.

1 The neoclassical model The decision-making process We need to begin our

analysis of decision-making by examining the process involved in most decisions. Of
course, some ‘decisions’ do not really appear to be decisions at all because we act
on instinct, without conscious thought. This is often because there is no time for
using the ‘cold’ analytical cognitive system described in the previous chapter as one
component of our dual system; instead we rely on the ‘hot’ affective system, which
is often dependent on instantaneous emotional or visceral responses to a situation.
Hence we may duck an object flying at our head, or lash out in anger, physically or
verbally, at someone who has offended us. However, even in cases where this ‘hot’
system is used, and we are not aware of making a conscious decision, there is still a
situation where some kind of stimulus results in an action. Whenever this situation
arises we deem that a decision-making process is involved. What is needed
therefore is a general model of the decision-making process that applies to
automatically reaching for the Oreo cookies at the supermarket, deciding where to
buy a lottery ticket, investing in a particular stock, or choosing a particular medical
treatment. Obviously these decisions are very different in nature and involve
different factors and even decision systems, but there are certain ele ments in
common. In general there are three fundamental characteristics of decision-making;
these are described below, along with their corresponding components in the
neoclassi cal model (NM) as expressed in equation (1.1), which is repeated here for
ease of reference: 63 PT• II FOUNDATIONS (1) max (2) t (3) p( st ) (4) U(xi t | st ).
x i t∈Xi t=0 s t ∈ St (1.1) repeated 1 Preferences – these are the rankings people
have over a set of options or gambles that are based on attitudes and values
related to the outcomes of these options (4). 2 Beliefs – these relate to the
probabilities with which people think various outcomes will occur, conditional on
available information (3). 3 Rationality – this involves all four components of the
standard model, referring to the ways in which people: • determine preferences
based on attitudes and values (4) • appropriately modify their beliefs in the light of
new information (3) • discount values of future outcomes (2) • succeed in choosing
optimal actions given their preferences and beliefs (1) In a comprehensive survey of
empirical findings in the field related to behavioral eco nomics, DellaVigna (2009)
examines deviations from the NM according to the above classification; he refers to
these deviations as non-standard preferences, non-standard beliefs, and non-
standard decision-making. Many of the examples he gives are discussed in this
chapter and the following chapters. The first of these deviations is examined in this
chapter, while the second deviation is examined in the next chapter. It should be
noted that there is significant interdependence and overlap between these three
categories: beliefs affect preferences, and both of these affect choices as reflected
in decision-making. Furthermore, rationality involves both beliefs and preferences;
thus aspects of rationality are discussed in both chapters, as well as in later
chapters in the book. There is also a final discussion of rationality in the concluding
chapter, after the topics of risk and uncertainty, intertemporal preferences, and
social preferences, have been discussed. Nevertheless, we shall see that the
beliefs–preferences–rationality classification is useful, in that the factors within each
category are involved at different stages of the decision-making process. Consumer
behavior We have already seen how the NM can be described in mathematical
terms in expression (1.1). This can be translated into a NM of consumer behavior
very simply. As Varian (2006) expresses it: ‘people choose the best things they can
afford’ (p. 33). This is essen tially a constrained optimization problem. The objects
of consumer choice are referred to as consumption bundles, and these relate to a
complete list of the goods and services that are involved in the particular choice
problem being considered. In a generalized situation we also need a description of
when, where, and under what circumstances these goods would become available.
People care not only about what goods are available now but about what will be
available at a later date; they also care more for a bottle of water if they are in the
middle of a desert than if they are in the Antarctic. Any bundle of goods can be
described in the simplest terms as (x1 , x2 ) or just X, where x1 denotes the amount
of one good and x2 the amount of another good, or the amount of all other goods.
By limiting the number of parameters to just two it is possible to use a graphical
method of representa tion and analysis. Preferences In the NM it is assumed that
consumers can rank bundles according to their desirability. If a consumer definitely
wants an x-bundle rather than a y-bundle then it is said that he 64 values,
PreferenCes anD ChoiCes CH•3 or she strictly prefers the x-bundle to the y-bundle.
This can be written as follows: (x1 , x2 ) ≻ (y1 , y2 ). Alternatively, if a consumer is
indifferent between the two bundles this means that they have no preference for
either bundle over the other. This relationship is usually described by the
expression: (x1 , x2 ) ~ (y1 , y2 ). Finally, if a consumer prefers or is indiffer ent
between the two bundles we say that he or she weakly prefers (x1 , x2 ) to (y1 , y2 )
and this is written: (x1 , x2 ) (y1 , y2 ). Preference relations are meant to be
operational notions. Thus it is also assumed in the NM that choice is determined by
preference; this is important since it is choice that is directly observable, not
preference. If a consumer chooses a particular bun dle, it is assumed that this
bundle is preferred to another bundle if that other bundle was both available and
affordable. This is what is meant by the concept of revealed preference. Indifference
curves The theory of consumer choice is often illustrated using a graphical approach
involv ing indifference curves. An indifference curve represents different
combinations of two goods between which the consumer is indifferent, that is,
which yield the same total utility. Figure 3.1 is an indifference map, showing a
number of indifference curves, with curves further away from the origin
representing bundles involving greater quantities of goods; these bundles are
therefore preferred to those on lower indifference curves. Thus any combination of
goods on curve I2 is preferred to any combination on curve I1 . However, nothing is
specified regarding how much more these goods are preferred; thus we can say
nothing regarding the relative sizes of the difference between I2 and the difference
between I3 and I2 . Indifference curves are normally drawn as downward-sloping
and with convex to the origin; this is because of certain assumptions that are
described in the next section. FIGURE 3.1 Indifference curve map y I3 I2 I1 x 65 PT•
II FOUNDATIONS Equilibrium Indifference curve maps can be used to illustrate the
notion of consumer equilibrium. It is assumed that consumers have a budget
constraint relating to the amount of money that they have available to spend on the
relevant consumption bundles. This budget constraint can be represented in terms
of the following inequality: px x + py y ≤ m where m is the available budget. The
budget constraint is shown in Figure 3.2 as the line going through points A, C and B.
Although the budget is sufficient to buy combinations A and B, these are not optimal
combinations since they are not on the highest indiffer ence curve that can be
reached. Combination C is the optimal combination, where the consumer purchases
the bundle (x1 , y1 ). This situation can be generalized: any optimal point of
consumption will occur at a point of tangency between an indifference curve and a
budget constraint line. Setting the slopes of these two curves equal gives the
condition that the consumer should spend so that the marginal utility of the last
dollar spent on each good is the same. This condition can be further generalized to
apply to any number of goods in a consumption bundle. The exposition of consumer
equilibrium in the NM often uses the term ‘marginal utility’, but in fact this is not
essential. The slope of an indifference curve can also be expressed in terms of the
marginal rate of substitution (MRS) of one good for another. The MRSxy represents
the amount of y that a consumer is prepared to give up in order to get one more
unit of x. Thus the equilibrium condition amounts to con suming a combination such
that the MRS is equal to the ratio of the prices: MRSxy = px /py . The advantage of
this form of exposition is that it avoids the thorny concept of utility. A detailed
discussion of utility is deferred until a Section 3.7. Applications of indifference curve
analysis related to behavioral anomalies are discussed in Angner (2016, ch. 3).
Figure 3.2 Indifference curves and consumer equilibrium y m/py y1 I3 A I1 I2 C B x1
m/py x 66 values, PreferenCes anD ChoiCes CH•3 3.2 Axioms, assumptions and
definitions Axioms In discussing the foundation of the NM, in particular in terms of
how it relates to con sumer preference, it is essential to distinguish between the
basic axioms of the model and other auxiliary assumptions that often accompany it.
While the former are fundamental to the model, the latter are somewhat flexible
and are varied in different circumstances. This distinction is relevant in discussing
the weaknesses of the NM in the following section, and relates to the Duhem-Quine
thesis discussed in the previous chapter. We will also see that there is a distinction
between descriptive and normative axioms. This aspect is discussed in Chapter 5 in
the context of risk and uncertainty. There are four main axioms relating to
consumer preference: 1 Completeness. A person can compare any two consumer
bundles, X and Y. Such a comparison must lead to one of three possible mutually
exclusive outcomes: a) Basket X is preferred to basket Y; we have seen that this can
be written: (x1 , x2 ) ≻ (y1 , y2 ). b) Basket Y is preferred to basket X; we have seen
that this can be written: (y1 , y2 ) ≻ (x1, x2 ). c) The consumer is indifferent
between the two baskets; we have seen that this can be written: (x1 , x2 ) ~ (y1 ,
y2 ). This is referred to as the completeness principle because the preferences
cover all possible outcomes. 2 Transitivity. If three different baskets, X, Y and Z, are
considered, a consumer who prefers basket X to basket Y, and who prefers basket Y
to basket Z, must also prefer basket X to basket Z. This can expressed as: if (x1 , x2
) ≻ (y1 , y2 ) and (y1 , y2 ) ≻ (z1 , z2 ), then (x1 , x2 ) ≻ (z1 , z2 ). Similarly, if a
consumer is indifferent between basket X and basket Y, and who is also indifferent
between basket Y and basket Z, must also be indifferent between basket X and
basket Z. 3 Reflexivity. Any bundle is at least as good as itself: (x1 , x2 ) (x1 , x2 ).
This axiom is generally regarded as being trivial. 4 Revealed preference. This axiom
can come in either weak or strong forms: If (x1 , x2 ) is revealed directly/indirectly
preferred to (y1 , y2 ), and the two bundles are not the same, then it cannot happen
that (y1 , y2 ) is revealed directly/indirectly pre ferred to (x1 , x2 ). Assumptions In
addition to the above axioms there are four other main assumptions, often referred
to as principles, which frequently accompany the NM. These assumptions are
particularly relevant in situations where there is uncertainty; they are therefore
discussed in more detail in Chapter 5 in relation to expected utility theory (EUT),
which is the aspect of the NM that is relevant in such situations. We will also see
that EUT involves certain additional assumptions, apart from the ones discussed
below. Tversky and Kahneman (1986) claim that the following substantive
assumptions can be ordered in terms of a hierarchy of increasing importance as far
as their normative appeal is concerned. Thus the 67 PT• II FOUNDATIONS
cancellation condition has been challenged by many theories, but the invariance
condition is absolutely necessary for any normative theory. 1 Cancellation. This is
the principle that any state of the world that results in the same outcome regardless
of one’s choice can be cancelled or ignored. It is sometimes referred to as a minimal
approach to decision-making, as opposed to a topical or com prehensive approach
(discussed in Chapter 5). Thus, if X is preferred to Y, then the prospect of winning X
if it rains tomorrow (and nothing otherwise) should be preferred to the prospect of
winning Y if it rains tomorrow, because the two prospects both result in the same
outcome (nothing) if there is no rain tomorrow. This assumption is also consistent
generally with the marginalist approach of the neoclassical economists. 2
Dominance. This condition is simpler and stronger than the first one, and therefore
is more fundamental to the standard model. It states that if option X is better than
option Y in one state and at least as good in all other states, then option X is
dominant over option Y and should be chosen. This is related to the reflexivity
axiom discussed earlier, but is not as strong a condition. 3 Extensionality. The NM
generally assumes that people have the same attitude to a particular object and
attribute the same value to it, however that object is described, given a certain
level of information about that object. Thus people should have the same attitude to
a particular kind of packaged meat, whether it is described as 5% fat or as 95% fat-
free. This is again related to the reflexivity axiom. 4 Invariance. This condition is
essential for any normative theory. It states that differ ent representations of the
same choice problem should yield the same preference. This therefore rules out
‘framing effects’, discussed at length in the next few chap ters. It states that the
relation of preference should not depend on the description of the options
(description invariance) or on the method of elicitation (procedure invariance).
Without stability across equivalent descriptions and equivalent elicita tion
procedures, a person’s preferences cannot be represented as utility maximization in
the standard EUT. There is also the assumption in the standard model that
consumers maximize utilities. This is usually taken as both a descriptive and a
normative statement: people do behave in this way, and ought to behave in this
way to maximize their welfare. This normative aspect is related to the concept of
rationality. Certain other assumptions are invoked in particular situations; these are
discussed in the relevant context in the remaining sections and chapters. However,
we need to be aware of the differing status of these assumptions; some are really
not a necessary component of the NM and can easily be relaxed without
compromising the model. For example, the NM is sometimes said to involve
assumptions regarding three characteristics of indifference curves. The first of these
has already been stated as an axiom, involving transitivity of pref erences. The
second characteristic is that the curves are downward-sloping; this implies that
more of a good is always preferred to less of a good. This principle is referred to as
the monotonicity of preferences, and can be expressed in technical terms thus: if
(x1 , x2 ) is a bundle of goods and (y1 , y2 ) is a bundle of goods with at least as
much of both goods and more of one, then (y1 , y2 ) ≻ (x1 , x2 ). This assumption
follows directly from the definition of a good, that it is commodity where more of it
is preferred to less. Some commodities, like pollution and garbage, are referred to
as ‘bads’. The assumption of monotonicity is not essential to the standard model;
for example, when people reach satiation the principle no longer applies. The third
characteristic of indifference curves is that they are convex to the origin. This is
sometimes referred to as the assumption of convexity of preferences, but it is not
really an essential assumption either, since it follows from the law of diminishing
marginal utility. This law states that, as more of a good is consumed within a certain
time 68 values, PreferenCes anD ChoiCes CH•3 period, additional units of
consumption will eventually yield less marginal utility. The law is in the nature of an
empirical regularity rather than being an assumption in analysis. It is this
characteristic of convexity which determines the inverse relationship between price
and quantity demanded in the conventional demand curve, although such a
relationship can also be derived by other means (Becker, 1976). Definitions Before
discussing the weaknesses of the standard model in terms of consumer preference,
it will help to clarify the situation if we give definitions of four terms that are
involved in the decision-making process: attitude, value, preference and choice. It
should be stated at the start that the definitions that follow are not universally
agreed upon or followed by either economists or psychologists, but they are widely
shared. 1 Attitude – This has been usefully defined as ‘a psychological tendency
that is expressed by evaluating a particular entity with some degree of favor or
disfavor’ (Eagly and Chaiken, 1996). This concept has a broader range of application
than the concept of preference, as will be seen shortly. Attitudes relate to any entity
that people can like or dislike. Entities include physical objects, living beings and
abstract concepts, involving anything that can elicit an affective response. It is
important to realize that objects of attitudes are mental representations, not
objective states of affairs. This means that attitudes are based on beliefs, discussed
in the next chapter. Realizing this helps us to understand the violations of the
invariance and extension principles discussed above. People have a different mental
representation of meat with 5% fat compared with meat that is 95% fat-free. A
related characteristic of attitudes is that they tend to involve judgment by proto
type. This refers to the phenomenon that global judgment of a category is
determined primarily by the relevant properties of a prototype, and is based on the
older concept of a representativeness heuristic (Kahneman and Tversky, 1972,
1973; Tversky and Kahneman, 1971, 1983). This heuristic is explained in the next
chapter, and leads to various kinds of extension bias. An example, discussed in
more detail in the next sec tion, involves duration neglect: people tend to recall
past experiences in terms of a prototypical moment, rather than as the total
experience. Thus they may ignore the length of an unpleasant experience like a
colonoscopy, recalling instead the moment of most intensive pain. 2 Value – This is
a problematical term, for it is widely used in two different senses. In one sense our
values determine our attitudes, while in the other sense our attitudes determine our
values. In the first sense values refer to tastes or likes/dislikes: we may have a taste
for cauliflower, a taste for roller coasters, or a taste for making charitable donations.
Thus values in this sense include moral values. One can of course take the question
further back, following a reductionist theme, and ask what determines a person’s
values; however, important as this issue is, it goes beyond the scope of this book
and into the details of evolutionary psychology. The second sense of the term value
refers to the quantitative evaluation that results from an attitude. The
determination of value involves judgment. This sense of the term value is what
economists conventionally mean by the term utility, whether refer ring to
experienced or decision utility. These terms will be discussed in more detail in the
following sections. 3 Preference – As we have already discussed, economists are
conventionally more concerned with preferences, in particular so-called ‘revealed
preferences’, while psychologists may be more concerned with attitudes. In the NM
it is assumed that attitudes determine preferences, but that is not necessarily true.
