivism. For non-philosophers, it is difficult to imagine what
they could mean. Metaphysics of science studies questions about fundamental
concepts, categories and definitions used in science. Logical positivism refers
to a philosophical movement to use observation together with logical methods
to generate and justify knowledge. Think also of different systems of language
used to describe evidence across science of science subfields, with results of dif-
ferent studies presented using for example statistical programming language,

Downloaded from [Link] by guest on 29 August 2025

computational language, language of conceptual analysis, mathematical for-
malism and other methodological languages. Researchers speak one or several
of these languages but not others. It will be key for researchers in science of sci-
ence to begin simplifying language, leaving out or simply defining field-specific
jargon in studies. For jargon constrains other researchers to understand studies
and build on research, in order to advance the common objective of the science
of science community of developing a coherent understanding of science.
At present the English language dominates science worldwide, including the
leading scientific journals and institutions. This presents a challenge as most
literature across science of science generally only studies articles published
in English and so can provide an incomplete picture of science. Large-scale
scientometric studies for instance commonly only capture articles in English.
If we want to understand for example what drives major discoveries, includ-
ing Nobel-Prize-winning discoveries, this can require using digital translation
programmes to extract necessary data since those discoveries were published
in over a dozen languages.
Language and writing systems—like scientific methods—are central think-
ing tools. The Western alphabet is viewed as structured analytically and a
natural tool for categorising. It functions as a model for classification systems,
and standard measures and weights. The Chinese writing system, on the other
hand, is largely pictographic and non-reductionist. It functions as a model for
viewing the world as continuous and holistic, and facilitates taking the broader
set of factors influencing a phenomenon into account holistically (Chapter 9).
(57,158)
Using a particular alphabet can thus shape the way we think and view
phenomena.
Another important area in which language shapes how we use and dissem-
inate science is the communication of science. (242) The Nobel-Prize-winning
physicist Ernest Rutherford stated, ‘An alleged scientific discovery has no
merit unless it can be explained to a barmaid.’ Findings of studies on topics
like climate change, nutrition, vaccinations and the coronavirus pandemic are
 LINGUISTICS OF SCIENCE 103

especially relevant to all of us, the general population. How science is com-
municated to the public, policymakers and other scientists is important as
it can affect their decisions and behaviour. Scientists need to ensure that a
study’s results cannot be easily misinterpreted, that results of political, social
and ethical relevance are presented sensitively and that uncertainty, risks and
values are communicated in a balanced way. Uncertainty is a feature of many
scientific fields. How uncertainty is communicated in studies (such as on envi-
ronmental issues and consumer behaviour) is crucial, given possible negative
effects on our choices.(243) Language used in social and digital media, news

Downloaded from [Link] by guest on 29 August 2025

outlets and popular science books can be especially susceptible to misuse, as
they do not undergo rigorous peer review but can still have a broad impact on
the public. (242)
In sum, our language, especially written language, helps lay part of the foun-
dation of science and our understanding of science by enabling us to reason,
acquire knowledge and use methods more complexly and cumulatively. It also
helps shape the way we express and disseminate our ideas and view the world,
while it can create barriers between fields laden in technical jargon. The lin-
guistics of science thus contributes to understanding part of the foundation of
the tower of science and the divisions we observe across the science of science
community. After discussing the 14 subfields of science and science, the next
chapter synthesises the evidence from across the subfields to be able to provide
a more comprehensive understanding of the origins, foundations and limits of
science—and thus a holistic picture of our tower of science.
 16
Science of Science: An Integrated and
Methods-Driven Understanding of Science

Downloaded from [Link] by guest on 29 August 2025

We have developed science by using our cognitive and sensory abilities that
have evolved within our environmental niche of the world and they face con-
straints, and we expand our vast knowledge of the world by developing new
scientific methods and instruments designed to reduce our constraints. This
universal and adaptive methodological toolbox of ours is at the centre of sci-
ence and enables us to do and advance science in new ways. Other factors also
influence science as we are social beings embedded in our scientific commu-
nity and its practices, socialised into a system of language and mathematics
used to express our knowledge, born into a historical context with particular
world views, abiding by scientific norms, principles and assumptions, moti-
vated by biological traits, subject to psychological biases, aided by computer
technology and influenced by recognition and ambition, funding and societal
objectives, public and economic institutions. Ultimately, science is a dynamic
system of human activities aimed at better understanding the world.
What drives this system are complex interactions between our evolved mind
and the methods we develop using our mind, on the one hand, and the world
and social institutions, on the other. Science is the outcome of arguably the
most cognitive and social activity that our species has undertaken. It is an
activity in which we interact with our natural and social environment to
develop sophisticated methods and instruments and bodies of knowledge that
we cumulatively build on over centuries. What we call science is thus an organ-
ised effort of accumulating knowledge by working together to develop and
apply increasingly complex scientific tools. These elements come together to
make our ideas, discoveries and scientific fields possible. The elements lay the
foundation, structure and form of our tower of science.
A more integrated and coherent understanding of the foundations and lim-
its of science is thus inseparable from methods and evidence in cognitive
science and methodology, biology and archaeology, computer science and
statistics, anthropology and psychology, sociology and philosophy, economics
and scientometrics, and history and linguistics. Scientists are members of a

Science of Science. Alexander Krauss, Oxford University Press. © Alexander Krauss (2024).
DOI: 10.1093/9780198937401.003.0017
 AN INTEGRATED AND METHODS-DRIVEN UNDERSTANDING 105

scientific community shaped by its methodology, its history, its sociology

and its philosophy. Artificial borders between disciplines have traditionally
characterised science and the study of science. What were once thought of
as independent disciplines—with tens of thousands of existing publications
studying science from their own disciplinary perspective and methodolog-
ical approach (Figure 1.1)—have been linked together here to account for
science holistically (Figure 1.2). What enables and constrains science can-
not be explained from a single disciplinary perspective alone. But almost all
articles studying science do precisely that. As one indication, less than 3%

Downloaded from [Link] by guest on 29 August 2025

of total publications across all subfields use the common terms inter-, cross-
or multi-disciplinary (Chapter 1). To address foundational questions about
science, some generally view the appropriate unit of analysis to be the indi-
vidual (psychologists and cognitive scientists). Others generally think it is the
group (anthropologists, sociologists, economists and linguists). Some gener-
ally study at the level of the species (biologists and evolutionary cognitive
scientists). Others generally investigate the past (historians, archaeologists and
some anthropologists). Some adopt a meta-level or methodological perspec-
tive (methodologists, scientometricians, computer scientists, statisticians and
philosophers).
Yet disciplinary isolation has given rise to simplified and at times polaris-
ing views. Leading scientometricians and network scientists, like Fortunato,
Wang and Barabási, have focused on and stressed the key role of publica-
tions and citations.(4,5) Leading historian of science, Kuhn, the changes in
scientific paradigms.(1,32) Leading philosopher of science, Popper, the eval-
uation of scientific theories.(14,15,31) Leading sociologists of science, Latour,
Woolgar and Bourdieu, the social practices of scientists,(10,11) and so on. This
common approach to studying science has led leading researchers to not yet
address the central questions of how important the particular ‘key’ factor they
study is and how it relates to other ‘key’ influencing factors identified by other
researchers across different fields that they do not study. It has led at times
to overinterpreting the role of the factor they study compared to other factors,
especially the foundational role of our scientific methods and instruments and
our mind in enabling and constraining science—which leading researchers do
not focus on (Chapters 2–15). Different researchers working at a different level
and on a different aspect of science of science are driven by the common aim
of understanding and improving science. But they generally do so in method-
ological and disciplinary silos, which has led to an incomplete understanding
of science.
106 SCIENCE OF SCIENCE

The unified account of the field of science of science presented here, by

bringing together the approaches and features across 14 disciplines, outlines
the evidence on science that is coherent across the natural, behavioural and
social sciences. Taking such a holistic approach represents the most compre-
hensive understanding we have of science for the following reason: the range of
disciplinary approaches apply different methods and focus on different features
of science, and there is coherence across the independent strands of evidence, in
particular in the role of methodological features in shaping science (Chapters
2–15).

Downloaded from [Link] by guest on 29 August 2025

To comprehensively understand for example how our mind enables us to
do science, it is not enough to just study our mind from one perspective.
We cannot just adopt a perspective from neuroscience using brain scans or
a perspective from psychology running experiments with individuals solving

Figure 16.1 An integrated account of the field of science of science.

 AN INTEGRATED AND METHODS-DRIVEN UNDERSTANDING 107

problems. Instead, we need to integrate such evidence together with evidence

about our universal methodological abilities of the mind, which we share with
other animals for observing and solving problems, which have evolved by
adapting to our niche of the world and which we expand by developing new
methodological innovations. We need to study how our mind is embedded
in a cultural and historical context that shapes our different approaches to
reasoning, as observed across scientific fields (Chapters 2–15).
As a simplified summary of the book, Figure 16.2 outlines the set of intercon-
nected abilities and conditions that enable and constrain science and, in doing

Downloaded from [Link] by guest on 29 August 2025

so, provides an integrated view of the field of science of science. It outlines
what the field looks like when we study science from all perspectives. We and
what we are methodologically capable of are at the centre of the foundations
and present limits of science. Understanding ourselves and our cognitive and
methodological constraints is the key to understanding science and the scien-
tific methods we develop to address those constraints and expand the research
frontier.
Explaining different aspects of the same phenomenon—science—at differ-
ent levels enriches our overall understanding, as Figure 16.2 illustrates. We
cannot comprehend the individual factors in isolation, because the parts (and
the groups) interact with each other to account for the greater whole. Internal
factors such as our cognitive abilities and senses shape which methods and
instruments we develop and improve to make scientific advances. External
factors such as science funding and the size of the scientific community influ-
ence which scientific methods and instruments we can develop, what research
we fund and how many researchers work together. Meta-level factors such
as our methodological limitations and assumptions shape the foundations of
our research and how we can advance science, as we go about addressing our
cognitive, methodological and instrumental constraints, and so on.
We cannot comprehensively understand the interconnected complexities
of science without integrating the disparate evidence from across fields. The
same applies to understanding scientific discoveries. Different aspects of the
discovery process have been studied by different researchers—with some
psychologists analysing imagination and analogies used in discoveries,(244)
scientometricians and network scientists assessing age dynamics and col-
laborations among discoverers,(4) computer scientists running simulations
of discoveries,(188) historians investigating theory change brought about by
discovery(1) and philosophers examining how discoveries are explained and
justified.(14,15,215) It seems a natural tendency and expected that studies com-
monly focusing on one topic highlight the importance of the particular topic
 Internal factors
Cognitive science Psychology
How we reason and acquire knowledge is grounded in our evolved methodological abilities for We face psychological biases (such as status quo bias and framing bias) and use simplified heuristics that
observing, solving problems, experimenting, causal reasoning and testing hypotheses; and memory, methods (such as statistics and randomisation) help reduce
learning and abstraction We are influenced by goals and interests

Downloaded from [Link] by guest on 29 August 2025

We are motivated by personality traits such as a drive for recognition and ambition
Our cognitive and sensory abilities are more constrained in developing knowledge as we get further away
from phenomena we can perceive in our niche in which we have evolved (phenomena not presently Linguistics
accessible to our mind) Language enables us to reason more complexly and use our methodological abilities more systematically
Biology With written language we can quickly acquire, through learning and teaching,
We have evolved abilities for vision, other senses and other related physiological functions that we share bodies of knowledge and methods
Language influences how we express our ideas and view the world and can divide us via technical jargon
with other animals and lay the foundation of our methodological abilities
We focus on meeting our biological needs and wants that are reflected in most questions we pursue
Sociology (society)
External factors
We thus aim, with our human perspective to the world, to improve our own lives and often study
We are trained in methods and instruments and socialised in a system of math
topics beyond human beings if they benefit us
and language used to express knowledge
Historical and cross-cultural factors We are embedded in a scientific community and cultural context
that influences the questions we pursue
Archaeology We can develop more knowledge through greater cooperation and population density
Cognitive abilities of early humans evolved over time to be able to develop Social institutions, objectives, norms and values guide research
increasingly sophisticated methods and tools (reflected in material artefacts) Economics (economy)
Making, e.g. hand axes, evidences our evolved abilities to systematically Our universal and adaptive Accumulating knowledge and methodological diversity are
observe, experiment, reason causally, test a hypothesis, imagine and plan methodological toolbox fostered through economies of scale, science policy and targeted funding
—abilities which we use in science today The reward system in science (priority to the first to publish an idea) incentivises innovation,
(to observe, experiment, develop demands research to be published publicly and fosters cumulative knowledge
Anthropology
Different cultures develop different methodological approaches to
statistics etc.) is at the centre of Economic and public institutions help plan, finance and manage
how knowledge is produced, distributed and used
reasoning about the world science and underlies factors
Different ways of thinking and worldviews are shaped by the social across all disciplinary perspectives Methodology
and natural environment we are exposed to We have developed methods (e.g. controlled experimentation) and instruments (e.g.
telescopes) that significantly expand our evolved cognition
History Methods however face assumptions, limitations and possible biases
We are born into a historical context with particular worldviews Meta-level and (like measurement and sampling bias)
Within this context, scientists share common methods and assumptions and methodological factors They have a set scope of which questions can be studied, which phenomena
these change over time modelled and which results derived
Methods we develop and our methodological abilities thus directly drive science and
Philosophy its present limits
We can justify our knowledge using methodological principles, such as induction Computer science
and falsification Computers make science much more efficient, particularly how we apply our methods,
We can use different methodological criteria for evaluating knowledge such as automate scientific processes, reduce human errors and process and analyse data
whether it accurately corresponds to reality, coheres with existing knowledge They largely determine how quick we access and share methods and knowledge,
and is useful and how well we are connected to other scientists globally
Assumptions underlie science and are reflected in our methods, definitions and concepts and shape
our views Statistics and mathematics
Scientometrics and network science Statistics and mathematics are two of the most widely used methods in science, with quantitative
High-impact science is characterised by conventional work that integrates new combinations measurement a central feature of scientific rigour
of topics or methods Statistics enables analysing millions of observations in a single dataset
Researchers are risk-averse—yet those willing to explore new, riskier areas are more We often collect data locally (in a lab or field) and the results are often context-specific
likely to expand a field with new knowledge Many phenomena cannot however be easily studied as we cannot make them amenable to quantitative
Large discoveries are generally made by younger researchers and explained analysis
by their higher productivity Common problems in producing reliable results are low sample size, p–hacking etc.

Figure 16.2 The origins, foundations and limits of science: the range of interconnected factors, and particularly our
methodological toolbox, that enable and constrain science.
Note: Other influencing factors can be included for example into economics of science (such as the political usefulness of research) or into philosophy
of science (such as ethical issues limiting what topics we can research and what experiments we can conduct). Regulatory bodies for instance place
limitations on research related to human cloning, aspects of gene editing and technologies fostering climate change. We need to view the classification
across the four areas loosely, with connections taking place across factors and areas.
 AN INTEGRATED AND METHODS-DRIVEN UNDERSTANDING 109

they study. However, when we instead adopt a unified approach and integrated
study, we can better evaluate the range of factors against each other, assessing
the role of methods and instruments developed to be able to make scientific
discoveries, the role of the traits of individual discoverers, the role of basic fac-
tors like collaboration and public funding, and so on. It is about assessing the
importance of the different factors simultaneously and better explaining how
we drive new advances.
A good analogy of the central challenge facing this field is that biologists,
historians and archaeologists can study different parts of an ecological system

Downloaded from [Link] by guest on 29 August 2025

and each provides valuable insights. But to understand the whole ecosystem
it is necessary for the different researchers—and in addition for ecologists—
to study the ecosystem in its entirety, from an integrated perspective. The
research focus of science of science has, to date, been like studying ecology
from different perspectives but not from ecology itself. The critical miss-
ing piece in science of science has been for researchers to study the scientific
ecosystem holistically, from an integrated perspective like ecologists do. There is
another way we can think about this central disparity in the field: the common
approach of understanding science, by studying an aspect of science from one
disciplinary perspective, is like trying to explain an ecosystem by only studying
trees, or the human body by only studying cells. We acquire much knowl-
edge, but that knowledge remains highly incomplete. To understand science,
we require adopting a holistic-disciplinary approach and methods, just as we
require doing so to understand other multidimensional phenomena, such as
the environment and climate change (as illustrated in Chapter 1), the replica-
tion crisis in science (Chapter 13), health pandemics, poverty, and so on.
Overall, there is no consensus among researchers on which proposed expla-
nation of what drives science is best and most accurate. Scientific consensus
is however a central feature of science and advancement in other scientific
disciplines, and for complex phenomena like science, consensus requires inte-
gration. To summarise, one central advantage of integrating the range of
influencing factors in a single study is that it enables assessing the intercon-
nected factors in a comparative way. This has not yet been possible in existing
studies focusing on an individual issue or field, and without an understand-
ing of the broader range of issues and fields and how they are connected.
The trend in science towards specialisation in a single discipline has gen-
erated much specialised knowledge. But it constrains us in understanding a
phenomenon as complex and diverse as science when not combined with a
meta-approach that pulls the pieces together and provides an overall picture
of science. The other central advantage of integrating the various factors is
110 SCIENCE OF SCIENCE

that it enables us to uncover which are most important and what central fac-
tor is shared in common across the different fields. In doing so, we find here
that our methodological toolbox underpins the different factors across all dis-
ciplinary perspectives and is the only factor that does so. In Figure 16.2 we
highlight the methodological features (in italics) in each of the subfields. This
shift in focus to methods is central to the integration of the field of science of
science. The degree to which other factors influence our scientific advances
and discoveries varies depending on the phenomenon we study. This inte-
grated approach enables us to develop a coherent understanding of science

Downloaded from [Link] by guest on 29 August 2025

and grounds the new-methods-drive-science theory: no factor plays as foun-
dational and ubiquitous a role in understanding the origins, foundations and
limits of science as new scientific methods and instruments we develop using our
mind’s methodological abilities (as each field in Figure 16.2 illustrates).
No other factor influencing science is relevant in all fields—with for example
linguistics and archaeology providing little, if any, insights into commonly
mentioned factors like scientific funding, incentive structures, the scientific
community and new theories. In ground-breaking scientific publications, we
observe that teams can be small or large, low or high funded, young or old, at
low and top ranked universities, or interdisciplinary or not. Money, collabora-
tions and a research community are basic factors that foster science, but alone
are not enough to break new ground. The central finding is that our universal
and adaptive methodological toolbox (that enables observing, solving prob-
lems, developing microscopes etc.) is the main mechanism through which we
directly drive science. All scientific advances and discoveries have been made
using our methodological toolbox in new ways.(64)
We can better understand science in light of the integrated new-methods-
drive-science theory presented here. This theory:

• places us, and the methods we develop using our mind, at the centre of
studying science;
• integrates evidence of the abilities and conditions that have enabled us
to develop science (biological, cognitive, social and methodological), the
abilities and conditions shaping the scope of science (including, in addi-
tion, historical, economic etc.) and, most importantly, the abilities and
conditions allowing us to expand the present limits of science (mainly
methodological and instrumental but also cognitive, sensory and social);
• combines thus insights into the origins and foundations of science (espe-
cially our evolved methodological abilities of the mind and the methods
 AN INTEGRATED AND METHODS-DRIVEN UNDERSTANDING 111

we develop) with insights into the present boundaries of science and how
to push them (especially addressing our methodological constraints);
• pools together the range of methods used to study science to provide an
integrated explanation that is consistent and better grounded in what is
already known across disciplines (Chapters 2–15; Figure 16.2).

