Rational Cognition Study

In a 1994 article in the journal Cognition, Eldar Shafir

describes a very straightforward rule from decision theory.
The rule, termed the sure-thing principle by Savage (1954),
says the following. Imagine you are choosing between two
possible outcomes, A and B, and event X is an event that
may or may not occur in the future. If you prefer prospect A
to prospect B if X happens and also you prefer prospect A to
prospect B if X does not happen, then you definitely prefer A
to B, and that preference is in no way changed by knowledge
of event X—so you should prefer A to B whether you know
anything about event X or not. Shafir calls the sure-thing
principle ‘‘one of the simplest and least controversial
principles of rational behavior’’ (p. 404). Indeed, it is so
simple and obvious that it hardly seems worth stating. Yet in
his article, Shafir reviews a host of studies that have
demonstrated that people do indeed violate the sure-thing
principle. For example, Tversky and Shafir (1992) created a
scenario where subjects were asked to imagine that they
were at the end of the term, tired and run down, and awaiting
the grade in a course they might fail and be forced to retake.
They were to imagine that they had just been given the
opportunity to purchase an extremely attractive vacation
package to Hawaii at a very low Is Dysrationalia Possible? :
125 price. More than half of a group of students who were
informed that they had passed the exam chose to buy the
vacation package, and an even larger proportion of a group
who had been told that they had failed the exam chose to
buy the vacation package. However, only one-third of a group
who did not know whether they had passed or failed the
exam chose to purchase the vacation. What these results
collectively mean is that, by inference, at least some subjects
were saying ‘‘I’ll go if I pass and I’ll go if I fail, but I won’t go if
I don’t know whether I passed or failed.’’ Shafir (1994)
describes a host of decision situations where this outcome
obtains. Subjects prefer A to B when event X obtains, prefer A
to B when X does not obtain, but prefer B to A when
uncertain about the outcome X—a clear violation of the sure-
thing principle. These violations are not limited to
hypothetical problems or laboratory situations. Shafir
provides some real-life examples, one involving the stock
market just prior to the Bush/ Dukakis election of 1988.
Market analysts were nearly unanimous in their opinion that
Wall Street preferred Bush to Dukakis. Yet subsequent to
Bush’s election, stock and bond prices declined and the dollar
plunged to its lowest level in ten months. Of course, analysts
agreed that the outcome would have been worse had Dukakis
been elected. Yet if the market was going to go down
subsequent to the election of Bush, and going to go down
even further subsequent to the election of Dukakis, then why
didn’t it go down before the election in response to the
absolute certainty that whoever was elected (Bush or
Dukakis), the outcome would be bad for the market? The
market seems to have violated the sure-thing principle. The
sure-thing principle is not the only rule of rational thinking
that humans have been shown to violate. A substantial
research literature—one comprising literally hundreds of
empirical studies conducted over nearly four decades—has
firmly established that people’s responses often deviate from
the performance considered normative on many reasoning
tasks. For example, people assess probabilities incorrectly,
they display confirmation bias, they test hypotheses
ine≈ciently, they violate the axioms of utility theory, they do
not properly calibrate degrees of belief, they overproject their
own opinions onto others, they display illogical framing
e√ects, they uneconomically honor sunk costs, they allow
prior knowledge to become implicated in deductive
reasoning, and they display numerous other information
processing biases (for summaries of the large literature, see
Arkes 1991; Baron 1994, 1998; Dawes 1998; Evans 1989;
Evans & Over 1996; Kahneman, Slovic, & Tversky 1982;
Nickerson 1998; Osherson 1995; PiattelliPalmarini 1994; Plous
1993; Shafir & Tversky 1995; Stanovich 1999; Tversky 1996).