69 PT• II FOUNDATIONS 4 Choice – This involves an action on the part of the
subject/consumer, involving some kind of decision. The standard model generally
assumes that choice is simply revealed preference. We will see however that
preferences and choices are not neces sarily identical and that choices need not be
a reflection of attitudes and judgments (Tversky, Sattath and Slovic, 1988; Tversky
and Griffin, 2000). There are other terms related to these that also need to be
clarified, such as pleasure, hap piness and well-being. In order to shed further light
on them, and on the four terms above, we need to examine the foundations of
utility in evolutionary biology. Only then can we understand why the goods we
consume give us utility at all. 3.3 The evolutionary biology of utility Nature and
functions of utility As explained earlier, it is necessary is to take a reductionist
approach in order to under stand the ultimate causes of human behavior. The brain
did not evolve in order to maximize utility, well-being or hedonic pleasure. The
forces of natural selection have caused the brain to be designed as a system that
maximizes biological fitness. This term can be understood as meaning our ability to
survive and reproduce. Evolutionary biolo gists also use the term inclusive fitness,
where the ability extends to our kin, since that increases the overall likelihood of
‘spreading our genes’. It is important to realize that this driving force behind
evolution operates at the genetic level strictly speaking, not at the level of the
individual. One main strength of neuroeconomic analysis is that it recognizes this
biological basis of behavior, not just for humans but for all animals. Historically
speaking, biological fitness is therefore closely linked to the number of offspring,
and this allows the construction of testable models of economic behavior, as we
shall see in following chapters. In particular these models involve the application of
evo lutionary game theory, involving a principal-agent situation where Nature is the
principal and the individual is the agent. The objective of the principal is to
maximize biologi cal fitness, which entails maximizing the number of viable
offspring, meaning offspring who are in turn likely to be able to reproduce
successfully. This is achieved through the competitive forces of natural selection. In
terms of economics, however, there are two fundamental and linked questions,
posed by Robson (2002): (1) Why is there utility at all? (2) Why is it defined on
intermediate goods rather than on offspring? The first question is easier to answer.
Utility exists as the criterion that humans, and other animals, use when selecting
actions in response to the variety of environmental situations they encounter. As
Robson (2002) states: ‘animals with the best repertoire of instinctive and learned
procedures for making these choices would be favored by evolu tion.’ Thus fear and
pain engender negative utility, urging animals to avoid actions that cause these
aversive sensations; animals with effective fear and pain disutility signals are likely
to have greater biological fitness. Likewise, nutritious foods engender high utility
signals, so that animals with effective food signals will again have greater biological
fitness. The second question involves a more complex answer. Nature imbues
rankings or preferences over consumption goods, which are really intermediate
goods as far as the ultimate goal of producing offspring is concerned. Why? Robson
(2002) suggests that a strategy whereby people or animals determine their
consumption habits according to personal experience or direct observation of the
number of offspring of others would not be effective, mainly because of the problem
of the small sample involved. People would only be able to observe a few relatives
or friends, and maybe one or more of these has an unhealthy diet but has a lot of
offspring, while another or others may have a healthy 70 values, PreferenCes anD
ChoiCes CH•3 diet but no offspring. The observer might then conclude on the basis
of this small sample that an unhealthy diet was best. A more effective strategy for
Nature to achieve biological f itness would be to imbue animals with preferences
over consumption bundles, in order to avoid this small sample error. Furthermore, it
is more efficient for individuals to have a utility function that applies to all bundles
rather than to determine consumption by comparing each pair of possible bundles.
When there is a very large number of possible pairs to be compared it would require
a larger brain to process and analyze the relevant information, and therefore
greater metabolic energy resources to build and maintain this organ. This would in
turn detract resources from other productive uses, like building the body, which
may be more effective at improving biological fitness. A utility function allows a
more efficient process of simply selecting the bundle with the highest utility rather
than making myriad comparisons. We can therefore state that the individual is
prompted to maximize hedonic pleas ure as a means to the ultimate end of
maximizing biological fitness. Actions that result in hedonic pleasure have in the
past generally improved biological fitness, while actions causing displeasure or pain
have been an indication that our biological fitness is threat ened. Furthermore, it is
important to understand that this hedonic pleasure relates not only to conventional
goods but also to what we can call moral sentiments, using the ter minology of
Adam Smith [1759] (2000), the first economist to inquire into their nature. Thus we
tend to take actions that make us feel proud or increase our self-esteem, while
avoiding those that cause us to feel guilt or shame. If we do not recognize that
moral senti ments also involve hedonic pleasure or pain we will become entangled
in the problem of dualism, which, as we will see in Chapter 11, still bedevils the
thinking of some neurosci entists, as well as economists and philosophers. Types of
selection process The mechanism of natural selection is responsible for producing
adaptations over time, to both bodies and brains, which ensure the maintenance of
biological fitness of a species in a particular environment. However, there are
various aspects of the environment that are relevant. We operate not only in a
physical environment, but also in a social environ ment. Therefore there are other
aspects to the selection process that are also important in understanding the nature
of utility. One aspect is referred to as sexual selection, and involves the selection of
individuals within a species who are most successful in finding and retaining mates
who will produce the most viable and healthy offspring. Since there is sexual
specialization as far as rearing offspring is concerned, in both humans and many
other animal species, the different sexes have different characteristics that are
particularly desirable to the opposite sex. Human males tend to be attracted to
females who are fer tile and healthy-looking, while females tend to be attracted to
males who promise to be good providers. Signaling is of great importance in this
selection process. Women often want to signal youth and beauty, thus the great
demand in the cosmetics industry, mainly (although certainly not exclusively) from
women. For men, conspicuous consumption is a vital signal, and this can be
displayed by buying designer clothing, accessories like watches and jewelry, luxury
cars and the like. A further aspect of the selection process is referred to as group
selection. Instead of operating at the level of the gene, or the individual, group
selection acts at the level of a group of individuals. The relative importance in
evolution of these three different levels of selection is still a highly controversial
issue in biology and sociology; it is not intended to enter this debate in this book,
merely to describe the process of group selection that has been proposed, and its
relevance in understanding utility. The theory of group selection essentially
proposes that if a group within a population possesses certain characteristics that
enable that group to reproduce faster than other groups, and that group is isolated
71 PT• II FOUNDATIONS reproductively from other groups for substantial periods,
then these characteristics may be able to spread within a population, even if
individuals possessing them are at a disad vantage within a particular group. The
main examples that are often given here relate to the characteristics of empathy
and cooperation, which can operate in many different ways. These characteristics
relate to social preferences; for example, an inequality in the distri bution of payoffs
among a group may give a person disutility. This aspect will be discussed in more
detail in Chapter 10. This does not mean that the existence of social preferences
relies on the process of group selection, merely that group selection may have
shaped our social preferences in our evolutionary past. Evolution and optimization A
common misconception regarding the neo-Darwinian synthesis (NDS) is that
‘survival of the fittest’, as natural selection is often called, leads ultimately to
optimization of design as far as species are concerned. This is very far from the
truth. As Richard Dawkins (1986) has explained, the evolutionary process resembles
a ‘Blind Watchmaker’, and has no ulti mate aim or objective. In philosophical terms
it is a mechanistic rather than a teleological process. Furthermore, the process can
only build on biological structures that already exist; thus our bodies and even our
brains often appear as components that have been cobbled together, in a similar
fashion to kludges in computer science and engineering. As far as the nature of
utility is concerned, there are two essential problems caused by the mechanistic
nature of the evolutionary process, causing our hedonic systems to be easily
hijacked: 1 Time lags. Lags exist between the optimal design and the demands of
the current envi ronment. A good example is our liking for sugary and fatty foods,
which improved our biological fitness in the past, but is now a serious threat given
their easy availability and the sedentary nature of many of our lives. 2 Use of
heuristics. These rules are useful shortcuts that simplify the decision-making pro
cess, particularly given time constraints and bounded rationality, but they are prone
to error. For example, we may be trying to choose a car to buy. Heuristics that could
be used here might be: only buy a make you have bought before; buy a domestic
make; buy the safest model within certain price bracket; buy the model with the
best fuel economy. These heuristics have again been shaped by the evolutionary
process, as adaptations, related in the above examples to different aspects of risk-
aversion. However, they are unlikely to maximize utility because each of them
ignores many relevant factors. After this summary of the underlying evolutionary
aspects of utility, we can now turn our attention to its neuroscientific basis, and this
will then allow a further discussion of ration ality and how this relates to the
predictions of the standard model. This will in turn enable us to identify various
different types of utility. 3.4 The neuroscientific basis of utility Techniques and
comparisons Over the last 25 years, and particularly in the last 15, there have been
many neuroeco nomic studies relating to utility, in terms of how it is correlated with
neural activity. These studies have important implications for economic theory.
Positron emission tomogra phy (PET) can detect changes in neurotransmitter
release, and although there are some technical problems here, these studies can
detect which areas of the brain are activated. The main advantage of PET studies is
that they can be used to detect neurochemical 72 values, PreferenCes anD ChoiCes
CH•3 changes related to neurotransmitters like dopamine. The other main
analytical technique, functional magnetic resonance imaging (fMRI), is not able to
do this, but fMRI is supe rior to PET in terms of providing greater temporal and
spatial resolution. This means that fMRI can pinpoint more precisely when a neural
change occurs in time, and where in the brain it occurs. The use of both techniques
in a complementary manner helps to confirm neuroeconomic hypotheses. Other
neural studies examine the effects of lesions or disruption of neural activity using
transcranial magnetic stimulation (TMS). Such studies are useful in identifying the
necessary (but not sufficient) conditions for a particular psychological effect or
economic behavior to occur. The main advantage here is the higher degree of
certainty with which these studies can draw conclusions compared with other
studies, but the main drawback concerns the limited nature of their conclusions.
The other main type of study that has been used relating to the concept of utility
involves single neuron studies. These are extremely accurate in pinpointing brain
areas involved in different functions, but their highly invasive nature currently
restricts their use to non-human primates. The major neuroscientific findings related
to utility that need to be discussed at this stage relate to (1) the nature of utility and
reference dependence; (2) loss-aversion; (3) measurement of utility; and (4) the
difference between reflective and reflexive brain modes. Some conclusions relating
to neuroeconomic evidence are then drawn. Nature of utility and reference
dependence Animal studies have for a long time indicated that reward was
associated with dopamine release, which creates a hedonic ‘high’. A major
development was a study by Schultz, Apicella, and Ljungberg (1993), later
supported by many other studies, that it was not consumption itself that stimulated
dopamine release, but the expectation of consump tion. For example, when
monkeys learned that the tone of a bell was likely to be followed by a reward of
juice, there was a release of dopamine at the tone but not at the later point of
consumption. Thus only unanticipated consumption led to dopamine release. These
studies led to the development of the ‘dopaminergic reward prediction error’ (DRPE
or just RPE) hypothesis concerning the encoding of utility in the brain. This
hypothesis proposed that it was the difference between how ‘rewarding’ an event is
and how rewarding it was expected to be that determined dopamine release and
therefore utility (in the hedonic sense). If the utility from consumption is fully
anticipated, then there is no prediction error and no dopamine release. The key
point here is that utility is reference-dependent. Although the RPE hypothesis is not
universally accepted within the neuroscience community, other theories relating to
dopamine are fairly similar in terms of also being reference-dependent, for example
Zink, Pagnoni and Martin (2003). Another example of a reference-dependent theory
proposes that dopamine encodes ‘incentive salience’, which differentiates between
how much something is wanted and how much it is liked (Berridge and Robinson,
1998). Caplin and Dean (2009) claim that various more recent experiments support
the basic RPE model compared with its rivals, but identify areas for further
development and expansion of the model. Although fMRI studies cannot test the
RPE model directly, they have been able to provide indirect evidence that at least is
consistent with it. Various studies have now shown that anticipated gain causes
increased blood flow in the nucleus accumbens (NAcc), whereas actual gain
outcomes cause increased blood flow in the medial caudate, MPFC, and posterior
cingulate regions (Knutson et al., 2001a, 2001b, 2003). This finding is what might
be expected if utility is reference-dependent, being determined by the difference
between what is expected and what is actually obtained. 73 PT• II FOUNDATIONS
There is another important implication of these studies. The fact that actual gains
activate different brain areas from expected gains suggests that those areas, like
the medial prefrontal cortex (mPFC), that are activated by actual gains serve as a
control center, checking to see if outcomes turned out as expected. Evidence
indicates that if the mPFC is damaged people are unable to learn from mistakes, for
example in the Iowa gambling task. Another aspect of outcomes and reference
points concerns opportunity cost and regret. Regret entails recognition that an
alternative, or counterfactual, action would have produced a more valued outcome
than the one achieved. In humans, the orbito frontal cortex (OFC) is active during
expressions of regret, and humans with damage to the OFC do not express regret.
In rats and non-human primates, both the OFC and the ventral striatum have been
implicated in reward computations, and Steiner and Redish (2014) have performed
experiments to induce sensations of regret in rats. Some fMRI studies have also
examined the prediction of choice. Knutson, Rick and Wimmer (2007) claim that
ventral striatal activation not only correlates with preference while viewing
products, but also predicts the choices of subjects better than self-reported
preference. Of course, traditional economists might not be too impressed with this
find ing, believing that self-reports are unreliable anyway, and that economists
should only be concerned with revealed preference or actual purchase decisions
and behavior. Single neuron studies also support the principle of reference
dependence, by indicating that multiple representations of value exist in the
primate brain (Platt and Padoa-Schioppa, 2009). More specifically, representations
of value in the OFC are abso lute (rewards from a particular good do not depend on
the alternatives available), whereas representations of value in the parietal cortex
are relative. Loss-aversion The main finding here is that gains and losses appear to
activate or deactivate different areas in the brain. For example, O’Doherty and
colleagues (2001) reports that gains and losses activate different areas of the
ventromedial prefrontal cortex (vmPFC). Another f inding of importance, from PET
studies, is that it appears that there is no detectable change in dopamine release
following unexpected loss, at least in the ventral striatum, which is the brain area
most affected by unexpected gain. This suggests that losses are encoded in a
different brain area, utilizing a different neural pathway. Even within the striatum,
the evidence for responsiveness to losses is weaker in the ventral striatum than in
the dorsal striatum (Knutson, Delgado and Phillips, 2009). Several studies, using
fMRI, implicate the insula as being involved in encoding losses, although other areas
such as the amygdala may also be involved. The significance of both PET and fMRI
studies is that their findings provide a neurological basis for the economic
phenomenon of loss-aversion. Measurement of utility The studies discussed so far
have been mainly concerned with the nature of utility and how and where it is
encoded in the brain. Other studies have examined the measurement of utility. A
number of these have shown that the OFC and dorso-lateral prefrontal cortex
(DLPFC) are important here. Plassman, O’Doherty and Rangel (2007) performed an
experiment with hungry subjects viewing snacks and the bids they made for them
after wards. They found that these bids correlated positively with activation in the
OFC. Other studies have shown that lesions or disruptions to the OFC or DLPFC
interfere with the ability to compute values and make consistent choices. Glimcher
(2009) has a particularly ambitious neuroeconomic program. The key con cept in his
analytical approach is ‘subjective value’, which is defined in neural terms rather
than in psychological terms, as is the case with utility. The distinction is a crucial
one, 74 values, PreferenCes anD ChoiCes CH•3 since subjective value obeys
different axioms to expected utility. Glimcher proposes that existing studies support
the hypothesis that subjective values ‘are equal to (or better yet defined as) the
mean firing rates of specific populations of neurons’ and ‘subjective values are
linearly proportional to the blood oxygen level dependent (BOLD) signal of fMRI as
measured in these same populations.’ He also proposes that subjective values
defined in this way have a reference-dependent anchoring point, called the baseline
firing rate. RPE is defined in terms of the difference between forecast and
experienced subjective value. The main implications of this approach are that
subjective value can be measured in cardinal terms (not just in terms of an ordinal
ranking of preferences as with some approaches to utility), and that subjective
values are always consistent with choice on a stochastic basis, again unlike
expected utility. The stochastic basis is necessary because in the Glimcher model of
subjective value there is a noise term or random element in terms of how we
measure value. This is consistent with the approach in terms of economic theory of
Butler and Loomes (2007), who emphasize the role of imprecision in causing
preference reversal. It is important to note that Glimcher does not reject outright
the expected utility model of traditional economic theory; he agrees that it can be a
very useful predictor of choice in many situations. Instead, his approach claims that
neuroeconomic insights can guide the development of economic models of choice
that are better predictors in situ ations where EUT produces anomalies. Glimcher is
also aware of the limitations of his subjective value model. It is essentially a model
of choice, not of well-being. Although subjective value is related to well-being, the
two concepts are not identical; thus maxi mizing well-being is not just a matter of
maximizing subjective value. For example, drug addicts could claim to be
maximizing subjective value at any point in time, but the concept of well-being
involves a longer-term state. Neuroscientists know much less about the neu ral
systems related to well-being compared to choice, but it appears that subjective
value is mediated by these other systems in order to determine well-being.