Methods are deeply embedded in our broader evolutionary, natural and social

Downloaded from [Link] by guest on 29 August 2025

context. Because our methods are at the centre of how we do science and make
discoveries, understanding the foundation of the methods we develop and use
is at the centre of science of science. Our methodological abilities and scientific
tools account for the foundation and different floors of our tower of science.
The fundamental importance of our methods is evident for all aspects of
science: conducting, evaluating and advancing science and also understand-
ing science. It is thus surprising that among all papers studying science, such a
small fraction analyse the nature of methods and instruments, their constraints
and how to improve them (Figure 1.1).
This account of science explains how our methods and instruments and our
human mind used to develop them set the scope within which we are able
to develop knowledge and science. Then, beyond nature and our cognitive
and methodological limitations in understanding nature, influences that are
economic, social, historical and the like, even if often less direct and important,
also shape the content and scope of the knowledge we create. Yet we cannot just
focus on scientific methods, as we would otherwise be partially committing the
mistake that traditional subfields make in focusing on one aspect of science—
though on one integrated aspect of science across all fields (methods) instead
of an isolated aspect of science in just one field (social norms, citation patterns
and so forth).
In Chapters 2–15 we have synthesised the evidence and contributions of
the different factors to our understanding of science. In light of this synthesis,
we can depict the level of scope that a given factor has in explaining science,
and the direct influence we have on that factor in shaping science (Table 16.1).
These are the two criteria used to assess each factor. Some factors (and fields)
mainly only help explain and shape the foundations of science (historical,
anthropological, archaeological and linguistic factors) and others can also
do so for the limits of science (biological, psychological and philosophical
factors). However, we cannot directly influence them to promote science.
These factors are thus classified as having lower scope to explain and influence
 Downloaded from [Link] by guest on 29 August 2025
Table 16.1 Scope of factors/fields in explaining and shaping the foundations, limits and advancement of science.
Explanatory scope: Direct influence:
factors/fields explaining the level of influence we
have on a factor in
foundations limits advancement shaping science
of science of of science
science
History Anthropology Archaeology Linguistics X Low

Biology Psychology Philosophy X X Low

Economics (economy) Sociology (society) Scientometrics/Network science X X X Medium

Statistics/Math Computer science Cognitive science
X X X Medium–High
Methodology
Note: Cognitive science broadly covers our evolved cognitive abilities and constraints (related to observation, memory and abstraction),
while psychology narrowly covers psychological biases and personality traits, as previously outlined.
 AN INTEGRATED AND METHODS-DRIVEN UNDERSTANDING 113

science. Other factors (and fields), such as the economy, society and sciento-
metric features, help explain and shape the foundations, limits and advance-
ment of science. And we can partially influence them to foster science. These
factors thus have medium scope to explain and influence science. Finally, other
factors (and fields), such as methodology, statistics/mathematics, computer
technology and cognition, also help explain and shape the foundations, lim-
its and advancement of science. And we can most directly influence them to
make new scientific advances. These factors thus have high scope to explain
and directly influence science (Table 16.1). Our scientific methods including

Downloaded from [Link] by guest on 29 August 2025

instruments are the only factor that underpins all 14 factors and fields (Figure
16.2) and that we are most directly able to influence to do and advance science.
 17
Science of Science: An Integrated Field
Grounded in the
New-Methods-Drive-Science Theory

Downloaded from [Link] by guest on 29 August 2025

We offer here a foundation for the integrated field of science of science that
studies science, and its foundations and limits, by combining methods and
evidence from across the sciences (Chapter 1). In Figures 1.2 and 16.2 we
outline what the unified field of science of science can look like. The field
needs to be approached just like other fields of science: evaluating evidence
cross-disciplinarily for consistency and coherence, applying methodologies
comparatively and studying experimentally. To establish the field, some sci-
entists of science and metascientists would need to be formally educated and
trained, just as statisticians, psychologists and biologists are. We should not
assume that some researchers across fields, who happen to develop an incli-
nation for foundational and methodological questions of science, will come
across such questions and find the time and resources needed to address them.
Training researchers in science of science and methodology of science would
help equip them with the necessary interdisciplinary and methodological
skills to address present shortcomings in the field. While some scientists of sci-
ence may focus on meta-level questions about science and scientific methods
in general, others may focus on more specialised subfields such as economics
of science or biology of science but do so in an integrated way coherent with
what is already known in other subfields rather than in isolation.
Establishing the field of science of science requires providing not only
an empirical foundation (Chapters 2–15) but also a theoretical foundation
for understanding science. The new-methods-drive-science theory presented
here can provide a unifying theory and foundation for the field that is
grounded in the powerful role of scientific methods which is the common
thread among this scientific community. The theory can integrate and unify
the disparate fields studying science as our methods and instruments are con-
nected to all features of science (Figure 16.2). Our evolved methodological

Science of Science. Alexander Krauss, Oxford University Press. © Alexander Krauss (2024).
DOI: 10.1093/9780198937401.003.0018
 NEW-METHODS-DRIVE-SCIENCE THEORY 115

abilities of the mind (our universal methodological toolbox) and sophisti-

cated methods and instruments we develop using our mind (our adaptive
methodological toolbox) are the main mechanism that directly enables us to
develop knowledge and science (Chapters 2–15). Our tools allow us to do sci-
ence and also set the present limits of what science we are able to do. The
theory describes how our methods have driven the origins, foundations and
present limits of science.
The new-methods-drive-science theory explains how we advance science
by developing new methods or refining existing methods that expand our

Downloaded from [Link] by guest on 29 August 2025

present cognitive, sensory and methodological reach to the world. New meth-
ods and instruments we create—such as novel statistical techniques, X-ray
methods and telescopes—enable us to make new breakthroughs by reduc-
ing our present constraints to studying the world in new ways. In contrast,
existing leading (competing) accounts of science are outlined throughout the
book—for example in the history of science by Kuhn who argued that science
goes through paradigm shifts in theories;(1,32) in scientometrics in which sci-
entists argue that career trajectories, team collaboration, research output and
networks of scientists are the central parameters driving science;(4,5,9,35,37,38)
in the sociology of science by Merton(130) and Latour and Woolgar(10) who
highlight the central role of social factors shaping science; and so on.
In describing the new-methods-drive-science theory, we define the cen-
tral terms here. Science is the study of the natural and social world by using
our cognitive abilities (including observation, experimentation and problem
solving) and the methods and instruments we develop (including statistical
techniques and algebra, and particle accelerators and electrophoresis) with
the aim of describing, explaining, predicting and controlling phenomena.
Scientific methods are systematic techniques and scientific instruments are
systematic tools used for scientific research and which are generalisable. In
general, if more scientific methods are created, then more scientific progress
will be achieved. The theory predicts that scientific progress will be brought
about by generating novel methods, and the theory can be directly tested. And
the theory is confirmed in a companion study assessing over 750 major sci-
entific discoveries.(64) An assumption of the theory is that basic factors are
in place, including our cognitive abilities and a minimal level of funding and
collaboration to generate methods and tools.
The theory connects our new scientific tools to scientific progress. For they
are what allow us to observe farther, process information better and mea-
sure phenomena more precisely, providing new perspectives to the world. Our
methods and instruments are how we experiment with and control different
116 SCIENCE OF SCIENCE

phenomena in the world and expand our scientific scope. They determine
how we design, implement and analyse studies and how we define and gather
evidence. In science we require measuring, conducting experiments and
analysing data on what we study. And to measure, conduct experiments and
analyse data we require methods and instruments. Without them, we are not
able to describe and explain most complexity in the world. The new-methods-
drive-science theory is thus a general theory of scientific advancement through
methodological advancement.
Yet where does the demand for developing new methods and instruments

Downloaded from [Link] by guest on 29 August 2025

arise? The demand in contemporary science is driven by scientists who
require a better method or instrument to better study a phenomenon. Chro-
matography, the Geiger-Müller counter, the spectroscope and electrophoresis
for example has each been applied to make at least a dozen Nobel-Prize-
winning discoveries, and scientists developed each of them.(138) How we
advance science can be categorised in five main stages. The methodologi-
cal constraint needs to first be identified, as researchers generally run into
a practical problem they cannot solve when using a method or instrument
to study a phenomenon. Next, methodological scanning is the process of
scanning one’s own field and related fields for a method or tool that can
address the given constraint—and if not found, moving to the next stage.
Methodological conceptualisation is the process of conceiving the new method
or instrument designed to address the methodological constraint we face.
Methodological development is the process of acquiring the needed resources
and creating the new method or instrument that enables us to measure and
observe the world in new ways. And new-methods-driven discoveries is the
final process of applying the new method or instrument to make a new
breakthrough.
Within this new-methods-drive-science framework, we can begin to strate-
gically plan and target efforts to create scientific tools that enable us to expand
the scientific frontier. There are two central ways for us to increase the speed at
which we develop new breakthroughs. One way is directing more attention to
expanding existing methods and instruments, recombining them in entirely
new ways and creating completely new ones that allow us to make scientific
advances. The other way is using scientific tools available in other fields to
address questions in novel ways and also sharing knowledge of new scientific
tools across fields immediately once they are created. To do so, leading journals
and institutions will need to incentivise researchers to develop and publish
methodological innovations. For some methods and instruments are often not
used in other fields for decades, such as RCTs that were applied in medicine
 NEW-METHODS-DRIVE-SCIENCE THEORY 117

long before psychologists and economists began applying the method that
revolutionised their fields.
A central point in this book that goes against the common view in science
and science policy is that science is and should be question driven. Scientific
publications are generally structured around a research question. Novel meth-
ods and tools of science have however often led to scientific breakthroughs by
conducting exploratory research using them—independent of any existing or
not-yet-formulated questions. Making scientific advances can be fostered by
focusing our attention not just on question- and hypothesis-driven science but

Downloaded from [Link] by guest on 29 August 2025

also on methods-driven science that is exploratory (without any predefined
question). Academic journals, grant committees and university hiring panels
need to equally recognise the development of methodological advances as cen-
tral to doing and advancing science. The hope of this book is to help spark such
a methodological shift in how we view, carry out and reward research.
Our methodological toolbox, through collective mind–method synergies,
has been and will continue to be the main mechanism we have to address
the big questions and challenges we face. These range from developing new
technologies that shape contemporary life, to extending the way we make
evidence-based biomedical, social and environmental decisions. So how can
we improve science and push the boundaries of science? The answer to this
question helps us better understand the science of the future—and the answer
lies in understanding the power of our methodological toolbox and in taking
steps to refine it. The answer involves more effectively tapping our abilities to
use and improve our methodological toolbox in innovative ways—the topic of
the remaining chapters.
A final important realisation and implication here is that some of our lead-
ing scientific methods, from statistical methods to RCTs, could have turned
out differently from how they did. And they will turn out differently from
how they currently are as we continue developing them over time. Placebos,
double-blinding, p-values and the like are not inevitable. Other methodolog-
ical solutions to the problems they aim to address are conceivable when
conducting controlled experimentation.(42) Once we become aware of this
fact, it opens our minds to the undetermined power of the methods we create
and the possibility of developing a range of previously unimaginable meth-
ods to address new problems and challenges we face. This would enable us
to continually expand and reshape our tower of science. And it could mark
the beginning of a methodological revolution in science, and enable us to
continually push back the present limits of science.
 18
The Limits of Science: An Overview

In the next three chapters we will now pull together the different evidence to
outline the present limits of science and, most importantly, how to push those

Downloaded from [Link] by guest on 29 August 2025

limits and expand our research frontier. Are we approaching the boundaries
of science? Can we continue advancing at the periphery of space, matter and
complex ecological, biological and economic systems? Are we close to reach-
ing the boundaries in fields like fundamental physics and brain science? These
are fundamental questions about science we have not yet addressed. We know
that the universe came into existence about 14 billion years ago, that funda-
mental forces govern physical reality (gravity, electromagnetism, the strong
force and the weak force), that the periodic table of elements represents the
chemical elements that make up the world, that the universe and life evolve,
and that DNA carries genetic information needed for biological organisms to
develop, function and reproduce. These make up essential pillars of science,
so they would unlikely be substituted by completely different breakthroughs
and theories that are as extraordinary.
Are we thus nearing the boundaries of science and significant break-
throughs? Since the turn of the millennium, many groundbreaking discoveries
have been made, such as the Nobel-Prize-winning discoveries of CRISPR gene
editing in 2012,(245) the Higgs particle also in 2012 and the existence of grav-
itational waves in 2015,(246) among other major discoveries that did not win
the Nobel Prize such as the mapping of the human genome in 2004. These
recent discoveries redefined the frontiers of genetics, physics and astron-
omy and it does not appear that science will stop expanding soon. But how
did we make these discoveries? Each new breakthrough advancement that
pushed the frontier has been propelled by a newly developed methodology
or instrument: in these cases, a new method of differential RNA sequencing
developed in 2010, the large hadron collider (particle accelerator) in 2008,
the upgraded LIGO laser interferometer in 2015 and an improved genome
mapping technique in 2001, respectively. To go beyond our present scien-
tific horizons, we must comprehend their present limits and the particular
constraints we need to tackle to extend those limits. We need to specifi-
cally understand the limits of our current methods and instruments that

Science of Science. Alexander Krauss, Oxford University Press. © Alexander Krauss (2024).
DOI: 10.1093/9780198937401.003.0019
 THE LIMITS OF SCIENCE: AN OVERVIEW 119

enable us to study, measure and perceive the world and that shape the lim-
its of our current knowledge. For they enable us to gauge our scientific
progress and when we reach the borders of how we presently study the
world.
At the forefront of research, many mysteries exist that we have not yet solved:
understanding the foundations of mathematics; the fundamental nature of
consciousness, memory and time; identifying new mechanisms for extending
our lifespan and mitigating climate change; comprehending what our universe
is comprised of, and how stars, planets and human cooperation evolve,(247)

Downloaded from [Link] by guest on 29 August 2025

how to eradicate cancer, and how to best use artificial intelligence and machine
learning to advance science and society.
What our present limits of science are and how to address those limits are
among the most important questions we can ask. While biology and medicine
have made vast strides in reducing disease and improving life expectancy,
many illnesses persist. While physics and chemistry have brought about vast
technological and industrial advancements that shape contemporary society,
they have also fostered problems such as climate change, environmental degra-
dation, warfare and mental health issues related to dependencies on electronic
devices, which we have not yet solved. The scientific community does not gen-
erally investigate our scientific boundaries as they are seen as difficult to study
systematically, and thus a comprehensive and interdisciplinary analysis does
not yet exist.
Researchers have generally studied what particular phenomena we are not
yet able to study—not how we push our boundaries of science to be able to
study new phenomena in the world. Science published a special issue that out-
lines 25 large outstanding questions facing science.(247) Researchers studying
the boundaries of science have in turn investigated the topic taking a perspec-
tive from physics,(248,249,250) mathematics(249,251,252) or particular fields,(253)
from the human mind(254,255,256) and philosophical aspects about scientific
laws and scientific induction.(220,252,257,258,259) In a book that aims to study the
limits of science, Gleiser surveys the history of physics and aims ‘to illumi-
nate a variety of scientific and philosophical viewpoints’ largely by exploring
conceptual shifts in theories of physical reality and their limits.(248) In The
End of Science, a highly influential and cited book on the boundaries of sci-
ence, Horgan pessimistically argues that science may be nearing its end as
the big questions have nearly all been addressed.(253) But given the recent
major scientific advances and big open questions outlined above, we can see
that this is not the case. An integrated explanation of the present bound-
aries of science yet remains elusive. Understanding our scientific boundaries
and how we can extend them requires however combining methods and
120 SCIENCE OF SCIENCE

evidence from different fields as they provide different pieces to the overall
narrative.
We will explain how our methods and instruments and our mind used to
develop them set the present limits of what we can know and what is possible
in science—and economic, social, historical and spatio-temporal influences
help shape what we study within those limits. Researchers inevitably conduct
science applying a given method or instrument and our cognitive abilities,
and do so from within our spatio-temporal context. Developing methods and
instruments enables us to continually push back those present limits and is

Downloaded from [Link] by guest on 29 August 2025

commonly the factor we can most directly and strongly influence to do so. By
creating new methods and instruments, we can reduce our present cognitive,
methodological and social constraints to investigating the world and vastly
expand our scope and understanding of the world. Science reaches its limits in
explaining reality as it reaches the limits of the tools of science we have created
thus far which enable perceiving and representing nature. Expanding science
takes place at the pace at which we design new methods and instruments
that enable us to do science in new ways. Extending our bodies of knowledge
represents the shifting scientific borders reshaped by extending our method-
ological toolbox. Methodological growth is necessary for major scientific
progress.
Viewing new methods and instruments we develop as the main mechanism
of how we advance science provides a new framework for addressing ques-
tions about expanding the limits of science. Understanding how we advance
our frontiers by developing new tools and how cognitive, social and economic
factors foster those new tools is the best way we have to understand what shapes
the scope of science and its limits and, most importantly, how we can extend
those limits. The paradox is that by understanding what constrains us we can
best overcome those constraints and continue to advance science. This new-
methods-drive-science mechanism here has wide applicability in advancing
science across fields.
In this chapter we first explore historically how we have extended the bor-
ders of science and we then provide a general conceptual description of the
scope and limits of science. In the next chapter we illustrate the particularly
powerful role of the present boundaries of our methods and instruments shap-
ing the present boundaries of science. And in the following chapter we describe
how we can make advances at the frontier quicker by describing the steps to
extend our scientific tools to study the world in novel ways. We also discuss
whether there are pre-established boundaries to some domains of knowledge
and we outline pathways of the future of science that we can take.
 THE LIMITS OF SCIENCE: AN OVERVIEW 121

Scientific progress over time: persistently pushing back

the borders of science

In the early 16th century, most regarded the physical universe as geocentric
and finite. However, advancements in scientific methodologies and instru-
ments emerged during the 17th century, including calculus, the telescope,
statistics and the barometer. These innovations facilitated novel approaches
to investigating and theorising about the world, ultimately leading to the
widespread acceptance of a heliocentric and infinite physical world by the

Downloaded from [Link] by guest on 29 August 2025

end of the century. In the world of medicine, a systematic understanding
of how diseases, treatments and their causes were connected was lacking in
the mid-19th century. However, by the early 20th century, the creation of
systematic controlled trials and the methods of randomisation, blinding and
placebos, which were combined with statistics, revolutionised experimental
practices. These methodological enhancements significantly improved our
ability to assess the causal effects of medical interventions, vastly enhancing
our well-being and lifespan. How we give rise to such new understanding of
the world reflects a general principle across science: developing new methods
and instruments of science is how we expand the research frontier in new ways.
We have stretched the periphery of science continuously throughout his-
tory. To reduce our limited vision, we created instruments such as the micro-
scope in 1590 and then the telescope in 1608 that revolutionised science.(260)
The advent of microscopes transformed our understanding of human disease,
offering unprecedented insight into viruses, bacteria and microorganisms. The
development of telescopes obliged us to reassess our knowledge about plan-
ets, our solar system, the size of the universe and our place in it. We thus
reshaped the boundaries of disciplines like biology and physics which were
then no longer set by the bounds of our sensory abilities but by the bounds
of these new tools. Such tools were refined and very powerful up to the 19th
century until we hit the blurred boundaries of their resolving power and thus
again the boundaries of science in different domains. In the decades up to
the 1930s, scientists could perceive many phenomena but not others, despite
awareness of their existence. However, the creation of the electron microscope
in 1933 marked a pivotal moment, enabling researchers to explore previously
unseen realms such as living cells, large molecules and crystals.(261) The elec-
tron microscope revolutionised multiple disciplines and forced a redefinition
of their boundaries, which were once again set by this groundbreaking tool.
To reduce our cognitive constraints, we developed randomised controlled
experimentation in 1948 that reduces biases in the scientific process and
122 SCIENCE OF SCIENCE

enables us to better assess how effective our best medical treatments and pub-
lic policies are. This method redefined the way we evaluate and understand
the effectiveness of our medications and economic policies.(42) We thereby
reshaped the boundaries of science across the biomedical, behavioural and
social sciences. The present bounds of science are largely synonymous with the
current bounds of our best tools we have developed. In this sense, only by con-
structing the telescope for example was Galileo able to discover Jupiter’s moons
in 1610. Only by developing calculus was Newton able to describe the laws of
motion in 1687. Only by creating X-ray diffraction methods were Franklin,

Downloaded from [Link] by guest on 29 August 2025

Crick and Watson able to detect the double helix structure of DNA in 1953.
Such discoveries would otherwise not have been possible.
Throughout history, the advancement of science has predominantly been
attributed to the efficacy of the tools we have created. This fundamental fact
has been largely overlooked, leading to the absence of a comprehensive the-
ory of science outlining the role of methods in driving scientific progress and
defining its limits.(220,247,248,249,250,251,252,253,255,259) The scientific periphery
has been extended without a comprehension of the pivotal role played by novel
methods and tools. We have redrawn the limits of science in an ad-hoc way, by
individual researchers who happen to expand a given tool that enables better
studying a given phenomenon.
New tools have thus driven our expanding scope of the world in a cumu-
lative process. In the 17th century, the emergence of six groundbreaking
methods and instruments triggered most scientific breakthroughs at the time:
the microscope developed in 1590, telescope in 1608, barometer in 1643, air
pump in 1659, statistics in 1663 and calculus in 1675. The leading scholars,
Galileo, Hooke, Boyle, Newton and their contemporaries, each leveraged one
or more of these new methodological innovations to expand our understand-
ing in astronomy, biology, physiology, pneumatics, mechanics and optics.
These tools and their refinements continue to occupy an essential role in sci-
entific research. Scientific progress is thus not driven by chance but guided by
developing innovative scientific tools (Figure 18.1).
As we expand science, pushing the frontier has required increasingly sophis-
ticated methods and knowledge over the centuries. Hooke could discover
cells in 1665 using a novel low-power microscope. Newton could explain
gravitation in 1687 using mathematical methods and direct observation.(223)
Today, to contribute to our existing body of knowledge on cells or grav-
ity we need to first learn how to apply the needed complex methods
and instruments. We must also study the existing body of knowledge on
cells (by Flemming, Claude and others) or existing theories of gravity
(by Newton, Einstein, quantum theorists and others) before being able to
 THE LIMITS OF SCIENCE: AN OVERVIEW 123

Methods/instruments developed → subsequent discoveries made

LIGO detector, improved (2015) → observation of gravitational waves (2015)

differential RNA sequencing (2010) → method for genome editing – CRISPR (2012)
spectroscopy, improved adaptive optics (2002) → supermassive compact object at the centre of our galaxy (2002)
starch–column chromatography (1948) → synthesis of polypeptide hormones (1953)
electron tube, improved (1913) → laws governing impact of electron on an atom (1913)
X-ray diffraction (1912) → structure of DNA (1953)
Pushing and spectrograph, improved (1859) → atomic light signatures (1859)
discharge tube (1875) → X-rays (1895)
redefining the
microscope, with silver staining technique (1873) → structure of nervous system (1873)
limits of leyden jar (1745) → nature of electricity (1752)
science with thermometer, mercury (1714) → oceans control global weather (1770)
new methods calculus (1675) → hydrodynamics (1734)
and statistics (1663) → field of demography (1663)
instruments improved microscope (1662) → cells (1665)
air pump (1659) → Boyle's law (1662)
barometer (1643) → Pascal’s law (1647)
telescope (1608) → Jupiter's moons (1610)

Downloaded from [Link] by guest on 29 August 2025

Limits of science and our understanding of the world in the…
future 2000s 1900s 1800s 1700s 1600s <1000s
1100s
1200s
1300s
1400s
1500s

Figure 18.1 How we comprehend the world expands at the rate we advance new
scientific methods and instruments.

identify ways to expand them. Galileo could identify the moons of Jupiter in
1610 using the recently developed telescope, but to establish which chemical
compounds make up the moons of Jupiter and how they came into existence
requires first developing highly advanced telescopes and spectrometers. With-
out such methods, instruments and knowledge, we would not be able to push
the research frontier and we would not know we have without first acquiring
the existing knowledge. Many low-hanging-fruit discoveries have been made
using two paradigmatic instruments—microscopes and telescopes—by simply
pointing the new instrument (with increasingly greater power) at phenomena
around us, without searching for them. Yet such discoveries have become less
common over the past centuries. For each generation it thus takes us longer to
reach the [Link].(171)
The success of science is commonly viewed as culminating in the great the-
ories in physics of relativity and quantum mechanics that explain the world
on both the smallest and largest scales. It is reflected in the great discoveries in
biology of the structure of DNA and the mechanisms of evolution that explain
124 SCIENCE OF SCIENCE

the living world and the secrets of [Link].(262) Such discoveries have generally
given scientists the impression that there may not be big missing pieces to
understanding the world. We cannot however understand how science will
progress in the future by just viewing existing discoveries. We need to under-
stand what drives discoveries. Here we illustrate that major breakthroughs are
mainly driven by methodologically inclined scientists and inventors—those
who design our electron microscopes, radio telescopes, spectroscopes, con-
temporary statistics, X-ray diffraction methods and a vast array of methods
that have extended our lens to the world. These are the heroes of science who

Downloaded from [Link] by guest on 29 August 2025

make studying the world in new ways possible.
Despite greatly expanding our boundaries of science, many unsolved mys-
teries remain that inspire us: Is there other intelligent life in the universe?
Where do black holes come from? What are the causes of many mental
disorders?(247) What drove the cognitive revolution in early humans? How
much of mathematics is explained by nature or culture? How do our brains’
electrochemical impulses transform into our cognition and emotions? How
can we reconcile gravity with electromagnetism? How can we best change
human behaviour to mitigate environmental degradation? May technological
intelligence one day exceed human intelligence (that is, is so-called singular-
ity attainable)? And so on. The unknown fascinates us. Providing answers
to fundamental questions also generally gives rise to new, not-yet-conceived
questions in an iterative process at the research frontier. The key factors
constraining and enabling us to address such fundamental questions are
our methods and mind, and also our social, biological and spatio-temporal
influences.

