The role of moral sentiments in experimental economics

Full text

Explanations of economic behavior often fall into one of two camps: norma-
tive and descriptive. Normative accounts typically consist of attempts to fit a
wide variety of phenomena into one broad theoretical framework consisting
of a single set of axioms governing rational decision making (e.g., von
Neumann & Morgenstern, 1944). Descriptive approaches, in contrast, con-
sist of attempts to accurately describe how individuals actually behave, re-
gardless of whether these accounts are consistent with any particular norma
-
tive approach to rational decision making (Gilovich, Griffin, & Kahneman,
2002; Hogarth & Reder, 1987). What makes this normative–descriptive dis
-
tinction so intriguing is the observation that many descriptive accounts of
economic behavior are often strikingly inconsistent with normative models.
It turns out that this contradiction between how people ought to behave, ac
-
cording to normative models, and how they actually do behave, as evidenced
by descriptive accounts, is a puzzle that has intrigued observers of human na
-
ture for centuries.
Eighteenth-century moral philosophers such as David Hume (1777) and
Adam Smith (1759) attempted to bridge the gap between what people
should do (normative principles) and what they actually do (descriptive ac
-
counts) by exploring the role of sentiments in decision making. Hume and
Smith argued that moral sentiments, in the form of emotions and passions,
exert such a powerful influence on judgment that they often compete with
more rational deliberations in determining our behavior. In essence, they ar
-
gued that emotions can provoke otherwise rational individuals to behave in a
6
The Role of Moral Sentiments
in Economic Decision Making
Timothy Ketelaar
New Mexico State University
97

manner that is inconsistent with normative principles of good judgment. The
present chapter expands on this 18th-century conception of moral senti
-
ments to explore the role of emotion in contemporary economic decision re
-
search. After discussing the historical role that emotions have played in nor
-
mative models of good judgment and decision making, I provide a brief
review of recent empirical work linking moral emotions to norm-obeying and
norm-enforcing behaviors in experimental economics. Finally, it is argued
that insights from evolutionary game theory can illuminate the possible
adaptive design of moral sentiment systems and, in doing so, may help us un
-
derstand why decision makers often do not behave in the strictly rational
manner that some normative models suggest that they should behave.
EMOTIONS AND NORMATIVE MODELS
OF GOOD JUDGMENT
Just as contemporary decision theorists often employ conflicting standards
(e.g., normative vs. descriptive) to explain economic behavior, 18th-century
moral philosophers once grappled with an equally conflicting set of standards
for discerning an individual’s moral character—namely, the standards of ab-
solute moral perfection versus mere adherence to social convention. In his
Theory of Moral Sentiments, for example, Adam Smith (1759) argued that:
when we are determining the degree of blame or applause which seems due to
any action, we very frequently make use of two different standards. The first is
the idea of complete propriety and perfection. . . . The second is the idea of
that degree of proximity or distance from this complete perfection, which the
actions of the greater part of man commonly arrive at. (pp. 29–30)
Although the two standards to which Adam Smith referred are strikingly
similar to the normative–descriptive distinction employed by contemporary
decision researchers, Smith and other moral philosophers of his era were
more interested in the moral (rather than economic) implications of behav
-
ior. By contrast, contemporary economists and psychologists are much less
interested in whether a given behavior approximates a particular standard of
moral propriety and perfection. Instead economists and psychologists are
more concerned with the degree to which descriptive accounts of judgment
and decision making are consistent with particular normative standards of
sound economic reasoning, such as those standards embodied in the norma
-
tive rules of inference of the logical calculus (von Neumann & Morgenstern,
1944). Just as 18th-century moral philosophers regretted that individuals do
not always behave as they should in the moral domain, so too contemporary
decision theorists lament that economic decision makers routinely enact
98 KETELAAR

preferences that are at sharp contrast with normative standards of proper
economic behavior. Regardless of whether one considers violations of nor
-
mative standards to indicate moral turpitude or merely economic irrational
-
ity, it is quite obvious that the distinction between what people should do
and what most people actually do looms as large today in the minds of econ
-
omists and psychologists as it did in the writings of 18th-century moral phi
-
losophers.
Is there any reason to believe that contemporary decision theorists have
anything to learn from 18th-century moral philosophers? Although one might
imagine that distinguishing the moral and logical implications of a particular
behavior is a relatively straightforward task, it turns out that moral and logical
implications are, in fact, quite easily confused. Nowhere is this more evident
than in our evaluations of behaviors that violate normative standards of appro
-
priate behavior. In this regard, behaviors that deviate from normative stan
-
dards of sound reasoning are not only viewed as illogical, they are often met
with judgments of moral disapproval that include strong emotional reactions.
Even the most casual observer of human nature will note just how easy it is to
conflate judgments of deviant behavior with judgments of immorality. Social
psychologists, for example, have shown that merely observing a person violat-
ing a culturally shared norm is enough to evoke strong moral emotions such as
anger and contempt (Haidt, 2003; Ketelaar et al., 2006; Rozin, Lowery, Imada,
& Haidt, 1999). Although the types of non-normative behaviors studied by so-
cial psychologists (e.g., stealing a purse from a blind person) are a bit far re-
moved from the sorts of norm violations that interest economists (e.g., viola-
tions of the invariance axiom), empirical evidence now suggests that morally
charged emotions such as anger and guilt are routinely evoked in economic
decision tasks that were once conceptualized as purely logical deliberations
among rational participants (see Fehr & Gächter, 2002; Ketelaar & Au, 2003;
Pillutla & Murnighan, 1996; Van Kleef, De Dreu, & Manstead, 2004; Zeelen
-
berg & Pieters, chap. 7, this volume).
Given the ubiquitous role that emotions appear to play in everyday deci
-
sions (Damasio, 1994; Elster, 2000; Nussbaum, 2004), might there be room
for a more formal treatment of the role of these moral sentiments in norma
-
tive accounts of economic decision making? Along these lines, the present
chapter argues that a greater appreciation of the role of emotion might be in
-
formative in regards to two questions of interest to economists and psycholo
-
gists. First, why are emotions so easily provoked in purely economic deci
-
sions? Second, why do emotions often compel individuals to pursue strategies
that do not appear to be in their immediate material self-interest? It is argued
that a large part of the puzzle concerning why irrational emotions so often in
-
trude on our most important decisions may lie in recognizing this link be
-
tween moral sentiments and judgments of normatively acceptable (and unac
-
ceptable) behavior.
6. EMOTIONS AND ECONOMIC DECISION MAKING 99