Reflective and reflexive brain modes There has been a large number of
neuroscientific studies performed that have indicated that the human brain has two
main modes of operation: a reflective mode for making cold, rational decisions like
accepting a job offer or buying a house, and a reflexive mode for making quick
decisions based on instinctive, emotional or visceral impulses, like the ‘fight or
flight’ reflex. This dichotomy is discussed at length in Kahneman’s book Thinking,
Fast and Slow (2012). Stress plays an important part in flipping the brain from reflec
tive to reflexive mode. It causes a release of the neurotransmitter catecholamine,
which reduces the rate of neuron firing in the prefrontal cortex (PFC), the most
recently evolved brain area, responsible for higher cognitive functions.
Simultaneously, the primary sen sory cortices, the amygdala and striatum are
stimulated, leading to reflexive action. From an evolutionary perspective such a
switch is certainly valuable as a survival aid, but it has been shown that excessive
stress over a prolonged period can lead to impaired functioning of the PFC, in turn
causing a deterioration in decision-making (Arnsten, 2015). Conclusions Various
conclusions can be drawn regarding the neural process of valuation, particularly
from the research conducted since 2000: 1 There are multiple brain components
and systems involved, which interact with each other in a complex and dynamic
way. 2 Different stages in this process recruit different striatal components. 75 PT•
II FOUNDATIONS 3 The brain responds differently during anticipation of incentives
than in response to incentive outcomes, an indication of reference dependence. 4
Processing of gains does not appear to be the opposite of processing of losses.
However, even the most ambitious researchers are generally agreed that, in spite of
sub stantial recent progress, much more work needs to be done in order to clarify
the neural anatomy and physiology involved in the functions of value measurement.
Further aspects of neuroeconomic studies of utility as far as they relate to risk and
uncertainty, intertemporal preferences, and social preferences, will be examined in
later chapters. It will also be seen in these later chapters that a vital aspect of
understanding neuroeconomic processes relates to the fact that they have evolved
over millions of years and are therefore largely shaped by the forces of natural
selection as they have operated in an environment very different from the one in
which most humans now function. 3.5 Broadening rationality Criteria for rationality
Now that the basic parameters of the standard model have been described, and
evolution ary and neuroscientific factors discussed, it is possible to consider criteria
for rationality. In the light of the discussion of rationality in Chapter 1 and at the
beginning of this chap ter, a particularly useful view of rationality involves the
following four criteria: 1 Attitudes and preferences should adhere to the basic rules
of logic and probability theory. 2 Attitudes and preferences should be coherent. 3
Attitudes and preferences should not be formed or changed based on immaterial or
irrelevant factors. 4 Attitudes and beliefs should not be incompatible with empirical
observations known to the individual, including their own conscious actions. The
first three of these criteria have been outlined and utilized by Shafir and LeBoeuf
(2002), and in conjunction with the fourth one they are useful in discussing the
different types of irrational behavior that are then described. Before moving on to
this, however, it is useful to note various similarities between these criteria and the
axioms of consumer preference described earlier in the chapter. In particular the
first three criteria above relate closely to completeness, transitivity, independence,
monotonicity (or dominance) and invariance. The one exception is the final criterion
for rationality, relating to compatibility with empirical observations. This is not
explicitly covered by either EUT or the scheme of Tversky and Kahneman (1986)
relating to cancellation, transitivity, dominance and invariance. It has to be said that
this is a controversial addition. For example, it does not coincide with Baumeister’s
description of irrational behavior as being self-defeating. People could violate this
fourth criterion, but still not necessarily engage in behavior that was self-defeating.
This aspect is further discussed in the next chapter, in the context of causes of
irrational behavior, in relation to cognitive dissonance. However, the inclusion of the
fourth criterion can be justified by comparing it with the second one. The second
criterion for judging attitudes and beliefs is whether they are coherent, in the sense
of being consistent with each other. This criterion can be regarded as relating to
internal consistency. The last criterion for judging the rationality of beliefs relates to
external consistency, meaning whether they are supported by empirical evidence
76 values, PreferenCes anD ChoiCes CH•3 known by that individual. An individual
may hold a set of beliefs that are mutually con sistent, but are contrary to known
empirical evidence. In this case it may be claimed that the person is acting
irrationally by clinging on to such beliefs. This aspect of rationality is the one
involved in the commonly used economic concept of ‘rational expectations’, and it
was also implied in Alan Greenspan’s often-quoted expression ‘irrational
exuberance’, in referring to the stock market boom in 1996. It is relevant that the
empirical evidence must be known by the individual, for otherwise the fault is
ignorance, not irrationality. However, it is not relevant that a person may not have
any ‘free will’ in choosing what to believe; such a choice can still be regarded as a
decision-making action. Of course this kind of phenomenon, which we will see may
be extremely widespread, raises the issue of why people would cling on to their
beliefs in the face of contrary empirical evidence. This aspect is again discussed in
the next chapter. We can now discuss aspects of attitudes, values, choices and
preferences that deviate from these criteria. This entails a consideration of both
anomalies and incompleteness in the standard model. As explained in the first
chapter, it is important to distinguish between these two problems. The first relates
to deviations, or situations where the standard model makes inaccurate predictions,
while the second relates to situations where the standard model has nothing to say,
making no predictions at all. As far as the first area is con cerned, DellaVigna (2009)
describes three main types of deviation related to: self-control; reference
dependence; and social preferences. However, many of these deviations involve
factors discussed in later chapters. Self-control aspects are mainly discussed in
terms of intertemporal choice in Chapters 7 and 8; many aspects of reference
dependence involve risk and uncertainty and are discussed in Chapters 5 and 6;
and social preferences are discussed in Chapter 10. The second area of alleged
weakness is more controversial, since some economists argue that these areas of
‘silence’ do not represent a weakness at all; as we have also seen in the previous
chapter, it is claimed that economics has no interest in such areas. This issue is not
a matter of testing a theory by empirical evidence; it involves a subjective value
judgment regarding what economics should be concerned about. The first area
discussed below involves one such area. Happiness is a three-act tragedy Perhaps
the most obvious weakness in terms of omission concerns the nature of happi ness.
It might be intuitively surmised that the objective of expected utility maximization
would involve the concept of happiness, although, as we shall see later in the
chapter, there are complications here in terms of reasons why happiness might not
correspond to utility (Clark, Frijters and Shields, 2008). The evolutionary
psychologist and psycholin guist Steven Pinker (1996) has described happiness as a
three-act tragedy. This claim has the following elements: 1 Happiness involves an
interpersonal comparison of one’s perceived well-being, or subjective well-being
(SWB) with that of others. Although self-reported happiness appears to increase
sharply with income at any point in time (Easterlin, 2001), studies in the USA (Myers
and Diener, 1995), Japan (Easterlin, 1995) and Eastern Europe (Easterlin, 2009)
indicate that self-reported happiness in general has not increased over several
decades, in spite of a several-fold increase in real income in some cases. It also
appears that similar relationships between material wealth and life satisfac tion
occur with children, but only for young males not females (Hudson, 2013); boys
reported greater life satisfaction with more wealth, but less satisfaction with more
peer wealth. A recent German study examined how household consumption was
affected by the consumption of households perceived to be richer than that
household 77 PT• II FOUNDATIONS (Drechsel-Grau and Schmid, 2014). It was found
that a 1% increase in consumption of households perceived to be richer was
associated with a 0.3% increase in a house hold’s consumption. This observation
therefore invokes the notion of a reference point, a key element of prospect theory
that is discussed in detail in Chapter 5. Pinker quotes the words of Gore Vidal: ‘It is
not enough to succeed. Others must fail.’ 2 Happiness also involves an
intrapersonal comparison of one’s perceived well-being with one’s previous well-
being. This again involves reference dependence, but in this case the reference
point is related to a previous self-state rather than a current other state, as in the
previous case. A frequently quoted study regarding this issue is that of Brickman,
Coates and Janoff-Bulman (1978), which found that, after a period of adjustment,
lottery winners were not much happier than a control group, and para plegics not
much unhappier. 3 Happiness and unhappiness are not symmetrical reflections of
gain and loss. The impact of losses is greater than equivalent gains. This
observation invokes the concept of loss-aversion, another key element of prospect
theory. Pinker quotes the tennis star Jimmy Connors: ‘I hate to lose more than I like
to win.’ As discussed in Chapter 2, this weakness does not represent an anomaly of
the NM, since economists generally issue disclaimers as far as the concept of
happiness is concerned, only being concerned with a particular definition of welfare.
In the NM there is much consideration of Pareto efficiency. This measure of effi
ciency is normally expressed that nobody can be made better off without making
anyone else worse off. However, it is normally assumed in the NM that Pareto
efficiency is a desir able goal. As Varian (2006) states: If there is some way to make
some group of people better off without hurting other people, why not do it? (p.
613) The second act of the happiness tragedy indicates why we might not want to
do it. Making some people better off will automatically make others worse off in
terms of happiness, if not in terms of welfare in the narrow sense. This has very
important political implications as far as government policy is concerned. The
growing gap between the pay of bosses and workers in the USA is a case in point. In
general the economic welfare of workers has not suffered over the last 20 years but
they feel worse off knowing that their bosses are so much better off than they are.
Although the disparity is largest in the USA, resentment at ‘fat cat’ CEO pay has
been expressed in many other countries also. Stable, well-formed preferences It is a
fundamental tenet of both the NM and the everyday notion of rationality that people
have stable, well-formed preferences, according to the second criterion related to
coherence described earlier. Without such preferences it would not be possible to
make normative statements in economics since there would be no meaningful
concept for people to optimize. However, going back to the 1970s, several
researchers have found that significant anomalies occur in this respect. Much of this
research centered on experiments involving what are called P-bets and $-bets. A P-
bet offers a near sure thing, meaning a relatively large probability of a modest sum
of money. A $-bet is a riskier gamble, offering a smaller probability of a
considerably bigger prize and a larger chance of zero. Respondents are asked first
to state a preference between the two bets, and most people choose the P-bet since
they prefer a nearly sure thing. However, when respond ents are told that they own
the bet and are asked to state the lowest price at which they are prepared to sell it,
the majority value the $-bet more than the P-bet (Lichtenstein and Slovic, 1973;
Grether and Plott, 1979). Grether and Plott initially attributed this 78 values,
PreferenCes anD ChoiCes CH•3 anomaly to the fact that the studies were laboratory
experiments, lacking in real-world application, mainly because with no stakes there
was a lack of incentives. This is a com mon criticism of behavioral studies which
contradict standard theories. Yet both the studies mentioned above replicated their
findings using real money, the Lichtenstein and Slovic study using a Las Vegas
casino, so the anomaly cannot be explained by a lack of incentives. A recent study
by Isoni and colleagues (2016) attempts to isolate the effects of endogenous and
exogenous price cues on shaping preferences from the effects of market discipline
on revealed preferences by designing an experiment to ‘switch off’ the effects of
market discipline. They find that market discipline does cause shaping to be
reduced, but that substantial shaping effects remain, indicating that market
discipline does not cause preferences to be revealed in an unbiased way, as
predicted in the NM. Desire for consistency More recent research on stable
preferences has revealed some even more startling find ings. A study by Johansson
and colleagues (2013) has found that not only do people often fail to notice
mismatches between their decisions and the outcome of their choice but they also,
in addition, endorse the opposite of their chosen alternative. Subjects were given
the task of choosing which of a pair of faces they found most attractive. Unknown to
them, the experimenters sometimes used a card trick to exchange one face for the
other. The standard model, and indeed our intuition, predicts that most people
would easily notice such a radical change in the outcome of a choice. But that was
not the case: no more than a third of the exchanges were detected by the
participants. The study also included a second round of choices using the same face
pairs, and two stages of post-choice attractiveness ratings of the faces. This
enabled the study to measure prefer ence strength both as choice consistency and
by looking at measures of rating differences between chosen and rejected options.
It was found that the initially rejected faces were chosen more frequently in the
second choice, and the perceived attractiveness of these faces was increased even
in uncoupled individual ratings at the end of the experiment. This finding reveals
another important factor underlying consumer preferences: the desire for
consistency. This desire for consistency means that a previous choice frequently
leads to the same or similar choice in a later situation, in spite of the first choice
being arbitrary, trivial, or non-optimal. The phenomenon can manifest itself in many
different situations. Cialdini (1984) discusses a number of cases, which vary
considerably in context. Examples include: American POWs writing down tenets of
Chinese communism and then sympathizing with the communist ideology; painful
hazing initiation ceremonies increasing the loyalty of members of a group; and
California home-owners signing a petition to keep their state beautiful and then
later allowing big and ugly ‘drive carefully’ signs to be erected on their lawns.
Another aspect of the desire for consistency involves cognitive ease, a recurring
psychological theme in behavioral economics. It simply requires less cognitive effort
to repeat a previous decision, like ordering the same meal from a restaurant menu
as before. This results in status quo bias. However, there is something of a confound
with this effect: the repeating of a previous choice may confer more utility, but it
also may be a result of estimating that the probability of a previous choice being a
good one is higher than if selecting a previously untried option. The existence of
status quo bias has important pol icy implications related to sales strategy and also
government intervention, as discussed in a later section. Other examples of the
desire for consistency involve confirmation bias, the resolution of cognitive
dissonance, commitment to a future action, or the establishment of social status,
and these aspects are discussed in later chapters. 79 PT• II FOUNDATIONS
Anchoring effects In one respect these have a common factor with both of the
phenomena described above. They relate to the effects of prior events which result
in unstable preferences. However, in this case these events do not necessarily
relate to actions taken by an agent, since they can involve any environmental
factor. More specifically, anchoring effects occur when people’s responses are
‘anchored’ to other phenomena in their consciousness, however irrelevant these
might appear to be. An interesting example is given in Nudge, the book by Thaler
and Sunstein (2009): the urinals at Amsterdam airport have houseflies etched into
them. This ‘anchor’ has apparently had the effect of reducing spillage by 80%. Sev
eral studies have also found some apparently very strange results here, in
particular two papers by Ariely, Loewenstein and Prelec (ALP) in 2003 and 2006. For
example, in one experiment a group of students was asked as a preliminary
question to write down the last two digits of their social security number (essentially
a random number between 00 and 99). They were then asked to value half a dozen
different products, including a box of chocolates, two different bottles of wine, a
cordless trackball, a cordless keyboard and a design book. The results showed
remarkable consistency in the sense that the students with higher-ending social
security digits valued all the products more highly. Those in the top 20% (from 80 to
99) bid highest, and the difference between their bids and those of the lowest 20%
(from 00 to 99) varied between 216% and 346%! Ariely, Loewenstein and Prelec
(2003) have explained this phenomenon in terms of a theory of arbitrary coherence.
This is described as follows: ‘valuations of goods and experiences have a large
arbitrary component, yet, after one valuation has been made, people provide
subsequent valuations that are coherent in the sense that they are scaled
appropriately relative to the first.’ The coherence aspect can be explained in terms
of another experiment which was conducted in the same study. In this case the
students were divided into two groups, one of which was initially asked if they were
prepared to pay $2 to listen to a poetry recital, while the other group was asked if
they were prepared to accept $2 to listen to the poetry recital. Only 3% were
prepared to pay, while 59% were willing to be paid the $2. However, when both
groups were asked if they would listen for free the proportion of the first group rose
to 35%, while for the second group the proportion fell to 8%. This demonstrates the
anchoring effect of the first question. It also illustrates two other factors. First, it
demonstrates an expectations effect, whereby people expect more value or utility
when they are cued with a question involving willingness to pay, whereas they
expect disutility if they are cued with a question involving willingness to accept. Sec
ond, it demonstrates that both groups are displaying a normal downward-sloping
demand curve, as expected by standard economic theory. The coherence aspect is
demonstrated further by responses to another question in the experiment, which
asked how much each group was willing to pay/be paid to listen for different
periods, of one minute, three min utes, and six minutes. The responses were
consistent again with economic theory, in that they showed that the paying group
were willing to pay more to listen for longer periods, while the paid group wanted to
be paid more to listen for longer periods. The conclusion of the study regarding
coherence was that in spite of the arbitrariness of the initial anchor, demand curves
would still be the normal downward-sloping shape and stable. A number of other
empirical studies have found evidence of anchoring effects in various different
contexts. For example, a study by Guéguen and Jacob (2013) found that handling
money at an ATM had the effect of causing people to be less helpful to others
immediately afterwards, either in terms of participating in surveys or prompting
them to pick up an item that they appeared to have accidentally dropped. Another
study found that the anchoring effect was context-dependent (Dogerlioglu-Demir
and Kocas, 2014). This applied to anchor numbers which can be interpreted in
different units depending on the context. For example, ‘99’ can be perceived as
$.99 in the context of a hamburger 80 values, PreferenCes anD ChoiCes CH•3 price
(cheap), or as $99 in the context of a meal (expensive). Thus different anchors may
be more or less successful in increasing sales depending on context. Many recent
researchers have endorsed the conclusions of the above studies regard ing the
importance of anchoring effects as an anomaly in the standard model of EUT
(Kahneman and Sugden, 2005; Bernheim and Rangel, 2007, 2009; Beshears et al.,
2008; Fehr and Hoff, 2011), suggesting necessary modifications. However, as
discussed in the previous chapter, there has been some criticism of the anchoring
studies in terms of their conclusions regarding the importance of the anchoring
effect. Maniadis, Tufano and List (2014) were unable to replicate the findings of ALP
in some of their experiments using the same protocol. These authors suggest a
number of possible reasons for this, and caution that the anchoring effect, while still
real, may not be as prominent or ubiquitous as some researchers have suggested.