The scope and limits of science: the border between

the known and the unknown

What and how we perceive the world is invariably shaped by the scope of
our methods and cognitive and sensory abilities—that is, our methodologi-
cal toolbox—but also often by our social, economic and historical influences
as well as our human needs and objectives (Figure 18.2). Together, they set the
boundaries within which we are able to observe, process and understand phe-
nomena in nature and society. We then push the present limits of science by
developing tools that reduce our cognitive, methodological, social and other
human constraints. Methods and instruments are a necessary condition for
major scientific advances. Again, among these different factors, developing
 Our methods and instruments

Downloaded from [Link] by guest on 29 August 2025

(Our adaptive methodological toolbox) Legend:
Further into Phenomena studied in science–
Methods/instruments used to
the macro level study them
Size and nature of universe–Hubble telescope
Source of globle economic stability–statistics
More vast More rapid or slow
Speed of light–interferometer
Gravitational waves–laser interferometer
Seafloor spreading–seismometer
Giant magnetoresistance–molecular beam epitaxy
More complex
More intangible Ecological or biological systems
Functioning of the mind–microelectrode –comparative ecological/biological analysis
Black holes–mathematical field equations Society or markets–statistics, controlled experiment

Human mind
Further into the past (Our universal
methodological toolbox) Further into the future Our methods and instruments
Events thousands or billions of years ago
such as origin of universe and planets– i.e. cognitive abilities for observation, Events in deep future such as evolution of our
mathematics (Hubble constant) problem solving, experimentation etc. planet or species–computational simulations
Origin of consciousness or language– Climate change–statistical modelling
abductive reasoning
* Present limits of
our cognition *
More distant
Earth’s core–seismometer Spatio-
Galaxies and pulsars–radio telescope Human mind temporal
influences

* Present limits
of our methods * Further into Social/ecomomic influences
the micro level
* Present limits Germs and viruses–microscopes Biological/human influences
of science * Photons at quantum level–lasers
Carbon dioxide–chromatography

Figure 18.2 The boundaries of our mind and methods (our universal and adaptive methodological toolbox)
shape the present boundaries of science.
Examples are provided for each category as illustrations, with the phenomena also studied using other methods and instruments.
Phenomena can fit into more than one category, with for example ecological systems being macro-level phenomena with greater levels of
complexity.
126 SCIENCE OF SCIENCE

methodological innovations is the feature of science that we can most directly

influence.
Beyond our methods and what our mind is methodologically capable of
developing, we have no alternative way to broaden our scope to the world.
Science and its boundaries are largely fixed in a circle that begins and ends
with our tools. But we continually reshift the lines of that circle by invent-
ing better tools. Comprehending the bounds of science is thus contingent
on recognising that we humans carry out science, and we face cognitive and
methodological constraints to doing science that delineate our present bounds

Downloaded from [Link] by guest on 29 August 2025

at any given point. Pushing the research frontier is about reducing these
constraints.
Science and knowledge can be understood on a spectrum: with knowledge
of phenomena that are directly observable at one end of the spectrum—such
as flora, fauna and our habitat. This is how we have acquired much of our
knowledge in fields like botany, anatomy and palaeontology. Then there is
knowledge of phenomena that lie beyond the directly observable conditions
in which our mind and senses have evolved but that we can access using our
scientific instruments and methods—such as microscopic cells, quanta, galax-
ies, statistical probabilities of diseases and global pandemics. And, at the other
end of the spectrum, there are theories presently developed about phenomena
that lie beyond those conditions and we lack sufficient empirical evidence of—
such as the size of the universe, the historical origin of life, string theory and
singularity (Figure 18.3). We cannot access, with our mind or methods, such
phenomena as we move further away from our cognitive and methodological
niche.
As the present boundaries of science are most strongly and directly shaped
by the boundaries of scientific methods that we develop using our mind, we
turn to this topic in greater detail in the next chapter. There we also demar-
cate when we, given the limits of our methods and mind, cannot verify our
scientific theories and thus reach the limits of science.
 THE LIMITS OF SCIENCE: AN OVERVIEW 127

Downloaded from [Link] by guest on 29 August 2025

How we acquire knowledge and do science
No means
String theory, the size of the universe, the
historical origin of life, the extinction of some
giant mammals, multiple universe theory etc.

Our reliable theories

Quantum theory, theory of evolution,
Non-observable,
theory of relativity etc.
verifiable and reliable
knowledge
Limits of
Our methods and instruments science
(supported cognition and perception)
Statistics, controlled experimentation,
Observable, microscopes, telescopes, computers etc.
verifiable and
reliable knowledge
(at present)
Our human mind
Observation,
problem solving,
causal reasoning etc.

Our scientific scope and limits are also shaped by:

• Social/economic influences
• Biological/human influences
• Spatio-temporal influences

Figure 18.3 The scope and limits of science.

 19
The Limits of Science: Grounded in the
Boundaries of Our Methods and Mind

Downloaded from [Link] by guest on 29 August 2025

We still poorly understand where we draw the boundaries of science and when
we reach the present limits of our scientific theories about the world. Influen-
tial theories about the size of the universe, superstrings, the historical origin of
life, the general causes of democratisation and the emergence of our conscious-
ness exist across science. But can these influential theories be scientifically reli-
able if we cannot yet empirically test them rigorously? We know a lot about the
foundations and boundaries of our mind,(21,24,50,74,81,87,102,104,106,109) but the
topic has been studied largely independently of the foundations and bound-
aries of science.(248,249,250,251,252) Biologists, psychologists and anthropologists
highlight the role of the human mind in enabling our ability to reason that has
been shaped by cultural processes over our evolutionary history.(21,52,88) Cog-
nitive scientists have argued that because our mind is the product of evolution
it must have a biologically endowed structure and particular limits.(254,255,256)
Noam Chomsky for example studies the limits of our language and mind and
how they may influence our understanding.(256) Mathematicians and physi-
cists studying the limits of science have explored the question focusing largely
on areas in mathematics, physics and their history.(248,249,250) Researchers
have not yet explored the foundations and limits of our mind and how they
specifically shape the foundations and present limits of our knowledge and
science, by integrating evidence of the evolutionary origins of our mind,
and the boundaries of science and scientific methods we develop using our
mind. We do so here to explain how our cognitive, sensory, methodological
and instrumental capacities and constraints drive the present limits of sci-
ence, influencing the theories about the world we are able to develop and test
and those we are not yet able to. We explain what types of phenomena are
accessible to scientific investigation and what phenomena are presently not.
Our cognitive, sensory, methodological and instrumental abilities generally
become more constrained and less reliable in acquiring knowledge and doing
science as we get further away from studying phenomena that we can per-
ceive within our environmental and cultural niche in which our mind and

Science of Science. Alexander Krauss, Oxford University Press. © Alexander Krauss (2024).
DOI: 10.1093/9780198937401.003.0020
 GROUNDED IN THE BOUNDARIES OF OUR METHODS AND MIND 129

senses have evolved. This is especially the case as scientists develop theories
about phenomena like the size of the universe, dark matter and the general
causes of democratisation that are far beyond that niche. The central argu-
ment here is that we reach the present limits of science, and what science itself
is, when our theories involve phenomena that are not observable and thus the
theories are not verifiable and empirically reliable using our mind, methods and
instruments. This is called here the OVER criterion of science ( for observability,
verifiability and empirical reliability). Thus, if real phenomena in the world
are not directly or indirectly observable using our mind, methods or instru-

Downloaded from [Link] by guest on 29 August 2025

ments, then we lack sufficient evidence to develop theories about them that
are verifiable and empirically reliable, which is when we reach the present
limits of science. These are the features of phenomena that fall beyond our
sensory range that we are not able to perceive and are not presently acces-
sible to our mind, including the methods we develop using our mind–from
x-ray devices and lasers to computational techniques. They reflect the con-
ditions in which we hit the boundaries of what we can reliably study. It is
defined as the point at which phenomena are disconnected from our direct
perception (unaided, using our cognition) and from our indirect perception
(aided, using methods and instruments we develop using our cognition). And
we cannot yet access such phenomena in the world. The unique feature of this
account of science is the focus on and evidence provided by the evolution of
our mind and especially the present boundaries of our mind, methods and
instruments. The OVER criterion of science thus reflects the particular fea-
tures of phenomena that, given the limitations of our mind and methods, reach
the boundaries of our scientific theories and science—as we outline in the
chapter.
This explanation provides a new foundation for grounding science and its
central evaluation criteria of empirical testability, verifiability and reliability.
These are the essential features for explaining and understanding the present
boundaries of science: whether we view evidence or a theory about the world
as reliable depends on whether our evolved cognitive system (including the
tools we create using it) is able to observe the given phenomenon and test
and verify that evidence or theory. In this chapter we focus on scientific theo-
ries and their limits. A scientific theory aims to explain a phenomenon in the
world;(263,264) to ‘explain how or why something happens’ in the natural or
social world.(265)cf.(266,267) If our theories involve phenomena that do not meet
the OVER criterion, if they are thus not observable, verifiable and empirically
reliable, then we are not able to explain the phenomena and we reach the limits
of our scientific theories.
130 SCIENCE OF SCIENCE

In this chapter we first provide a novel empirical explanation for the present
limits of science, namely how our human mind, methods and instruments
constrain the theories we are presently able and not able to develop—that is,
theories about non-OVER phenomena. We then provide insight into the ori-
gins and foundations of science’s central evaluation criteria that are grounded
in the OVER criterion of science. Overall, we outline the foundations and
limits of scientific theories and thus science in light of the foundations and
limits of our mind and the methods and instruments we develop using our
mind. What emerges is the first account of a criterion of science that demar-

Downloaded from [Link] by guest on 29 August 2025

cates the foundations and limits of scientific theories and thus science which
is grounded in evidence on the evolutionary origins of our mind and on the
limitations of our cognitive and methodological abilities. We will then draw
implications for how the OVER criterion of science can help clarify ongo-
ing debates about influential scientific theories—such as on multiple universes
and superstrings—that have not yet been verified.

The borders of science are shaped by our human mind and

methods—our universal and adaptive methodological toolbox

While our mind and senses have evolved over human history largely to ensure
our basic needs and survival within our observable niche of the world, by
extension they have evolved in more recent history—especially over the past
70,000–100,000 years—in ways that enable us to formulate theories about
many phenomena that are not directly observable. We have accurate theories
about many phenomena we have direct (generally observational) experience
of and which we can thus empirically test and verify. These include flora,
fauna and habitats within our evolutionary niche. This is also the case for
many phenomena that are not directly observable, such as in chemistry
and biology, but which we can indirectly access and empirically test and
verify using our developed methods and instruments. In science, attaining
knowledge about real phenomena in the world is thus only possible for
phenomena we can get into contact with using our senses, mind and the tools
that our mind is capable of conceiving.
We face, in contrast, overwhelming constraints when it comes to formu-
lating and reliably assessing theories about phenomena that do not fulfil the
OVER criterion of science, such as multiple universes, the historical origin of
life, dark matter, the origin of the moon and the evolution of conscious expe-
rience. For they go beyond our spatio-temporal niche accessible to us using
 GROUNDED IN THE BOUNDARIES OF OUR METHODS AND MIND 131

our mind and methods. What such theories have in common is namely insuf-
ficient empirical data to prove them. The best we can do is make inferences to
the best explanation by interpreting and extrapolating from (related) available
evidence. These particular theories, if sufficient empirical evidence may arise
in the future, may turn out to be correct. But without such evidence, we are
not able to provide reliable and replicable knowledge. In many cases we can-
not reconstruct a phenomenon or restart a historical process to observe how
it has evolved.
The point here is not to isolate or criticise particular theories that are not yet

Downloaded from [Link] by guest on 29 August 2025

or may not become verifiable. But it is to highlight that the features of such the-
ories importantly illustrate when our knowledge cannot be reliable and where
the border of science lies: a border drawn by the OVER criterion of science.
While the present limits of science are dynamic and we can push them back
as we develop new methods and evidence, we reach those limits, at any given
point in time, when we formulate theories that are not yet testable, verifiable
and empirically reliable (non-OVER theories).
In Figure 18.2 we outlined how we are able to study phenomena using our
universal methodological toolbox, with abilities for observing, solving prob-
lems and experimenting that we are all born with. And how we are able to do so
using our adaptive methodological toolbox, with statistics and X-ray methods
that we collectively develop using these abilities and pass along within groups
and scientific communities. Together, they are what enable us to acquire our
vast bodies of knowledge about the world. Our universal methodological tool-
box reflects these evolved methodological abilities of the mind and allows us
to directly observe those parts of the world that our mind has largely adapted
to (our human domain or niche). Using our universal methodological tool-
box we can thus create sophisticated methods (our adaptive methodological
toolbox) to broaden our scope to the world. This enables us to access and
make sense of different phenomena (from chemical to immunological phe-
nomena) that lie beyond the scope of our universal methodological toolbox
and thus our human mind. We can think of the two as our core and extended
methodological toolbox.
Our early ancestors were likely not able to explain mysteries such as light-
ning, shooting stars, fire and many diseases. Today, we understand them well
thanks to the methods and instruments we have developed such as telescopes,
spectrometers and controlled experimentation. But our present methods only
enable us to partially understand phenomena like cancer, the origin of life,
many functions of the brain and the global economy. And we poorly under-
stand phenomena such as the size and nature of the universe and the evolution
132 SCIENCE OF SCIENCE

of our consciousness and language, as we have not yet created the methods
and instruments needed to shed light on them. Methods and instruments we
develop have enabled us to continually push back and redefine the present
limits of our scientific theories and thus science, and have driven scientific,
medical and technological advances.
It is with the present boundaries of the tools we have developed thus
far where we reach the present boundaries of what we can observe, test
and verify and thus what is reliable knowledge. Using our evolved mind we
gather information through our vision, perception of temperature and sense

Downloaded from [Link] by guest on 29 August 2025

of weight, time and speed. To stretch our sensory and cognitive abilities, we
create instruments like particle accelerators, electron microscopes and chro-
matography that have vastly increased our knowledge of the world of atoms,
microorganisms and chemical substances. To extend our cognitive abilities to
process that information, we develop methods like statistics and controlled
experimentation that have greatly improved our understanding of phenom-
ena such as effective medical treatments and human behaviour. Methods
and instruments help make phenomena that are not directly accessible to
our senses indirectly accessible to us and allow us to better make sense of
those parts of reality we can perceive directly with our bare senses. Think
of what is needed to study the world with thousands or even millions of
observations using regression analysis, for example of climate conditions or
medical patients. We were not able to do so until we developed contempo-
rary statistical methods and computers in the second half of the 20th century
(Chapters 12 and 13). Combining statistical methods with computers rev-
olutionised many fields of science, from experimental physics to medicine
and psychology. For it made large-scale statistical analysis possible for the
first time. This has been one of our greatest methodological revolutions. It
has also helped reduce researcher biases through mechanical and automated
processes.(30,42)
As we have seen in previous chapters, the purpose of developing a new
method or instrument is to improve our understanding or solve a problem
by reducing particular cognitive, sensory or methodological constraints. To
this end, microscopic and telescopic tools enhance the scope of our retinas to
reveal previously unimagined worlds of life within a drop of blood or water,
and moons around planets. Our mind is also not generally able to control
for, or understand well, the complex processes of many multivariate factors,
observable or unobservable, operating together at different levels within bio-
logical, physical or social systems. Statistics enable us to better understand such
complex systems (Chapter 13) as they allow us to reason inferentially with a
vast amount of data and process and analyse those data in complex ways and
 GROUNDED IN THE BOUNDARIES OF OUR METHODS AND MIND 133

over long periods of time. Our mind is not, without methods and instruments,
able to do such tasks.
Methods and tools are not just means of expanding our mental abilities.
They at times also represent a different mode of reasoning and theorising by
providing a different (often quantitative) way to view phenomena and a dif-
ferent scope to the world. Our universal and adaptive methodological toolbox
sets—at any given point in time—the present boundaries of what phenomena
we are able to observe, measure, theorise about and understand, and what phe-
nomena we are not yet able to. It makes up our world, our empirically verifiable

Downloaded from [Link] by guest on 29 August 2025

world.
Our evolved, universal methodological toolbox reflects our senses and intu-
itions that are most well adapted and we are most comfortable with. Without
developing highly complex tools, we would not have otherwise developed the-
ories about most phenomena beyond our directly observable realm—theories
about chemical compounds, our biosphere, infrared light and gravitational
waves. To do so, only methods and instruments can extend our reach. Our
basic perception and experience have been and remain at the centre of how
we develop our understanding of the [Link].(107) All scientific fields once
began using only our universal methodological toolbox. And fields like botany,
zoology and anthropology still mainly only use that toolbox, while other fields
like chemistry, molecular biology and astrophysics mainly study phenomena
using our adaptive methodological toolbox. Pushing the boundaries of science
is about collectively stretching our cognitive and methodological resources. In
a nutshell, that is the core of science and how we expand science.
We generally face more challenges in understanding phenomena the fur-
ther we move away from our human domain—the further we look back or
forward in time, the smaller and larger the phenomena we study, and so on
(Figure 18.2). We reach the present boundaries of science when it is no longer
possible to gather new data—or interpret and extrapolate from existing data in
new ways. Quantum mechanics is a peculiar exception as it is not very intuitive
but provides reliable predictions of the behaviour of atoms. In general, the rea-
son why our theories often become less reliable as we move further away from
our human domain is because we generally have less empirical evidence to
back them up.(34,107) We access a part of the world concentrated in our human
domain and accessible with our methodological toolbox. This is evident when
we think about all potential knowledge of the world—particular phenomena
at the micro and macro level (that presently lie beyond our cognitive, sensory
and methodological abilities), with high levels of complexity (that presently lie
beyond our cognitive and methodological abilities), over the span of history
(that presently lie beyond our spatio-temporal means of data collection) and
134 SCIENCE OF SCIENCE

outside our anthropocentric viewpoint (that presently lie beyond our biolog-
ical reach). When we study phenomena with increasing complexity, or with
increasing depth at the micro or macro level, the present limits of science are
generally shaped by human limits—our cognitive and methodological limits.
Yet when we study phenomena further back or forward in time, the present
limits of science are shaped by those human limits but also by greater diffi-
culties in accessing and gathering data as we move further into the past or
future.
Our methodological toolbox is thus the driving force of science and its

Downloaded from [Link] by guest on 29 August 2025

present limits. This basic fact is evident given that species, including our own,
do not evolve the cognitive and sensory abilities to be able to perceive phenom-
ena beyond what is directly observable, such as quarks and other galaxies, and
given that it is only through the tools we humans have recently created that we
are able to observe and verify such phenomena. Yet we need to understand the
present bounds of science loosely, changing and evolving as we expand and
improve our adaptive methodological toolbox.

The present limits of science are grounded in the present limits

of our methods and mind: observability, verifiability and
empirical reliability

We can access and verify phenomena directly (such as plants and animals) and
indirectly (such as molecules and bacteria). And our evidence and theories
about both types of phenomena are more concrete, reliable and replicable
(using multiple methods) than our theories about phenomena like the size
of the universe and dark matter. For these are presently not penetrable with
human sight or our instruments. We poorly understand such unobservable
phenomena and we face difficulties in attempting to develop theories about
them. This is partly explained by the basic fact that they are not the kinds of
phenomena that our mind and sense organs have evolved to be able to perceive
and process, or that the methods and instruments we create can deal with. In
this sense, the reliability of our best theories and the lack of reliability of our
non-verifiable theories are inseparably linked to our cognitive and method-
ological abilities. While we for example cannot directly observe the evolution
of our human mind itself, we have many different forms of empirical evidence
to test and verify it and thus also the theory of evolution. Anatomy describes
how species share similar traits that were present in a common ancestor.
Genetics demonstrates the degree to which species are related. Archaeology
 GROUNDED IN THE BOUNDARIES OF OUR METHODS AND MIND 135

studies fossils to illustrate how species, including our own, change over time
and how extinct species relate to living species and intermediate species.(268)
Theories reaching the present limits of science, such as on the fundamental
structure of reality, the historical origin of life, dark matter and the size of the
universe, can be called unverifiables. They are qualitatively of a different level
of abstraction than theories that deal with not directly observable entities like
molecules, cells and viruses but that we can, in contrast, indirectly observe
using for example electron microscopes. With unverifiables—defined as phe-
nomena that are disconnected from our direct or indirect perception—we do

Downloaded from [Link] by guest on 29 August 2025

not have sufficient empirical evidence to verify theories about them. We are
captivated by these deep questions concerning distant origins, vast complexity
and fundamental structures, but addressing them eludes us given our current
limitations in perception and experimentation. This is where science stands
still and we reach the threshold of our knowledge. Theories like those about
superstrings and multiple universes lack empirical evidence. Without good
empirical evidence, theories do not stand the test of time. We can thus classify
theories about different phenomena into one of three types (Table 19.1).
No way exists for us to perceive and verify phenomena in the world and
deem them reliable except by using our mind and what the mind is method-
ologically capable of. It is thus a basic fact of science that our present limits
of science are primarily defined by our present cognitive and methodological
limits. Social organisation, including institutions and funding, can also influ-
ence aspects of the limits of science, such as financial resources and education
levels needed to develop particular tools and knowledge (Chapters 6 and 7).
The OVER criterion of science establishes when we reach the boundaries
of our scientific theories. The criterion can be applied to assess any scientific
theory in any scientific field and research programme that aims to explain real
phenomena in the world—such as physics, chemistry, biology, medicine, earth
sciences, economics, psychology and agriculture, including the empirical and
theoretical branches of these fields. It thus refers to any theory in empirical
and theoretical physics, empirical and theoretical chemistry etc. It refers to
real phenomena we study and the theories of the world we develop about them,
and whether they represent observable, verifiable and reliable knowledge. It
does not refer to abstract methods we create to study such phenomena. These
include mathematics and logic that are not observable in an immediate sense
and are not real phenomena in the world independent of us—in any sense
comparable to scientific phenomena we study in the real world. All sciences—
from natural to social sciences—reflect empirical or theoretical fields or both,
whereas fields like mathematics, logic and computer science (at times labelled
 Downloaded from [Link] by guest on 29 August 2025
Table 19.1 The present limits of science: three types of theories about phenomena in the world.
The OVER criterion of science
Phenomena Theories Methods and
are observable (about evidence (on which
Type of theories about Accessibility to phenomena
Examples (directly or phenomena) theories are based)
phenomena are empirically
indirectly) are verifiable
at present at present reliable at present