A BRIEF HISTORY OF MORAL SENTIMENTS
AND DECISION MAKING
The role of emotion in judgment and decision-making research has perhaps
been most strongly shaped by the views of David Hume for reasons having to
do with his provocative treatment of the question regarding whether our
moral judgments have their basis in our capacity for reason or instead in our
capacity for moral sentiment (Gardner, 1987; Norton, 1993). In his Treatise
on Human Nature, Hume (1740/1978) initially espoused the quite radical
view that reason was the “slave of the passions,” but later in his Enquires
Concerning Human Understanding and Concerning the Principles of Morals
(1777), he modified his position to portray moral sentiment and moral rea
-
soning on more equal footing. Thus, although he eventually came to admit
that rational deliberation played an important part in moral decision making,
what is most striking about Hume’s view is that he immediately recognized
that emotion played a central role in the decision process.
By contrast, 20th-century psychology began with a relatively balanced
treatment of emotion and reason (Gardner, 1987), but quickly became domi-
nated by a strong strain of rationalism soon after the cognitive revolution of
the 1960s. This rationalist surge could not have come at a worse time for re-
searchers interested in exploring the role of emotion in decision making. Just
as von Neumann and Morgenstern’s (1944) Theory of Games and Economic
Behavior was beginning to inspire social scientists to model social encounters
as strategic interactions, emotions were being banished from normative mod-
els of strategic decision making. Consistent with this rationalist slant, emo-
tions were soon relegated to the status of disruptive influences on judgment
and decision making. For the next several decades (into the late 1980s), the
ideas of Kant and Locke were alive and well, whereas the views of Hume and
Smith were dead or dying (see Haidt, 2001, for a historical review).
The growing rationalism of postcognitive revolution psychology gave a
much different twist to the age-old question of why individuals sometimes
strayed from, and at other times upheld, the normative path of good judg
-
ment. In contrast to the moral philosophers of the 18th century, the cog
-
nitively oriented psychologists of the 20th century emphasized moral reason
-
ing rather than moral sentiments as the basis for adherence to normative
standards of good behavior (Haidt, 2001). In this light, the ideas of Kohlberg
(1971) and Piaget (1932/1965) loomed large, whereas the views of Hume
and Smith receded to the shadows. Individuals who acted in a normatively
(morally) acceptable manner were believed to be employing a deliberate and
conscious process of sound (moral) reasoning, whereas individuals who fell
short of rational enlightenment and moral perfection were seen as being in
-
fluenced by unnecessary and disruptive moral sentiments (Kohlberg, 1971).
Although this rationalist view of emotions as disruptive influences was strong
-
100 KETELAAR

est in the moral reasoning literature, rationalism managed to extend its reach
into the more general literature on social judgment and decision making (see
Clore, 2005; Ketelaar & Clore, 1997). As one emotion researcher observed
(Scherer, 1984):
Today the image of man is no longer that of an individual enslaved by his pas
-
sions, but rather that of a philosopher making decisions on the basis of logical
deduction and inference. In this tradition, emotion is seen as a regrettable flaw
in an otherwise perfect machine. (p. 293)
The view from economics has not been much rosier (Elster, 1995):
The standard view of the relation between rationality and emotions is, of
course, that emotions interfere with rationality. They are, as it were, sand in
the machinery of action. (p. 1394)
Although emotionally influenced decisions often appear to be at odds
with normative principles of rational inference (von Neumann & Morgen-
stern, 1944), perhaps consistency with the cold rules of logic is the wrong ref-
erence point from which to view these sentiments (Clore, 2005; Ketelaar &
Clore, 1997; Ketelaar & Todd, 2001). In this regard, recent research on
moral judgment suggests that emotions have more in common with relatively
automatic processes such as intuition and perception than they do with more
calculated processes such as rational deliberation and logical inference
(Haidt, 2001). To the degree that our judgments, moral or economic, are of
-
ten based on moral intuitions and emotional perceptions, rather than logic
and rational deliberation, one might suspect that the impact of emotion on
economic decision making has been greatly underestimated. Participants in
studies of moral judgment, for example, have no difficulty generating reasons
for their judgments. Yet further analysis reveals that many of these reasons
are actually post hoc confabulations, rather than veridical recollections of
the actual thought processes that led to their judgment (Haidt, 2001). More
-
over, participants in studies of moral judgment often report that they “don’t
really know” why a particular behavior is morally wrong, it just is! (Haidt,
2001). Although these findings do not necessarily establish that moral rea
-
soning is enslaved by the passions, they do suggest that many decision pro
-
cesses that were previously thought to be the product of logical inference and
rational deliberation are instead more accurately described as being the prod
-
uct of relatively automatic emotional intuitions and perceptions. The re
-
mainder of this chapter explores how this insight might be integrated into
our understanding of certain types of economic decisions.
6. EMOTIONS AND ECONOMIC DECISION MAKING 101

EMPIRICAL EVIDENCE FOR EMOTIONAL
INFLUENCES ON ECONOMIC
DECISION MAKING
Emotion researchers have found it useful to conceptualize passions and senti
-
ments in terms of their intrapersonal and interpersonal functions (Bowlby,
1969; Van Kleef et al., 2004). Intrapersonal functions refer to the impact that
emotions have on individual decision making, such as when postdecision re
-
gret motivates you to pursue an economic opportunity that you had previ
-
ously rejected (Zeelenberg, 1999). Interpersonal functions, in contrast, focus
on the impact that emotions have on social decision making, such as when
anger motivates you to punish a selfish contributor in a public goods game
(Fehr & Gächter, 2002). In relation to these intra- and interpersonal func
-
tions, emotion theorists have identified at least two important clusters of
moral sentiments: (a) self-focused sentiments such as shame, embarrassment,
regret, and guilt; and (b) other-focused sentiments such as contempt, anger,
disgust, and schadenfreude
1
(Haidt, 2003; Lewis, 1993; Ortony, Clore, & Col-
lins, 1988). Self-focused emotions emphasize an evaluation of the appropri-
ateness of one’s own behavior, whereas other-focused emotions tend to di-
rect one’s attention to the appropriateness of others’ behavior. Although
there exists a great variety of ways in which the relation between emotions
and economic decisions can be conceptualized (see Zeelenberg & Pieters,
chap. 7, this volume), the remainder of this chapter focuses on two areas in
which recent theoretical and empirical developments have been quite illumi-
nating: (a) self-focused emotions and norm-obeying behaviors, and (b)
other-focused emotions and norm-enforcing behaviors.
Self-Focused Emotions and Norm-Obeying
Behavior
Scottish philosopher Adam Smith (1759) was among the first to develop a
theory of how self-focused emotions can motivate an individual to obey (or
disobey) certain normative standards of good judgment. Smith argued that
when we experience certain moral sentiments, such as guilt, they intuitively
compel us to do what is morally appropriate, despite the fact that rational
economic calculation might suggest otherwise. The sentiment of gratitude,
for example, can compel an individual to repay an act of kindness even when
the cost of repayment exceeds the benefit initially bestowed on the actor
(Hirshleifer, 1987). So strong are these emotional influences on behavior
102 KETELAAR
1
1
Schadenfreude is a German term that refers to the experience of positive emotions while ob
-
serving another’s misfortune.