Framing effects These effects, which are discussed further in the next two chapters,
are one of the most important phenomena in behavioral economics, violating the
invariance principle. Numerous studies have found that people’s responses, in
terms of values, attitudes and preferences, depend on the contexts and procedures
involved in eliciting these responses. For example, when subjects have been asked
to rate their overall level of happiness, their responses have been influenced by a
prior question regarding the number of dates they have had in a recent time period.
While this can be seen as an example of an anchoring effect, it might be considered
that the prior question in this case is relevant to the second; however, if the first
question influences the answer to the second, it is an example of pro cedural
invariance, and therefore a violation of the standard model of EUT. Framing effects
are particularly important since they account for a high incidence of preference
reversal (Slovic and Lichtenstein, 1983; Tversky, Slovic and Kahneman, 1990). This
phenomenon relates to situations where people favor option A when a question or
problem is posed or framed in one way but favor option B when the same problem
is posed in a different way. Evidence from both the field and from experiments
suggests that framing effects are widespread, occurring in many different
situations. For example, there is evidence that products are evaluated more
favorably, and chosen more frequently, when the surrounding environment contains
more perceptually or conceptually related cues (Berger and Fitzsimons, 2008).
There is much evidence that people’s eating habits, especially the quantity
consumed, can be influenced by the size of plates, packages or serving bowls used
(Wansink, Just and Payne, 2009), even though people deny this (Wan sink and
Cheney, 2005). Large sizes can increase the amount served and consumed by 15%
to 45% (Wansink, 2006). This principle appears to operate even when cues are not
intrinsically related to the product. An interesting example is that when NASA
landed the Pathfinder spacecraft on Mars in 1997, there was an increase in sales of
Mars Bars, even though the Mars Bar takes its name from its company founder, not
the planet. There is also evidence that framing effects even extend to experimental
economists themselves, at least at a junior level (Gächter et al., 2009). Framing
effects relate to the social and insti tutional context of choice too in many instances,
but theories of social framing are still at a relatively early stage of development in
behavioral economics and sit uneasily with the individualist psychological focus that
dominates the field (see Klaes, 2002, 2008). Menu effects There is another type of
framing effect that is sometimes referred to as a ‘menu effect’, since it refers to
how people choose from several options on a menu, rather than how each option is
described. There are a number of different types of menu effect, which 81 PT• II
FOUNDATIONS all involve different choice heuristics; seven are discussed here and
another two will be discussed in Chapter 6 on mental accounting, since they involve
aspects of risk and uncer tainty or ambiguity. 1 The ‘attraction effect’ This is a
prominent example of a menu effect, and involves the principle of refer ence
dependence mentioned earlier (Huber and Puto, 1983). Ariely (2008) refers to the
effect as a ‘decoy effect’, since it has become a much-used marketing practice. He
opens his first chapter of Predictably Irrational with an example relating to sub
scriptions to the Economist magazine. There are three options on the menu: (1) 1-
year online subscription for $59 (2) 1-year print subscription for $125 (3) 1-year
online and print subscription for $125 Ariely suggests that the second option is
offered merely as a decoy. The underlying psychology is that when offered only
options 1 and 3, consumers are not sure which one to choose, because there is a
trade-off between price and quality, and they may prefer the cheaper and less
profitable option 1. In Ariely’s survey of MIT MBA stu dents 68% chose this option.
However, when all three options are presented, options 2 and 3 are more easily
comparable because they are the same price, and only differ on quality. Since
option 3 offers more in quality, it is said to dominate option 2; thus it is a ‘no-
brainer’ that most people will prefer option 3 to 2, and this comparison will tend to
cause them to prefer this over the first option as well. Thus he found that when
presented with all three options 84% of the students chose option 3, another
example of preference reversal. These effects have significant marketing
implications. The ‘decoy effect’ or attraction effect is a widespread practice in
marketing strategy, offering consum ers decoys that firms do not really want them
to buy. We can generalize about the practice by saying that if a firm has two main
offerings on the menu, A and B, but A is more profitable, the firm can encourage
consumers to buy A by extending its offerings to include a decoy, which we can
label A– (following Ariely, 2008). A– is obviously inferior to A, in terms of being a
dominated choice. Hedgcock and Rao (2009) have proposed that the underlying
psychological explanation for the attraction effect involves trade-off aversion, and
they have con ducted a neuroeconomic study to test this theory. The fMRI
technique was used, and this detected a number of differences in brain activation
when subjects faced a three item menu with a decoy compared with a two-item
menu. In particular these included a decreased activation in the amygdala, an area
of the brain associated with negative emotion, and an increased activation of the
DLPFC, an area of the brain associated with the use of decision rules. The authors of
the study are cautious in interpret ing these results, bearing in mind the problems
of reverse inference discussed in Chapter 2, but suggest that the introduction of a
dominated option in the three-item menu allows people to avoid a stressful
evaluation of a trade-off and instead apply the simple heuristic of choosing the
dominating option. There is evidence that firms are becoming more conscious of
different types of menu effect in their marketing practices. Not surprisingly, fast-
food chains are among the foremost to take an interest. Domino’s Pizza announced
that it has asked its media agency, Arena Media, to integrate behavioral economics
into its planning process (Marketing Magazine, 2011). This actually involves much
broader aspects of marketing than simply menu effects, as will be seen later. Many
fast-food chains are coming under pressure from government agencies to become
more socially respon sible and try to encourage people, particularly children, to eat
healthier foods. This is a very challenging task, given people’s propensity to like
starchy, fatty and sugary 82 values, PreferenCes anD ChoiCes CH•3 foods. Simply
adding healthier options to the menu will not necessarily be effective, especially
given the vicarious consumption effect discussed later in this subsection. Some
more recent studies have suggested that the attraction effect is not as important or
widespread as claimed in earlier studies (Frederick, Lee and Baskin, 2014; Yang and
Lynn, 2014), in particular only applying to situations where all the relevant variables
can be described quantitatively. These studies suggest that most comparisons
involve non-quantitative factors, like quality of hotel bedrooms or res taurant food,
and the Yang and Lynn study was unable to detect attraction effects in more than
11 cases out of 91 attempts. On the other hand Huber and colleagues (2014)
maintain that the attraction effect is still robust, since it can be replicated under
similar experimental conditions. 2 Preference for the salient Evidence suggests that
people simplify complex decisions by choosing a salient option. This may apply to
supermarket shoppers, for example when faced with a large shelf filled with
different brands, although the factor of limited attention (discussed in the next
section) is also important here. In the case of financial markets, Barber and Odean
(2008) show that investors prefer to buy stocks of companies that are cur rently in
the news, even if the news is bad. Being first on a list of options is often a
considerable advantage. In 1981 American Airlines found that travel agents booked
the first flight that appeared on their computer screen 53% of the time, and a flight
that appeared somewhere on the first screen almost 92% of the time. This caused
the airline to manipulate the order of flights as they appeared in the booking
system in order to increase profits, and this ultimately led to complaints from other
airlines and intervention from federal government. This ‘first on the list’ heuristic
also applies in voting in the political arena. Ho and Imai (2008) conducted a study in
California, where the order of candidates on the ballot is randomized, and found
that there was a significant advantage for a candidate in being first on the list. The
advantage was greater for minor party candidates compared with those from major
parties, suggest ing that voters use irrelevant information when they lack other
informational cues. The order of items on a list can be relevant in more complex
ways. For exam ple, Suk, Lee and Lichtenstein (2012) report that when differing
brand options are listed in descending price order, people tend to choose higher-
priced items, while when options are listed in ascending price order people tend to
choose lower-priced items. The authors suggest that the reason for this
phenomenon lies in the association between price and perceived quality for
branded products combined with reference dependence, where consumers compare
items with others earlier on the list. 3 The compromise effect This effect refers to
the finding that people frequently choose intermediate options in terms of price.
This finding has been reported in a number of different settings, such as
contributing to savings plans. A recent study by Pinger, Ruhmer-Krell and
Schumacher (2016) indicates that in a restaurant setting this phenomenon appears
to be quite robust, and persists in the presence of many other influencing factors,
such as size of choice set and opportunity to choose familiar options. 4 Choice
avoidance This is another counterintuitive finding, sometimes referred to as the
paradox of choice. Marketing managers may feel that they are both maximizing
profits and ben efiting consumers by offering them a greater range of choices, but
the end result may be that consumers avoid the choice altogether, which often
means not purchasing any item in the range. For example, Iyengar and Lepper
(2000) compare the behavior of consumers who were offered the opportunity to
taste 6 jams (the simple-choice treatment) with consumers who were offered the
opportunity to taste 24 jams (the 83 PT• II FOUNDATIONS difficult-choice
treatment). They find that, although more consumers stop to sample jams when
there is more choice, substantially fewer actually buy jams (4 compared with 31
customers). Choi, Laibson and Madrian (2009b) report the same paradox in f
inancial decision making, in that a smaller number of investment options increases
participation in a 401(k) plan. Beshears and colleagues (2013) find that offering a
binary choice, with a pre-set contribution level and asset allocation versus the
status quo, increased enrollment in retirement savings plans by 10–20%. Kida,
Moreno and Smith (2010) find a similar effect for inexperienced investors, but the
opposite effect for experienced investors, who were actually less likely to invest
when faced with a limited choice set. What could explain the psychology underlying
the paradox of choice? Evi dence suggests that making complex decisions is
stressful and people may try to avoid this stress. Sagi and Friedland (2007) propose
a theory that post-decisional regret is related to the comparison between the
alternative chosen and the union of the positive attributes of the alternatives
rejected. This of course contradicts the standard model’s view of opportunity cost,
which considers only the positive attrib utes of the next best alternative rejected.
However, the Sagi–Friedland theory does explain the paradox of choice, since a
greater number of alternatives would increase post-decision regret, and the authors
find that their theory was supported by four experiments. A flip side to the above
situation, which again represents a paradox, is that peo ple are averse to single-
option offers. Mochon (2013) finds that the choice share of an option increases
(relative to deferral) when a competing attractive option is added to the menu.
When only a single option is offered, DVD players in this study, people are more
inclined to defer purchase. This may occur because people are more likely to
consider a larger set of alternatives, with more positive attributes, when a single
option is presented, whereas the addition of a single competing option may serve to
limit the alternatives and attributes considered. 5 The momentum effect This effect
occurs when an initial purchase provides a psychological impulse that enhances the
purchase of a second, unrelated product (Dhar, Huber and Khan, 2007). These
authors report experimental evidence that the purchase likelihood for a second
‘target’ item (a key chain) increases with the purchase incidence of an ini tial,
unrelated ‘driver’ item (an educational CD). This increased likelihood was not
caused either by complementarity between the two items or by a consequent reduc
tion in transaction costs. It was proposed that the effect is explained by Gollwitzer’s
(1990) theory of implementation and deliberation mindsets, where an initial
purchase moves the consumer from a deliberative to an implemental mindset, thus
driving sub sequent purchases. The existence of this effect has important marketing
implications, for example related to the use of loss leaders to get people into stores.
6 The vicarious consumption effect A study related to choice of food items has
shown that adding a healthy item to the list of available options has the perverse
effect of causing people to choose less healthy food items than otherwise (Wilcox et
al., 2009). Apparently ‘the mere presence of the healthy food option vicariously
fulfills nutrition-related goals and provides consum ers with a license to indulge’. It
would be interesting to follow up this research and observe if supermarkets that
display fruit and vegetables near the entrance actually sell more of these items. 7
Confusion A final behavioral deviation noted by DellaVigna (2009) concerns
confusion. This does not reflect a preference, but is really an effect of cognitive
failure. Examples 84 values, PreferenCes anD ChoiCes CH•3 include mistaken
trades of stocks (confusing MCI with MCIC), reported by Rashes (2001), and
mistaken voting in elections, where votes are placed for candidates whose names
are adjacent to the intended candidate on the ballot, reported by Shue and Luttmer
(2009). Discrepancies between objective causes and subjective effects There
appear to be discrepancies between objective measures of sources of comfort/
discomfort and reported measures of subjective feelings. Again, many defenders of
the NM would argue that reported measures of subjective feelings are not economic
phe nomena, and that therefore they are of no concern to economists (Gul and
Pesendorfer, 2007). However, when such feelings do or can affect later decisions,
this is of relevance to economics. A study by Redelmeier and Kahneman (1996) of
patients undergoing colonos copies illustrates this phenomenon well. The patients
were asked to report the intensity of current pain on a scale of 0 to 10 at minute
intervals over a period up to 69 minutes. However, those patients who suffered
more pain for longer periods did not necessarily have a worse recollection of the
experience as a whole. Instead it seemed that the most important determinant of
post-experience evaluation was a combination of the maximum pain suffered at any
point and the mean pain suffered during the last three minutes. This f inding has
become known as the peak-end rule. It has been confirmed by later studies on the
basis of evidence from a range of different areas, in the context of both negative
and positive experiences, and also extending to the experience of material goods
(Do, Rupert and Wolford, 2008). More recent research has indicated that the peak-
end rule is not just confined to humans. Egan Brad and colleagues (2016) have
conducted experiments involving capu chin monkeys as well as humans where
subjects could select the sequence of food intake, f inding that capuchins also
experience peak-end effects. The authors of the study derive two important
conclusions from their experiments. First, the fact that the peak-end effect is found
in a species that diverged from the human one 35 million years ago is an indicator
that this effect, and many other decision-making biases discussed in behavioral
economics, ‘do not necessarily emerge as a result of specific cultural experiences
and human-specific selective pressures—instead, our human biases might be the
result of evolved cognitive strategies or the result of shared cognitive limitations,
ones present in our primate lineage for considerable phylogenetic time.’ The second
main conclusion of the study relates to policy implications, and involves a further
anomaly as far as the standard economic model of rationality is concerned: both
capuchin monkeys and humans seem incapable of learning how to implement the
peak end rule to maximize hedonic payoffs. This issue is discussed further in the
section on policy implications, in terms of saving the best to last. Expectations
effects Another problem with the NM concerns the effect of expectations. There is
some evi dence that high expectations of happiness can lead to disappointment.
This emerges in particular from the study by Schooler, Ariely and Loewenstein
(2003) regarding peo ple’s plans for the millennium celebration of 2000. Those
people who spent the most time, effort and money tended to be the least satisfied.
It seems that the reference point phenomenon is again relevant here. Of course in
this case it can be argued that the disap pointment, or lower utility, after the event
may be more than offset by the higher utility associated with the anticipation of the
event. This leads to a consideration of a further related omission from the NM,
relating to anticipatory utility. This issue is discussed fur ther in the next section,
but the essential point is that anticipation of pleasure can itself 85 PT• II
FOUNDATIONS be pleasurable, with the result that people may defer the
pleasurable experience in order to prolong the anticipatory utility. A study by
Richins (2013) entitled ‘When wanting is better than having’ finds that materialistic
people experience a hedonic high before pur chase, due to the expectation that a
particular product will transform their lives in some significant manner, and after
purchase there is a hedonic decline. Such a pattern in utility was not found to occur
with people low in materialism. It is this anticipatory factor that may at least partly
explain the saying that revenge is a dish that is better served cold (it is also better
for being planned rather than hastily delivered in an emotional state, as will be
explained shortly). There is another aspect of expectations effects, which is
described well in Ariely’s book Predictably Irrational (2008) with the Chapter 9
subtitle ‘Why the mind gets what it expects’. The examples he reports all show that
the utility that we derive from activities or consumption depends on our
expectations, as does our behavior in general. It is worth considering three
examples here to give a flavor of the phenomenon involved. Bargh, Chen and
Burrows (1996) performed an experiment which showed that after a group of
subjects were given a word-unscrambling task involving words related to the
concept of the elderly, like ‘Florida’, ‘bingo’ and ‘ancient’, their walking speed was
substantially slower than a control group that had not been primed with such words.