Using our universal methodological

Direct observables Plants, animals and habitats toolbox (cognition and senses) to observe, Yes Yes Yes
(Direct OVER phenomena) experiment and solve problems
Molecules, chemical Using our adaptive methodological
Indirect observables compounds and viruses toolbox (methods and tools)
(Indirect OVER including microscopes and statistics that Yes Yes Yes
Mental states
phenomena) we develop using our mind and that reduce
and economic markets
cognitive and sensory constraints
The size and nature of the
Not currently accessible using cognition,
universe, dark matter and
senses or methods, but with direct
general causes of
evidence possible
democratisation
Present
Not currently accessible using cognition, limits of
The historical origin of life, the
senses or methods, and additionally science
Unverifiables moon and our evolved language, No No No
constrained given spatio-temporal data
(Non-OVER phenomena) the evolution of consciousness,
collection limitations. That is, no direct
and the future evolution of the
evidence is possible, as phenomena
planet
lie in the past or future
Pseudoscientific phenomena Phenomena have been empirically tested
such as homeopathy, astrology using scientific methods and proven not to
and classic psychoanalysis be reliable
 GROUNDED IN THE BOUNDARIES OF OUR METHODS AND MIND 137

formal science) are especially used as formal methods or language tools that
we create and, in contrast, apply within each of those fields of science. And
our focus here has been on scientific theories that aim to explain real phenom-
ena in the world—not on such methods we use to develop or evaluate those
scientific theories or on other aspects of science.
Theoretical science accounts for a small share of all publications across
scientific fields, as illustrated via the Scopus database (the largest database
of scientific journals).(66) Publications on ‘applied/empirical/experimental
physics’ (which do not include the terms ‘theoretical/formal physics’ in

Downloaded from [Link] by guest on 29 August 2025

the publications) account for 94% of publications, with 6% of publi-
cations on ‘theoretical/formal physics’ (which do not include the terms
‘applied/empirical/experimental physics’ in the publications). This share of
applied, empirical or experimental research accounts for 98% of publications
in psychology, 97% in economics, 95% in medicine and 88% in biology (using
the identical search terms with Scopus’s default search function). Especially in
theoretical science we are more likely to hit the limits of our theories and it is
here where they are often most disputed, such as string theory. It is when the-
ories involve phenomena that do not fulfil the OVER criterion. This criterion
enables us to assess the reliability of all scientific theories across empirical sci-
ence (that makes up most of science) and theoretical science (that make up a
small share of science). Any researcher can apply the OVER criterion to assess
any scientific theory that aims to explain a phenomenon across the physical,
biological, behavioural and social sciences. Science is a pragmatic enterprise
and the OVER criterion can thus be applied to likely over 99% of scientific
theories, which is often as good as we can get in scientific practice. The small
share of mathematicians, theoretical computer scientists and logicians who
may not view their fields as methodological fields (which involve methods and
tools applied across the natural and social sciences) do not need to apply the
criterion.
We need to also view these features of science together. This is because when
phenomena are not directly or indirectly observable, the theories we formulate
about them are not verifiable, and the methods and evidence they are based
on are not empirically reliable. How these features interact is through the sci-
entific process of observing phenomena (or attempting to), formulating and
verifying theories about them (or attempting to) and establishing whether the
methods and evidence on which they are founded are reliable and replicable.
This allows us to make predictions about theories. When we cannot fulfil the
OVER criterion of science, we reach our present scientific limits. Ultimately, if
phenomena in the world (P) are not directly or indirectly observable (O) via
138 SCIENCE OF SCIENCE

our universal or adaptive methodological toolbox (M), then we lack sufficient

evidence (E) to develop theories about them (T) that are verifiable (V ) and
empirically reliable (R), which is when we reach the present limits of science
(LS). Theories in science can be assessed using this limits-of-science rule:

If P is not O via M ⇒ insufficient E for T that is not V and R = LS.

There is also a distinction we need to make among unverifiable theories. It

is namely between cases in which we do not know how to develop the right

Downloaded from [Link] by guest on 29 August 2025

theory about some difficult-to-comprehend phenomenon and cases in which
we may have developed the right theory but we are not certain. In both cases
the challenge is that we are not able to gather sufficient empirical data and
verify them using our cognitive and methodological resources.
We are not directly able to perceive phenomena like DNA and quanta,
but we eventually developed instruments and carried out experiments that
enabled collecting sufficient empirical data to be able to test and verify them.
Yet DNA and quantum phenomena were beyond the margins of science before
we could collect data to confirm them. These are the distinguishing features of
theories, namely: between those about presently non-observable phenomena
that are non-testable and non-verifiable, and those that may become testable
and verifiable as we develop new methods and experiments. The point is that
these distinguishing features draw the defining lines between theories that
are not reliable and those that may become reliable. In the case of DNA and
quanta, we resketched the edges of the biological and physical sciences in
the 20th century. We resketched the edges by developing new methods and
instruments that enabled us to gather evidence about such phenomena that
moved from being unobservable and unverifiable to indirectly observable and
verifiable. For theories, or more precisely hypotheses, like those about super-
strings and multiple universes, we do not (yet) have methods and instruments
to empirically test them and they are thus outside the periphery of reliable
science.
Yet given our remarkable methodological abilities of the mind, where does
the mind end and the methods we create begin? We use our mind’s internal
abilities to develop complex external methods, and must interact with our
natural and social environment to do so. Methods are, once created, external
material artefacts in the world that can be shared and used by others—such as
statistical and computational programmes. Sophisticated methods we develop
facilitate the scientific process, such as generating hypotheses (using machine
 GROUNDED IN THE BOUNDARIES OF OUR METHODS AND MIND 139

learning methods), collecting and cleaning data (using database systems),

analysing data and simulating experiments (using statistical programmes).
Yet we always use our mind’s methodological abilities to develop such pro-
grammes and evaluate and interpret results, draw conclusions and assess
ethical and policy implications. All research and discoveries in contemporary
science require using both our universal and adaptive methodological toolbox.
While direct empirical evidence for phenomena is often lacking and abstrac-
tion can at times improve understanding about them, we eventually, when
phenomena are less and less observable, arrive at the bounds of science. Ulti-

Downloaded from [Link] by guest on 29 August 2025

mately, the foundation of science is thus empirical: if we cannot empirically
test a scientific theory about real phenomena, if it is unprovable, then it cannot
get beyond the realm of abstraction. While we increasingly require abstraction
and imagination to study indirect observables, we often rely on high levels of
abstraction and imagination when dealing with unverifiables. Yet in science,
using abstraction and imagination to study the unobservable, when empirical
testing is not possible, often remains largely that—abstraction and imagina-
tion. Our strongest theories about the world are founded on vast empirical
evidence and reinforced using a variety of methods and data sources collected
over time.
Consider a hypothetical scenario of science in which humans are conceived
as the pinnacle of a meticulous designer’s creation, rather than the result of
an evolutionary process within our ecological and cultural context. Our visual
capacity would be flawless, allowing us to perceive phenomena at the atomic
level, the vast universe and beyond, rendering microscopes, telescopes and
similar aids redundant. Our memory would be infallible, retaining precise
recollections of all observations without reliance on datasets. Our process-
ing capacity would enable rapid computation of thousands or millions of
observations, making obsolete the need for computational tools and statis-
tical methods to analyse complex phenomena. We would be endowed with
an omniscient understanding of reality that surpasses the limitations of our
human context. Such flawless abilities would facilitate unhindered scien-
tific research and making flawless predictions about the future. However,
the reality is that we humans have adapted evolutionarily to our ecologi-
cal and cultural context, which has inevitably shaped our cognitive capaci-
ties, perception and the methodologies we devise to comprehend the world.
Mind–method synergies set the present limits of science by shaping the
methods that our mind is presently capable of developing to study the
world.
140 SCIENCE OF SCIENCE

Science’s central evaluation criteria of testability, verifiability

and reliability are grounded in our evolved mind and methods

How reliable our evidence and conclusions are is a question that all scientists
must constantly deal with. This account of science here also explains how our
evolved cognitive system lays the foundation for science’s central evaluation
criteria. Whether we view evidence or a theory about the world scientifically
reliable depends on whether we, using our mind, are able to perceive the phe-
nomenon, test and verify that evidence or theory and develop methods and

Downloaded from [Link] by guest on 29 August 2025

instruments that are commonly needed to do so. Only with our human mind,
methods and instruments can we do this. Science’s central evaluation crite-
ria of empirical testability, verifiability and reliability are embedded in our
cognitive and methodological architecture.
Our cognitive and methodological limitations thus provide an empirically
grounded explanation for how we evaluate science. It is an evolutionarily
grounded and seemingly natural explanation. And the central features are cap-
tured in the OVER criterion of science that takes us further than other existing
criteria of science, which we discuss below. When we study the extremely
miniscule or large, the extraordinarily complex, or the very distant past or
future, theories about phenomena are eventually no longer verifiable and
reliable (Figure 18.2).
It is the theories of research programmes that move from being unestab-
lished to established science, depending on whether we can eventually empir-
ically verify them or not—and not just a property or feature but the entire
theory. The reliability and accuracy of our theories depend on how well they
fare empirically, with evidence in the world. Some are more reliable and accu-
rate than others. We increase the reliability of our evidence and theories when
they are consistent with one or more established areas of evidence and theories
and when they are consistent using other methods.
This account places our methodological toolbox at the foundation of sci-
ence and its evaluation: if theories about phenomena are not empirically
testable and verifiable using our methodological toolbox, then they cannot be
reliable—and thus cannot be consistent and replicable using different meth-
ods and evidence across different contexts. Yet why can’t we just say that ‘string
theory is elegant mathematical formalism and does not yield testable predic-
tions’ or that ‘we currently have no good theories of dark matter’? Why do we
need the OVER criteria to support these widely held views among scientists?
The OVER criterion provides a common measure to explain and justify why
such theories are not yet reliable and it helps us settle debates in science over
which evaluation criteria are necessary.
 GROUNDED IN THE BOUNDARIES OF OUR METHODS AND MIND 141

Defining the boundaries of science: shifting how we demarcate

science from what is and is not scientific, to what is and is not
presently scientifically reliable

Distinguishing between scientific and non-scientific theories is, for particular

cases especially at the research frontier, not feasible nor would it be mean-
ingful. We need instead to shift the debate to distinguish between reliable
and presently non-reliable theories. This reflects a pragmatic shift from
thinking about science in general idealised terms to doing so in practical

Downloaded from [Link] by guest on 29 August 2025

measurable terms. This cognitive and methodological explanation here of
the foundations and limits of science provides a new perspective to the
demarcation debate—the debate about the boundary between the scientific
and non-scientific. This question of how to demarcate science is one of
the central long-standing debates in philosophy of science. Historians,
philosophers and sociologists of science have attempted to address this
question by arguing that what is scientific and what is not depends on
whether it is or is not experimentation-based,(214) logically verifiable,(107)
falsifiable (refutable),(15,33) puzzle-solving and hypothesis-revising,(208)
guided by certain epistemic norms,(130) advancing a research programme(16)
or reflected in systematic knowledge(269) —or they have argued that no crite-
rion is needed.(17) The proposed criteria are narrow, each reflecting a single
criterion, and they are also commonly based on theoretical and conceptual
arguments—not empirical evidence.(31,259)
The present account is the first to be grounded in empirical evidence on
the limitations of human cognitive, methodological and instrumental capaci-
ties. It provides a more empirical, evidence-based and scientific foundation for
distinguishing between what is and is not presently scientifically reliable.(ibid.)
This methods-driven perspective here provides a new way to demarcate the
boundaries of science by placing the central role of our sophisticated scientific
methods, instruments and methodological abilities at the centre of focus. We
place here the powerful role of our methodological abilities and sophisticated
tools and their constraints and potentials at the centre of defining what sci-
ence is, its present limits and how to push those limits. This shifts the focus
from such theoretical definitions to an active process of understanding and
fostering methodological innovation to push our scientific borders. We con-
nect here our mind’s methodological abilities and sophisticated tools with
empirical exploration and verification of phenomena they enable into a unified
framework.
The central argument here is that we can best distinguish what is and is
not presently scientifically reliable by evaluating when our theories involve
142 SCIENCE OF SCIENCE

phenomena that are not yet (directly or indirectly) observable and thus the
theories are not verifiable and empirically reliable using our mind, methods
and instruments. This OVER criterion of science is always needed for getting
science off the ground—to be able to do science. It is as close as we may get to a
universal definition or unified scientific methodology for evaluating theories
that aim to explain real phenomena in the world. This criterion clarifies the
distinction of the borders of science for old theories that have led researchers
astray (such as geocentric theories of the universe, theories of spontaneous
generation and the steady-state theory of the universe), for pseudoscientific

Downloaded from [Link] by guest on 29 August 2025

theories (such as on homeopathy, astrology and classic psychoanalysis) and
for particular historical theories in fields like biology, geology and cosmology
for which we lack sufficient empirical evidence in the deep past. This crite-
rion helps distinguish between reliable and not presently reliable theories in
over 99% of cases in science. We may have to abstain from labelling particular
theories developed by some research programmes at the research frontier as
reliable knowledge until they may become empirically testable and verifiable.
Ultimately, the OVER criterion helps distinguish between established science
and those research programmes that have not yet been empirically established
(that are at the research frontier) or that may not become established (that are
beyond the scientific periphery). We shift the focus here from whether theo-
ries can in principle be tested to whether they can be tested in practice with
our current or foreseeable methods and technologies.
In sum, the bounds of science, and what science itself is, can be better
explained and understood in light of this account of what we are able to
observe, test and verify, and what we are presently not able to, by applying our
evolved mind and tools. The foundations and boundaries of our mind and
methods thus shape the foundations and boundaries of our knowledge. We
are only able to acquire knowledge about those phenomena in the natural and
social world that fall into our present sensory, cognitive and methodological
range that we are able to perceive and access. Where our mind, methods and
instruments provide us evidence of the world, we can have scientific knowl-
edge. Where they have not yet provided us evidence, we do not yet have
scientific knowledge. The history of science is a history about all the ways we
have transcended the limits imposed on us by evolution and by our methods
and instruments, at any given point, that we developed thus far. This account
of science is thus grounded in a seemingly natural explanation for both the
foundations and limits of our scientific theories and thus science—namely the
OVER criterion of science. By providing a new foundation for evaluating our
scientific theories and science that is grounded in our evolution, history and
 GROUNDED IN THE BOUNDARIES OF OUR METHODS AND MIND 143

empirical evidence, this criterion of science can resolve ongoing debates about
influential theories. This includes debates about scientific theories that do not
(yet) fulfil the criterion (such as on multiple universes and superstrings) and
that have been proven not to fulfil the criterion (such as on homeopathy and
classic psychoanalysis).
Most importantly, gaining deeper insight into the foundations and bound-
aries of science facilitates expanding science by directing our attention to our
particular cognitive, sensory and methodological constraints and establishing
ways to mitigate them. Each scientific method and tool, like our senses, has a

Downloaded from [Link] by guest on 29 August 2025

given range and limitations, and we can extend that range by understanding
those limitations and developing methods and tools designed to tackle them.
We thus push the limits of science by creating new tools that expand our scope
to the world.
 20
The Limits of Science: Expanding the
Limits by Expanding Our Methodological
Toolbox

Downloaded from [Link] by guest on 29 August 2025

In every era, there exist scientists who believe that we have already largely
solved the grand enigmas of the universe. Isaac Newton elucidated the laws
of motion and gravitational force, while James Clerk Maxwell delineated the
properties of electricity, magnetism and light. Dmitri Mendeleyev assembled
the fundamental components of the periodic table that form the cornerstone
of chemistry. Charles Darwin and Alfred Wallace outlined the evolutionary
principle underpinning biology. Adam Smith explained the principles of the
Wealth of Nations and the ‘invisible hand’ of capitalism: division of labour,
freedom of trade and pursuit of personal interest driving societal benefit.
These contributions laid the foundations for physics, chemistry, biology and
economics, and many then regarded these foundations as essentially complete.
As the 19th century closed, the laws governing motion, gravity, electric-
ity and magnetism, as well as conservation of energy and thermodynamics
seemed to explain our entire physical reality. Physics seemed to have unrav-
elled the fundamental laws that govern the world. In 1900, Lord Kelvin,
a pre-eminent figure in the field of physics, asserted during a meeting at
the British Association of Science: ‘There is nothing new to be discovered
in physics now. All that remains is more and more precise measurement.’
However, the early 1900s witnessed the emergence of revolutionary theo-
ries such as relativity and quantum mechanics, which unveiled significant
gaps in our understanding of our physical universe at both macro and micro
scales. Einstein’s relativity theory in 1905 introduced a novel way to perceive
time and space, and in combination with quantum theory developed in sub-
sequent years they revolutionised the field of physics. These advancements
posed a challenge: Can quantum theory, which concerns the behaviour of
particles at the subatomic level, be reconciled with relativity theory, which
concerns the behaviour of massive objects on cosmic scales? The unification of
electromagnetism with gravity remains an unresolved puzzle. That no major

Science of Science. Alexander Krauss, Oxford University Press. © Alexander Krauss (2024).
DOI: 10.1093/9780198937401.003.0021
 PUSHING THE LIMITS OF SCIENCE 145

breakthroughs in science remain to be made is a view held by many scientists

at any point in history.
In this chapter we will describe how we can make advances at the scientific
frontier quicker by describing the steps to extend our scientific tools to study
the world in novel ways. We will also discuss whether there are pre-established
boundaries to some domains of knowledge, and we outline pathways of the
future of science that we can take.

Downloaded from [Link] by guest on 29 August 2025

Extending the present borders of science by extending our
methods and tools

We continually make new scientific advances through our new methodological

and technological innovations, and the pace of making such innovations has
not decreased. In the past few decades, physics has been continually expanded.
In 2002 a supermassive compact object at the centre of our galaxy was dis-
covered by applying a new high-powered spectroscope, which revamped how
we understand our galaxy and our place in it.(270) In 2015 gravitational waves
were detected by applying a newly refined laser interferometer—100 years
after Einstein developed his general theory of relativity in 1915 that predicted
gravitational waves.(246) Biology and chemistry have also been continually
broadened. In 2000 the mapping of ribosomes was completed by applying
a newly upgraded electron density map, which fosters producing antibiotics
to combat the bacterial infections we face. In 2012 the gene editing method
CRISPR was discovered by applying a new differential RNA sequencing tech-
nique, which reshaped the life sciences, enables us to alter the DNA of plants
and animals and facilitates the fight of cancer.(245) These major breakthroughs
each earned a Nobel Prize and were brought about by inventing these new
methods and instruments. A general explanation does not however exist for
how we extend our scientific periphery which holds for all new major scientific
advances.(220,247,248,249,250,251,252,253,255,259)
Here we provide a general explanation of how developing new methods
and instruments is the central driving force of new scientific advances by
enabling us to study the world in novel and innovative ways. We have observed
earlier how science can be shaped by different factors (Chapters 2–15). Yet
while our mind, methods, scientific community and broader economic and
historical context influence science, the factor we can most directly influence
is developing and applying new methodological innovations that enable
making scientific advances. Newly created methods and tools are always
146 SCIENCE OF SCIENCE

2. Identify the
1. To expand our 3. Improve an
underlying constraint 4. Scientific
scientific boundaries existing or develop a
that is methodological advance/discovery
we commonly need new method
and at times also (made by applying
to address a major or instrument
cognitive etc. to the new method or
question, problem to address the
addressing the instrument)
or challenge constraint
question

Figure 20.1 General steps commonly taken to expand the frontiers of

science.

applied and are the factor we commonly require investing the greatest effort

Downloaded from [Link] by guest on 29 August 2025

into (the time dedicated to developing and applying them).
Science’s most powerful tools—from electron microscopes and X-ray meth-
ods to electrophoresis—were each developed in one field (such as chemistry)
but later used to make discoveries within different scientific fields (such as
biology, physics and medicine). These discoveries were not anticipated by the
inventors of these powerful tools and were not the purpose of developing them.
This highlights the direct causal link of new tools triggering scientific advances
that they were not made for.
While developing new major tools generally updates our understanding
about a part of the world, scientists generally concentrate on doing research
and not on studying and conceiving how we expand our tools and the needed
steps to do so. Yet this is required for us to expand our means of studying
and understanding the world. Designing new methodologies and instruments
that are better than existing ones can however often be challenging. As new
methodological innovations drive new advances at the frontier, we need to
address a critical question: What drives new methodological innovations and
how do we make them?
Pushing and redefining the limits of science is achieved through major
scientific advances—not by conducting conventional research that applies
existing methods to create knowledge. To stretch our existing boundaries in
a field and address a major question, we commonly have to identify and tackle
the constraints to studying the question. Constraints are methodological and
can simultaneously be cognitive, social, economic and related to other human
features. We need to thus improve an existing method or design a new method
to address the constraint, often iteratively through trial and error. We push
science through this methodological process (Figure 20.1). The discovery pro-
cess can at times begin with just this mechanism: breakthroughs are at times
brought about by a new method or instrument that we use in an exploratory
way, with no question being tested. Awareness of this mechanism enables us
to take steps to address our constraints and extend our scientific reach.
 PUSHING THE LIMITS OF SCIENCE 147

Think of the spectrometer for instance. Kirchhoff and Bunsen created this
instrument in 1859 which allows us to analyse wavelengths of electromagnetic
radiation that we were otherwise constrained in perceiving. This powerful
tool stretched our ability to investigate the structures of atoms and molecules
and the chemical composition of planets and stars, pushing back the borders
of the scientific periphery. Kirchhoff was thus labelled ‘as Bunsen’s great-
est discovery.’(271) Yet the limitations of current spectrometers have become
evident over time, including relatively low sensitivity,(272) often requiring com-
bining them with other methods like chromatography to better probe more

Downloaded from [Link] by guest on 29 August 2025

complex samples, and also computational constraints.(273) The spectrometer
is just one example among leading scientific methods and instruments, and we
discuss others later.
In general, when designing any scientific tool, we are asking how we can
answer a question by improving our existing cognitive or methodological
abilities to study the world. Establishing the methodological and human con-
straints we face is essential, as we cannot otherwise design a scientific tool that
can mitigate a constraint that we are not yet fully aware of. We need to thus
first generally detect a problem or limitation in the way we perceive or process
phenomena in the world that we cannot currently explain, and then identify
the basis of that problem or limitation.
The best researchers at the frontier are generally those who can best describe
the gaps and complexities in our knowledge we have not yet figured out
and then design and experiment with new methods and instruments that
can fill those gaps, or collaborate with those able to do so. Making break-
throughs is about generating the right new methods and instruments that
enable redrawing the lines at the edge of science.
This central point can be demonstrated with for instance the method of
RCTs. Prior to conceiving RCTs in 1948 our ability to explain the causes
of many medical diseases was limited. The RCT method—as most scientific
methods—is the result of continual refinement over time of previously used
methods and techniques. RCTs are, as outlined, an experimental method in
which people are distributed randomly into treatment and control groups to
test if a treatment or intervention may be effective. The creation of RCTs intro-
duced innovative methodological techniques such as randomisation, aimed at
mitigating confounding variables by randomly assigning experimental groups
within a sample to ensure balance across those variables between the groups.
Techniques of blinding were introduced to counteract biases like observer
effects and confirmation biases by withholding information that can inad-
vertently influence the behaviour of participants and researchers in the study.
148 SCIENCE OF SCIENCE

Placebos were also introduced to address the issue of participant expectations

of a treatment impacting experiments by providing a group of participants
with a placebo resembling the treatment but lacking a therapeutic effect.(42)
Through iterative refinement and integration of these techniques into the RCT
design, we gradually developed this composite methodology over time.
The RCT method has become the leading method across the biomedi-
cal, behavioural and social sciences for assessing how effective treatments
and interventions are and the method has redefined the limits of knowl-
edge in these fields. In this sense, the experimental setup of RCTs serves as

Downloaded from [Link] by guest on 29 August 2025

a powerful success story of our ability to alleviate constraints and deliver life-
saving evidence. And we will continue developing techniques to reduce the
human constraints and biases we face.(42) The invention of randomised con-
trolled experimentation exemplifies well how we devise methods and tools that
broaden the horizons of scientific research.
The invention of novel methodologies and instruments commonly requires
a point of reference rooted in existing tools. In our quest to expand compre-
hension across domains such as the complexities of the mind, the intricacies
of life or the vastness of the universe, we require methodological scaffolding.
This entails a process of exploration and experimentation aimed at innova-
tively building on and reconceiving our available tools to better explain a given
phenomenon. The transition for instance from rudimentary trials in early
medicine to the systematic evaluation of medical treatments through sophisti-
cated controlled trials exemplifies the culmination of successive methodolog-
ical enhancements and breakthroughs over time. These are closely linked to
developing the RCT method. The same applies to the technological advance
from horses to automobiles as a means of transportation, with our initial
motor-driven vehicles bearing much resemblance to horse carriages.
Screening major discovery-making papers we can establish the ways
new methods and instruments have been developed that sparked scientific
advances. We observe that improving existing and developing new tools gen-
erally involves recombining and extending features of existing tools. We have
extended the frontiers of science by expanding our methodological scope to
the world through different pathways:

• employing methods including tools from near and distant scientific

fields;
• integrating methods within and across scientific fields;
• experimenting with ways to upgrade our methods to address new ques-
tions and challenges;
 PUSHING THE LIMITS OF SCIENCE 149

• recognising the assumptions and limitations inherent in our methods and

devising strategies to mitigate them;
• inventing completely novel methods for tackling new kinds of challenges
we face;
• establishing our evolved cognitive, sensory and social constraints, and
enhancing our methods to mitigate them;
• establishing strategies to address features of our spatio-temporal
constraints.