that certain sentiments, such as greed or lust, can even motivate an individ
-
ual to disobey these same moral standards:
The man who acts according to the rules of perfect prudence, of strict justice,
and of proper benevolence, may be said to be perfectly virtuous. But the most
perfect knowledge of those rules will not alone enable him to act in this man
-
ner; his own passions are very apt to mislead him—sometimes to drive him, and
sometimes to seduce him, to violate all the rules which he himself, in all his sober and
cool hours, approves of. (Smith, 1759, p. 349; italics added)
In essence, Smith (1759) argued that our capacity to obey normative stan
-
dards of appropriate behavior was more often determined by our experience
of certain moral sentiments than our employment of rational deduction and
logical inference. Depending on the nature of the particular emotion that
one experienced (e.g., greed vs. guilt), Smith argued that moral sentiments
could compel an individual to pursue or avoid a particular course of action
even when this choice ran contrary to one’s immediate economic self-
interest.
Although contemporary economists and psychologists are more con-
cerned with earthly rewards than heavenly utilities, Smith’s insights regard-
ing moral sentiments can be applied to questions that interest contemporary
decision theorists. For example, why do emotions often compel individuals to
pursue economic strategies that do not appear to be in their immediate ma-
terial self-interest? Along these lines, there is an ever-expanding body of the-
oretical and empirical research on self-focused emotions and economic deci-
sion making that is consistent with Smith’s insights regarding the influence
of moral sentiments on decision making (Ketelaar, 2004; Ketelaar & Au,
2003; Van Kleef et al., 2004). For example, economist Robert Frank’s (1988,
2004) theory of emotions as commitment devices essentially argues that the
experience of self-focused emotions such as guilt can explain why some indi
-
viduals forgo the immediate rewards associated with cheating in favor of
more cooperative strategies that pay their benefits, not immediately, but
rather in the long run. According to this view, moral sentiments can:
compete with feelings that spring from rational calculations about material
payoffs. . . . Consider, for example, a person capable of strong guilt feelings.
This person will not cheat even when it is in her material interests to do so.
The reason is not that she fears getting caught but that she simply does not
want to cheat. Her aversion to feelings of guilt effectively alters the payoffs she
faces. (Frank, 1988, p. 53)
Because moral sentiments are experienced as powerful and unignorable
moral intuitions, an individual experiencing an emotion like guilt has two
concurrent sources of information that are taken into account when deciding
6. EMOTIONS AND ECONOMIC DECISION MAKING 103

how to behave (Ketelaar & Clore, 1997). One source of information (from
deliberate economic calculations) informs the individual about the immedi
-
ate material consequences of a given strategy, and the second source (from
the intuitive moral emotions) informs the individual about the future (in this
case, negative) consequences of that particular strategy choice. In this man
-
ner, the mental pain of guilt can compel an individual to not cheat even
when it is in their immediate material interest to do so (Frank, 1988).
For over one decade, this view of moral sentiments (Frank, 1988) has
been so compelling that it has been accepted at face value without being
subject to empirical test. Recently, research at the intersection of social psy
-
chology and experimental economics has begun exploring the role of moral
sentiments in economic decision making (Ketelaar & Au, 2003; Van Kleef et
al., 2004). Consistent with this emotions-as-commitment devices view
(Frank, 1988, 2004; Hirshleifer, 1987), several studies have shown that indi
-
vidual differences in the capacity to experience feelings of guilt can translate
into rather large differences in the magnitude of monetary offers made in the
final round of a repeated ultimatum game (Haley & Fessler, 2004; Ketelaar
& Au, 2003). In one study, over 91% of individuals who felt guilty after pro-
posing an unfair monetary offer in the first round of play were later observed
to make a generous offer in the second round (Ketelaar & Au, 2003). By
contrast, less than 22% of individuals who reported no feelings of guilt (after
proposing a similarly unfair offer) were observed to make a generous offer in
the second round. Similar findings have been obtained in a repeated Pris-
oner’s Dilemma game, where it was observed that individuals who felt guilty
after behaving noncooperatively subsequently displayed 25% more coopera-
tion compared with those who felt no guilt (see Ketelaar & Au, 2003).
Finally, in studies of postdecision regret in the ultimatum game, it has been
shown that feelings of regret are a good predictor of subsequent offers when
the game is repeated with another partner (Zeelenberg & Beattie, 1997).
In one study, participants reported their level of regret after receiving
feedback informing them that their offer was either (a) only slightly higher
than the responders’ minimal acceptable offer, or (b) much higher than the
responders’ minimal acceptable offer. Not surprising, participants who re
-
ceived feedback that their offer was much higher than the responders’ mini
-
mal acceptable offer reported significantly more regret than participants who
were informed that their offer was only slightly higher than the responders’
minimal acceptable offer. Interestingly, when analyzed together, only feelings
of regret (and not feedback about minimal acceptable offers) were able to
predict the magnitude of proposers’ offers in the subsequent Ultimatum
Game.
Taken together, research on guilt and regret suggests that individuals
differ in their capacity to experience particular moral sentiments, and these
individual differences are predictive of subsequent decision behavior
104 KETELAAR