This experiment therefore indicates the importance of subliminal messages in
prompting expectations, as well as the effect of expectations on behavior. We shall
return to this at the end of the chapter in the discussion of policy implications. A
second example of expectations effects also has important policy implications,
related to pricing. A study by Waber and colleagues (2008) involved using a placebo
for reducing pain. Subjects were administered electric shocks in two consecutive
treatments, but given a ‘drug’ purported to be a painkiller before the second
treatment. Sure enough, the subjects reported less pain in the second treatment
compared to the first, even though the ‘drug’ was actually a vitamin C capsule. The
interesting point was that in a following test the subjects reported a very different
response according to the advertised price of the pill; at a price of $2.50 almost all
the subjects experienced pain relief, but at a discounted price of 10 cents only half
of them did. Whereas the study described above involved examining self-reports of
pain, a subjec tive measure, another study by Shiv, Carmon and Ariely (2005)
examined the behavioral effect of different prices in an objective way. This study
involved the use of an energy drink, SoBe Adrenaline Rush, in order to see if the
price paid had an effect on actual performance. Performance in this case was
measured in terms of the ability to solve anagram word puzzles. It was found that
not only did the student subjects drinking the regular-priced beverage report less
fatigue subjectively, they were also able to perform significantly better on the word
puzzles than a group who drank the same beverage dis counted to only about a
third of the regular price (although no better than a control group who did not
consume the drink). Again, the policy implications of these findings will be
discussed later. Another area where expectations effects are important is with the
performance of branded products. There are conflicting findings here. A study by
Garvey, Germann and Bolton (2015) concludes that the use of status brands
improves performance by consum ers, as might be expected from the findings of
the studies described above. There was a twist to this outcome: the consumers took
the credit for the improved performance them selves; this is an interesting example
of overconfidence, a phenomenon discussed in more detail in the next chapter. The
authors speculate that the psychology underlying this effect is that the status of the
brand improves self-esteem, which in turn reduces task-induced stress, which then
leads to an improvement in the performance of many consumers. How ever, the
findings of this study conflict with those of another study by Gosline, Banker and
Lee (2013). This study finds that, although consumers may rate branded products
more 86 values, PreferenCes anD ChoiCes CH•3 highly and be prepared to pay
more for them, their performance may be lower. In this case the authors speculate
that consumers rate their performance relative to other users, and that using the
abilities of high-status brand consumers as a reference point can lead to low
expectations regarding one’s own abilities. The jury is therefore still out regarding
the effect of status brands on performance, and there appear to be a number of
mitigating fac tors involved, in particular the preexisting beliefs of the consumer
regarding self-efficacy in the relevant domain. Addiction and abstention In the
conventional model it is assumed that more consumption of a good gives more total
utility. The NM does take into account ‘bads’, like garbage or pollution, where more
con sumption decreases total utility, but these are phenomena where increasing
consumption is monotonically bad, meaning increasingly bad throughout the range
of consumption. For some people, however, there are goods that give too much
pleasure, and excessive consumption is associated with various problems in terms
of health, and time and money spent. Addiction is a major factor in this context. The
phenomenon of addiction can cover a wide range of goods: alcohol, tobacco and
other recreational drugs are the most com monly cited examples, but one can also
include food in general (or particular types of food like junk food), gambling, sex,
computer games and indeed any activity involving a significant degree of
excitement. The psychological and physiological mechanisms relating to enjoyment
of these goods are complex, being of a double-edged nature. However, one factor
that deserves mention at this point is the concept of diagnostic utility. A number of
studies have found that people infer their happiness from their actions in a self-
signaling manner (Campbell and Sawden, 1985; Elster 1985a, 1989; Bodner and
Prelec, 1997, 2001). The last study quotes as an example a person who takes a
daily jog in spite of the rain, who may view that activity as a gratifying signal of
willpower, dedication or future well-being. Bodner and Prelec continue: ‘For
someone uncertain about where he or she stands with respect to these dispositions,
each new choice can provide a bit of good or bad “news”.’ One impli cation of this
concept is that people who fear that they may be, or may become, addicted to a
good may be better off, and feel themselves better off, abstaining from
consumption completely. To indulge even slightly may reveal themselves to indeed
have an addictive personality, and the acknowledgment of such weakness may
make it impossible to break the addiction. These concepts of diagnostic utility and
self-signaling may well help to explain the unfortunate and all-too-common
phenomenon of ‘falling off the wagon’, with its vicious circle of low self-esteem and
compensatory indulgence. The issue is discussed in more detail in Case 3.2.
Endowment effects These effects are discussed in more detail in Chapter 5, but the
essence of the phenom enon is that utility is not independent of possession. Those
people who have acquired a good in some way, through either purchase or gift,
tend to value it more highly than others. Some researchers, notably Plott and Zeiler
(2007), object to the use of the term ‘endow ment effect’ since it suggests a
particular theory for the phenomenon to be explained, and prefer to use the term
‘exchange asymmetry’ for the phenomenon itself; this is an important distinction,
since we shall see that there is some controversy regarding theories underlying the
effect. The effect was first noted in a study by Knetsch (1989). He performed an
experi ment which divided subjects randomly into three groups: one group was
endowed with a mug, another with a candy bar and the final group was not
endowed with either. 87 PT• II FOUNDATIONS The first two groups were allowed to
trade their good for the other one, while the third group was offered the choice of
either good. If preferences were independent of endowment one would expect the
proportions favoring one good over the other to be the same for each group.
However, while 56% of the third, non-endowed, group favored the mug over the
candy bar, 89% of the mug-endowed group preferred the mug, being unwilling to
trade, and only 10% of the candy-endowed group preferred the mug, wanting to
trade. The main psychological factor underlying endowment effects was originally
claimed to be loss-aversion, discussed in Chapter 5 in the context of prospect
theory. More recent explanations examine the role of the possession-self link
(Dommer and Swaminathan, 2013) and self-threat involved with sale (Chatterjee,
Irmak and Rose, 2013). We will also see that the evidence regarding endowment
effects is mixed (List, 2004; Plott and Zeiler, 2005, 2007; Knetsch and Wong, 2009).
Attention We have already encountered the concept of bounded rationality in the
previous chapter, seeing that people tend to simplify complex decisions by using
heuristics; one implication is that they process only a subset of the information
available. The size and type of this subset depends on the importance of the
decision, the salience of signals relating to the decision, and the number of
competing signals. There is evidence of inattention in sev eral field studies in
different situations. DellaVigna (2009) states that limited attention helps explain
neglect of (1) shipping costs in eBay auctions; (2) non-transparent taxes, like
indirect state taxes not included in the price; (3) complex information in rankings,
like those of hospitals and colleges; and (4) earnings news, especially before
weekends, on days with more competing news, news related to linked companies,
or news related to events several years in the future. Related to evidence of limited
attention is the find ing by Frederick and colleagues (2009) that consumers often do
not consider alternative purchases and the opportunity costs involved in purchase.
Malmendier and Lee (2011) f ind dramatic evidence of this in online auctions, where
for a particular item (a board game) auction prices exceeded the fixed price for the
same item on the same web page over 40% of the time. Paradoxically, there
appears to be another effect related to attention that acts in an opposite manner to
the one described above. There is some evidence that suggests that excessive
attention, in terms of deliberating over a choice, can result in a poorer deci sion; for
example, preference consistency has been found to be reduced (Nordgren and
Dijksterhuis, 2009). This may be caused by the confusion factor described above.
3.6 Visceral factors Nature There is a general consensus among behavioral
scientists that emotions are evolved tools for dealing with the challenges of human
life. The term ‘emotion’ refers to ‘some change in subjective experience, autonomic
responses (e.g. heart rate, respiration, electrodermal activity), physical action (or an
increased likelihood to perform an action, such as facial muscle movements,
skeletal muscle movements), as well as some perception, thought, or judgment of
the surrounding world’ (Lindquist et al., 2013). Thus emotions involve both a
psychological dimension and an underlying physiological state. There is much
evidence that emotions like anger, fear, joy, surprise, anxiety, envy and pity affect
our behavior in 88 values, PreferenCes anD ChoiCes CH•3 significant ways. We
have already seen that ‘drives’, like hunger, thirst and sex, as well as cravings and
pain, affect behavior. Psychologists tend to use the term ‘visceral factors’ to refer to
the combination of all of these feelings. At sufficient levels of intensity, these feel
ings tend to cause people to behave in ways contrary to their long-run self-interest,
often with the full awareness that they are doing so. This behavior violates various
aspects of the standard model: (1) people do not maximize any kind of function; (2)
people miscalcu late probabilities; (3) people do not discount exponentially, at a
constant rate; (4) people measure utilities inconsistently. In this chapter we are
primarily concerned with the first and fourth aspects, while the other aspects are
discussed in later chapters. Salience Visceral factors like drives tend to be recurring
states, which increase in intensity until they are assuaged, when they temporarily
fall to a low level before rising again. This inev itable roller-coaster ride has a
number of important implications for decision-making. First of all, as they increase
in intensity, they narrow our attention onto the satisfaction of the drive, making it
salient. Starving people become obsessed with food, prison inmates become
obsessed with sex, drug addicts become obsessed with getting a ‘fix’; all other
desires fade into obscurity. In economic terms the MRS between the desired object
and other goods approaches zero. Furthermore, the increase in intensity also
focuses attention on the present at the expense of the future, causing a lapse in
self-regulation, at least as far as that particular visceral factor is concerned. Short-
sighted decisions tend to occur in such circumstances as future consequences are
ignored. These aspects are examined in Chapter 8, related to hyperbolic
discounting and modifying instantaneous utility. Another kind of narrowing of
attention occurs as individuals experiencing intense levels of visceral factors tend to
become more selfish. They are less likely to cooperate with others, unless they see
that such cooperation is likely to achieve the satisfaction of their needs. We have
already discussed a relatively mild example of this in the context of anchoring
effects: simply handling money, as in visiting an ATM, has the immediate effect of
reducing cooperation, in terms of not being willing to participate in a brief sur vey,
or failing to prompt others to pick up an item that they appeared to have
accidentally dropped. The other side of this coin is that people are also more likely
to ‘defect’, in game-theoretic terms. For example, under duress like interrogation
they are more likely to betray friends and family. In moderation visceral factors tend
to prompt us to take sensible actions, but as they increase in intensity and narrow
our attention excessively they tend to cause us to make self-defeating choices, as
Loewenstein (1996) has noted. Thus extreme fear may produce a panic that causes
people to ‘freeze’ rather than adopting a more healthy ‘fight or flight’ reaction
(Janis, 1967). Likewise, extreme anger can result in impulsive and destructive
behavior that is regretted soon afterwards. Sometimes even moderate anxiety may
cause us to make irrational decisions, by focusing our attention unduly on certain
factors. For example, it has been observed that earthquake insurance purchases
rise after earthquakes, when the objective probability is probably at a low-point
(Palm et al., 1990). Similarly, purchases of flood and earthquake insurance are
influenced more by whether friends have experienced the event than by the
experience of one’s immediate neighbors, even though the experience of neighbors
should provide a better guide to the probability of experiencing flood or earthquake
(Kunreuther et al., 1978). We tend to give friends more attention than neighbors.
Thus the problems relating to salience involve both the estimation of utility, or
disutility, and to the calcula tion of subjective probabilities, an issue discussed in
the next two chapters. 89 PT• II FOUNDATIONS Consciousness Visceral factors,
particularly when they are intense, tend to affect behavior directly, with out any
conscious deliberation process (Bolles, 1975). Brain centers are activated, either
chemically or electrically, which often bypass conscious or cognitive mediation, and
action results. This mechanistic language indicates once again the irrelevance of
volition as far as at least some behavior is concerned. An extreme example is
people falling asleep at the wheel; nobody makes a conscious decision to do this,
but an intense desire for sleep can override the instinct for survival in this case. The
extreme sensitivity of the brain’s pleas ure centers to stimulation is a vital factor in
understanding drug addiction. It has long been known that laboratory animals will
continue to administer electrical stimulation to pleasure centers, in preference to
food, water and sex until the point of collapse and even death (Olds and Milner,
1954). This does not imply that we are not conscious of the operation of visceral
factors, or that we are not able to regulate them in terms of their effects. Regulation
is particularly important when the visceral factor changes in intensity gradually, like
the onset of hunger, and this has important policy implications in terms of well-
being that are considered later in the chapter. Impulsivity Visceral factors also play
an important role in influencing impulsivity. In general terms we tend to think of
impulsivity as relating to situations where people depart from prior decision plans.
Usually the departure is prompted by some trigger factor, and this has an
immediate effect. Impulsivity is often explained in terms of non-constant
discounting, dis cussed in Chapter 8, but there are certain aspects that cannot be
easily explained in such terms. These aspects relate particularly to the effects of
visceral or emotional states such as hunger, sexual desire, anger or fear, which are
frequent causes of impulsive behavior. We will see that modification of the
instantaneous utility function may be necessary in order to account for such factors.
We can also consider impulsivity in terms of the distinction drawn by Loewenstein
(1996) between actual and desired value, which parallels the distinction of
Kahneman, Wakker and Sarin (1997) between predicted and decision utility
discussed in the next section. As the intensity of the relevant visceral factor
increases, this increases the desired value (or decision utility), and thus the
discrepancy between actual and desired value, increasing the probability of
impulsive behavior. Another way of describing the situation is that we want
something more than we expect to like it. This distinction between wanting and
liking is important in a number of different contexts. For example, a smoker may
want to give up, meaning he has the motivation to give up, but does not like giving
up, because he gains no hedonic pleasure from giving up. It has been found that
depressive people want self-esteem, but do not like pursuing it; similarly, they have
been found to want, but not like, alcohol and friendship (Bushman et al., 2012). The
implications of this in terms of different types of utility are further discussed in the
next section. Effects of environmental factors The intensity of visceral factors, and
the resulting impulsivity, are also strongly influenced by situational factors.
Temporal or physical proximity, or sensory contact (sight, sound, touch or smell),
can elicit visceral cravings. In a series of experiments carried out by Mis chel (1974)
and Mischel, Shoda and Rodriguez (1992) children were placed in a room by
themselves and taught they that they could summon the experimenter by ringing a
bell. They would then be shown a superior and inferior prize and told that they
would receive 90 values, PreferenCes anD ChoiCes CH•3 the superior prize if they
could wait successfully for the experimenter to return. One main f inding was that
the children found it harder to wait for a delayed reward if they were made to wait
in the presence of either one of the immediate or delayed reward objects. This
finding is particularly important since it provides evidence for the visceral factor
theory as against the non-constant discounting theory. According to the latter
children should be more willing to wait in the presence of the superior delayed
reward. Visceral cravings may even be relevant in decision-making situations where
the relevant drive or emotion may not be directly related to the decision. For
example, people showing their homes to prospective buyers may do well to bake
bread or cakes beforehand to create a more ‘homely’ environment, even though
buying a home is not normally considered to be an impulsive purchase. There is
substantial evidence that environmental factors like the above can significantly
affect mood and thus behavior, representing another type of anchoring effect. The
weather is a good example. People tip more at restaurants on sunny days (Rind,
1996). There tends to be a negative rela tionship between cloud cover and
aggregate stock returns (Saunders, 1993; Hirshleifer and Shumway, 2003).