Downloaded from [Link] by guest on 29 August 2025

A comprehensive explanation and framework has eluded us, until now,
for how to expand our scientific frontiers by expanding our scientific
tools.(220,247,248,249,250,251,252,253,255,259) We need to take such steps to broaden
our powerful methodological toolbox. These pathways hold enormous poten-
tial to understand parts of the world currently beyond our grasp by offering
novel perspectives. Much cutting-edge science entails experimenting with and
using available methodologies from across disciplines and devising, probing
and refining innovative methodologies, while applying much trial and error.
The question of how to maximise our cognitive and sensory capacities can
appear strange, but addressing this question is essential to mitigate our cogni-
tive and sensory limitations by developing tools that offer novel perspectives
and processing capabilities. Identifying which constraints remain unaddressed
is paramount because it opens uncharted research territories and enables mak-
ing advances by strategically examining and establishing those constraints
and experimenting with methods to mitigate them. Improved methods enable
studying greater complexity, deeper into the micro or macro levels, further
into the past, and so forth. Making advances requires continually reducing our
methodological constraints. We have not yet been able to establish the best
methodological means needed to address a given problem or its underlying
constraint for many phenomena. Consider big data methods or network anal-
ysis methods used to study highly complex phenomena, which can involve as
much noise and constraints as they hope to reduce. Tackling the existing limi-
tations of computational power, machine learning, big data methods, network
analysis methods and the like can play a significant role in fostering scientific
progress across disciplines.
Next, we outline common methods and instruments of science and the main
constraints they decrease, as well as their remaining constraints that need to
be addressed to further expand science (Table 20.1). Making major scientific
advances involves always reducing methodological constraints and at times
simultaneously other constraints (cognitive, social, economic etc.). Shifting
150 SCIENCE OF SCIENCE

greater attention to such constraints and reducing them will enable us to access
new phenomena and drive new advances.
Thus far we have outlined the scope and boundaries of science and can now
better delineate what they mean. How we expand the frontiers of science is
defined here as follows:

Pushing the present limits of science is a cycle in which we have fewer meth-
ods and instruments available as we get closer to the boundaries of science,
such as the electron microscope and statistical simulations. Here we have to

Downloaded from [Link] by guest on 29 August 2025

identify which scientific methods and instruments we need to improve or
newly develop to reduce our present cognitive, sensory and methodologi-
cal constraints to studying and accessing parts of the world. And once we
develop the new method or instrument—by building on and extending exist-
ing methods and instruments—we can access, measure and understand the
given phenomenon in the world in new ways. We thus push back the bound-
aries of science, and the cycle begins again in an iterative process in which
we continually expand science.

When researchers increasingly depend on abstraction and imagination such as

in parts of theoretical physics and theoretical economics, they are more likely
to hit against the limits of their field in the absence of empirical evidence. As the
rate at which we generate complex computational, statistical and technological
means is expanding quickly, science has the potential to continue expanding
quickly. What are novel ways we can create scientific methods? We can for
example increasingly carry out some studies using large-scale online collab-
orations with researchers linked via computers across the world and we can
continually update such studies in real time (similar to the online encyclopae-
dia Wikipedia) and without a predetermined result. Such studies can provide a
highly collaborative and iterative process in which replication and peer review
are integrated in the research design. We can also increasingly use automated
computational programmes to reduce some human biases in the scientific pro-
cess. This includes routine processes that take much time and applies across all
fields. This allows directing greater attention to creative aspects of the scientific
process.
About everyday knowledge, Ludwig Wittgenstein stated that ‘the limits of
my language are the limits of my world.’ About science and discovery, it is
mainly the limits of our scientific methods and instruments that are the limits
of our world. For our scientific tools are the present boundaries of how we are
able to measure and analyse our world. We hit the limits of science as we hit the
 Downloaded from [Link] by guest on 29 August 2025
Table 20.1 Creating new scientific tools advances science by mitigating our constraints to studying the world (examples of common
methods and instruments).
Method or Phenomena that the method or Current main constraints of method or
Year Constraints (methodological,
instrument instrument enables us to study or instrument (that we must overcome to,
first cognitive, social etc.) that the
developed to reduce explain (i.e. without which we would not once again, push the present limits of
developed method or instrument reduces
constraint be able to) science)
Observing microscopic phenomena such as Light microscopes have an ultimate resolution of
Our limited visual capability to magnify
Modern microscope 1873 cells, viruses and minerals; biomolecular ~250 nm, related to the diffraction-limited
and perceive very small objects
interactions resolution etc. (274)
Analysing large-scale data of phenomena across
Our limited cognitive storing and Limited computational speed per unit of energy/at
1940s/ most fields, complex computational analysis,
Computer processing capacity; mechanical fixed power supply; limited storage per unit of
1950s such as on climate modelling, genome analysis
processing of information etc. space; computation quantum speed limit etc. (275)
and econometric modelling
Our limited ability to control, randomise Bias can arise due to sample bias (as some people
Estimating how effective medical treatments
1948 and blind participants to study them in a refuse to participate), selection bias (as some are
RCTs and public policies are; more robust causal
controlled experiment; biases such as partially blinded/unblinded) and measurement bias
estimates in studies
observer effects and confirmation effects (as some are treated for different durations) etc. (42)
Machine learning algorithms for rapid data
Human error and bias; influence of cleaning, statistical analysis and identifying patterns Automation bias; generally requires obtaining very
Machine learning 1957 individual objectives and interests; limited in datasets without human influence; used in large datasets; ethical considerations when applied
processing capacity medical practice, speech recognition and data to medical issues etc. (276)
mining of population-level data
152 SCIENCE OF SCIENCE

limits of our best telescopes, particle detectors and statistical methods. It may
thus appear obvious but an important insight that has not been given much
attention in science is the following: the scope and limits of our present scien-
tific tools are the extent we currently reach in explaining our scientific reality;
they largely define the present scope and limits of science. Developing more
sophisticated tools has enabled us to continually push back the present bound-
aries in all fields that would have otherwise been unimaginable (Figure 18.2).
We cannot say much more about the scientific world than what our best tools
enable us to say. The central message here is: the key to expanding the limits of

Downloaded from [Link] by guest on 29 August 2025

science is designing and experimenting with better methods and instruments—it
is extending our methodological scope to the world.

Do we face pre-established boundaries in science?

Our species faces inherent limitations in perception and information process-

ing and has evolved to meet basic needs and reproduce, not specifically to
understand the complexities of life, the mind and our physical reality. Does
this imply predetermined boundaries to knowledge and science? Devising
novel methods and instruments enables us, method by method, to extend
the horizons of science. We go beyond our current methodological grasp by
combining, expanding and devising improved methods. While our biological
evolution, our cognitive limitations, our economic constraints and our spatio-
temporal context delineate the realm of the world accessible to us, they do not
dictate the kinds of tools we can conceive to mitigate those constraints and
better comprehend the world.
The universe is finite. Life is not eternal. The speed of light is not infinite.
The number of existing species of plants and animals is not limitless. The
combinations of chemical elements are not boundless. The planet’s natural
resources we consume are limited. No organised complex phenomena, like
the human mind or the universe, could exist or develop if there were no limits
to what is possible in nature and if there was not much stability in the world.
In our finite world, making major discoveries and opening scientific fields may
thus also not be limitless and endless. However, our cognition seems less con-
fined by explicit boundaries given our vast human creative imagination that
we can tap to conduct science and devise new methodologies.
Generalising about the present limits of science can be difficult. Across dif-
ferent fields the complexity of phenomena we study and the presently available
methods and instruments vary widely. Different fields of science face dif-
ferent data limitations. Are there for example limits to theoretical physics?
 PUSHING THE LIMITS OF SCIENCE 153

There are because our descriptions of phenomena are abstract and not testable
when theorising about superstrings and multiple universes. Are there limits to
palaeontology? There are since we can only investigate and reconstruct those
past forms of life on earth for which fossils remain. Are there limits to cognitive
science? There are given that we cannot recreate the origin of the mind or the
evolution of consciousness, language and altruistic behaviour, and we study
the mind from a self-referential point of view using our mind. Are there limits
to cosmology? There are due to insufficient empirical data and a point of refer-
ence for events prior to the big bang. Are there limits to comparative biology?

Downloaded from [Link] by guest on 29 August 2025

There are as we do not have systematic means of studying how other ani-
mals, with different sensory and cognitive abilities, perceive us and the world.
There are limits to what we can predict across science, such as predicting well
complex phenomena like ecosystems, economies and human behaviour.
Establishing major new fields that are as foundational as physics, chemistry
and biology is also improbable—just like we are today not going to come
across major new mountain ranges and rivers on our planet. Groundbreaking
advances that generate a new basis for a major field are likely to decrease over
time. Yet we will still generate new major breakthroughs and open smaller
fields especially related to the range of phenomena on our planet. But much
more is unknown beyond our planet and solar system. Is our universe finite
or infinite? Do other intelligent beings exist beyond our solar system? Would
different physical principles apply on different planets? Would different
mechanisms of evolution exist for their species? Answers to such questions
are currently beyond our methodological reach. For now, nearly all of science
is based on a single dataset, our own planet, and its particular ecosystem,
characteristics and evolutionary processes. This dataset basically accounts for
our entire body of knowledge about the world. Coming across other species on
other planets and galaxies and being able to conduct experiments there would
be a big revelation for science—to be able to test, compare and update our best
theories about the world. Science itself would no longer have just one experi-
mental group but entirely new experimental and control groups (other planets
and galaxies). Some knowledge we have acquired up to now would become
more robust, if the evidence would be the same for physical phenomena and
species on other planets, while other knowledge would need to be revised.
A paradox of science is that each advance brings us into deeper complexi-
ties that are presently beyond our reach. As our knowledge expands, generally
so does our awareness of what we do not yet know. At the border of science,
developing new advances and tackling problems generates new problems and
puzzles. At least in some fields, the pace at which we push the research frontier
154 SCIENCE OF SCIENCE

will likely eventually become slower and slower. An indication of when a

scientific field may partly near its present limits is when it becomes increas-
ingly saturated with abstract theories that cannot be empirically verified. This
appears to be the case in parts of theoretical physics (such as string theory, the-
ories of multiple universes and dark matter) and theoretical economics (such
as the dominance of rational choice theory). String theory is for example a
long-sought attempt of a unified theory of physics. It aims to combine quantum
mechanics with the general theory of relativity, in which particles are modelled
as string-like entities rather than point-like particles. Yet as it is largely a math-

Downloaded from [Link] by guest on 29 August 2025

ematical construct and not verifiable with experimental evidence, it is more
an idea than a testable theory.(277) String theory is a classic example of how we
hit our present limits of science and how science breaks down, in the absence
of empirical evidence. Another indication of when a scientific field may partly
near its present limits is when major discoveries are not made for long periods
of time despite many scientists and much research in the field.
Our attention in science will always remain on those aspects of the world
that our mind and the tools we can devise with our mind enable us to
access. Their boundaries are our current boundaries that we continuously
reshift. Some questions and parts of the world—independent of the amount
of researchers, methods and resources we direct towards them—may remain
beyond our cognitive and methodological scope. These include for example
what came before the origin of our universe and the size of the universe. Ulti-
mately there will also be things we will not know. But we will not know that we
do not know them. There are thus unknowns that we are confident exist but
do not yet have the needed tools to uncover them—such as dark matter. There
are imaginables that we can imagine but cannot yet test empirically—such as
life beyond our planet. And there may also be unknowables that lie beyond
both our methodological reach and imagination.
A methodological barrier in science is that it investigates phenomena that
we can observe and are methodologically tractable and mostly quantifiable.
Consequently, phenomena like the nature of time, knowledge, human con-
sciousness and experience, freewill, politics or love pose considerable chal-
lenges for scientific investigation. In fields like philosophy, epistemology and
ethics, what we study often lies beyond the scope of empirical evidence.
While our tools largely dictate the existing barriers of scientific disciplines,
they are not able to elucidate all aspects at the scientific frontier. The barriers
are also partly influenced by the allocation of policy attention and funding
to particular fields and tools, especially relevant in subfields of astronomy
and physics (Chapter 7). The barriers are also partly delineated by ethical
 PUSHING THE LIMITS OF SCIENCE 155

considerations, including decisions of regulatory agencies on the legality

of human cloning, gene editing, artificial intelligence, nuclear energy, mil-
itary research and technologies impacting climate change. The boundaries
of scientific exploration vary across different fields, with some more limited
than others, and the extent to which we continue expanding those bound-
aries mainly depends on the ways we identify to extend our methodological
scope. Ultimately only those researchers optimistic about the prospect of new
breakthroughs are incentivised to explore and broaden our methodological
horizons and scientific scope.

Downloaded from [Link] by guest on 29 August 2025

The future prospects of science: three pathways

We are confronted with more uncertainty the greater we look into the future,
and we can either choose to not say anything about the future given the
degree of uncertainty or we can discuss possible pathways of the future of
science. Discussing such pathways can foster better planning the future of
science. We opt here for the latter, which moves us into deeper philosophical
terrain. We outline here three pathways of the future of science that we can
take to different degrees in different fields:

• the filling-in-the-details view of future science;

• the empirically limitless but theoretically limited view of future science;
• the methodologically limitless view of future science.

The first possible pathway is the filling-in-the-details view of future science

that ultimately, in possibly a few hundred years, nearly all major methods,
instruments and discoveries would have been made and science would mainly
involve smaller but important details and refining existing findings. The fun-
damental puzzles and grand breakthroughs would have been resolved—as
currently with quantum mechanics, relativity, evolution, DNA, the periodic
table and the big bang—leaving little room for us to make new foundational
discoveries that entirely replace them. Debates on human evolution would for
example reduce to debates about details of particular mechanisms of evolu-
tion. Measurements and results would however continue to be updated as we
improve our methods and instruments.
The second possible pathway is the empirically limitless but theoretically
limited view of future science that science would eventually focus largely on
new empirical applications to changing environmental, medical, technologi-
cal and social challenges of our time. All major theoretical discoveries would
156 SCIENCE OF SCIENCE

have eventually been made—as some already [Link].(253) Theoretical sciences

would eventually decelerate, since establishing foundational theories, in fields
such as physics and chemistry, makes establishing new theories less likely.
Constantly creating grand theories that explain reality well and that constantly
replace themselves would be inherently contradictory and is highly unlikely.
The theory of evolution for example serves as the cornerstone of biology and
is deeply embedded in independent strands of evidence using methods from
molecular biology, palaeontology, primatology, cognitive science and archae-
ology. The periodic table of elements accounts for the elements that make

Downloaded from [Link] by guest on 29 August 2025

up the world and thus the foundation of chemistry. This makes it highly
unlikely that novel fundamental theories will completely replace the theory
of evolution, periodic table of elements or other central theories of science.
Just as we can only discover once our genetic code or that DNA has a dou-
ble helix structure and transmits genetic information, we will not limitlessly
make breakthroughs about the foundation of biology (evolution), chemistry
(the periodic table) or physics (relativity and quantum mechanics). But they
would remain similar to our current well-predictable theories, though become
more detailed. Major new tools and breakthroughs have been and will con-
tinue to be developed among those grand theories and advances—such as
CRISPR gene editing, mapping the human genome and the existence of the
Higgs particle—that were made since 2000 and have earned Nobel Prizes.
Researchers across fields like medicine, biology, psychology and economics
will likely explore new questions as long as our species exists. The enduring
importance of such research is driven by the demand for novel treatments
to combat emerging viruses, diseases and cancers, and changes in environ-
mental and social conditions over time and context. Researchers in fields like
agricultural science, environmental and mechanical engineering, computer
science and technology-related fields will likely open new questions indefi-
nitely, given their pivotal roles in enhancing human welfare and addressing
our changing human, geographic and social challenges. Science and tech-
nology are highly interconnected. The perpetual quest for enhanced health,
technological advancements and improved human well-being ensures con-
tinued growth in applied sciences. In the same way that the inventions of the
internet, space shuttles and atomic energy were not foreseeable a century ago,
many future scientific and technological developments remain equally unpre-
dictable. Although some may believe we are reaching a plateau in scientific
growth given that major theories are not likely to be abandoned,(253) the trajec-
tory of applied sciences and technological progress shows no signs of slowing
down in the future, and we observe continual advances. Embedded within
 PUSHING THE LIMITS OF SCIENCE 157

this unified perspective of scientific and technological advancement lies the

pivotal insight: it is often methodological and technological advances—such
as enhanced telescopes, large-scale spectrometers, innovative machine learn-
ing methods and novel particle detectors—that have driven and continue to
drive new major breakthroughs.
The third possible pathway is the methodologically limitless view of future
science that we will continue to create currently unimaginable methods and
instruments and integrate them in seemingly limitless ways across diverse
domains. Generating new tools will enable us to continually expand exist-

Downloaded from [Link] by guest on 29 August 2025

ing and create new domains of science and discoveries given our endless
imagination. It would seem at first glance contradictory for the world to be
finite, but not our knowledge of the world. Yet we investigate the observable
world by applying our flexible cognition and tools. The nature of knowledge
is cognitive and methodological and thus variable, with scientists applying
creativity and critical assessments that are not inherent in the world out-
side. Using human imagination, we can expand available tools to devise
new tools that are seemingly unbounded by predetermined constraints in
light of our seemingly boundless creativity. Thus even if the world is not
boundless, science to some extent is, in light of our elastic cognition and
tools.
The future of science will likely encompass, to different degrees in differ-
ent fields, all three pathways that each fundamentally relies on the scientific
methods and instruments we are able to develop. Elements of the three path-
ways will thus likely take place in all fields. The future of science depends on
what we can methodologically imagine and create. Endless major theoretical
innovations are unlikely in all three pathways. How we answer the question of
whether we may be getting closer to the boundaries of science and what the
future of science holds relies on the part of science we turn our attention to:
whether we view scientific progress as more closely tied to new major empirical
breakthroughs and the application of science, or to the creation of new major
methods and tools that trigger breakthroughs, or to fundamental theoretical
breakthroughs, or all of them. Theories are not however able to explain or jus-
tify themselves but always rest on empirical evidence and using tools and they
generate questions that only empirical advances can tackle. Novel questions
and tools we create generate in turn further questions and tools and oppor-
tunities for novel knowledge. Science will be around as long as our species
is around, since our societal needs change and new challenges emerge. We
always have novel medical and technological problems to solve. Much of sci-
ence will not ever be finished. Overall, these are different pathways for how we
158 SCIENCE OF SCIENCE

can develop our tower of science in the future—that is, the pathways for the
future of science.
A question of history of science is whether our theories of the world will
appear as incorrect to our descendants in a few hundred or thousand years as
Aristotle’s theories appear to us? Yet the difference is that Aristotle’s theories
of motion, four elements (earth, water, fire and air) and biological theories
could not accurately explain and predict our physical, chemical and biological
reality. But today’s methods and theories enable us to do so with high precision,
so differences in accuracy will likely be much smaller between the theories of