The reader need not be familiar with all these principles of
rational 126 : keith e. stanovich thinking. It is su≈cient to
appreciate that many of them are as fundamental as the
sure-thing principle just discussed. It is also important to
point out that these reasoning errors do cash out in real-life
behaviors that are decidedly suboptimal and unpleasant for
those displaying these processing biases. Because of the
failure to follow the normative rules of rational thought—
because of the processing biases listed above—physicians
choose less effective medical treatments (McNeil et al. 1982;
Redelmeier & Tversky 1990, 1992; Sutherland 1992); people
fail to accurately assess risks in their environment
(Lichtenstein et al. 1978; Margolis 1996; Yates 1992);
information is misused in legal proceedings (Saks & Kidd
1980–1981); millions of dollars are spent on unneeded
projects by government and private industry (Arkes & Ayton
1999; Dawes 1988, pp. 23–24); parents fail to vaccinate their
children (Baron 1998); unnecessary surgeries are performed
(Dawes 1988, pp. 73–75); animals are hunted to extinction
(Baron 1998; Dawkins 1998); billions of dollars are wasted on
quack medical remedies (Gilovich 1991); and costly financial
misjudgments are made (Belsky 1995; Belsky & Gilovich
1999; Fridson 1993; Thaler 1992; Tversky 1996; Willis 1990).
Many of these examples concern what philosophers call
pragmatic, or practical, rationality—how well individuals
maximize the satisfaction of their desires, given their beliefs
(Audi 1993; Harman 1995; Nathanson 1994). This is often
contrasted with epistemic rationality, which is concerned with
the consistency of a person’s network of beliefs and how well
it represents the external world (the so-called theoretical
rationality of philosophy) (Audi 1993; Foley 1987; Harman
1995). Smart People Doing Dumb Things: Resolving the
Paradox The findings from the reasoning and decision making
literature and the many real-world examples of the
consequences of irrational thinking (e.g., Belsky & Gilovich
1999; Gilovich 1991; Piattelli-Palmarini 1994; Shermer 1997;
Sutherland 1992; Thaler 1992) create a seeming paradox.
The physicians using ine√ective procedures, the financial
analysts making costly misjudgments, the retired
professionals managing their money poorly—none of these
are unintelligent people. The experimental literature is even
more perplexing. Over 90 percent of the subjects in the
studies in the literature are university students—some from
the most selective institutions of higher learning in the world
(Tversky and Shafir’s subjects are from Stanford). Yet these
are the very people who have provided the data indicating
that a substantial proportion of people can sometimes violate
the most basic stric- Is Dysrationalia Possible? : 127 tures of
rational thought, such as transitivity or the sure-thing
principle. It appears that an awful lot of pretty smart people
are doing some incredibly dumb things. How are we to
understand this seeming contradiction? The first step in
understanding the seeming paradox is to realize that the
question ‘‘How can so many smart people be doing so many
dumb things?’’ is phrased in the language of folk psychology.
The issue of how to interpret folk psychology is a topic of
immense interest in cognitive science at present, and it is the
subject of much controversy (Christensen & Turner 1993;
Churchland & Churchland 1998; Davies & Stone 1995;
Greenwood 1991; Stich 1996). Positions vary from those who
think that most folk psychology needs to be eliminated from
the terminology of scientific psychology to those who think
that folk psychology should be the very foundation of a
scientific psychology. My concern here is not with these
classic issues but with how concepts in cognitive science can
be used to make folk usage more precise in ways that serve
to dissipate seeming paradoxes.∞ I propose to do just this
with the ‘‘smart but dumb’’ phrase. In this chapter, I identify
the folk term ‘‘smart’’ with the psychology concept of
intelligence (defined as an amalgamation of cognitive
capacities). The acts that spawn the folk term ‘‘dumb’’ I
identify with violations of rationality as that term is
conceptualized within cognitive science, philosophy, and
decision science (Baron 1993a; Harman 1995; Je√rey 1983;
Kleindorfer, Kunreuther, & Schoemaker 1993; Nathanson
1994; Nozick 1993). This mapping does not immediately
solve the problem, because there are several di√erent ways
of parsing the concepts intelligence and rationality—
especially within psychology. Thus, I present one such
partitioning that I think is useful in contextualizing the ‘‘smart
but dumb’’ phenomenon and dissolving its seemingly
paradoxical status. The partitioning that I prefer relies heavily
on distinguishing levels of analysis in cognitive theory.