(Schwartz, Ward, Monterosso, Lyubomirsky, White, & Lehman, 2002;
Tangney, 1991). If self-focused moral emotions (e.g., regret, guilt) are a uni
-
versal solution to the so-called commitment problem (Frank, 1988), one might
wonder why we observe these individual differences in the capacity to expe
-
rience guilt and regret? Nowhere is this puzzle regarding individual differ
-
ences in moral sentiments more evident than in research examining other-
focused emotions and norm-enforcing behavior.
Other-Focused Emotions and Norm-Enforcing
Behavior
The existence of individuals willing to enforce norms is an intriguing para
-
dox for any model of decision making that treats individuals as purely self-
interested agents. This is the case because many norm-enforcing behaviors,
ranging from specific acts of punishment to credible threats directed at po
-
tential norm violators, appear to be altruistic acts (Fehr & Gächter, 2002).
The logic behind this claim is straightforward: Individuals who display
threatening signals and/or perform acts of punishment must directly incur
the costs of carrying out these actions, yet the benefits accrued by these
norm-enforcing behaviors are often distributed across the entire social group.
In this manner, one might argue that norm-enforcing behaviors constitute a
form of second-order public good (Fehr & Fischbacker, 2004). As such, one
expects that most self-interested agents will refuse to contribute to this pub-
lic good (fail to threaten or punish violators) and instead will free ride on the
actions of a few punisher types who enforce social norms for the entire group.
Interestingly, experimental economists have demonstrated that a significant
portion of individuals are often willing to incur a cost to inflict punishment
on others who violate social norms such as undercontributing in a public
goods game (Fehr & Fischbacker, 2004; Fehr & Gächter, 2002). This section
explores the claim that emotions operate as the proximate psychological
mechanisms underlying these norm-enforcing behaviors (Fehr & Gächter,
2002).
Just as certain moral sentiments seem to play a central role in motivating
individuals to obey norms, other sentiments appear to play a central role in
motivating individuals to enforce these norms. In particular, emotion re
-
searchers have identified three other-focused emotions—contempt, anger,
and disgust—that are activated by detecting violations of normative stan
-
dards (Rozin, Lowery, Imada, & Haidt, 1999). In addition to being associ
-
ated with distinct facial signals, these three moral sentiments appear to be
triggered by distinct types of norm violations (Haidt, 2003). Contempt, for
example, has been shown to be associated with violations of normative rules
regarding community standards and customs, as when one observes someone
failing to carry out his or her duties in the community or social hierarchy.
6. EMOTIONS AND ECONOMIC DECISION MAKING 105

Anger, on the other hand, is associated with violations of normative stan
-
dards governing one’s personal rights, as when one directly experiences
another person infringing on his or her personal liberties. Finally, disgust ap
-
pears to be triggered by violations of normative standards governing purity and
divinity, as when one observes another person disrespecting culturally shared
sacred beliefs or religious traditions. Moreover, these three moral sentiments
appear to be excellent candidates for relatively universal, culturally shared re
-
actions to these particular types of norm violations (Rozin et al., 1999).
This linkage between specific moral sentiments and specific types of norm
violations suggests that the same event can evoke any one of these three
moral sentiments depending, of course, on how the event is perceived. Con
-
sistent with this view, a sample of registered Democrats were observed in one
study to spontaneously display a range of moral emotions (contempt, anger,
and disgust) when asked how they felt about the very same event: the recent
presidential election in which the Republican candidate had won (Ketelaar,
2005). In this context, one supposes that those individuals who reacted with
contempt displays (62.5% of the sample) were focusing on the impact that
the election results had on their social group or community. Similarly indi-
viduals who reacted with anger (12.5%) may have been focusing on the im-
pact that the election results had on their personal rights, whereas individu-
als who reacted with disgust (25%) were presumably focusing on the impact
that the election results had on fundamental beliefs that they held sacred.
Such an interpretation would be consistent with the view that contempt, an-
ger, and disgust involve subjective perceptions that can rise to systematic ex-
periences of specific moral emotions in response to violations of specific types
of normative standards of appropriate behavior. The clear empirical evidence
for several distinct other-focused moral sentiments (contempt, anger, dis-
gust; see Haidt, 2003) leads one to ask why an individual would possess pow
-
erful emotions to condemn behaviors in others that they (the observer) are
rarely or never motivated to perform? Before discussing the intriguing ques
-
tion of whether it is in one’s material self-interest to experience these other-
condemning emotions, I briefly review empirical research that suggests that
these other-directed moral sentiments can indeed exert a strong influence on
economic decision making.
Several studies have demonstrated a clear link between anger and puni
-
tive behavior in economic bargaining games. In one study of investment be
-
havior in a public goods game, punishment took the form of the ability to de
-
duct monetary earnings from other members of the collective after each
round of play (Fehr & Gächter, 2002). Interestingly these acts of punishment
could only be made at a cost to the punisher and resulted in no monetary
gain for the punishing party. Consistent with the view that anger acts as a
moral sentiment that compels an individual to punish individuals who vio
-
late normative standards governing one’s personal rights, these punishing
106 KETELAAR

acts were significantly correlated with self-reports of anger and were typically
executed by above-average contributors and inflicted on below-average con
-
tributors (Fehr & Gächter, 2002).
Anger has also been shown to motivate individuals to punish norm viola
-
tors by spitefully rejecting monetary offers that appear to violate norms of
fairness. In one study of unfair ultimatum offers, it was observed that self-
reported feelings of anger were a better predictor of rejections of unfair offers
than actual perceptions of unfairness (Pillutla & Murnighan, 1996). One
reason that angry reactions may not be synonymous with perceptions of un
-
fairness may be due to the fact that some unfair offers are not the result of
intentional actions, and thus fail to elicit the appropriate attributions of re
-
sponsibility and blameworthiness that some theorists would claim are neces
-
sarily but not sufficient conditions for producing anger (Lerner, Goldberg, &
Tetlock, 1998; Ortony, Clore, & Collins, 1988; but see Kuppens, Van
Mechelen, Smits, & de Boeck, 2003). Along these lines, other studies have
shown that feelings of anger are strongly correlated with attributions of re
-
sponsibility and blameworthiness directed toward one’s negotiation partner
(Allred, Mallozzi, Matsui, & Rai, 1997). Consistent with the claim that an-
ger is a norm-enforcing emotion, other studies have manipulated partici-
pants’ perceptions of the emotional state of their interaction partner (angry
vs. happy) and found that participants make greater concessions when they
believe their negotiation partner is angry than when they believe their part-
ner is happy (van Kleef et al., 2004). In short, there appears to be a clear link
between the experience of anger and the motivation to punish individuals
who have violated norms of appropriate behavior, even when these acts of
punishment entail a cost for the angry enforcer. Because anger is just one of
several other-directed moral sentiments, future research might explore the
possible norm-enforcing role of additional other-focused emotions such as
contempt, disgust, and schadenfreude.
Game-Theoretic Insights Into Emotion-Based
Strategy Types
From the perspective of normative models of rational decision making, it is
clear that emotions are often associated with a number of peculiar economic
behaviors. Guilt-prone individuals often forgo monetary rewards by obeying
norms that prescribe cooperation rather than selfishness (Ketelaar & Au,
2003), and angry individuals often enforce norms by punishing norm viola
-
tors even when doing so provides a benefit to the group at a significant cost
to the punisher (Fehr & Gächter, 2002). As paradoxical as these norm-
obeying and norm-enforcing acts may appear to be, it must be emphasized
that not all individuals engage in such behavior. Thus, one might argue that
a complete understanding of the influence of emotion on norm-obeying and
6. EMOTIONS AND ECONOMIC DECISION MAKING 107