Downloaded from [Link] by guest on 29 August 2025

today and the future. This is a mark of the great scientific progress we have
made thus far.
Vast growth has characterised science over the past century. Yet is it possi-
ble for scientific progress to continue as rapidly in the future? After making
a discovery, such as the electron, particle accelerator and DNA molecule,
they can no longer be discovered again. Yet such major breakthroughs gener-
ate new disciplines that cumulatively grow over time. Numerous novel fields
of science are continually arising, such as computational biology, environ-
mental science, artificial intelligence/machine learning, quantum chemistry,
nanotechnology, genetic engineering and science of science. Such fields hold
promise for novel scientific breakthroughs beyond our current imagination,
by integrating, specialising and adapting to new conditions in society and our
environment. At the forefront of science, progress appears promising. On the
one hand, the prospect appears positive because the rate at which new pivotal
breakthroughs and fields have emerged has not shown signs of decelerating
in recent decades. On the other hand, the systematic development and dis-
semination of new methodological advances has yet to be carried out in a
coordinated and targeted way across science despite their pivotal role in driv-
ing scientific advances. Awareness among the scientific community about this
methods-driven nature of science has been lacking. The growth of science will
be intricately tied to the continual expansion of our methodological toolbox.
In sum, we humans have evolved astonishing methodological abilities that
enable us to do science and understand that we and our mind are the result of
an evolutionary process and live on a planet revolving around a star within
one among millions of galaxies in a vast and expanding universe. We have
not yet fully addressed foundational questions about the nature of life, matter,
the universe and science, but we do know a lot about them. Science pos-
sesses an inherent capacity for self-renewal; science will not culminate in a
completed endeavour. We have achieved the majority of our major break-
throughs across science only since the 20th century, and the future is much
 PUSHING THE LIMITS OF SCIENCE 159

longer and holds many more scientific advancements through powerful new
tools we devise. The horizon of scientific exploration and future knowledge
knows no definitive boundaries, and we have outlined here three pathways of
the future of science. Ultimately, understanding everything about the universe
would imply that we not only have answered all existing questions but also
have established all possible questions and devised the needed methodolo-
gies and instruments to tackle them. The forefront of science is characterised
by posing the right questions and conceiving the right tools—both small and
large. Pursuing novel questions characterizes the common approach at the

Downloaded from [Link] by guest on 29 August 2025

frontier, which is observed in scholarly publications, research initiatives and
funding proposals. This common strategy, when attention is also directed to
methods and instruments, can help us conceive and develop the methods
and instruments needed to address such questions. An alternative strategy has
been at least as successful, if not more: focusing our attention strategically on
upgrading our existing tools and creating new ones that answer our existing
questions and can answer completely new questions not yet raised. The cen-
tral conclusion is that the best way we have to achieve new scientific advances
is through the incredible power of developing new methods and instruments
that have continually reset our present scientific boundaries. Elucidating here
this methodological mechanism propelling scientific exploration puts us in a
better position to achieve new breakthroughs and achieve them quicker. This
central insight has not been given sufficient attention in studies in science of
science or the limits of science to date.(220,247,248,249,250,251,252,253,255,259)
In general, we humans are what set the present scope and limits of science:
the methods and instruments that we have been able to develop thus far, using
our mind and within our broader social context. We ourselves are, through
our methodological toolbox, the origin and boundaries of science—the ori-
gin and boundaries of what we can perceive, measure and understand in the
world. Methodological advances allow us to greatly extend those boundaries
by reducing our human and methodological constraints and broadening our
scope to the world. Predicting the rate of new breakthroughs hinges on the rate
of new methodological innovations.
This methodological mechanism of expanding our scientific frontier has
vast ramifications: scientists need to focus greater attention and time to detect-
ing our present methodological shortcomings to investigating the world and
tackling those shortcomings to spark new advances. We have described here
how we can push the current limits of science more rapidly and in a more struc-
tured way by outlining the steps we need to take to extend our tools of science.
This shift in our focus needs to be combined with education and university
160 SCIENCE OF SCIENCE

systems also placing a greater focus on students better studying the meth-
ods and instruments used in science, their present limits and how to improve
them. We need to train future scientists to better detect problems with our
best tools of science and how to address those problems by experimenting
with new methods, tools and techniques. Targeted public funding schemes
for experimentation in and advances in methods and instruments are also
needed to foster this shift across science. Strategically fostering this methods-
driven mechanism of science could mark the beginning of a methodological
revolution in science that changes the way we understand and do science and

Downloaded from [Link] by guest on 29 August 2025

accelerates the way we advance science.
 Conclusion and Implications

Viewing the field of science of science through an integrated lens can pro-
vide answers to fundamental questions about science: its origins, foundations

Downloaded from [Link] by guest on 29 August 2025

and limits and how to push those limits. To date, these questions have com-
monly been studied by adopting a perspective from an individual discipline.
Scientometricians, historians of science, psychologists of science and other
researchers carrying the title ‘… of science’ have thus not yet been able to pro-
vide a comprehensive and integrated account of science. The central challenge
of the field of science of science has been accounting for and integrating the
different empirical and theoretical knowledge across disciplines into a holis-
tic field and uncovering the general mechanism driving science across fields.
This book has aimed to tackle this challenge and offer a foundation for this
integrated field by combining methods and evidence from across the natu-
ral, behavioural and social sciences (Figures 1.2 and 16.2). Adopting such a
holistic approach has enabled us to file down the often inflated role of a single
factor and assess which factors are most important and how they fit together,
to then be able to develop a deeper and more coherent understanding of sci-
ence (Figure 16.2). Science of science, conceived here as an integrated field,
provides a unified understanding of science and how to improve science by
identifying the abilities and conditions that drive and constrain science. The
different subfields of science of science use different methods and study dif-
ferent aspects of science, and the evidence from the independent strands are
coherent with what is already known across fields (Chapters 2–15).
The central conclusion is that the powerful role of our universal and adap-
tive methodological toolbox has been identified across fields as the main
mechanism driving science that we can directly influence. Our method-
ological toolbox underlies the different factors across disciplinary perspec-
tives. We observed that the central factors that have been proposed as the
most important single factor explanation of science—namely the paradigm
shifts that define scientific theories,(1) the principal evaluation criterion of
science,(14,15,215) the key social influences on scientists(10,11) and so on—are
not able to explain as much and do not have direct influence on the foun-
dations, limits and advancement of science compared to other factors. These

Science of Science. Alexander Krauss, Oxford University Press. © Alexander Krauss (2024).
DOI: 10.1093/9780198937401.003.0022
162 CONCLUSION AND IMPLICATIONS

factors proposed by the most cited researcher studying science within a partic-
ular field—namely Kuhn in history of science, Popper in philosophy of science,
Latour and Woolgar in sociology of science and so on—need to be left in
the background. Our methodological toolbox needs to be brought into the
forefront of how we understand, study and advance science.
So what do the economist of science, scientometrician, philosopher of sci-
ence and other individual researchers gain that they did not have before
the integration of these fields into a holistic science of science? For some
researchers it is a shift towards a joint research focus on methodology and

Downloaded from [Link] by guest on 29 August 2025

better understanding and improving our best methods and instruments that
drive science and discovery. For other researchers it is addressing the frac-
tured approaches and filing down the rough disciplinary explanations about
an often overemphasised role of power by sociologists, the truth of theories by
philosophers, citations by scientometricians and so on to develop more inte-
grated explanations. For all researchers it is a shift to studying science in an
integrated way, coherent with already common knowledge in other subfields,
rather than in isolation, whether they study meta-level questions about sci-
ence or specialised questions. As making major discoveries has required us
to apply new scientific methods and instruments, we cannot adopt an under-
standing of how we make discoveries that is independent of our scientific tools.
But existing accounts of science (scientometric, historical etc.) have not yet
incorporated the essential role of methods and tools into such accounts to
date.
This new-methods-drive-science theory presented here illustrates how our
mind and sensory abilities (to observe, experiment and process informa-
tion) make doing science possible but also shape what and how we observe
and reason, as they have evolved within our environmental niche. Our sci-
entific methods and instruments (from new statistical methods to electron
microscopes) enable us to study a much broader range of phenomena, but
they also have constraints to how we measure and perceive phenomena and
express our theories. Institutions, funding and societal challenges help influ-
ence what knowledge and research methods we produce, distribute and use.
Scientific norms and methodological assumptions shape the way we evaluate
our evidence, among other influences (Figure 16.2). Taken together, this the-
ory explains how sophisticated methods and instruments we develop using
our mind’s methodological abilities set the scope and present limits of what
we can know and what is possible in science—and economic, social and
historical influences help shape what we study within that scope and those
limits.
 CONCLUSION AND IMPLICATIONS 163

The origins, foundations and limits of science can be better explained,

understood and advanced in light of the new-methods-drive-science theory.
It illustrates that our scientific methods and instruments are the only factor
that underpins all 14 subfields. With the foundation, boundaries and advance-
ment of science largely shaped by our methods, a central focus on methods
across all subfields is also crucial to the integration of the science of science.
This methods-driven understanding of science can help spark a methodolog-
ical revolution in science. We have thus seen how the tower of science we have
reconstructed is made possible by our evolved methodological abilities (that

Downloaded from [Link] by guest on 29 August 2025

account for its foundation) and our scientific methods and instruments (that
account for the different floors of its structure). Together, they largely deter-
mine how far we can perceive, measure and understand the world. Overall,
four main implications arise from this meta-approach to studying science.

1 We need to measure the success of the field of science of science by estab-

lishing a society, journals, conferences and interdisciplinary institutes that
adopt a truly integrated approach to studying science.

These would help tackle constraints confronting the interdisciplinary

nature of the field. They include methodological obstacles that involve
researchers acquiring skills in multiple methods and collaborating in teams
adopting multiple methods. They include institutional constraints that involve
collaborations within cross-disciplinary institutes and greater access to estab-
lished sources of research funding. While the status quo of isolated fields study-
ing science has not led to integration, establishing a society, journals, confer-
ences and institutes (or centres) in science of science would institutionalise
collaboration networks and a forum of integrated debate. Establishing which
disciplinary fields and factors are most important in helping advance science
is key to the field (Table 16.1). It can enable scientists and scientific institu-
tions to better direct their attention to particular features of science. Scientific
institutions like the European Commission, National Science Foundation,
governments and other funding agencies need to begin incentivising inte-
grated meta-scientific research and new methodological research that is as
important as other established funding areas in advancing science and dis-
coveries. Leading scientific journals need to publish not only new scientific
breakthroughs but also new methodological breakthroughs, as they are what
drive new scientific breakthroughs—at times independent of pre-existing
questions or hypotheses. We need to also revise the mainstream view of
science and science policy that conceives scientific research as question driven.
164 CONCLUSION AND IMPLICATIONS

Shifting much more attention from question- and hypothesis-driven science

to methods-driven science that is exploratory can help foster new advances.

2 We need to better train researchers studying science and better conduct

research in a more interdisciplinary way.

Basically all researchers studying science currently pursue disciplinary spe-

cialisation that can serve an important function, namely the division of cog-
nitive labour by allowing researchers to concentrate deeply on a given topic.

Downloaded from [Link] by guest on 29 August 2025

In many research domains this has been a very successful strategy—think of
particle physics, molecular spectroscopy or brain surgery without extensive
specialisation. But using this strategy to study a phenomenon as multidimen-
sional and complex as science is the reason why we, to date, have not yet
developed a coherent general theory of science and how science advances—
the central question in science of science. There is no way around it: if we
want to better understand science, we cannot resort to just saying that time
and resources constrain us in working more interdisciplinarily. This has come
at the cost of at times contradicting evidence across subfields on the same
topics—that is, non-replicable evidence across fields.
We need to make a better effort and reform the way that science has been
studied to date. The vision of science of science outlined here would ide-
ally consist of perhaps roughly three-fourths of researchers studying science
by continuing to pursue disciplinary specialisation but all researchers would
begin to be trained in and spend a share of their time to be broadly informed
about existing research on the same topic across other subfields of science of
science. This is the only way to ensure a coherent understanding of science
that is consistent, not conflicting across neighbouring subfields. And impor-
tantly, we would otherwise not even know that we have gained a coherent
understanding of science, unless we compare the findings and evidence to
assess if they are coherent across the subfields of science of science that use
different methodologies to address the same question. The remaining perhaps
roughly one-fourth of researchers would ideally then adopt a meta-perspective
that truly integrates fields. This however requires that some science of sci-
ence researchers would be formally educated and trained, through university
departments and institutes, just as biologists or statisticians are trained—
with degrees in science of science. We cannot assume that some researchers
who happen to be inclined to adopt an integrated big-picture approach to
science will come across such questions and find the time to dedicate them-
selves in a full-time capacity to addressing them. Just as there are general
 CONCLUSION AND IMPLICATIONS 165

meta-researchers in environmental sciences, we also need to begin training

such general meta-researchers in science of science, if we want greater depth
and breadth in understanding science.

3 By better understanding the set of constraints we face in science, we can

reduce them to advance science.

We can mitigate constraints and biases facing our cognitive and sensory
abilities by making efforts to develop methods that expand these abilities. We

Downloaded from [Link] by guest on 29 August 2025

can decrease constraints and biases facing our methods by improving them,
applying multiple methods and developing new methods. We can reduce
some social and contextual influences we face by conducting and comparing
studies in different contexts and time periods. We can address psychological
biases we face by automating some processes using computers and by differ-
ent scientists replicating studies and thus confirming or refuting the results
of others. We can mitigate our assumptions by expanding our methods, and
so on. Essential to improving science is an awareness of the importance of
developing new methods and instruments designed to reduce our human and
methodological constraints to studying the world (Figure 16.2). With such an
awareness, we can strategically target and address our constraints to advancing
science. To push the limits of science, we need—as the central driving force—
to extend existing methods and instruments and develop entirely new ones
(Chapter 20).

4 We need to revisit our best methods for assessing and measuring sci-
ence and discoveries and adopt a broader set of empirical methods from
neighbouring fields to better address foundational questions about science.

To date, researchers studying science (including scientometricians, network

scientists and psychologists of science) commonly do so descriptively study-
ing one factor. Most studies in subfields of science of science, and nearly all
studies in scientometrics, are descriptive studies that use observational data.
Such observational and big data studies can uncover important and at times
strong relationships among factors. But a critical step to move the field forward
will be conducting studies that can assess causal effects of central factors driv-
ing science and their interrelationships, as in most human sciences.(4,5,35,39)
Improving our understanding of science requires turning to methods that
enable us to measure causal relationships better—such as those widely used
in economics, public health and the medical sciences (outlined in Chapter 7).
166 CONCLUSION AND IMPLICATIONS

We need to begin conducting for example longitudinal studies that follow

scientists over their lifetime to assess long-term changes over their careers:
including the effects of their discoveries, new collaborations and relocating to
top universities.
To better understand the range of factors that shape science and its foun-
dations and boundaries, we will also have to begin applying a wider range
of methods from across subfields: longitudinal observational studies, instru-
mental variable methods, natural experiments, randomised controlled experi-
ments, institutional analyses and so forth. Science of science is about applying

Downloaded from [Link] by guest on 29 August 2025

science to understand science, and to do so better we have to move beyond
descriptive big-data methods—as just one method among other methods. We
cannot narrowly view science of science as the ‘field that relies on big data
to unveil the reproducible patterns that govern individual scientific careers
and the workings of science’ by studying primarily citations,(4)cf.(5,35) because
big data can for example rarely provide causal knowledge that will require
us to begin applying such experimental methods of medical statisticians and
economists. For a number of big questions in science of science we will have to
rely on observational studies and will not be able to apply experimental studies.
But we will have to begin applying observational studies in a more integrated
and interdisciplinary way.
Causal understanding is more difficult to attain in science of science than
in other fields, given the nature of science as a complex phenomenon with
multiple influencing factors at different levels. This makes it more difficult
to compare how important different factors are against each other (such as
newly developed experimental methods, cognitive biases, public funding, and
team and community size). This is because they are captured at different lev-
els using different types of data (such as experimental, individual, institutional
and country-level data). Causal identification can thus be partly constrained
when studying particular large-scale phenomena like scientific communities,
the evolution of scientific fields or the efficiency of research institutions, or
predicting the future of science. It can be constrained in addressing questions
beyond well-defined, quantifiable interventions at the level of the individual
scientist. The more complex and bigger the phenomenon we study, the greater
the uncertainty we often have—think of networks of scientists and institutions,
and novel breakthroughs in science. For such complex phenomena, we can-
not generally apply randomisation techniques that are used to identify causal
relationships, as we do not usually have a comparable counterfactual. Scien-
tists cannot be easily randomised into treatment and control groups to assess
for example the effect of making a major discovery or receiving a Nobel Prize
 CONCLUSION AND IMPLICATIONS 167

on one’s career or on science. For the sample would have to be near univer-
sal of the entire scientific community and thus include millions of scientists.
Measuring science better requires that we better conduct studies assessing
causation. There is no way around it: for science of science to reach its poten-
tial it will have to move beyond its strong focus on observational studies and
conduct studies that assess the causal effects of interventions driving science.
Measuring science better will also require us to go beyond the common use of
citations as the key impact metric of science(4,5,39) that needs to be viewed as
just one among other measures of scientific impact and advancement—such as

Downloaded from [Link] by guest on 29 August 2025

scientific discoveries, methodological advances and social and policy impact.
Qualitative studies (that integrate insights from across disparate fields) are
also important. However, constraints of such studies include quantitatively
measuring factors studied across different fields (norms, institutions, practices,
assumptions etc.); balancing more breadth across more fields with inevitably
less depth on any single field; and synthesising across fields, as each field has its
own methodological preferences and complexities. As highlighted up front, a
companion book The Motor of Scientific Discovery assesses over 750 major sci-
entific discoveries including all nobel-prize-winning discoveries and provides
statistical evidence of how we drive science and discovery through new meth-
ods and instruments — and complements this book with such quantitative
data and grounds this theoretical framework.(64)

More generally, improving science, the structure of science and fostering

interdisciplinary science are important goals for advancing knowledge and
addressing complex problems, whether in science of science or other fields
across science. What are general steps we can take to improve and advance
science, and also do so in interdisciplinary ways? One, we need to continually
foster and adopt emerging technological advancements and digital tools that
can enhance the way we conduct scientific research, data analysis and dissem-
inate knowledge. These include high-performance computing, data analytics,
imaging techniques and other cutting-edge technologies that can bridge dis-
ciplinary boundaries. Two, we need to better incentivise interdisciplinary
collaboration and provide resources for researchers from diverse fields to bet-
ter work together to address complex challenges. Fostering interdisciplinary
research centres can play a key role in bringing together scientists from across
fields and provide a platform for interdisciplinary collaboration, knowledge,
resource sharing and problem solving. We need to support cross-disciplinary
funding programmes, training opportunities, conferences and workshops,
and reward interdisciplinary advances. We require also developing better
168 CONCLUSION AND IMPLICATIONS

educational programmes that expose students and researchers to interdisci-

plinary approaches early on. Yet interdisciplinary research can generally be
more complex and require more time, effort and resources.(278,235,72) Three,
we need to ensure research funding supports a broad portfolio of science
projects, not just mainstream question-driven projects but also exploratory,
higher-risk, interdisciplinary or longer-term projects that are less commonly
funded. Four, we need to better support early-career researchers through
career development opportunities and university mentorship programmes
that connect young researchers with experienced researchers to provide guid-

Downloaded from [Link] by guest on 29 August 2025

ance. Five, we require fostering efforts to improve science communication
and its impact by bridging the gap between scientific studies and their use
among policymakers and the general public. Promoting simple language in
publications would increase public understanding and engagement.
These different measures require collective efforts from researchers, institu-
tions, funders and policymakers to continually make our scientific ecosystem
more robust and effective. Implementing these measures can enhance the
scientific system and create an environment conducive for interdisciplinary
science, whether in science of science or other scientific fields.
Ultimately, the integrated field of science of science holds a vast potential for
addressing fundamental questions about the origins, foundations and limits
of science. It holds a vast potential for better understanding how we drive new
scientific advances and methods that open new and unknown frontiers.
 Appendix

A brief clarification is provided here for how the share and number of publications across
the subfields of science of science have been calculated (Figure 1.1). Publications, which
include articles, books and other scientific formats, were identified by searching a given
term in the title, abstract or keywords (Scopus’s default search function). Data were col-
lected for the fields that contribute to understanding the origins, foundations or limits of
science; data reflect the outcome of searches in early 2024 using the terms ‘biology of sci-

Downloaded from [Link] by guest on 29 August 2025

ence’ or ‘philosophy of biology’ (656 publications), ‘cognitive science of science’ or ‘philoso-
phy of cognition/mind’ (3084), ‘psychology of science’ or ‘philosophy of psychology’ (456),
‘linguistics of science’ or ‘language of science’ (638), ‘sociology of science’ or ‘philosophy
of sociology’ (1604), ‘economics of science’ or ‘philosophy of economics’ (319), ‘(scien-
tific) methodology of science’ or ‘philosophy of (scientific) methodology’ (300), ‘computer
science of science’ or ‘philosophy of computer science’ (50), ‘statistics/mathematics of sci-
ence’ or ‘philosophy of statistics/mathematics’ (1225), ‘scientometrics’ or ‘network science
of science’ (5418), ‘philosophy of science’ or ‘metaphysics/ontology/epistemology of sci-
ence’ (10,524), ‘anthropology of science’ or ‘philosophy of anthropology’ (137), ‘history of
science’ (8268), ‘archaeology of science’ or ‘philosophy of archaeology’ (23), and ‘science
of science,’ ‘metascience’ or ‘metaresearch’ (956).
Here we also briefly outline common methods used in each subfield (Figure 1.2). Biol-
ogy studies science using methodological approaches such as comparative analysis of the
abilities of human and non-human animals and an evolutionary theoretical framework.
Cognitive science studies our cognitive and sensory abilities for perceiving and reasoning
about reality. Psychology conducts experimental studies of reasoning, biases and moti-
vation and uses direct observation in labs and surveys of scientists. Linguistics analyses
the communication of science and language as a thinking tool in science and as a tool to
accumulate knowledge. Sociology carries out quantitative and qualitative surveys of the
scientific community, practice and norms and carries out institutional analysis. Economics
conducts quantitative analysis of productivity, the reward system, research funding, science
policy and institutional analysis. Methodology covers methodological analyses and stud-
ies the drivers (such as methods and instruments) of new discoveries. Computer science
analyses computational methods (used to process and analyse data) and their foundations,
including artificial intelligence. Statistics/mathematics analyses the foundations of statisti-
cal and mathematical methods and their constraints and biases. Scientometrics including
network science conducts citation and publication analysis and uses large-scale data. Phi-
losophy covers theoretical and normative analysis of science and its methods and analysis
of assumptions, concepts, causation and laws. History carries out case studies of scientists
and discoveries (autobiographies, notebooks) and historical analysis of scientific progress.
Anthropology involves case studies and past and present cross-cultural studies (e.g. on rea-
soning). Archaeology analyses fossil records and material artefacts and tools (e.g. related to
our ancestors’ cognitive abilities).
 References

1. Kuhn, Thomas. The Structure of Scientific Revolutions. Chicago: University of

Chicago Press (1st edition 1962; 3rd edition 1996).
2. Daston, Lorraine, Peter Galison. Objectivity. New York: Zone Books (2007).
3. Daston, Lorraine, Michael Stolleis. Natural Law and Laws of Nature in Early Modern

Downloaded from [Link] by guest on 29 August 2025

Europe: Jurisprudence, Theology, Moral and Natural Philosophy. Milton: Routledge
(2017).
4. Wang, D., A. Barabási. The Science of Science. Cambridge: Cambridge University
Press (2021).
5. Fortunato, S., C.T. Bergstrom, K. Börner, et al. Science of science. Science, 359 (2018).
6. Zeng, An, Zhesi Shen, Jianlin Zhou, et al. The science of science: from the perspective
of complex systems. Physics Reports, 714–715, 1–73 (2017).
7. Sinatra, R., D. Wang, P. Deville, C. Song, A. Barabási. Quantifying the evolution of
individual scientific impact. Science, 354, aaf5239 (2016).
8. Azoulay, P., J. Graff-Zivin, B. Uzzi, et al. Toward a more scientific science. Science,
361 (6408), 1194–1197 (2018).
9. Park, M., E. Leahey, R.J. Funk. Papers and patents are becoming less disruptive over
time. Nature, 613, 138–144 (2023).
10. Latour, Bruno, Steve Woolgar. Laboratory Life: The Construction of Scientific Facts.
Princeton: Princeton University Press (1979).
11. Bourdieu, Pierre. Science of Science and Reflexivity. Chicago: University of Chicago
Press (2004).
12. Pickering, Andrew, Keith Guzik. The Mangle in Practice: Science, Society and
Becoming. Durham, NC: Duke University Press (2009).
13. Collins, H.M. Scientific knowledge, sociology of. In: Wright, J.D. (ed.), International
Encyclopedia of the Social and Behavioral Sciences. Amsterdam: Elsevier, 2nd ed.,
Vol. 21, pp. 308–312 (2015).
14. Popper, Karl. Normal science and its dangers. In: Imre Lakatos, Alan Musgrave (eds),
Criticism and the Growth of Knowledge. Cambridge: Cambridge University Press
(1970).
15. Popper, Karl. Zwei bedeutungen von falsifizierbarkeit. In: Helmut Seiffert, Gerard
Radnitzky (eds), Handlexikon zur Wissenschaftstheorie. Munich: Ehrenwirth, 2nd
ed., pp. 82–85 (1994).
16. Lakatos, Imre. Falsification and the methodology of research program. In: Imre
Lakatos, Alan Musgrave (eds), Criticism and the Growth of Knowledge. Cambridge:
Cambridge University Press, pp. 91–197 (1970).
17. Feyerabend, Paul. Against Method. London: New Left Books (1975).
18. Ladyman, J. Understanding Philosophy of Science. London: Routledge (2002).
 REFERENCES 171

19. Chalmers, Alan. What Is This Thing Called Science? Indianapolis: Hackett Publish-
ing Company, 4th ed. (2013).
20. Cosmides, L., J. Tooby. Cognitive adaptations for social exchange. In: J. Barkow,
L. Cosmides, J. Tooby (eds.), The Adapted Mind. Oxford: Oxford University Press
(1992).
21. Tomasello, Michael. Becoming Human: A Theory of Ontogeny. Cambridge, MA:
Harvard University Press (2019).
22. Proctor, R., E. Capaldi. Psychology of Science: Implicit and Explicit Processes.
Oxford: Oxford University Press (2012).
23. Feist, Gregory, Michael Gorman. Handbook of the Psychology of Science. New York:
Springer Press (2013).