International soccer matches also adversely impact daily stock returns for the losing
country (Edmans, Garcia and Norli, 2007). Simonsohn (2010) found that students
were more likely to enrol at colleges known for their academic rigor when they
visited on cloudy, rather than sunny, days. The author suggested that this was
because on sunny days the opportunity cost of studying was more salient, since
other outdoor activities were more likely to be considered. Likewise, temperature
also has significant effects on emotional response. Bruno, Melnyk and Völckner
(2017) con ducted an experiment that found that physical coldness enhances
consumers’ attitudes towards emotionally warm ads, while physical warmth
decreases consumers’ responses to emotionally warm ads and instead enhances
attitudes towards emotionally cold ads. This finding has policy implications for
advertising themes being linked to seasonal and geographical factors. In some
cases the environmental factors that affect behavior are very general. For example,
it has been shown that investor sentiment is reflected in the tenor of status updates
on Facebook, so that daily stock returns are correlated with the Facebook Gross
National Happiness Index (FGNHI). Using data from millions of such status updates
on an international basis, Siganos, Vagenas-Nanos and Verwijmeren (2014) were
able to show a positive causal relationship, with returns following upticks in
sentiment in the short term, but then reversing and returning to normal in the long
term. This trend essen tially is the same as predicted by De Long and colleagues
(1991), when financial markets have a large population of ‘noise’ traders. It is also
noteworthy that the effects of these environmental factors tend to be reduced when
people’s attention is drawn to them. Affective forecasting We tend to be poor at
predicting visceral states. For example, we tend to overestimate how long our
current emotional state, whether happy or sad, angry or fearful, will last. This also
applies to hunger; hence the advice: don’t go to the supermarket when you are
hungry. Furthermore, when the visceral drive has been satisfied, we tend to
underestimate its strength in the future; for example, we think we will be able to
control our hunger pangs and not eat all the junk food we just bought in a big binge.
Meyvis, Ratner, and Levav (2010) report five studies that indicate that we tend to
mispredict our emotional states, and then misremember them; these errors and
biased recall involve events such as a Superbowl loss, a presidential election and an
important purchase. The investigators propose that the biased recall is caused by
an anchoring effect related to the subject’s cur rent emotional state. However, the
consequence of this is that we fail to learn from past mispredictions, and there is a
compounding of errors over time. 91 PT• II FOUNDATIONS Theoretical foundations
Some psychologists have proposed general theories relating to the relationships
between emotions and decision-making. A prominent recent example is the
Appraisal-Tendency Framework (ATF) of Han, Lerner and Keltner (2007). This theory
addresses how and why specific emotions carry over from past situations to
influence future judgments and choices. It differs from previous theories using a
valence-based approach, which assumed that positive moods had typical effects on
judgment and decision-making, which con trasted with the typical effects of
negative moods. Han, Lerner, and Keltner, following Lerner and Keltner (2001),
demonstrated that two mood states such as fear and anger can lead to different
judgmental effects, even though both have negative valence. They also f ind that
other negative emotions do not have identical effects on decision-making. Dis gust
reduces buying prices or willingness to pay (WTP) whereas sadness increases them.
The effect is reversed for selling prices, or willingness to accept (WTA). Construal
level theory (CLT) has also been frequently applied to explain differ ent emotional
responses. CLT proposes that people think about objects and events on a dimension
of concreteness or abstractness according to their psychological distance. This
distance can be temporal, spatial or social. Anger, for example, tends to be directed
towards specific targets, and therefore involves a more concrete construal level
compared with shame, which is a more general attitude, and therefore more
abstract. The signifi cance of this distinction is that, while anger and shame are both
negative emotions, angry people are more likely to hold on to prior preferences,
while shameful people are more accepting of new information and prone to change
(Han, Duhachek and Agrawal, 2014). Similarly, sad people, having a more concrete
construal level, tend to be more inclined to choose high-risk/high-reward options,
while anxious people, with a more abstract con strual level, tend to choose
low-risk/low-reward options. A strong negative emotion that has already been
mentioned earlier in the context of different responses is fear. People commonly try
to use fear as a means of persuading others to take specific actions because of its
intensity. However, this strategy can eas ily backfire. For example, anti-smoking
campaigns often make use of cancer statistics or even shocking pictures and
photos, as in Australia. There is much controversy regarding whether government
policy has been effective here, but research has shown that people tend to erect
strong psychological defences when presented with disturbing and disgust ing
information and images. The emotion of disgust tends to lead to avoidance for
sound evolutionary reasons. This issue is discussed further in the next chapter in
the context of the theory of cognitive dissonance. However, one further research
finding is relevant here. It has been found that mixing humor with fear can be
effective in advertising, since humor can reduce the fear tension, preventing
consumers from building defences against the advertising message (Mukherjee and
Dubé, 2012). Regret can also be considered as a visceral factor and regret theory
has an important role in consumer choice. This is discussed in more detail in
Chapter 5, but one aspect can be touched on here, relating to the opportunity cost
of choice and counterfactuals. If we decide to buy an item, we may later suffer
regret in the form of buyer’s remorse. On the other hand, if we decide not to buy
than we may also suffer from regret, this time in the form of missing an opportunity.
One study has suggested that when people buy a material good the first type of
regret (from action) is more important, while when people buy an experience then
the second type of regret (from inaction) is more relevant (Rosenzweig and Gilovic,
2012). The authors contend that not only do people tend to get more lasting
satisfaction when they buy experiences rather than material goods, but also they
suffer less from regret. This is because there are more obvious substitutes for
material goods, whereas experiences tend to be more unique, making comparisons
more difficult, and the opportunity cost is less salient. 92 values, PreferenCes anD
ChoiCes CH•3 Conflict can also cause negative emotions, such as depression and
anxiety. Some past research has indicated that these emotions result in poorer
decision-making (Shah and Kruglanski, 2002) or choice deferral (Iyengar and
Lepper, 2000). Anxiety may cause us to ignore certain new information, or not seek
it; for example, we may choose not to visit a doctor when we have a significant
medical symptom. This may increase our short-term utility, but cause us to
misestimate probabilities. However, a recent study has suggested that conflict can
aid decision-making, since a conflict mindset can lead to more systematic
processing using more available information (Savary et al., 2015). The flip side of
the above situation is that a state of happiness may cause decision making to
become more difficult. When choosing between various options the tendency is to
focus on differences or unique qualities. A positive mood tends to increase this
focus, making the choice more difficult (Pocheptsova et al., 2015). Positive emotions
like pride and compassion also have different effects on consump tion when
bandwagon effects are involved. This has policy implications, since undesirable
social tendencies like teenage binge drinking may be reduced by appealing to
people’s pride, or desire to be different. 3.7 Types of utility In order to shed light on
the different aspects of judgment and decision-making described above we now
need to examine more closely the concept of utility. This involves a discus sion of
the concept’s evolution over time, the issue of measurement, and the different
types of utility that are relevant in decision-making. Historical evolution The concept
of utility is one of the most basic building blocks in economic theory. In par ticular it
underlies the theory of consumer choice. In this chapter, however, we will only be
concerned with riskless choice; the effects of risk and uncertainty will be discussed
in the next two chapters. The assumption that the objective of consumers is
expected utility maximization is the most fundamental single component of the NM,
dating back to Jeremy Bentham [1789](1948). It should be noted that, strictly
speaking, the term ‘expected’ implies an ele ment of risk or uncertainty, the
discussion of which is deferred until the next two chapters. In Bentham’s original
usage of the term utility referred to the experiences of pleasure and pain, which
‘point out what we ought to do, as well as what we shall do.’ Thus utility has a
hedonic characteristic, which later researchers, notably Kahneman (2000), refer to
as experienced utility. We shall see later in this section that the meaning of the
term utility has changed since Bentham’s time, and is now regarded as outmoded
by many economists, who tend to favor the concept of decision utility. As discussed
in more detail in Chapter 5, this meaning of utility refers to the weight assigned to
an outcome in a deci sion, and is revealed by people’s choices. It is this revealed
preference meaning of utility that is generally used in the NM. This modern concept
of utility appears to have two obvious advantages over Bentham’s concept. First, it
is easier to measure, since decision utility can be inferred from the choices and
actions that people take. Second, it no longer implies a commit ment to a hedonistic
philosophy. Sen (1987) in particular has been at pains (excuse the hedonistic pun!)
to point out that the maximization of experienced utility is not always what people
are trying to achieve, and his opinion is shared by many economists and
psychologists. Indeed, some economists would go further and say that the study of
peo ple’s objectives is outside the realm of economics and belongs in psychology or
even 93 PT• II FOUNDATIONS philosophy. As stated in the introductory chapter, this
is not a view that we share, since it is not consistent with our reductionist approach,
which advocates consilience between different disciplines. So where does this leave
us as far as the concept of utility is concerned? The main point is that Bentham’s
hedonistic concept of utility may well still be appropriate in terms of determining
our choices and actions, as well as conferring the additional advantage of allowing a
more parsimonious model of behavior. This argument will become clearer as we
examine the various types of utility later in the section. Cardinal and ordinal utility
Early economists believed that utility could be measured quantitatively, in terms of
an arbitrary unit called ‘utils’, using a ratio scale with a zero point. Thus if
consumption of basket A yielded 10 utils and basket B yielded 20 utils, then it could
be said that basket B yielded twice as much utility as basket A. Some economists
even considered that utility could be added interpersonally, meaning that a utility
for John of 10 utils could be added to a utility for Jane of 20 utils, yielding a total
utility of 30 utils. Some economists tend to disfavor any cardinal measure of utility,
meaning a meas ure involving an interval scale. Instead they favor an ordinal
measure, where baskets of commodities are simply ranked according to preference.
This view implies that state ments like ‘basket A has twice as much utility as basket
B’ are meaningless, and certainly interpersonal additions of utility are invalid. The
use of an ordinal measure as opposed to a cardinal measure has the advantage of
involving fewer assumptions regarding the nature of utility. We have already seen
that the equilibrium condition relating to the behavior of consumers in terms of
utility maximization can be expressed in terms of ordinal utility using the concept of
the MRS. The law of diminishing marginal utility can also be expressed in ordinal
terms. However, the NM as it has been stated in this text does use a cardinal
measure of utility, since this has the advantage of being far more tractable
mathematically than any model expressed in ordinal terms. Further more, we will
see that neuroeconomic evidence supports the existence of a cardinal utility
function. We will now see that the concept of utility is a complex one, having many
differ entiated meanings and determinants. By examining these we will attain a
much greater understanding of observed behavior, and in particular the types of
judgment and choice described in the previous section. Decision utility This is the
type of utility usually discussed by economists, since it is easiest to measure in
terms of revealed preference. It is important to note that decision utility does not
there fore necessarily reflect attitudes or judgments. A study by Tversky and Griffin
(2000) illustrates this point. Sixty-six undergraduate students were presented with
the following information: Imagine that you have just completed a graduate degree
in communications and you are considering one-year jobs at two different
magazines. (A) At Magazine A, you are offered a job paying $35,000. However, the
other workers who have the same training and experience as you do are making
$38,000. (B) At Magazine B, you are offered a job paying $33,000. However, the
other workers who have the same training and experience as you do are making
$30,000. 94 values, PreferenCes anD ChoiCes CH•3 Approximately half the students
were asked which job they would choose, while the other half were asked which job
would make them happier. The first question relates to deci sion utility while the
second relates to hedonic, or experienced utility. In this case the experienced utility
is expected in the future, and it appears that people try to imagine what it would
feel like to experience those states, involving the formation of an attitude. However,
when people are asked to make a choice or decision, they tend to search for
reasons or arguments to justify their choice. This difference was reflected in the
survey results: 84% of the subjects chose (A), the job with the higher absolute
salary and lower relative position, but 62% of the subjects thought that (B), the job
with the lower absolute salary and higher relative position, would make them
happier. In this sense then it could be claimed that people preferred (B), thus
indicating a distinction between choice or revealed preference on the one hand and
actual preference in terms of happiness on the other. A more recent study by
Comerford and Ubel (2013) reports a similar discrepancy between preference and
choice when it comes to effortful jobs. They find that people express a preference
for jobs requiring more effort, but choose jobs that involve less effort. In a
laboratory experiment where subjects were assigned to work at a job based on the
wage they set themselves, they also find that those whose wage demands led them
to be assigned to the effortless job experienced lower enjoyment than those who
were assigned to the effortful job. Thus we observe a discrepancy between
preference and choice, and also between decision utility and experienced utility.
Why do people choose an option that they think will make them less happy? There
are various possible reasons that we will have to explore. Another discrepancy
between choice (decision utility) and attitude (experienced utility) arises when the
object of choice/attitude has many attributes. For example, the consideration of a
car may involve the attributes of safety, fuel economy, size, durability and
performance (to name just a few). The standard decision-making approach involves
two steps: (1) determine values for each attribute, using some kind of scale, and (2)
deter mine weights for each attribute, in order to compare them. For example, a
loss of one mile to the gallon in terms of fuel economy may be equivalent to two
cubic feet of boot space. There may also be certain minimum requirements for each
attribute. However, it has been shown, for example by Tversky, Sattath and Slovic
(1988), that this proce dure is more relevant in determining attitudes or judgments;
when it comes to preference the most important attribute is weighted more heavily,
presumably because it is a more convenient rationale for choice (Tversky and
Kahneman 1973). This bias is sometimes referred to as the prominence effect. This
effect may also be explained by the Somatic-Marker Hypothesis (SMH), pro posed
by the neuroscientist Damasio (1994). He points out the problems involved in the
conventional utility maximization model, both in terms of the amount of time
involved and the other cognitive difficulties described by Tversky and Kahneman,
and concludes that people can still make good, and quick, decisions based on ‘gut
feeling’. In essence Damasio is referring to the existence of visceral factors
discussed in the previous section, but in this case proposing that they can lead to
better rather than worse decisions. His SMH proposes that these factors create a
‘somatic marker’, an unpleasant gut feeling when a bad outcome connected with a
given response option comes to mind. The somatic marker: forces attention on the
negative outcome to which a given action may lead, and functions as an automated
alarm signal which says: Beware of danger ahead if you choose the option which
leads to this outcome. The signal may lead you to reject, immediately, the negative
course of action and thus make you choose among other alternatives. The
automated signal protects you against future losses, without fur ther ado, and then
allows you to choose from among fewer alternatives. (Damasio, 1994, p. 173;
emphasis in original) 95 PT• II FOUNDATIONS Thus this mechanism is a first stage in
the decision-making process, which Damasio believes probably increases both the
accuracy and efficiency of the process. He has observed that patients with damage
to the PFC of the brain, where this process appears to occur, tend to be
handicapped in many real-life decisions, for example the choice of who to befriend,
who to have as a marriage or business partner, and what to pursue as a career. By
using only ‘pure reason’ as opposed to ‘practical reason’ they take too long to make
many decisions, and end up making many mistakes that ultimately reduce their
well-being. In conclusion it can be said that sometimes decision utility, as illustrated
by revealed preference, will cause us to make bad decisions in terms of our
happiness. However, in other circumstances, taking into account the SMH, and
indeed any subconscious factors involved in decision-making, people may make
good decisions even though utility maximi zation may not be involved as a
conscious mental process at all. Of course this particular phenomenon does not
contradict the NM; people may be performing in an ‘as if’ manner. Experienced
utility So far experienced utility has been described as a unified concept. However,
Kahneman (2000) draws a useful distinction between remembered utility, after the
experience, and real-time utility, during the experience. These are measured in
different ways, and are relevant for different purposes. 1 Remembered utility is
measured using a memory-based approach; this involves a ret rospective
evaluation of past experience. The concept is therefore subject to bias, in particular
the application of the Peak-End rule. It has been observed in colonoscopy studies by
Redelmeier and Kahneman (1996) that a prolongation of the duration of pain at a
lower level of intensity can lead to a more favorable evaluation, violating the
principle of dominance. This can also be relevant in decision-making, if people base
decisions on remembered past utility, or disutility. Cowley (2008) refers to this
process as retrospective hedonic editing, and draws attention to its dangers in
terms of decision-making. In particular it can result in the justification of past
indulgences, like gambling or over-eating, thus leading to a continuation of self-
harming behavior. This phenomenon is related to cognitive dissonance, discussed in
the next chapter, where people adjust their attitudes to be consistent with their
behavior. 2 Real-time utility is measured using a moment-based approach; this is a
more difficult procedure to implement, since it involves a continuous monitoring of
the subjects. For example, in the colonoscopy studies the subjects were asked to
rate their pain on a scale of 0 (no pain at all) to 10 (intolerable pain) every 60
seconds. An example relating to two patients is given in Figure 3.3. This moment-
based approach can also be used to derive what Kahneman refers to as total utility,
which in turn can be used as a measure of ‘objective happiness’, under certain
assumptions, as will be discussed in the next section. There is another aspect of
lack of unity as far as the concept of experienced utility is con cerned, and this
concerns the distinction between ‘wanting’ and ‘liking’ (Berridge, 2007). ‘Wanting’
is associated with the motivation aspects of reward, which can be dissociated from
‘liking’, which is concerned with the hedonic aspects of reward. Thus we can want
not to like something, like smoking. There is neuroeconomic evidence for this
distinction in terms of different neural systems being involved; it appears that
dopamine is associated with the motivation aspects of reward, and disruption of the
dopamine system does not impair the hedonic aspects, which appear to be
mediated by opioid systems in the ventral striatum and palladium (Fox and
Poldrack, 2008). We shall examine the neuroeconomic aspects of utility in more
detail in the next section. 96 values, PreferenCes anD ChoiCes CH•3 Figure 3.3
Moment utility of two colonoscopies Patient A Pain intensity 10 8 6 4 2 0 0 10 20
Patient B Pain intensity 10 8 6 4 2 0 Time in minutes 0 10 20 Time in minutes
Endowment and contrast effects An important theory regarding happiness and well-
being (not necessarily the same thing as we will see) is that people judge their level
of these based on the concepts of endowment and contrast effects (Tversky and
Griffin, 2000). It should be noted that the endowment effect in this context is not
the same kind of endowment effect referred to earlier, where the acquisition of
goods causes the acquirer to value the goods more highly than expected. In the
words of Tversky and Griffin: The endowment effect of an event represents its direct
contribution to one’s hap piness or satisfaction. Good news and positive
experiences enrich our lives and make us happier; bad news and hard times
diminish our well-being. (p. 709) The contrast effect is an indirect effect that works
in the opposite direction: A positive experience makes us happy, but it also renders
similar experiences less exciting. A negative experience makes us unhappy, but it
also helps us appreciate subsequent experiences that are less bad. (p. 709)
Contrast effects are particularly important in societies where people are generally
becom ing more affluent over time, and in this context are often referred to as
treadmill effects. They are responsible for the second act of Pinker’s three-act
tragedy, described earlier. There are two main theories explaining the existence of
treadmill effects. The old est theory explains them in terms of adaptation (Helson,
1964; Brickman and Campbell, 1971). This is a hedonic response, which can be
most easily seen in physiological terms. When subjects immerse one hand in cold
water and the other in hot water for a period of 97 PT• II FOUNDATIONS time, and
then immerse both hands in the same container of lukewarm water, they experi
ence the strange sensation of one hand feeling warm (the one that was previously
in cold water) and the other hand feeling cold (the one that was previously in hot
water). This theory has been used to explain why both lottery winners and
paraplegics appear to adjust rapidly to their changes in circumstances. However,
not everyone is convinced by the adaptation theory as far as such expla nations are
concerned. Frederick and Loewenstein (1999) have proposed a number of reasons
why reported happiness may not be reliably measured on the scales used.