norm-enforcing behavior must also explain why only a certain portion of the
population behaves in this manner. In this final section, I review several evo
-
lutionary game-theoretic insights regarding how individual differences in the
capacity to experience certain moral sentiments could be functionally linked
to individual differences in strategic economic behavior (Ketelaar, 2004).
One of the most compelling explanations for the apparently irrational in
-
fluence of emotion on economic behavior centers on the claim that certain
emotions function as commitment devices that compel individuals to maxi
-
mize long-term payoffs at the expense of sometimes forgoing immediate re
-
wards (Frank, 1988, 2004; Hirshleifer, 1987). However, this commitment de
-
vice view of emotion does not explain why certain individuals (e.g., guilt-free
individualists and emotionless sociopaths) lack these important emotional
commitments. In other words, if one wishes to explain the disruptive influ
-
ence of emotion on economic decision making by saying that certain emo
-
tions, such as guilt or gratitude, serve important functions, one can ask why
all individuals do not experience these putatively functional sentiments. In
one study of the Ultimatum Game, for example, only 57% of individuals re-
ported experiencing feelings of guilt after proposing an unfair split of the
money (Ketelaar & Au, 2003). However, these individual differences in
emotion resulted in quite striking differences in ultimatum offers in the sec-
ond round of play. When the Ultimatum Game was repeated 1 week later,
over 91% of individuals who felt guilty (the previous week) now made gener-
ous offers, whereas only 22% of individuals who reported no feelings of guilt
made similarly generous offers.
Such strategically relevant individual differences are not limited to studies
of emotional influences on norm-obeying behavior; they also appear to play a
prominent role in studies of norm-enforcing behavior. In one study of altruis-
tic punishment, for example, approximately 16% of the sample never pun
-
ished other group members even when they contributed significantly less
than the group average in a public goods game. Moreover, only one third of
the sample consistently (across several rounds of play) punished group mem
-
bers who made deviant contributions, and these costly acts of punishment
were shown to be strongly linked to individual differences in the experience
of anger directed toward low contributors (Fehr & Gächter, 2002). In sum
-
mary, not all individuals are compelled by feelings of guilt to resist the temp
-
tation to defect (Ketelaar & Au, 2003), and not everyone becomes angry
enough to incur the cost of punishing norm violators (Fehr & Gächter,
2002). Although the emotions-as-commitment-device approach is able to
explain how emotions can compel individuals to obey norms or punish norm
violators, such models leave open the question of why not everyone behaves
as if they are emotionally committed to these norm-obeying and norm-
enforcing actions.
108 KETELAAR

Strategies as Decision Rules
Although Adam Smith (1759) explained individual differences in the capac
-
ity to experience moral sentiments in terms of a failure to exercise self-
command over these emotions, one can also utilize evolutionary game theory
to explore another, more intriguing explanation for these individual differ
-
ences. Specifically, one can examine the possibility that individual differ
-
ences in the capacity to experience moral sentiments may represent a rela
-
tively stable (frequency-dependent
2
) distribution of emotion-based strategy
types (Ketelaar, 2004; Mealey, 1995). One way to think about individual dif
-
ferences in emotion-based strategy types is to think of strategies as analogous
to computer-programmed decision rules (Binmore, 1998). According to this
way of thinking, there are at least two ways that multiple strategies (i.e., indi
-
vidual differences in strategic behavior) can manifest themselves in a popula
-
tion of agents. First, a population may contain multiple behavioral strategies
because the population consists of several distinct strategy types, each corre
-
sponding to a distinct decision rule governing each agent’s behavior (e.g.,
compare “IF you cooperate, THEN I will cooperate” to “IF you cooperate,
THEN I will defect”). A second reason that one might observe different be-
havioral strategies in the same population concerns the presence of certain
types of reactive strategies (see Tooby & Cosmides, 1990a, 1990b). In other
words, a population may consist of just one strategy type, but this strategy
type may correspond to a complex decision rule that generates different be-
havioral phenotypes contingent on the phenotypic characteristics of the
other agents that it encounters. For example, imagine a decision rule of the
form “IF I am weaker than you AND IF you cooperate, THEN I will cooper-
ate; ELSE IF I am stronger than you AND IF you cooperate, THEN I will de-
fect.” A reactive strategy of this form appears cooperative when interacting
with strong opponents and noncooperative when encountering weak oppo
-
nents. The presence of such a reactive strategy might lead one to the errone
-
ous conclusion that their exists two strategy types in the population (cooper
-
ators and noncooperators), when, in reality, the population is comprised of
just one strategy type that reacts differentially to particular phenotypic char
-
acteristics (e.g., relative strength) that vary across agents.
Finally, there exists a third means of obtaining a diversity of strategy types
in a single population of agents, and I argue that this explanation may be key
to understanding the influence of emotions on economic decision making.
This third explanation for multiple strategies is a hybrid of the two types of
6. EMOTIONS AND ECONOMIC DECISION MAKING 109
2
2
Frequency-dependent selection occurs when the evolutionary success of a strategy type is
dependent on the relative frequency of certain other strategy types in the population (Maynard
Smith, 1982).