Downloaded from [Link] by guest on 29 August 2025

24. Heyes, Cecilia. Cognitive Gadgets: The Cultural Evolution of Thinking. Cambridge,
MA: Harvard University Press (2018).
25. Stephan, Paula. How Economics Shapes Science. Cambridge, MA: Harvard Univer-
sity Press (2015).
26. Stephan, Paula. The economics of science. Journal of Economic Literature, 34 (3), pp.
1199–1235 (1996).
27. Lane, Julia. Assessing the impact of science funding. Science, 5932 (324), 1273–1275
(2009).
28. Partha, Dasgupta, Paul David. Toward a new economics of science. Research Policy,
23 (5), 487–521 (1994).
29. Greenland, S., S.J. Senn, K.J. Rothman, et al. Statistical tests, P values, confidence inter-
vals, and power: a guide to misinterpretations. European Journal of Epidemiology, 31,
337–350 (2016).
30. Ioannidis, J. Why most published research findings are false. PLoS Medicine, 2 (8)
(2005).
31. Hansson, Sven Ove. Science and pseudo-science. Stanford Encyclopedia of Philos-
ophy Archive, [Link]
(2017).
32. Kuhn, Thomas, edited by Bojana Mladenovic. The Last Writings of Thomas S. Kuhn:
Incommensurability in Science. Chicago: University of Chicago Press (2022).
33. Popper, Karl. Conjectures and Refutations: The Growth of Scientific Knowledge.
Melbourne: Hassell Street Press (2021).
34. Popper, Karl. The Logic of Scientific Discovery. London: Routledge, 2nd ed.
(1959/2005).
35. Clauset, Aaron, Daniel Larremore, Roberta Sinatra. Data-driven predictions in the
science of science. Science, 355, 477–480 (2017).
36. Hu, J. Studying the science of science. Science, March 28 (2016).
37. Sugimoto, Cassidy. Scientific success by numbers. Nature, 593 (2021).
38. Xu, Fengli, Lingfei Wu, James Evans. Flat teams drive scientific innovation. Proceed-
ings of the National Academy of Sciences, 119 (23) e2200927119 (2022).
39. Editorial. Broader scope is key to the future of ‘science of science’. Nature Human
Behaviour, 6, 899–900 (2022).
172 REFERENCES

40. Ioannidis, J., D. Fanelli, D. Dunne, S. Goodman. Meta-research: evaluation and

improvement of research methods and practices. PLoS Biology, 13 (10), e1002264
(2015).
41. Moher, D., S. Hopewell, K. Schulz, et al. CONSORT 2010 Explanation and elabo-
ration: updated guidelines for reporting parallel group randomised trials. BMJ, 340,
c869 (2010).
42. Krauss, Alexander. Why all randomised controlled trials produce biased results.
Annals of Medicine, 50 (4), 312–322 (2018).
43. Audretsch, David, Barry Bozeman, Kathryn Combs, et al. The economics of science
and technology. Journal of Technology Transfer, 27 (2), 155–203 (2002).
44. Gil, Y., M. Greaves, J. Hendler, H. Hirsh. Artificial intelligence. Amplify scientific

Downloaded from [Link] by guest on 29 August 2025

discovery with artificial intelligence. Science, 346 (6206), 171–172 (2014).
45. Titano, J.J., M. Badgeley, J. Schefflein, et al. Automated deep-neural-network surveil-
lance of cranial images for acute neurologic events. Nature Medicine, 24, 1337–1341
(2018).
46. Alkhateeb, Ahmed, Aeon. Can scientific discovery be automated? The Atlantic, April
25 (2017).
47. Nersessian, Nancy. The cognitive basis of model-based reasoning in science. In:
Peter Carruthers, Stephen Stich, Michael Siegal (eds), The Cognitive Basis of Science.
Cambridge: Cambridge University Press (2002).
48. Carruthers, Peter, Stephen Stich, Michael Siegal. Introduction: what makes science
possible? In: Peter Carruthers, Stephen Stich, Michael Siegal (eds), The Cognitive
Basis of Science. Cambridge: Cambridge University Press (2002).
49. Giere, Ronald. Explaining Science: A Cognitive Approach. Chicago: University of
Chicago Press (1988).
50. Mercier, Hugo, Dan Sperber. The Enigma of Reason. Cambridge, MA: Harvard
University Press (2017).
51. Darwin, Charles. On the Origin of Species by Means of Natural Selection, or the
Preservation of Favoured Races in the Struggle for Life. London: John Murray (1859).
52. Laland, Kevin. Darwin’s Unfinished Symphony: How Culture Made the Human
Mind. Princeton: Princeton University Press (2017).
53. Hull, D. Science and Selection: Essays on Biological Evolution and the Philosophy of
Science. Cambridge: Cambridge University Press (2001).
54. Pigliucci, M., J. Kaplan. Making Sense of Evolution: The Conceptual Foundations of
Evolutionary Theory. Chicago: University of Chicago Press (2006).
55. Lévi-Strauss, Claude. The Savage Mind. Chicago: University of Chicago Press (1966).
56. Blurton-Jones, N., M. Konner. !Kung knowledge of animal behaviour. In: R. Lee,
I. DeVore (eds), Kalahari Hunter–Gatherers. Cambridge: Cambridge University
Press (1976).
57. Nisbett, R., K. Peng, I. Choi, A. Norenzayan. Culture and systems of thought: holistic
vs. analytic cognition. Psychological Review, 108 (2), 291–310 (2001).
58. Edwards, Jeanette, Penny Harvey, Peter Wade. Anthropology and Science: Episte-
mologies in Practice. Oxford: Berg Publishers (2007).
59. Hess, David. Social movements and the anthropology of science and technology.
American Anthropologist, 109 (3), 463–472 (2007).
 REFERENCES 173

60. Mithen, Steven. Human evolution and the cognitive basis of science. In: Peter
Carruthers, Stephen Stich, Michael Siegal (eds), The Cognitive Basis of Science.
Cambridge: Cambridge University Press (2002).
61. Toth, N., K. Schick, S. Savage-Rumbaugh, R. Sevcik, D. Rumbaugh. Pan the tool
maker: investigations into the stone tool-making and tool-using capabilities of a
bonobo (Pan paniscus). Journal of Archaeological Science, 20 (1993).
62. D’Errico, F. A new model and its implications for the origin of writing:
the La Marche antler revisited. Cambridge Archaeological Journal, 5, 81–91
(1995).
63. S. Schneegans, T. Straza, J. Lewis (eds). UNESCO Science Report: The Race Against
Time for Smarter Development. Statistical Annex. Paris: UNESCO (2021).

Downloaded from [Link] by guest on 29 August 2025

64. Krauss, Alexander. The Motor of Scientific Discovery. Forthcoming (2025).
65. Oxford English Dictionary. Science. [Link]
tab=factsheet#23959055 (2022).
66. Scopus. Scopus database. Amsterdam: Elsevier, [Link] (2024).
67. Znaniecki, Florian. The subject matter and tasks of the science of knowledge. In:
Bohdan Walentynowicz (ed.), Polish Contributions to the Science of Science (trans-
lation, 1982). Dordrecht: Reidel Publishing Company (1923).
68. Skalska-Zlat, M. Nalimov and the Polish way towards science of science. Scientomet-
rics, 52, 211–223 (2001).
69. Galton, F. English Men of Science: Their Nature and Nurture. London: MacMillan
(1874).
70. de Solla Price, Derek. Little Science, Big Science. New York: Columbia University
Press (1963).
71. Zuckerman, H. Scientific Elite: Nobel Laureates in the United States. New York: Free
Press (1977).
72. Boersema, Jan. Lucas Reijnders. Principles of Environmental Sciences. Dordrecht:
Springer (2009).
73. Thorndike, E. Animal Intelligence. Darien, CT: Hafner Publishing Company (1970).
74. de Waal, Frans. Are We Smart Enough to Know How Smart Animals Are? New York:
W.W. Norton & Company (2016).
75. McGrew, W. Chimpanzee Material Culture. Cambridge: Cambridge University Press
(1992).
76. Boesch, C., H. Boesch. Possible causes of sex differences in the use of natural hammers
by wild chimpanzees. Journal of Human Evolution, 13, 415 (1984).
77. Morell, V. Watch this genius bird plan for its ‘handyman’ job. Science, February 10
(2023).
78. Herrmann, Esther, Josep Call, Maráa Victoria Hernàndez-Lloreda, Brian Hare,
Michael Tomasello. Humans have evolved specialized skills of social cognition: the
cultural intelligence hypothesis. Science, 317, 1360 (2007).
79. Garcia, J., B.K. McGowan, F.R. Ervin, R.A. Koelling. Cues: their relative effectiveness
as a function of the reinforcer. Science, 160 (3829), 794–795 (1968).
80. Miller, R., S. Owens, B. Rørslett. Plants and colour: flowers and pollination. Optics
and Laser Technology, 43, 282–294 (2011).
174 REFERENCES

81. Sullivan, P. Objects limit human comprehension. Biology & Philosophy, 24, 65–79
(2009).
82. Krauss, Alexander. Homo methodologicus and the origin of science and civilisation.
Heliyon, 9, 10 (2023).
83. National Science Foundation (NSF). Federal Funds for R&D series. American Asso-
ciation for the Advancement of Science (AAAS), [Link]
d-budget-and-policy/research-science-and-engineering-discipline (2019).
84. Harmand, Sonia, Jason Lewis, Craig Feibel, et al. 3.3-million-year-old stone tools
from Lomekwi 3, West Turkana, Kenya. Nature, 521, 310–315 (2015).
85. Susman, R. Who made the Oldowan tools? Fossil evidence for tool behaviour in Plio-
Pleistocene hominids. Journal of Anthropological Research, 47 (2), (1991).

Downloaded from [Link] by guest on 29 August 2025

86. Gowlett, J. Mental abilities of early man: a look at some hard evidence. In: R. Foley
(ed.), Hominid Evolution and Community Ecology. Cambridge, MA: Academic Press
(1984).
87. Sterelny, Kim. The Evolved Apprentice: How Evolution Made Humans Unique.
Cambridge, MA: MIT Press (2012).
88. Henrich, Joseph. The Secret of Our Success: How Culture Is Driving Human Evo-
lution, Domesticating Our Species, and Making Us Smarter. Princeton: Princeton
University Press (2016).
89. Carruthers, Peter. The roots of scientific reasoning: infancy, modularity and the art
of tracking. In: Peter Carruthers, Stephen Stich, Michael Siegal (eds), The Cognitive
Basis of Science. Cambridge: Cambridge University Press (2002).
90. Tomasello, Michael. The Cultural Origins of Human Cognition. Cambridge, MA:
Harvard University Press (1999).
91. Currie, Adrian. Rock, Bone, and Ruin: An Optimist’s Guide to the Historical Sciences.
Cambridge, MA: MIT Press (2018).
92. Diamond, J. Guns, Germs, and Steel: The Fates of Human Societies. New York: W.W.
Norton & Company (1997).
93. Banning, E. The Neolithic period: triumphs of architecture, agriculture and art. Near
Eastern Archaeology, 61 (4), 188–237 (1998).
94. Williams, L. Pearce. History of science. Encyclopaedia Britannica, [Link]
[Link]/science/history-of-science (2017).
95. Freeden, Willi, Clemens Heine, Zuhair Nashed. An Invitation to Geomathematics.
Cham: Birkhäuser, Springer Nature (2019).
96. Wilkins, Robert. Neurosurgical Classics. Park Ridge, IL: American Association of
Neurological Surgeons, 2nd ed. (1992).
97. Chalmers, I., E. Dukan, S. Podolsky, G. Davey Smith. The advent of fair treatment
allocation schedules in clinical trials during the 19th and early 20th centuries. Journal
of the Royal Society of Medicine, 105, 221–227 (2012).
98. Peirce, Charles S. Illustrations of the logic of science. Popular Science Monthly,
November (1877).
99. Atran, Scott, Douglas Medin. Introduction. In: The Native Mind and the Cultural
Construction of Nature. Cambridge, MA: MIT Press (2008).
100. Pinker, S. The cognitive niche: coevolution of intelligence, sociality, and language.
Proceedings of the National Academy of Sciences, 107 (2010).
 REFERENCES 175

101. Shettleworth, S. Cognition, Evolution and Behaviour. Oxford: Oxford University

Press, 2nd ed. (2010).
102. Gigerenzer, Gerd, Henry Brighton. Homo heuristicus: why biased minds make better
inferences. Topics in Cognitive Science, 1, 107–143 (2009).
103. Tooby, J., L. Cosmides. The theoretical foundations of evolutionary psychology. In:
D.M. Buss (ed.), The Handbook of Evolutionary Psychology, 2nd ed., Vol. 1. New
Jersey: Wiley (2016).
104. O’Brien, M., K. Laland. Genes, culture, and agriculture: an example of human niche
construction. Current Anthropology, 53 (4), 434–470 (2012).
105. Pinker, S. How the Mind Works. New York: W.W. Norton & Company (2009).
106. Hutmacher, Fabian. Why is there so much more research on vision than on any other

Downloaded from [Link] by guest on 29 August 2025

sensory modality? Frontiers in Psychology, 10, October 4 (2019).
107. Carnap, Rudolf. Scheinprobleme in der Philosophie (translation, Pseudoproblems in
Philosophy). Berlin: Bernary, pp. 301–343 (1928).
108. Bohr, Niels. The quantum postulate and the recent development of atomic theory.
Nature, 121, 580–590 (1928).
109. Kahneman, Daniel. Thinking Fast and Slow. New York: Farrar, Straus and Giroux
(2011).
110. Tversky, A., D. Kahneman. Judgment under uncertainty: heuristics and biases. Sci-
ence, 185, 1124–1131 (1974).
111. Zvereva, E.L., M.V. Kozlov. Biases in ecological research: attitudes of scientists and
ways of control. Nature Scientific Reports, 11, 226 (2021).
112. Kukucka, Jeff, Saul Kassin, Patricia Zapf, Itiel Dror. Cognitive bias and blindness: a
global survey of forensic science examiners. Journal of Applied Research in Memory
and Cognition, 6 (4) (2017).
113. O’Sullivan, E.D., S.J. Schofield. A cognitive forcing tool to mitigate cognitive bias—a
randomised control trial. BMC Medical Education, 19, 12 (2019).
114. Bishop, Dorothy. The psychology of experimental psychologists: overcoming cogni-
tive constraints to improve research. Quarterly Journal of Experimental Psychology,
73 (1) (2019).
115. Bishop, Dorothy. How scientists can stop fooling themselves over statistics. Nature,
584 (9) (2020).
116. Kerr, N. HARKing: Hypothesizing After the Results are Known. Personality and
Social Psychology Review, 2 (3), 196–217 (1998).
117. Rosenthal, R., R. Rosnow. Essentials of Behavioral Research: Methods and Data
Analysis. New York: McGraw-Hill, 2nd ed. (1991).
118. Feist, G. Psychology of science as a new subdiscipline in psychology. Current Direc-
tions in Psychological Science, 20 (5), 330–334 (2011).
119. Krauss, Alexander. Science’s greatest discoverers: a shift towards greater interdis-
ciplinarity, top universities and older age. Nature, Humanities and Social Sciences
Communications, 11, 272 (2024).
120. Merton, R. Behavior patterns of scientists. Leonardo, 3 (2), 213–220 (1970).
121. Henry, John. The Scientific Revolution and the Origins of Modern Science.
Houndsmill: Palgrave Macmillan, 3rd ed. (2008).
122. Watson, James. The Double Helix. New York: New American Library (1969).
176 REFERENCES

123. Strauss, Mark. Nobel laureates who were not always noble. National Geographic,
October 6 (2015).
124. Planck, Max. Scientific Autobiography and Other Papers (translation, Frank Gaynor).
New York: Philosophical Library (1949).
125. Usher, I., P. Hellyer, K.S. Lee, et al. “It’s not rocket science” and “It’s not brain
surgery”—“It’s a walk in the park”: prospective comparative study. BMJ, 375 e067883
(2021).
126. Ericsson, K., R. Hoffman, A. Kozbelt, et al. The Cambridge Handbook of Expertise
and Expert Performance. Cambridge: Cambridge University Press (2006).
127. Gascoigne, Robert. The historical demography of the scientific community,
1450–1900. Social Studies of Science, 22 (3), 545–573 (1992).

Downloaded from [Link] by guest on 29 August 2025

128. Tweney, R., M. Doherty, C. Mynatt. On Scientific Thinking. New York: Columbia
University Press, pp. 325–326 (1981).
129. Harwood, J. Styles of Scientific Thought: The German Genetics Community
1900–1933. Chicago: University of Chicago Press (1993).
130. Merton, Robert. Science and technology in a democratic order. Journal of Legal and
Political Sociology, 1, 115–126 (1942).
131. Merton, R. The Matthew effect in science. Science, 159, 56–63 (1968).
132. Sokal, A. Transgressing the boundaries: towards a transformative hermeneutics of
quantum gravity. Social Text, 46/47, 217–252 (1996).
133. Gross, P., N. Levitt. Higher Superstition: The Academic Left and Its Quarrels with
Science. Baltimore: Johns Hopkins University Press (1994).
134. Encyclopaedia Britannica. Jocelyn Bell Burnell. Encyclopaedia Britannica, https://
[Link]/biography/Jocelyn-Bell-Burnell (2023).
135. Reichenbach, H. Experience and Prediction: An Analysis of the Foundations and the
Structure of Knowledge. Whitefish, MT: Literary Licensing, LLC (1949).
136. Chen, C. On the shoulders of giants. In: Mapping Scientific Frontiers: The Quest for
Knowledge Visualization. London: Springer (2003).
137. Yin, Y., Y. Dong, K. Wang, et al. Public use and public funding of science. Nature
Human Behaviour, 6, 1344–1350 (2022).
138. The Nobel Prize. The Nobel Prize. [Link] (2024).
139. Krauss, Alexander. External Influences and the Educational Landscape: Analysis of
Political, Economic, Geographic, Health and Demographic Factors in Ghana. New
York: Springer, pp. 1–143 (2013).
140. Derry, Gregory. What Science Is and How It Works. Princeton: Princeton University
Press (2002).
141. Federal Reserve Bank of St. Louis. FRED Economic Data. Invest-
ment in government fixed assets. [Link]
tables?rid=356&eid=145951#snid=145952 (2020).
142. Yao, L., Y. Li, S. Ghosh, J. Evans, A. Rzhetsky. Health ROI as a measure of misalignment
of biomedical needs and resources. Nature Biotechnology, 33, 807–811 (2015).
143. Boudreau, K., T. Brady, I. Ganguli, P. et al. A field experiment on search costs and the
formation of scientific collaborations. The Review of Economics and Statistics, 99 (4),
565–576 (2017).
 REFERENCES 177

144. Bol, T., M. de Vaan, A. van de Rijt. The Matthew effect in science funding. Proceedings
of the National Academy of Sciences, 115 (19), 4887–4890 (2018).
145. Azoulay, Pierre, Christian Fons-Rosen, Joshua S. Graff Zivin. Does science advance
one funeral at a time? American Economic Review, 109 (8), 2889–2920 (2019).
146. Walsh, E., M. Rooney, L. Appleby, G. Wilkinson. Open peer review: a randomised
controlled trial. British Journal of Psychiatry, 176 (1), 47–51 (2000).
147. Kirk, Geoffrey, John Raven, Malcolm Schofield. The Presocratic Philosophers: A Crit-
ical History with a Selection of Texts. Cambridge: Cambridge University Press, 2nd
ed. (1983).
148. Zirkle, Conway. Natural selection before the ‘Origin of Species’. Proceedings of the
American Philosophical Society, 84 (1), 71–123 (1941).