Kahneman (2000) proposes a different mechanism by which treadmill effects can be
explained, using the term ‘satisfaction treadmill’. This explains the phenomenon in
terms of an aspiration effect. Kahneman explains this effect by using an example of
a graduate student who is constrained by her income to eating mediocre dishes
when she goes to a restaurant. When she takes a lucrative job she can afford to
consume food of a higher quality, and her overall utility increases for a time.
However, after a transition period, we observe that her satisfaction returns to its
previous level. Her aspiration level has increased, and her utility is influenced by her
aspiration level; other things being equal, the higher the aspiration level, the lower
the utility. Alternatively, we can say that there is no longer a contrast between the
dishes she is consuming now and the ones that she was consuming a while ago, just
after her income had increased. Two points should be noted regarding this
aspiration effect. Kahneman refers to the satisfaction treadmill as being distinct
from the hedonic treadmill. However, while the mechanism involved is different, it
can also be claimed that the aspiration effect is still hedonic in nature. The second
point is that the aspiration effect may involve a ratchet effect. Further research
needs to be done in this area, but it may be that people find it easier to adjust their
aspirations upwards rather than downwards. The concept of loss aversion is
relevant here. The endowment and contrast theory (ECT) has some interesting
applications. For example, room surroundings have been found to affect people’s
general satisfaction and their satisfaction with their current housing situation in a
way that is incompatible with conventional theory but compatible with the ECT
(Schwarz et al., 1987). Subjects were required to spend an hour either in an
extremely pleasant room (spacious, nicely furnished and decorated with posters
and flowers) or in an extremely unpleasant room (small, dirty, smelly, noisy and
overheated). Subjects who were placed in the pleasant room reported higher overall
life satisfaction than those in the unpleasant room, showing the dominance of the
endowment effect as far as general satisfaction or well-being is concerned.
However, subjects placed in the unpleasant room reported higher satisfaction with
their housing than those in the pleasant room, showing the dominance of the
contrast effect in the context of a relevant standard of comparison. Thus Tversky
and Griffin (2000) conclude: A specific event, therefore, is likely to have a significant
contrast effect in the domain to which it belongs, and little or no contrast effect in
others. (p. 719) Anticipatory utility As mentioned earlier, people gain hedonic utility
from the anticipation of events in the future, for example by looking forward to a
holiday or dreading a visit to the dentist. This anticipatory utility is based on a
person’s expected or predicted utility, meaning their belief about the future
experienced utility of an event. Again endowment and contrast effects are relevant.
Playing the lottery presents an interesting application, since this type of behavior is
not readily explained by EUT, as discussed in more detail in the next two chapters.
Unrealized hopes and fears can give rise to positive or negative endowment in
terms of anticipatory utility. The probability of winning a lottery is very low, which
98 values, PreferenCes anD ChoiCes CH•3 means that the failure to win does not
cause much disappointment. Therefore, as Tversky and Griffin (2000) state, ‘the
dream of becoming an overnight millionaire could produce enough pleasure to
offset the mild disappointment of not winning the lottery’. When the positive
endowment effect outweighs the negative contrast effect people can enjoy playing
the lottery even when they do not win. Evidence regarding this anticipatory utility
comes from a study by Kocher, Krawczyk and van Winden (2014), who conduct an
experiment with real lottery tickets and find that a substantial minority of subjects
prefer delayed resolution of risk, meaning they prefer to wait longer before finding
out the result. They also preferred to spread outcomes, in terms of preferring two
tickets in separate draws rather than in the same draw, thus prolonging the period
of anticipation. It should be noted that the contrast effect appears to be highly
sensitive to the prob ability of winning or losing. As the probability of winning
increases, the costs of the disappointment of losing (the contrast effect) appear to
increase more quickly than the benefits of hope of winning (the endowment effect).
The implication of this is that, for a given expected value, people should tend to
prefer long-odds situations rather than short-odds – they have sweeter dreams and
milder disappointment. Another possible explanation of people playing the lottery is
given in Chapter 5, in terms of probabilities and decision-weighting. Residual utility
Whereas anticipatory utility looks forward to future events, residual utility relates to
pleasure or pain felt at later periods of time in separate episodes. This phenomenon
arises because utility profiles may be concatenated or disjunctive. For example, a
person may gain anticipatory utility regarding going on holiday in Hawaii for a
month before the actual event, then enjoy the holiday for a week, and then maybe
suffer a contrast effect when they return to work. In addition, maybe a month later,
they may feel another ‘utility boost’ related to the same holiday experience, when
they reminisce with friends. These later episodes may be repeated at various
intervals after the original experience to which they relate. Residual utility can also
be negative if someone dwells on a bad experience in the past, for example thinking
about the time when they had a match point in a tennis championship but then
went on to lose the match. Diagnostic utility This aspect of utility has already been
mentioned as an anomaly in the NM. Diagnostic utility refers to the situation where
people infer their utility from their actions. It was seen that the phenomenon that is
relevant here is the process of self-signaling, which is particu larly important for
people who are uncertain where they stand in terms of certain personal attributes,
for example the possession of strong willpower. Thus when we consider the
situation of someone deciding whether they should have an alcoholic drink, we
should not just consider the experienced or hedonic utility of the good consumed,
we should also consider the utility to be inferred from the action of consumption, in
terms of signaling the vice of a weak will or the virtue of a strong will. It may be that
the negative diagnostic util ity of an action may outweigh the positive expected
experienced utility related to the good consumed. In this case the person will
abstain from consumption. The significance of this is that temptation should not just
be seen as a cost; it can also bring a benefit (Dhar and Wertenbroch, 2012). Thus a
dieting person may experience hunger as disutility, but simul taneously gain the
benefit of positive diagnostic utility in terms of improved self-esteem if they refrain
from eating. If the dieter feels no hunger, for example if they have had gastric band
surgery, they will not gain such diagnostic utility. It will be seen in the first case
study at the end of the chapter that this concept of diagnostic utility has
widespread applications. 99 PT• II FOUNDATIONS Transaction utility The NM frames
the net value of the purchase in terms of benefit minus cost. In reality the
phenomenon of loss-aversion makes this coding of the purchase hedonically
inefficient. Both Kahneman and Tversky (1984) and Thaler (1985) therefore reject
the idea that costs are necessarily viewed as losses. Thaler proposes instead that
there are two types of utility that consumers gain from a transaction: 1 Acquisition
utility – this represents the value of the good obtained relative to its price,
equivalent to the concept of consumer surplus. 2 Transaction utility – this
corresponds to the perceived value of the ‘deal’, in other words the difference
between the reference price and the price paid. The reference price may often be
the price that the consumer expected to pay for the good. Thaler notes two
important implications of the transaction utility component. The first is that people
are often tempted to buy ‘deals’, where transaction utility domi nates acquisition
utility; we then often find that these items are seldom used. Marketing strategies
skillfully manipulate the framing of offers, using reference prices and emphasiz ing
savings (‘silver linings’). Examples of such goods are clothing, household gadgets
and health club memberships (Wilkinson, 1996, 2003). In the last case self-control
factors are also relevant, and these will be discussed in the next section. The other
implication relates to the opposite situation, where people forgo goods that have
the potential to benefit the consumer in terms of acquisition utility, but are rejected
because of a high perceived transaction disutility. Thaler gives the example of a
thirsty beer drinker who will pay $4 for a beer from an expensive resort, but refuse
to pay $2.50 for the same beer from a grocery, on the grounds that he has a
reference price of only $2 in the latter case. We will see later, in Chapter 10, that
there is an additional dimension that may also be involved in these sorts of
decisions: our notion of fairness may be violated. More recent research has added
some nuances to the concept of transaction utility. For example, it has been
reported that consumers perceive the deal to be better if they believe that the offer
is more valuable than the marketer intended (Sela, Simonson and Kivetz, 2013).
The flip side of this phenomenon is that customized offers that are tailor made for
particular consumers may be perceived as less valuable. The authors of the study
concluded that competitive consumers gain utility if they believe they are
outsmarting the market. This has important counter-intuitive policy implications for
marketers in terms of targeting promotions, a strategy normally regarded as
desirable. 3.8 Policy implications As we have already seen, the factors discussed in
this chapter have a number of implica tions for managerial and public policy, which
may not be intuitive, and often contradict current practices and government policies
in many countries. Some examples have already been mentioned, for example
menu effects and the paradox of thrift; other examples that have been investigated
by empirical studies are discussed below, and some are discussed in more detail in
the case studies. The first four situations refer to manage rial implications, the next
two to personal implications, and the last three to government policy. The desire for
consistency and commitment There is a sales tactic referred to as ‘lowballing’,
where a salesperson offers a customer a deal at a particularly low price, but the
deal is not genuine, since the dealer never intends 100 values, PreferenCes anD
ChoiCes CH•3 it to go through. It is commonly used by car dealerships, the purpose
being to cause a customer to make a buying decision for a certain car. Once the
decision has been made, various activities ensue that are designed to cement a
sense of commitment to the pur chase item. For example, forms may be filled out,
the customer’s credentials checked, f inance terms arranged, and the customer
may be encouraged to use the item on a trial basis for a certain time (preferably in
public). During this time the dealer knows that the customer will develop a number
of new reasons for buying the product. Once these props are in place, the original
trigger for buying the product is then removed by the dealer. This is frequently done
by detecting some kind of ‘error’; maybe the salesperson overlooked the pricing of
some option on the car, or, better, some external agent like the finance company
may have discovered the ‘error’, thus shifting the blame. The customer then faces
the original price, or a price not too much higher than the lowball, and at this point
is inclined to take the new offer. Sometimes the lowball is applied to the price
offered for a trade-in, in this case an excessively generous offer. Again this offer is
later withdrawn, after the ‘error’ has been spotted. In both kinds of lowball the
psychology involves creat ing a sense of commitment by the buyer, so that even
when the original trigger is removed, the additional reasons for buying the product
that the consumer has developed in pursuit of consistency drive the purchase
(Cialdini, 1984). Companies can also take advantage of customer loyalty and status
quo bias. Whereas many companies reward regular customers for their loyalty, for
example airlines and coffee shops, others take the opposite strategy, increasing
their rates and relying on customers renewing existing contracts. This applies to
utility and insurance companies in particular. In 2016 the Competition and Markets
Authority (CMA), Britain’s antitrust agency, concluded that 70% of customers of the
Big Six energy suppliers—Centrica, SSE, Scottish Power, Npower, [Link] and EDF—
had failed to switch from the highest standard variable tariffs (SVTs). On average, in
2011–15 they paid 11% more for their electricity than those on other tariffs.
Essentially, the suppliers boosted their profitabil ity by taking advantage of inactive
customers, who tend to be poor, under-educated and elderly. The CMA found no
evidence of collusion between the Big Six, and concluded that price-rigging would
not be beneficial since it would stifle competition and reduce incentives for
consumers to switch suppliers. However, in April 2017 the Conservative government
ignored this advice and decided to incorporate a price cap policy on the Big Six in
their election manifesto for the upcoming election in June, following the Labour
Party’s pledge to do the same thing. A nudge policy aimed at increasing the
availability of relevant information for both suppliers and customers may actually be
more benefi cial here. Another tactic, also aimed at establishing consistency and
creating commitment, but less questionable from an ethical viewpoint, is sometimes
used in hotels that want to reduce their laundry expenses, and at the same time be
environmentally friendly. This specifically targets towel use, and asks guests to
hang towels for reuse. One experiment, reported by Baca-Motes and colleagues
(2013), involved a hotel requesting guests to wear a lapel pin signaling their
commitment to be environmentally friendly by reducing towel usage. The authors
find that this had the effect of increasing the number of towels hung by over 40%.
There is another interesting finding in this experiment, relating to the desire for
conformity, which will be described in the next chapter. Deterring unhealthy
behavior Governments frequently want to deter behavior which is unhealthy, both
for the individu als involved and for others, for example smoking, eating junk food,
and driving under the influence of drugs. Usually a combination of approaches is
used. Australia is perhaps a leading example as far as deterring smoking is
concerned. The government has set a target 101 PT• II FOUNDATIONS of reducing
the proportion of adults who are daily smokers to 10% by 2018. In order to achieve
this the following measures have been implemented: • A regime of large yearly tax
increases, 12.5% per year, to reach a price of AUD$40 by 2020. • Plain packaging
rules, implemented since 2012, with no branding or logos. These require 75% of the
front of the cigarette pack to be covered by a health warning, and 90% of the back.
• Graphic health warnings with a strong visceral impact, showing images of throat
can cer and other physiological harm. Rousu and colleagues (2014) find evidence
that this policy is more effective than a text-only health warning in reducing
purchase behavior. Although some smokers use an avoidance strategy, it has been
found that such smokers are more likely to quit later. • Restrictions on locations
where smoking is allowed. This includes many public places, like bars and
restaurants, and also places where people tend to congregate, like bus stops, taxi
ranks and train platforms. Smoking is also prohibited within 10 meters of
playgrounds and in vehicles if children are present. Heavy fines, up to AUD$2,000
can be levied. The Australian government has reported that the prevalence of
smoking has fallen by 0.55% since 2012. Encouraged by this fall, the UK
government has also adopted some of the above policies, for example the
requirement for all cigarettes to be sold in plain packaging with large and graphic
health warnings. One important aspect of these policies is that a social norm is
developing whereby smokers are becoming marginalized. This is created in
particular by the restrictions on locations, so that smokers have to seek places to
smoke that are generally undesirable, inconvenient or uncomfortable. Further exam
ples of government policy and deterrence are illustrated in the case study on
obesity in Chapter 10, since this also involves social norms. The pricing of medicines
Superficially it might appear that the demand for medicines obeys the general law
of demand that more is bought at lower prices. However, this superficial finding for
most medicines obscures an interesting socio-psychological phenomenon. Samper
and Schwartz (2013) have found that the price of a medication influences
consumers’ beliefs about their own disease risk. This is because they believe that
government policy generally favors access to goods that are necessities; therefore
they perceive a low price as a signal that a medicine must be an essential lifesaving
product. This in turn causes consumers to per ceive a greater threat to their health
if they do not buy the product, thus leading to greater consumption. The opposite
effect occurs with higher prices. However, this phenomenon only appears to apply
to situations where the health threat is perceived is to be relevant to the individual,
and indicates that consumers make inconsistent assumptions about risk and need.