explanation discussed earlier. According to this hybrid view, all agents pos
-
sess essentially the same reactive decision rule for determining strategic be
-
havior (e.g., IF X THEN Y, but only IF B
C), yet each agent differs in
terms of the actual values they assign to certain parameters in this decision
rule. To appreciate this idea, imagine an evolved decision rule comprised of a
number of parameters (X, Y, B, C, etc.), some of which are invariant across
individuals and others that vary systematically across individuals. Consider,
for example, a decision rule that essentially translates into the following
form: “IF you fail to contribute to the public good by some factor X, THEN I
will punish you by some factor Y, but only IF the benefit B of inflicting this
punishment is greater than the cost C associated with punishing you.”
3
Fur
-
ther imagine that the amount of punishment Y that is inflicted on a norm vi
-
olator is a function of the degree of norm violation entailed by X and, more
-
over, that this parameter is some fixed function [Y = f(X)] shared by all
agents in the population. Thus, all agents who encounter a given instance of
norm violation will necessarily punish the same amount as determined by the
function Y = f(X).
However, now imagine that the two cost–benefit parameters (C and B in
this hypothetical decision rule) are determined by natural selection accord-
ing to their evolutionary success. Because agents are constantly interacting
with each other across time and space, one might imagine that natural selec-
tion could eventually converge on a single algorithm that determines how all
agents will calculate these cost–benefit parameters (C and B). Because the
algorithm that calculates these cost–benefit parameters would essentially
correspond to a sort of threshold for determining whether an agent will pun-
ish a particular instance of norm violation, if all agents possess the same algo-
rithm for calculating these parameter values, all agents will essentially pos-
sess the same behavioral phenotype in regard to punishing norm violators. By
contrast, if different agents possessed somewhat different algorithms for cal
-
culating these parameters (costs and benefits of punishing), this would nec
-
essarily generate several distinct behavioral phenotypes (i.e., punishers and
nonpunishers). Therein lies the rub, or at least a possible evolutionary source
for individual differences in punishment behavior or any other strategic be
-
havior for that matter.
4
The only difference between this explanation for
multiple strategies and the reactive strategy explanation (discussed earlier) is
the subtle point that several distinct strategy types can emerge merely as a
result of subtle differences in how certain parameter values are calculated in
the agent’s decision rule. How might this work?
110 KETELAAR
3
3
This isa purelyhypothetical decisionrule thatis merelyused toillustrate howcertain parame
-
tersof decisionrulesmightbesubject tofrequency-dependentselection.Itis notmeanttobetaken
literally as a model of the determinants of norm-enforcing behaviors (e.g., punishment).
4
4
This same logic could be used to explain individual differences in the proclivity to reward
other agents.

If agents are analogous to preprogrammed decision rules and evolutionary
selection pressures do not converge on just one method for calculating pa
-
rameter values for all agents, the population, by definition, will consist of
several different strategy types. For example, if evolutionary dynamics al
-
lowed for different agents to possess different values for the cost–benefit pa
-
rameters contained in the hypothetical punishment decision rule “IF X
THEN Y, but only IF B
C,” this could result in a diversity of strategy types
in regards to their propensity to punish norm violators. Depending on how
these cost–benefit parameters varied, some versions of this rule would dictate
that certain agents never punished (punishment would always be perceived
as too costly), whereas other versions of this rule, with somewhat different
cost–benefit parameters, would dictate that certain agents would invariably
punish every instance of norm violation that they encountered (it would al
-
ways be perceived as beneficial to punish).
Again one could imagine that evolutionary forces might invariably con
-
verge on just one method of calculating these cost–benefit parameters for all
agents, whereby the population of agents (decision rules) that emerges is at
some sort of game-theoretic equilibrium. Yet it turns out that the ecological
conditions under which such decision rules often operate (i.e., indefinitely
repeated interactions) lend themselves to a vast number of possible equilibria
and, thus, an equally vast number of successful decision rules. The implica-
tion of this insight—known as the folk theorem—is that evolutionary selec-
tion pressures are capable of generating a vast number of successful distribu-
tions of strategy types (each consisting of a distinct decision rule) and, thus,
one must consider the possibility that human populations may be comprised
of a multitude of decision rules (not just one) governing the implementation
of norm-obeying and norm-enforcing behaviors.
Evolutionary game theory gives us several hints at how dynamic forces
(such as those encapsulated in the concept of frequency-dependent selection
and the folk theorem) could give rise to more than one equilibria in a single
population of agents (Binmore, 1998; Boyd, 1989; Boyd & Richardson, 1992;
Maynard Smith, 1982). For example, if one considers strategy types as analo
-
gous to finite decision rules (finite automata) that encounter one another in
indefinitely repeated games, an interesting phenomenon emerges. Spe
-
cifically, one observes that:
any two finite automata playing each other for long enough in a repeated game
will eventually end up cycling through the same sequence of plays. A player’s
per-game payoff can therefore be calculated simply by taking the average pay
-
off in a cycle. (Binmore, 1998, p. 301)
For example, if the finite strategy of tit for tat interacts with itself (or any
other finite strategy that is never the first to defect), these two strategies will
6. EMOTIONS AND ECONOMIC DECISION MAKING 111

end up in a continuous cycle of mutual cooperations and obtain whatever
payoffs are associated with this particular cycle. By contrast, tit for tat inter
-
acting with a finite strategy of always defect will end up in a continuous cycle
of mutual defections and obtain a necessarily different set of payoffs associ
-
ated with this very different cycle. The folk theorem emerges from the simple
fact that, because finite strategies invariably end up infinitely cycling through
a particular sequence of outcomes
5
(mutual cooperation, mutual defection,
etc.), this sets up an incentive for either agent (i.e., each finite decision rule)
to locate and shift to an alternative pairing of strategies that can generate a
different cycle with a greater payoff than that obtained in their current cycle
(Binmore, 1998). It follows that indefinitely repeated games that occur in a
diverse population of strategy types can allow for a great variety of equilibria
simply because this context (indefinitely repeated interactions, different
strategy types) allows for a vast number of more profitable alternative cycles
on which various strategy types can coordinate (see Binmore, 1998; Lom
-
borg, 1996, for a fuller treatment of this so-called folk theorem). This argu
-
ment depends, of course, on agents being able to maintain credible commit-
ments to sticking with a particular alternative cycle of payoffs (Lomborg,
1996). The claim here is that emotions are a plausible source of these credi-
ble commitments (Frank, 1988, 2004; Hirshleifer, 1987, 2002; Ketelaar,
2004). Because individuals who experience different emotions (e.g., guilt vs.
no guilt, anger vs. no anger) often end up making quite different choices in
economic bargaining games, it may be useful for future research to explore
the possibility that individual differences in moral sentiments constitute one
source of these individual differences in strategy types.
CONCLUSIONS
Moral sentiments—such as feelings of anger—are easily labeled as irrational
when they are observed to motivate individuals to engage in behaviors (e.g.,
altruistic punishment) that benefit others, but are costly to the individual
who performs them. To the degree that the psychological mechanisms un
-
derlying our emotions have been shaped by natural selection, one wonders
how adaptively designed moral sentiment systems could routinely generate
preferences that compel individuals to pursue outcomes with lower (rather
than higher) payoffs. This chapter argues that an intriguing answer to this
question may lie in the realization that not all individuals confront identical
112 KETELAAR
5
5
The claim is not that a cycle will invariably end up repeating just one type of outcome over
and over again,such asastring ofcontinuous mutualdefectionsor astring of continuousmutual
cooperations. These cyclesmay correspond to agreat variety of ratherrandom-looking, but ulti
-
mately repeating, sequences of outcomes (see Binmore, 1998, for a nice exposition of this argu
-
ment).