Downloaded from [Link] by guest on 29 August 2025

149. Bowler, Peter. Evolution: The History of an Idea. Berkeley: University of California
Press, 3rd ed. (2003).
150. Darwin, Erasmus. Zoonomia. New York: AMS Press (1794).
151. Keynes, Richard. Charles Darwin’s Zoology Notes and Specimen Lists from H.M.S.
Beagle. Cambridge: Cambridge University Press (2000).
152. Boughn, Stephen. Hasenöhrl and the equivalence of mass and energy. European
Physical Journal H (2013).
153. Bartocci, Umberto. Albert Einstein e Olinto De Pretto. La Vera Storia della For-
mula Più Famosa del Mondo (translation, Albert Einstein and Olinto De Pretto.
The True History of the Most Famous Formula in the World). Bologna: Andromeda
(1999).
154. Janssen, Michel, Matthew Mecklenburg. From classical to relativistic mechanics: elec-
tromagnetic models of the electron. In: Vincent F. Hendricks, Klaus Frovin Jørgensen,
Jesper Lützen, Stig Andur Pedersen (eds), Interactions: Mathematics, Physics and
Philosophy, 1860–1930. Dordrecht: Springer (2007).
155. Einstein, Albert. Das prinzip von der erhaltung der schwerpunktsbewegung und die
trägheit der energie. Annalen der Physik, 20 (8), 627–633 (1906).
156. Einstein, Albert. Über die vom relativitätsprinzip geforderte trägheit der energie
(translation, On the inertia of energy required by the relativity principle). Annalen
der Physik, 23, 371–384 (1907).
157. Needham, Joseph. The Grand Titration: Science and Society in East and West.
Toronto: University of Toronto Press (1969).
158. Logan, R. The Alphabet Effect. New York: Morrow (1986).
159. Needham, Joseph. Science and Civilisation in China. Cambridge: Cambridge Uni-
versity Press, Vol. 3, pp. 154, 167–168 (1956).
160. Needham, Joseph. Foreword. In: D. Raven, W. Krohn, R.S. Cohen (eds), Edgar Zilsel,
The Social Origins of Modern Science. Dordrecht: Kluwer Academic Publishers
(2003).
161. Sheldrake, Rupert. Science Set Free: 10 Paths to New Discovery. New York City:
Harmony/Rodale (2012).
162. Baker, M. 1,500 scientists lift the lid on reproducibility. Nature, 533, May 26 (2016).
163. Shi, F., J. Foster, J. Evans. Weaving the fabric of science: dynamic network models of
science’s unfolding structure. Social Networks, 43, 73–85 (2015).
178 REFERENCES

164. Li, J., Y. Yin, S. Fortunato, D. Wang. Scientific elite revisited: patterns of productivity,
collaboration, authorship and impact. Journal of the Royal Society of Interface, 17,
20200135 (2020).
165. Rzhetsky, A., J. Foster, I. Foster, J. Evans. Choosing experiments to accelerate collec-
tive discovery. Proceedings of the National Academy of Sciences, 112, 14569–14574
(2015).
166. Uzzi, B., S. Mukherjee, M. Stringer, B. Jones. Atypical combinations and scientific
impact. Science, 342, 468–472 (2013).
167. Shi, F., J. Evans. Surprising combinations of research contents and contexts are
related to impact and emerge with scientific outsiders from distant disciplines. Nature
Communications, 14, 1641 (2023).

Downloaded from [Link] by guest on 29 August 2025

168. Azoulay, P., J. Graff Zivin, G. Manso. Incentives and creativity: evidence from the
academic life sciences. The RAND Journal of Economics, 42, 527–554 (2011).
169. Jones, B., E. Reedy, B. Weinberg. Age and scientific genius. In: D.K. Simonton (ed.),
The Wiley Handbook of Genius. Chichester: John Wiley & Sons, pp. 422–450 (2014).
170. Simonton, D. Creative productivity: a predictive and explanatory model of career
trajectories and landmarks. Psychological Review, 104, 66–89 (1997).
171. Jones, Benjamin. The burden of knowledge and the “death of the Renaissance Man”:
is innovation getting harder? The Review of Economic Studies, 76, 1, 283–317 (2009).
172. Wuchty, S., B. Jones, B. Uzzi. The increasing dominance of teams in production of
knowledge. Science, 316, 1036–1039 (2007).
173. Wu, L., D. Wang, J.A. Evans. Large teams develop and small teams disrupt science
and technology. Nature, 566, 378–382 (2019).
174. Danús, Lluís, Carles Muntaner, Alexander Krauss, Marta Sales-Pardo, Roger
Guimerà. Differences in collaboration structures and impact among prominent
researchers in Europe and North America. EPJ Data Science, 12 (2023).
175. Krauss, Alexander, Lluís Danús, Marta Sales-Pardo. Early-career factors largely deter-
mine the future impact of prominent researchers: evidence across eight scientific
fields. Nature, Scientific Reports, 13, 18794 (2023).
176. Li, W., T. Aste, F. Caccioli, et al. Early coauthorship with top scientists predicts success
in academic careers. Nature Communications, 10, 5170 (2019).
177. Clarivate. History of citation indexing. [Link]
essays/history-of-citation-indexing/ (2024).
178. Martin, A., R. Synge. A new form of chromatogram employing two liquid phases.
Biochemistry Journal, 35 (12), 1358–1368 (1941).
179. Gordon, J., H. Zeiger, C. Townes. Molecular microwave oscillator and new hyperfine
structure in the microwave spectrum of NH3 . Physical Review, 95, 282 (1954).
180. Knoll, M., E. Ruska. Das elektronenmikroskop. Zeitschrift für Physik, 78, 318–339
(1932).
181. Svedberg, The, Herman Rinde. The ultra-centrifuge, a new instrument for the deter-
mination of size and distribution of size of particle in amicroscopic colloids. Journal
of the American Chemical Society, 46 (12), 2677–2693 (1924).
182. Britannica, The Editors of Encyclopaedia. “Yongle dadian.” Encyclopedia Britannica,
[Link] (2015).
 REFERENCES 179

183. Shannon, Claude. A mathematical theory of communication. The Bell System Tech-
nical Journal, 27, 379–423, 623–656 (1948).
184. Haven, Kendall. 100 Greatest Science Discoveries of All Time. London: Libraries
Unlimited (2007).
185. Krenn, M., R. Pollice, S.Y. Guo, et al. On scientific understanding with artificial
intelligence. Nature Reviews Physics, 4, 761–769 (2022).
186. Hollon, T.C., B. Pandian, A.R. Adapa, et al. Near real-time intraoperative brain
tumor diagnosis using stimulated Raman histology and deep neural networks. Nature
Medicine, 26, 52–58 (2020).
187. The Royal Society. The AI Revolution in Scientific Research. London: The Royal
Society and The Alan Turing Institute (2019).

Downloaded from [Link] by guest on 29 August 2025

188. Langley, P., A. Arvay. Scientific discovery, process models, and the social sciences. In:
M. Addis, P. Lane, P. Sozou, [Link] (eds) Scientific Discovery in the Social Sciences.
Cham (Switzerland): Springer (2019).
189. Schmidt, M., H. Lipson. Distilling free-form natural laws from experimental data.
Science, 324, 81–85 (2009).
190. Džeroski, S., L. Todorovski. Equation discovery for systems biology: finding the struc-
ture and dynamics of biological networks from time course data. Current Opinion in
Biotechnology, 19, 360–368 (2008).
191. Simon, H. A., Langley, P., & Bradshaw, G. L. Scientific discovery as problem solving.
Synthese, 47, 1–27 (1981).
192. Khoury, M., J. Ioannidis. Big data meets public health. Science, 346, 6213 (2014).
193. Fisher, R.A. Statistical Methods for Research Workers. Edinburgh: Oliver & Boyd
(1925).
194. Ioannidis, J., G. Witte. Reproducibility in crisis: sample quality and the importance
of early and ongoing analysis. Science, March 14 (2018).
195. Munafò, M., B. Nosek, D. Bishop, et al. A manifesto for reproducible science. Nature
Human Behaviour, 1 (0021) (2017).
196. Allison, D., Andrew W. Brown, Brandon J. George, Kathryn A. Kaiser. Reproducibil-
ity: a tragedy of errors. Nature, 530, 27–29 (2016).
197. Goldacre, B. Make journals report clinical trials properly. Nature. 530, 7 (2016).
198. Krauss, Alexander. Multi-studies: a novel approach to addressing irreplicability in
RCTs. In: Margaret MacDougall (ed.). A Medical Educator’s Guide to Thinking Crit-
ically about Randomised Controlled Trials: Deconstructing the “Gold Standard”.
Cham (Switzerland): Springer (2024).
199. Frenken, Koen. Geography of Scientific Knowledge: A Proximity Approach. Working
Papers. Eindhoven: Eindhoven Center for Innovation Studies (2010).
200. Leonelli, Sabina. Data-Centric Biology: A Philosophical Study. Chicago: University
of Chicago Press (2016).
201. Cartwright, N. What are randomised controlled trials good for? Philosophical Stud-
ies, 147, 59 (2010).
202. Peters, Uwe, Alexander Krauss, Oliver Braganza. Generalization bias in science.
Cognitive Science, 46, e13188 (2022).
203. Godfrey-Smith, P. Theory and Reality: An Introduction to the Philosophy of Science.
Chicago: University of Chicago Press (2003).
180 REFERENCES

204. Okasha, Samir. Philosophy of Science. A Very Short Introduction. Oxford: Oxford
University Press (2002).
205. Curd, M., J. Cover, C. Pincock. Philosophy of Science: The Central Issues. New York:
W.W. Norton & Company, 2nd ed. (2012).
206. Hume, David. A Treatise of Human Nature. London: John Noon (1739).
207. Schurz, Gerhard. Hume’s Problem Solved: The Optimality of Meta-Induction. Cam-
bridge, MA: MIT Press (2019).
208. Kuhn, Thomas. Logic of discovery or psychology of research? In: P.A. Schilpp (ed.),
The Philosophy of Karl Popper. The Library of Living Philosophers, Volume xiv, Book
II. La Salle: Open Court, pp. 798–819 (1974).
209. van Fraassen, Bas. The Scientific Image. Oxford: Oxford University Press (1980).

Downloaded from [Link] by guest on 29 August 2025

210. van Fraassen, Bas. Scientific Representation: Paradoxes of Perspective. Oxford:
Oxford University Press (2008).
211. Kant, Immanuel. Critique of Pure Reason (translation, 1929). London: MacMillan
(1781).
212. Plato. The Theaetetus of Plato, with a revised text and English notes by Louis
Campbell. Oxford: Oxford University Press (1861).
213. Wittgenstein, Ludwig. Tractatus Logico-Philosophicus. London: Kegan Paul (1922).
214. Fowler, Thomas (ed.). Bacon’s Novum Organum. Oxford: Clarendon Press (1878).
215. Thornton, Stephen. Karl Popper. Stanford Encyclopedia of Philosophy Archive,
[Link] (2017).
216. Mumford, Stephen, Matthew Tugby (eds). Metaphysics and Science. Oxford: Oxford
University Press (2013).
217. Dilworth, Craig. The Metaphysics of Science: An Account of Modern Science in
Terms of Principles, Laws and Theories. Dordrecht: Springer (2006).
218. Schrenk, Markus. Metaphysics of Science: A Systematic and Historical Introduction.
London: Routledge (2017).
219. Earman, John, Clark Glymour, Sandra Mitchell (eds). Ceteris Paribus Laws. Springer,
Erkenntnis (2002). [Link]
220. Cartwright, Nancy. The Dappled World: A Study of the Boundaries of Science. New
York: Cambridge University Press (1999).
221. Donahue, William. New Astronomy (1609) (translation, Astronomia Nova). Cam-
bridge: Cambridge University Press (1992).
222. Stillman, Drake. The Assayer (1623) by Galileo Galilei (translation, Il Saggiatore).
Philadelphia: University of Pennsylvania Press (1960).
223. Newton, Isaac. Philosophiæ Naturalis Principia Mathematica. Pepys: London (in
English 1728) (1687).
224. Einstein, Albert. Does inertia of a body depend on its energy content? (translation).
Annalen der Physik, 18, 639–641 (1905).
225. Encyclopaedia Britannica. Scientific method. [Link]
scientific-method (2022).
226. Collins Dictionary. Science. [Link]/dictionary/english/science
(2022).
227. Cambridge Dictionary. Science. [Link]
english/science (2022).
 REFERENCES 181

228. Encyclopaedia Britannica. Science. [Link]

(2022).
229. Sober, Elliott. Ockham’s Razors: A User’s Manual. Cambridge: Cambridge University
Press (2015).
230. Weinberg, S. Toward the final laws of physics. In: R. Feynman, S. Weinberg (eds), Ele-
mentary Particles and the Laws of Physics. Cambridge: Cambridge University Press
(1987).
231. Feynman, R.P., R.B. Leighton, M. Sands. The Feynman Lectures on Physics. Reading,
MA: Addison-Wesley (1963).
232. Hawking, S. Is the End in Sight for Theoretical Physics? Cambridge: Cambridge
University Press (1988).

Downloaded from [Link] by guest on 29 August 2025

233. National Academy of Sciences. Science, Evolution, and Creationism. Washington,
DC: The National Academies Press (2008).
234. Einstein, Albert. The World As I See It. London: John Lane, p. 29 (1935).
235. Larivière, V., S. Haustein, K. Börner. Long-distance interdisciplinarity leads to higher
scientific impact. PLoS ONE, 10, e0122565 (2015).
236. Gintis, H. A framework for the unification of the behavioral sciences. Behavioral and
Brain Sciences, 30 (1), 1–61 (2007).
237. Merriam-Webster Dictionary. Scientific method. [Link]/
dictionary/scientific_method (2022).
238. Krauss, Alexander. Redefining the scientific method: as the use of sophisticated
scientific methods that extend our mind. PNAS Nexus, 3 (4), 112 (2024).
239. Chang, Hasok. Inventing Temperature: Measurement and Scientific Progress.
Oxford: Oxford University Press, p. 238 (2004).
240. Kaplan, Abraham. The Conduct of Inquiry: Methodology for Behavioral Science. San
Francisco: Chandler Publishing Company (1964).
241. Spelke, E., S. Tsivkin. Language and number: a bilingual training study. Cognition,
78, 45–88 (2001).
242. Bucchi, Massimiano, Brian Trench. Routledge Handbook of Public Communication
of Science and Technology. London: Taylor and Francis, 2nd ed. (2014).
243. Gustafson, Abel, Ronald Rice. The effects of uncertainty frames in three science
communication topics. Science Communication, 41 (6), 679–706 (2019).
244. Dunbar, K., L.A. Baker. Goals, analogy, and the social constraints of scientific
discovery. Behavioral and Brain Sciences, 17, 538–539 (1994).
245. The Nobel Prize. The Nobel Prize in Chemistry 2020. [Link]
prizes/chemistry/2020/press-release/ (2020).
246. LIGO Laboratory. Facts. Pasadena: California Institute of Technology, [Link]
[Link]/page/facts (2022).
247. Kennedy, D., C. Norman (eds). What don’t we know? Science, July 1, 78, 309 (2005).
248. Gleiser, Marcelo. The Island of Knowledge: The Limits of Science and the Search for
Meaning. New York: Basic Books (2014).
249. du Sautoy, Marcus. What We Cannot Know: Explorations at the Edge of Knowledge.
London: 4th Estate (2016).
250. Barrow, John. Impossibility: The Limits of Science and the Science of Limits. Oxford:
Oxford University Press (1999).
182 REFERENCES

251. Dewdney, A.K. Beyond Reason: 8 Great Problems That Reveal the Limits of Science.
Hoboken, NJ: Wiley (2004).
252. Yanofsky, Noson. The Outer Limits of Reason: What Science, Mathematics, and Logic
Cannot Tell Us. Cambridge, MA: MIT Press (2013).
253. Horgan, John. The End of Science: Facing the Limits of Knowledge in the Twilight of
the Scientific Age. New York: Basic Books (2015).
254. Fodor, J. The Modularity of Mind: An Essay on Faculty Psychology. Cambridge, MA:
MIT Press (1983).
255. Chomsky, N. New Horizons in the Study of Language and Mind. Cambridge: Cam-
bridge University Press (2000).
256. Chomsky, N. Science, mind, and limits of understanding. In: A.J. Gallego, R. Martin

Downloaded from [Link] by guest on 29 August 2025

(eds), Language, Syntax, and the Natural Sciences. Cambridge: Cambridge University
Press (2018).
257. Williamson, Timothy. Knowledge and Its Limits. Oxford: Oxford University Press
(2002).
258. Rescher, Nicholas. The Limits of Science. Pittsburgh: University of Pittsburgh Press
(1999).
259. Pigliucci, Massimo, Maarten Boudry (eds). Philosophy of Pseudoscience: Reconsid-
ering the Demarcation Problem. Chicago: University Chicago Press (2013).
260. Bardell, David. The invention of the microscope. Bios, 75 (2), 78–84 (2004).
261. The Nobel Prize. The Nobel Prize in Physics 1986. [Link]
prizes/physics/1986/press-release/ (1986).
262. Gribbin, John. Britannica Guide to 100 Most Influential Scientists. London: Robinson
Publishing (2008).
263. Popper, Karl. The Logic of Scientific Discovery. New York: Harper (1959).
264. Lakatos, Imre, edited by John Worrall and Gregory Currie. The Methodology of
Scientific Research Programmes: Philosophical Papers. Cambridge: Cambridge Uni-
versity Press (1978).
265. Macmillan Dictionary. Theory. [Link]/dictionary/british/
theory (2022).
266. Oxford English Dictionary. Theory. [Link]
theory_n?tl=true (2022).
267. Collins Dictionary. Theory. [Link]/dictionary/english/theory
(2022).
268. Dawkins, Richard. The Greatest Show on Earth: The Evidence for Evolution. New
York: Free Press (2009).
269. Hoyninengen-Huene, Paul. Systematicity: The Nature of Science. Oxford: Oxford
University Press (2013).
270. The Nobel Prize. The Nobel Prize in Physics 2020. [Link]
prizes/physics/2020/press-release/ (2020).
271. Weeks, Mary Elvira. The discovery of the elements. XIII. Some spectroscopic discov-
eries. Journal of Chemical Education, 9 (8) (1932).
272. Emwas, A.H.M., R.M. Salek, J.L. Griffin, et al. NMR-based metabolomics in human
disease diagnosis: applications, limitations, and recommendations. Metabolomics, 9,
1048–1072 (2013).
 REFERENCES 183

273. Del Sole, R., P. Chiaradia. Surfaces, optical properties of. Encyclopedia of Condensed
Matter Physics, pp. 144–150 (2005).
274. Robson, Annie-Louise, Paul C. Dastoor, Jamie Flynn, et al. Advantages and limitations
of current imaging techniques for characterizing liposome morphology. Frontiers in
Pharmacology, 9 (80) (2018).
275. Sinitsyn, Nikolai. Computing with a single qubit faster than the computation quan-
tum speed limit. Physics Letters A, 382 (7), 477–481 (2018).
276. Fardouly, J., R.D. Crosby, S. Sukunesan. Potential benefits and limitations of machine
learning in the field of eating disorders: current research and future directions.
Journal of Eating Disorders, 10, 66 (2022).
277. Greene, Brian. String theory. Encyclopaedia Britannica, [Link]

Downloaded from [Link] by guest on 29 August 2025

com/science/string-theory (2022).
278. Frodeman, Robert, Julie Thompson Klein, Roberto Carlos Dos Santos Pacheco (eds).
The Oxford Handbook of Interdisciplinarity. Oxford: Oxford University Press, 2nd
ed. (2017).
 Acknowledgements

I am thankful for comments from Corinna Peters, Uwe Peters, Michael Stuart, Nikolas
Schöll, Samuli Reijula, Martin Zach, J.P. Grodniewicz, Christopher Evans and Dan Taber.
I am also grateful for funding received by the European Commission (Marie Curie pro-
gramme) and the Ministry of Science and Innovation of the Government of Spain (grant

Downloaded from [Link] by guest on 29 August 2025

RYC2020-029424-I and PID2021-126200NB-I00).
 Index

abstraction 28, 139 mind 33–35

age 75 Merton, Robert 39–40, 43
Anthropology of science 60–65 metaphysics 93–97
Archaeology of science 27–32 methodological abilities 7, 27, 31, 33, 35, 66,

Downloaded from [Link] by guest on 29 August 2025

artificial intelligence 80–81 134–135, 138–139
Methodological toolbox 7–10, 19f , 66, 108f ,
biases, psychological 38–39 125f , 130–134, 136t, 140, 161–162
biases, statistical 86–87, 89, 147–148 Methodological tower of science 7–11
Biology of science 23–26 Methodology of science 66–73
methods 7–11, 13–14, 35, 49–50, 66–73, 77,
causal inference 31, 51–52, 95–96, 166–167 104–113, 114–117, 120–126, 128–143,
citations 4, 74, 76–77 145–160, 161–163
civilization 29–31
Cognitive science of science 33–37 New-methods-drive-science theory 114–116,
collaborations 51, 75 162–163
Computer science of science 79–82 norms 41–42

discoveries 67–69, 75, 107–109, 122–123, 146 observation 35–37, 63–64, 91–92
OVER criterion of science 128–131, 135–138,
Economics of science 48–52 140–142
experimentation 31, 84
paradigm shifts 3, 53–54
falsification 92–93 personality traits 39–40
funding 25–26, 48–51 Philosophy of science 91–100
future science 155–160 Popper, Karl 3–4, 92–93
Psychology of science 38–40
gender 45–46 publications 16–17

History of science 53–59 randomised controlled trials (RCT) 88, 95, 117,
Homo sapiens 28–29 147–148, 151t
replication 86–87, 89
impact 74–77 reward system 48
induction 92
institutions 48 science 13–14
instruments 7–11, 13–14, 35, 49–50, 66–73, 77, science of science 1–168
104–113, 114–117, 120–126, 128–143, scientific method 95–96
145–160, 161–163 Scientometrics and network science 4, 74–78
Sociology of science 41–47
Kuhn, Thomas 3, 53–54, 55–56 Statistics and mathematics of science 83–90

limits of science 118–160 teams 75

Linguistics of science 101–103 tools see instruments

Matthew effect 43, 51 Zuckerman, Harriet 42–43

Downloaded from [Link] by guest on 29 August 2025