The conclusion is that greater price transparency for medications may not nec
essarily lead to better choices; lower prices may encourage unnecessary
consumption, a serious problem in the case of antibiotics, while higher prices may
discourage the use of highly beneficial products. Using framing effects to increase
sales Many empirical studies discuss the policy implications of framing effects in
different con texts. For example, many firms use non-rounded numbers when it
comes to pricing, in particular odd numbers such as $19.95 as opposed to $20, and
much evidence suggests that this tactic is successful in increasing sales. However, a
study by Wadhwa and Zhang (2015) 102 values, PreferenCes anD ChoiCes CH•3
suggests that the success of this tactic depends on the nature of the purchase
decision. They provide evidence that non-rounded prices may be more attractive for
purchasers driven by cognition, since they require more mental processing;
however, for purchasers driven by feelings or visceral factors, rounded prices may
be more attractive since they involve less mental processing. A different, and
counter-intuitive, policy implication arises when consumers who perceive eating
healthy food as mandatory sampled a food framed as healthy and later reported
being hungrier and consumed more food than those who sampled the same item
framed as tasty or those who did not eat at all (Finkelstein and Fishbach, 2010). The
authors of this study suggest that the psychology here is that eating the healthy
food signals that the health goal was sufficiently met, and thus it increases the
strength of the conflicting motive to fulfill one’s appetite. This kind of signaling,
which can even be achieved through vicarious consumption, has been found to
occur in other situations regarding the consumption of healthy items, as is seen in
Case 3.3. Another framing effect related to increasing sales involves the use of
alliteration. A study by Davis, Bagchi and Block (2012) theorizes that alliterative
offers increase sales because the repetition of phonological units in alliterative
prices ‘sounds’ better, which in turn influences perceptions of the deal presented.
For example, the offer ‘9 for $.90’sounds better than ‘9 for $.88’, and ‘3 Threybles
$30’ sounds better than ‘3 Fables $30’. The authors provide evidence from several
studies that supports this contention. Sometimes consumers make decisions
through inaction rather than action. This dis tinction can be important in the context
of choice options where the difference in utility is large or small. A study by
Evangelidis and Levav (2013) suggests that when the options involve a large
difference in utility consumers prefer to make an active choice. However, if the
difference in utility is small, meaning there is considerable uncertainty regarding
the optimal outcome, consumers may prefer an inaction frame, using some kind of
default mechanism as a tie-breaker. Framing effects are examined in more detail in
Case 3.3, where the psychology of supermarket shoppers is considered. Use of
default options Firms can also use default options as a means of manipulating
consumer choices. The default option may be chosen for various reasons, including
cognitive ease, an impli cation that it is the norm or that it is recommended.
Sometimes firms may actually have the best interests of their consumers at heart,
for example when the default option on a restaurant menu is a healthy one, like
salad instead of chips as a side dish, to encourage customers to eat more healthily.
Colby, Li and Chapman (2014) show that this may not always achieve the desired
results. Although customers may be more likely to choose the healthy dish if it is a
default, they may also be less likely to return to the restaurant. This latter effect
may be caused by the finding that people feel less virtuous when they choose the
healthy dish as a default than when they choose it when it is not the default. This is
a good example of a framing effect, once again violating the invariance principle of
EUT. Bonus packs and discounts Both bonus packs and discounts are common sales
promotion techniques in marketing. The issue that many firms face is to determine
which method is most effective in particular circumstances. Mishra and Mishra
(2011) propose that the relevant factor here is whether the product is regarded as a
‘virtuous good’ or a ‘vice good’. Most goods are virtuous goods, being good for us in
the long run, and in this case Mishra and Mishra conclude that bonus packs are the
more effective method for promoting additional sales. The reason 103 PT• II
FOUNDATIONS is that price discounts can be harmful to brand image if they are
used on a regular basis. However, for vice goods the conclusion is that price
discounts may be a better method. The reasoning here relates to post-purchase
guilt. Consumers may find it difficult to jus tify to themselves buying additional
amounts of goods that are bad for them in the long run, so bonus packs may not be
an effective promotional tool in this case. With a price dis count the consumer may
end up buying more of a good than otherwise over a particular time period, but the
salience of this is not so great as when they buy a bonus pack. There are additional
behavioral factors that are relevant in this context, as we shall see in later chapters,
for example consumer reactions to product bundling, and whether consumers are
naïve or sophisticated. Saving the best to last As seen earlier, peak-end effects have
been observed not only in humans but also in capuchin monkeys (Egan Brad et al.,
2016), suggesting that they are not simply a cul tural artifact but represent an
evolved psychological adaptation. However, this study also found that both humans
and capuchins were unable to construct optimal sequences of consumption that
apply the peak-end rule. An optimal sequence involves leaving the best until last.
This presents something of a puzzle, since, in the context of food consumption,
adult humans and capuchins are able to delay gratification in order to gain greater
future rewards, and humans at least are able to select sequences of rewards, for
example income profiles, that implement the peak-end rule in laboratory
experiments. The authors sug gest that the reason for the human (and capuchin)
failure to implement the peak-end rule in real-life situations is that when it comes to
constructing sequences, as opposed to choosing among preselected sequences, the
salience of visceral factors causes people to act impulsively and consume the best
first. Once again, in spite of millions of years of evolutionary development of the
PFC, we seem to lack the insight to appreciate and apply a rule that maximizes
hedonic well-being. There is perhaps an evolutionary reason for this, however: in
real-life situations animals are often under pressure to eat quickly, either because of
competition from other members of their own species or to avoid predators. In this
situation it may be optimal to eat the best parts first. Regulating the emotions As
with the phenomenon above, the regulation of emotions relates to an area of
behavior where individuals often do not act in optimal ways, and in this case are
also frequently aware of this after the fact. The situation becomes even more
problematical when the relevant behavior is repeated over and over again, with the
individual failing to learn from experience. An example of this is where a person
tends to become aggressive and abusive after consuming alcohol, comes to realize
this, but fails to alter their behavior, either in terms of the aggression or the
drinking. Research indicates that the regulation of emotional arousal depends on
the innate qualities of a person and on the strategies, conscious or unconscious
used by the indi vidual. Gross and John (2003) have identified two main strategies
that are relevant here, reappraisal and suppression, which tend to have very
different outcomes in terms of opti mality. The reappraisal strategy demands a
certain amount of self-awareness, so that the person can implement it as the
emotion starts to develop. Reappraisal refers to cogni tively changing a situation’s
meaning in a way that alters its emotional impact. Using reappraisal strategy has
shown to efficiently alter the entire subsequent emotion trajec tory, especially when
down-regulating negative emotion (Hariharan et al., 2015). This may well involve
thinking about the positive side of a negative incident or putting it into 104 values,
PreferenCes anD ChoiCes CH•3 the broader context. For example, if we are beaten
in a competitive game we may reas sure ourselves that the winner was a highly
rated opponent and that the game served as a useful learning experience. Thus the
negative emotions related to loss of self-esteem and disappointment may be
softened. Reappraisal might therefore be an adaptive strategy. In contrast,
suppression comes relatively late in the process of emotion development, and
modifies the behavioral aspect of emotion response tendencies, referred to as a
response focused strategy (Gross, 1998). Suppression refers to influencing
physiological, experiential or behavioral responding as directly as possible, and may
involve efforts to hide what one is feeling, or attempts to inhibit what one is feeling
(Hariharan et al., 2015). In the previous example of being beaten in a competitive
game it may involve dwelling on defeat but trying to hide the loss of self-esteem by
an outward display of bravado. Use of suppression has been shown to be correlated
with increased negative affect, poorer recovery from changes in negative affect,
and decreased self-efficacy for managing future emotions (Gross, 2002). Jury
awards of punitive damages The psychology underlying these awards was
investigated by Kahneman, Schkade and Sunstein (1998). Three main questions
were asked about scenarios relating to compensa tory damages in product liability
cases: 1 How outrageous was the defendant’s behavior? 2 How severely should the
defendant be punished? 3 How much should the defendant be required to pay in
punitive damages? The first two questions required a rating on a seven-point scale,
while the third question required a dollar amount to be stated. It is notable that
correlations between the evalua tions are high, always at least 0.80. Since the
outrage rating appears to be a direct measure of the affect evoked by cases of
personal injury, the implication is that the amount of puni tive damages awarded is
determined largely by the outrage factor, determined by attitude rather than
economic preference. In addition to this general finding, the same authors found
that respondents experi enced a degree of outrage that was independent of the
amount of harm caused. Thus it seems that we judge behavior in terms of
outrageousness regardless of its consequences. However, when it comes to judging
punitive intent and assessing damages, the second and third questions, the
consequences of the behavior are of great importance. This finding does seem
intuitive, since punishment contains a retributive element based on the harm
caused. A further finding of the study was that the defendant’s ability to pay was
relevant as far as the third question is concerned. Large firms with more resources
were penalized more heavily in terms of damages, although size of the firm had no
effect on evaluations of outrageousness or punitive intent. This finding again seems
plausible, since a $10 mil lion payment represents a large amount for a small firm
whereas it is small change for a large firm. A further finding illustrates the
importance of context-dependence, which can cause preference reversals. It is
commonly observed that there is a high correlation between punitive awards and
compensatory damages awarded. When there is a large amount of f inancial harm
compensatory damages tend to be high, leading to high punitive damages. This is
likely to be caused by the anchoring effect of the high compensatory damages. On
the other hand, in cases of personal injury punitive damages tend to be low, since
compensatory damages are also low. It should be noted that in real life these cases
are determined in isolation, by different juries. However, it is very probable that
outrage is higher in cases of personal injury (such as a child being burned by a
faulty product) than 105 PT• II FOUNDATIONS in cases of business fraud. This
presents a possibility for testing the theory that degree of outrage is relevant in
evaluating the size of punitive award. An experiment by Kahneman, Schkade and
Sunstein (1998) placed subjects in the position of having to judge two cases
together as well as independently; in both situations compensatory damages had
already been awarded, $500,000 for the personal injury case and $10 million for the
financial harm. As predicted, subjects who only judged one case awarded more
punitive damages in the financial case (median = $5 million) than in the personal
injury case (median = $2 million). However, a large majority (75%) of the
respondents who judged the two cases together assessed larger awards in the
personal injury case, resulting in a significant pref erence reversal (median of $2.5
million for personal injury, $0.5 million for the financial harm). This result confirms
two important theories in behavioral economics: the existence of context-
dependence and anchoring effects, and the influence of affect on attitudes and
evaluations. In response to these anomalies, Sunstein, Kahneman and Schkade
(1998, p. 2079) have proposed certain reforms which ‘would require jurors to do
what they can do well, not what they can do poorly’. Their conclusion was that
jurors are good at combining normative evaluations with empirical facts, meaning
that they have sound intuitions about the appropriate severity of punishment.
However, their ability to translate these intuitions into dollars appears to be weak;
therefore the authors proposed that the jury make graded recommendations
relating to the severity of punishment to the judge, who would then perform the
task of translating this intent into a dollar amount. The contingent valuation method
and public goods The contingent valuation method (CVM) is often used as a means
of eliciting the value the people place on public goods, including non-use goods
such as the continued existence of rare species. The valuation produced is then
used as a basis for public policy decisions. CVM sometimes relies on asking people
to report their stated willingness to pay (SWTP) to achieve various results, although
more recently superior indirect techniques of valua tion have been used, which may
reduce some of the problems discussed below. Although asking people to value
public goods, like cleaning up a lake, may superficially seem quite dissimilar to
asking jurors to estimate punitive damages, there are important similarities and
problems. Studies have shown similar results in terms of high correlations between
various evaluation methods when attitudes towards protecting the environment are
involved (Kahneman and Ritov, 1994; Payne et al., 2000). For example, Kahneman
and Ritov (1994) used four different measures to evaluate an intervention to protect
the peregrine falcon from pollution: 1 Stated willingness to pay (SWTP) for the
intervention 2 Degree of political support for the intervention 3 Personal satisfaction
expected from making a voluntary contribution 4 Importance of the problem as a
public issue The last three measures were all based on a rating scale. As with the
punitive damages situation, the high correlations between the different measures
suggest that the measures are all a reflection of an underlying attitude or affect.
There are two important implica tions of this relating to the use of SWTP in terms of
obtaining biased and unreliable results: 1 Anchoring effects. As we have already
seen, these are prominent in several situations, for example the estimation of
punitive damages. A further example is given in Case 3.3 involving environmental
protection. 106 values, PreferenCes anD ChoiCes CH•3 2 Insensitivity to scope. This
is a form of extension bias. Kahneman (1986) found that Toronto residents were
willing to pay only a little more to clean up all the polluted lakes in Ontario than to
clean up polluted lakes in a particular region of Ontario. Jones-Lee, Loomes and
Philips (1995) found that the SWTP of UK respondents to a programme to reduce
the risk of non-fatal head injuries increased by only 29% when the number of
prevented injuries was increased by 200%. Further examples relating to
environmental protection are again given in Case 3.3. An important consequence of
this bias is that when CVM uses the technique of asking for people’s SWTP it is
highly misleading to use the add-up principle, whereby the values of different
results are simply added together. For example, one cannot reliably conclude that
the value of saving two species, A and B, from extinction is the value of saving
species A plus the value of saving species B, even in the situation where the two
events may seem quite independent from each other (the species being unrelated).
Some improvements to the SWTP technique of CVM have been proposed, nota bly
the referendum protocol. This involves asking people simply to vote on an issue; the
wording of the question for a particular respondent would therefore only quote a
single value, for example: ‘Would you vote for a proposition requiring you to pay
$20 to clean up Lake Ontario?’ Different groups of respondents are given different
values to respond to, and the responses allow a distribution of WTP to be estimated.
However, although this survey protocol may alleviate the problem of anchoring
effects, the general problems of framing effects, extension bias and insensitivity to
scope remain. When values are based on attitudes, the standard rules of economic
theory regarding preference no longer apply. Kahneman, Ritov and Schadke (2000)
propose a somewhat similar solution in general terms to the problem of CVM to the
solution proposed for jury awards: elicit and measure public opinion regarding their
attitudes to the relevant issue using psychometric criteria, and then use expert
opinion to convert these judgments into monetary terms. Placebo effects We have
seen a number of examples of ‘the mind getting what it expects’, for example with
aspirin and energy drinks. There are some far-reaching policy implications of this,
which Ariely (2008) points out. These relate in particular to government policy
regarding health. Many governments have been trying to reform their health
policies, which are often seen as inadequate, inefficient and excessively costly. The
USA currently spends a greater proportion of GDP on health care than any other
Western nation, though much of this is accounted for by the private sector, whereas
in Europe medicine tends to be more ‘social ized’, with state provision being more
prominent. Since 2008 state provision in particular has been threatened by
recession and government cutbacks forced by the need for deficit reduction (large
deficits themselves largely being caused by recession and financial bail outs).
Expensive drugs for diseases like cancer and AIDS have to be rationed and are often
not available via state systems. Many people are scandalised by this, particularly
since there is a fairness issue involved; drugs may be provided in some areas or
regions but not others. Ariely poses the question: ‘How do we deal with the fact that
expensive medicine (the 50-cent aspirin) may make people feel better than cheaper
medicine (the penny aspirin)?’ In view of the psychological effects involved this is
not an easy issue to address. We can indulge in people’s irrationality, raising the
costs of health care, or we can insist that people get the cheapest generic drugs,
which are in objective terms identical to the more expensive ones, but in practical
terms are less effective. Of course, these are stark alterna tives; there may be
variations of both that are judged to be preferable. Other ethical issues arise with
respect to placebo effects. It is illegal in many coun tries to hype products with false
claims on labels or in advertising. Yet it has been found 107 PT• II FOUNDATIONS
that such claims can actually increase the perceived quality of the product, and
even the effectiveness of the product in objective terms. For example, when it was
(falsely) claimed that there was evidence from 50 scientific studies that the SoBe
energy drink increased mental functioning, the subjects who were notified of this
claim did significantly improve their scores on the anagram word puzzle test
compared to those who were not notified of the claim. So was the false message
hype after all?