payoffs even when they find themselves in the same situation (see Kelley,
1984; Kelley & Thibaut, 1978). For example, an agent who embodies the an
-
gry punisher type may experience a quite different set of payoffs associated
with inflicting punishment as compared with a relatively angerless, non
-
punishing type. In this manner, if one is forced to select just one set of pay
-
offs as the rational way of viewing the utilities associated with punishing and
not punishing, by definition, one of these two agents (strategy types) will be
viewed as possessing irrational preferences. Yet if the payoffs associated with
punishment are determined not by one theoretical gold standard, but rather
by virtue of the evolutionary (including frequency-dependent) success of the
specific parameter settings contained in an agent’s decision rule, one might
wonder whether the specific parameter values, and the corresponding “payoff
estimates” that they generate, must necessarily be identical for all agents.
Just as the predicted payoff for approaching elephant dung ineluctably de
-
pends on whether you are a human or a dung beetle, it may turn out that the
ultimate future payoff associated with punishing norm violators depends on
which strategy type you are, as well as the particular distribution of other
strategy types in the same environment. In a population comprised largely of
defectors, for example, it is more costly to be a punisher than in an environ-
ment comprised largely of norm-obeying agents. Along these lines, it was ar-
gued that individual differences in the capacity to experience certain moral
sentiments might correspond to frequency-dependent individual differences
in the settings (values) of certain decision rule parameters that influence
norm-obeying and norm-enforcing behavior. The obvious implication is that,
by virtue of their ability to generate individual differences in how payoffs are
assigned to outcomes, emotion-based strategies might play a significant role
in explaining why some decision makers often do not act in the strictly ra-
tional manner that some normative models suggest that they should behave.
Along these lines, the present chapter argued that a large part of the puzzle
concerning why irrational emotions so often intrude on our most important
decisions lies in recognizing this important link between moral sentiments
and judgments of normatively acceptable and unacceptable behavior.
REFERENCES
Allred,K. G., Mallozzi,J.S., Matsui, F.M.,&Rai, C.P.(1997).The influence ofangerandcom
-
passion on negotiation performance. Organizational Behavior and Human Decision Processes,
70, 175–187.
Binmore, K. (1998). Game theory and the social contract: Volume 2. Just playing. Cambridge, MA:
MIT Press.
Bowlby, J. (1969). Attachment and loss: Volume 1. Attachment. New York: Basic Books.
Boyd, R. (1989). Mistakes allow evolutionary stability in the repeated prisoner’s dilemma game.
Journal of Theoretical Biology, 136, 47–56.
6. EMOTIONS AND ECONOMIC DECISION MAKING 113

Boyd, R., & Richardson, P. J. (1992). Punishment allows the evolution of cooperation (or any
-
thing else) in sizeable groups. Ethology and Sociobiology, 13, 171–195.
Clore, G.L. (2005).For loveor money:Some emotionalfoundations ofrationality. TheChicago-
Kent Law Review.
Damasio, A. R. (1994). Descartes’error: Emotion,reason andthe humanbrain. NewYork: Grosset/
Putnam.
Elster, J. (1995). Rationality and the emotions. The Economic Journal, 116, 1386–1397.
Elster, J. (2000). Strong feelings: Emotion, addiction, and human behavior. Cambridge, MA: MIT
Press.
Fehr,E., &Fischbacker,U. (2004). Third-partypunishmentand socialnorms.Evolution and Hu
-
man Behavior, 25, 63–87.
Fehr, E., & Gächter, S. (2002). Altruistic punishment in humans. Nature, 10, 137–140.
Frank, R. H. (1988). Passions within reason: The strategic role of the emotions. New York: W. W.
Norton.
Frank, R. H. (2004). Introducing moral emotions into models of rational choice. In A. S. R.
Manstead, N.H. Frijda,& A.H. Fischer(Eds.), Feelings and emotions: The Amsterdam sympo
-
sium (pp. 422–440). New York: Cambridge University Press.
Gardner, H. (1987). The mind’s new science: A history of the cognitive revolution. New York: Basic
Books.
Gilovich, T., Griffin,D., & Kahneman,D. (2002).Heuristicsand biases:Thepsychology ofintuitive
judgment. Cambridge: Cambridge University Press.
Haidt, J. (2001). The emotional dog and its rational tail: A social intuitionist approach to moral
judgment. Psychological Review, 108, 814–834.
Haidt, J. (2003). The moral emotions. In R. J. Davidson, K. R. Scherer, & H. H. Goldsmith
(Eds.), Handbook of affective sciences (pp. 852–870). Oxford: Oxford University Press.
Haley, K. J., & Fessler, D. M. T. (2004, July). The effects of guilt induction on decision-making
in two economic games. In T. Ketelaar (Chair), Emotion in experimental economics. Sympo-
sium presented at the annual meeting of the Human Behavior and Evolution Society
(HBES), Berlin, Germany.
Hirshleifer, J. (1987). On the emotions as guarantors of threats and promises. In J. Dupré (Ed.),
The latest on the best: Essays on evolution and optimality (pp. 307–326). Boston: Bradford
Books–MIT Press.
Hirshleifer,J.(2002).Game-theoretic interpretations ofcommitment.InR.M.Nesse(Ed.),Evo
-
lution and the capacity for commitment. New York: Russell Sage Foundation.
Hogarth, R. M., & Reder, M. W. (1987).Rational choice: The contrast betweeneconomics and psy
-
chology. Chicago: The University of Chicago Press.
Hume, D. (1740/1978). A treatise of human nature. Oxford: Clarendon Press.
Hume, D. (1777/1996). Enquiries concerning human understanding and concerning the principles of
morals. Oxford: Oxford University Press.
Kelley, H. H. (1984). Affect in interpersonal relations. Review of Personality & Social Psychology,
5, 89–115.
Kelley, H. H., & Thibaut, J. W. (1978). Interpersonal relationships. New York: Wiley.
Ketelaar, T. (2004). Ancestral emotions, current decisions: Using evolutionary game theory to
explore the role of emotions in decision making. In C. Crawford & C. Salmon (Eds.), Evolu
-
tionary psychology, public policy and personal decisions (pp. 145–168). Mahwah, NJ: Lawrence
Erlbaum Associates.
Ketelaar, T. (2005). The display of moral sentiments in response to disappointing election results.Un
-
published manuscript, New Mexico State University.
Ketelaar, T., &Au, W. T.(2003). The effectsof guiltyfeelingson thebehaviorof uncooperative
individuals in repeated social bargaining games: An affect-as-information interpretation of
the role of emotion in social interaction. Cognition & Emotion, 17, 429–453.
114 KETELAAR

Ketelaar, T., Au, W. T., Hu, W. P., Tomala, M., Kopins, B., & Steiner, R. (2006). A cross-
cultural study ofmoral emotions as reactions to violations ofnormative standards of appropriatebe
-
havior. Manuscript in preparation, New Mexico State University.
Ketelaar, T., & Clore, G. L. (1997). Emotions and reason: The proximate effects and ultimate
functions of emotions. In G. Matthews (Ed.), Personality, emotion, and cognitive science (pp.
355–396). Amsterdam: Elsevier Science Publishers (North-Holland).
Ketelaar, T., & Todd, P. (2001). Framing our thoughts: Ecological rationality as evolutionary
psychology’s answer to the frame problem. In H. R. Holcomb III (Ed.), Conceptual challenges
in evolutionary psychology (pp. 179–211). Dordrecht: Kluwer Academic Publishers.
Kohlberg, L. (1971). From is to ought: How to commit the naturalisticfallacy and getaway with
itin the studyofmoral development. InT.Mischel(Ed.), Cognitivedevelopmentandepistemol
-
ogy (pp. 151–235). New York: Academic Press.
Kuppens,P., VanMechelen,I., Smits,D.J. M.,&de Boeck, P.(2003).The appraisalbasisof an
-
ger: Specificity, necessity, and sufficiency of components. Emotion, 3, 254–269.
Lerner, J. S., Goldberg, J. H., & Tetlock, P. E. (1998). Sober second thought: The effects of ac
-
countability, anger, and authoritarianism on attributions of responsibility. Personality and So
-
cial Psychology Bulletin, 24, 563–574.
Lewis, M.(1993). Self-consciousemotions: Embarrassment,pride, shame,and guilt.In M.Lewis
& J. M. Haviland (Eds.), Handbook of emotions (pp. 223–236). New York: Guilford.
Lomborg, B. (1996). Nucleus and shield: The evolution of social structure in the iterated Pris
-
oner’s Dilemma. American Sociological Review, 61, 278–307.
Maynard Smith, J. (1982). Evolution and the theory of games. Cambridge: Cambridge University
Press.
Mealey,L. (1995).Thesociobiology of sociopathy:Anintegrated evolutionary model.Behavioral
and Brain Sciences, 18, 523–599.
Norton, D. F. (1993). Hume, human nature, and the foundations of morality. In D. F. Norton
(Ed.), The Cambridge companion to Hume. Cambridge: Cambridge University Press.
Nussbaum, M.(2004). Emotionsas judgmentsof valueand importance.In R. C. Solomon (Ed.),
Thinking about feeling (pp. 183–199). Oxford: Oxford University Press.
Ortony, A., Clore, G. L., & Collins, A. (1988). The cognitive structure of emotions. New York:
Cambridge University Press.
Piaget, J. (1932/1965). The moral judgment of the child (M. Gabain, Trans.). New York: The Free
Press.
Pillutla, M.M., &Murnighan, J.K. (1996).Unfairness, anger,and spite: Emotional rejections of
ultimatum offers. Organizational Behavior and Human Decision Processes, 68, 208–224.
Rozin, P., Lowery, L., Imada, S., & Haidt, J. (1999). The CAD triad hypothesis: A mapping be
-
tween three moral emotions (contempt, anger, disgust) and three moral codes (community,
autonomy, divinity). Journal of Personality and Social Psychology, 76, 574–586.
Scherer, K. R. (1984). On the nature and function of emotion: A component process approach.
InK.R. Scherer &P.Ekman(Eds.),Approachesto emotion(pp.293–318).Hillsdale,NJ:Law
-
rence Erlbaum Associates.
Schwartz, B., Ward, A., Monterosso, J., Lyubomirsky, S., White, K., & Lehman, D. R. (2002).
Maximizingversus satisficing:Happinessis a matterofchoice. JournalofPersonality and Social
Psychology, 83, 1178–1197.
Smith, A. (1759/2000). The theory of moral sentiments. New York: Prometheus.
Tangney,J.P.(1991). Moral affect:Thegood,thebad,andtheugly. Journal ofPersonalityandSo
-
cial Psychology, 61, 598–607.
Tooby, J., & Cosmides, L. (1990a). On the universality of human nature and the uniqueness of
the individual: The role of genetics and adaptation. Journal of Personality, 58, 17–67.
Tooby, J., & Cosmides, L. (1990b). The past explains the present: Emotional adaptations and
the structure of ancestral environments. Ethology and Sociobiology, 11, 375–424.
6. EMOTIONS AND ECONOMIC DECISION MAKING 115

Van Kleef, G.A., DeDreu,C. W.W., &Manstead,A. S.R. (2004). Theinterpersonal effectsof
emotion innegotiations: Amotivated information processing approach. Journalof Personality
and Social Psychology, 87, 510–528.
von Neumann, J., & Morgenstern, O. (1944). The theory of games andeconomic behavior. Prince
-
ton, NJ: Princeton University Press.
Zeelenberg, M. (1999). Anticipated regret, expected feedback and behavioral decision-making.
Journal of Behavioral Decision Making, 12, 93–106.
Zeelenberg, M., & Beattie, J. (1997). Consequences of regret aversion: 2. Additional evidence
for effects of feedback on decision making. Organizational Behavior and Human Decision Proc
-
esses, 72, 63–78.
116 KETELAAR