Professional Documents
Culture Documents
Emotions and Economic Theory Jon E PDF
Emotions and Economic Theory Jon E PDF
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms
American Economic Association is collaborating with JSTOR to digitize, preserve and extend
access to Journal of Economic Literature
This content downloaded from 200.12.184.4 on Tue, 13 Nov 2018 01:50:33 UTC
All use subject to https://about.jstor.org/terms
Journal of Economic Literature
Vol. XXXVI (March 1998), pp. 47-74
JON ELSTER
Department of Political Science, Columbia University
1 am7i gratefiul to Robert Frank, George Loewenstein, and an. anonymtiouis referee for thei-
comments orn. ani. earlier draft of this article. Many of the issue.s raisecl here are discussed at
greater length in Jon, Elster (forthcoining).
47
This content downloaded from 200.12.184.4 on Tue, 13 Nov 2018 01:50:33 UTC
All use subject to https://about.jstor.org/terms
48 J ournal of Economic Literature, Vol. XXXVI (March 1998)
solved questions in emotion theory ration, and liking.2 Second there are
(Paul Ekman and Richard Davidson various counterfactual emotions gener-
1994), the only one that is related to ated by thoughts about what might have
behavior is "What is the function of happened but didn't: regret, rejoic-
emotions?" A question which is lacking ing, disappointment, elation (Jonathan
from their survey-not because it has Baron 1994, pp. 367-72). Third, there
been answered, but because it has not are emotions generated by the thought
even attained the status of an unre- of what may happen: fear and hope.
solved problem-is the following: "How Fourth, there are emotions generated
can emotions help us explain behavior by good or bad things that have hap-
for which good explanations seem to be pened: joy and grief. Fifth, there are
lacking?" This is the main focus of the emotions triggered by the thought of
present article. the possessions of others: envy, malice,
I shall proceed as follows. In Section indignation, and jealousy. Finally there
2, I briefly summarize the features of are cases that do not fall neatly into any
emotions emphasized by psychologists. special category, such as contempt, dis-
In Section 3, I address the question gust, and romantic love or "limerence"
whether emotions can be chosen, be it (see 5.5). Borderline or controversial
for their intrinsic value or for their in- cases include surprise, boredom, inter-
strumental usefulness. In Section 4, I est, sexual desire, enjoyment, worry,
discuss the converse issue, whether and frustration. All of these emotions
emotions can help us make choices also allow for innumerable variations
when rationality by itself is insuffi- and nuances, depending on the exact
cient. In Section 5, I consider what is nature of the beliefs that trigger them.
probably the most important issue in Thus if I believe that another has vio-
this area, viz. how emotions may com- lated my interest, I may feel anger; if I
bine with other motivations such as believe that in doing so he has also vio-
rational self-interest to produce be- lated a norm, I feel indignation.
havior. Section 6 offers a brief conclu- Whether these emotions are universal
sion. or culture-specific remains an unre-
solved issue (see for instance the essays
2. What Are Emotions? in Richard Shweder and Robert LeVine,
eds. 1984). A plausible (but unproved)
There is a large degree of consensus intermediate position is that all or most
in the scholarly literature on what emo- of the enumerated emotions are univer-
tions there are, and a quite good agree- sal in the sense that their typical physi-
ment on what emotions are. ological and behavioral expressions are
2.1 What Emotions There Are found in all societies, but that some so-
cieties may lack a cognitive label for a
Among the states that unambiguously given emotion. Thus it has been argued
qualify as emotions we may first list that the notion of guilt did not exist in
various social emotions: anger, hatred, classical Greece (Bernard Williams
guilt, shame, pride, pridefulness, admi-
This content downloaded from 200.12.184.4 on Tue, 13 Nov 2018 01:50:33 UTC
All use subject to https://about.jstor.org/terms
Elster: Emotions and Economic Theorq 49
1993), that in Western Europe the no- causal histories. This being said, the
tion of romantic love did not exist until features to be enumerated seem to de-
the eleventh century (Clive S. Lewis fine a roughly homogeneous class for
1936), that it is only recently that bore- the purpose of explaining behavior.
dom has come be viewed as an involun-
2.3 Cognitive Antecedents
tary mental state rather than as a sin
(Patricia M. Spacks 1995), and that Ta- Emotions differ from other "visceral
hitians diagnose what we would call de- factors" (Loewenstein 1996) such as
pression as mere fatigue (Robert Levy pain, hunger and drowsiness in that
1973). Conceptualization of a state as they are triggered by beliefs. Although
an emotion is a necessary condition for some emotions may be triggered by sen-
that emotion to trigger meta-emnotions sory signals rather than by cognitions
as when we feel shame of envy or guilt (LeDoux 1996), and pain as well as hun-
about our anger. It is also a necessary ger may be influenced by beliefs, the
condition for triggering what one might contrast remains broadly valid. Whereas
call second-party emotions as when we studies of other visceral motivations, in-
feel contempt for another person's fear cluding addictive cravings, can learn
or hope that our love is requited. much from the study of animal behav-
ior, the central role of beliefs in the for-
2.2 What Emotions Are
mation of emotions implies a limited
The emotions listed above are typi- role for animal studies. Objects of be-
cally distinguished from each other, as liefs may be observable or unobserv-
well as from non-emotional mental able, physical or mental, and real or
states, by a number of observable fea- imagined. Animals can only form beliefs
tures. I shall use six features to define about real, physical objects. These need
the emotions: cognitive antecedents, in- not be observable. We know from
tentional objects, physiological arousal, many studies that animals are capable
physiological expressions, valence, and of forming mental representations of
action tendencies. These were already physical objects that are absent from
noted by Aristotle and remain central in the present sensory field (Elster 1983a,
modern discussions (e.g., Nico Frijda pp. 132-33). But there is no evidence
1986). There is not a perfect fit be- that animals can form beliefs about
tween the definition and the phenom- mental states such as beliefs, emotions
ena enumerated in 2.1. In fact, there and motivations, or that they can form
does not seem to be a single feature counterfactual beliefs.
that is invariably found in all emotional
2.4 Intentional Objects
phenomena. Joseph LeDoux (1995, pp.
1049-50; 1996, pp. 126-27) suggests Emotions also differ from other vis-
that the idea of "emotion" may in fact ceral factors in that they have an inten-
not be a useful scientific category, in tional object: they are about something.
the sense of not forming what philoso- The object of an emotion may be a per-
phers of science call "a natural kind." son (envy of a wealthier person) or a
When we know more about the neurobi- state of affairs (indignation with world
ology of the various emotions, we may poverty). Usually, the object of the
come to see them as being made up of emotion is closely tied to its cognitive
several classes that relate to each other antecedent. The object of anger is the
as whales to sharks or birds to bats-su- person whom we believe to have hurt
perficially similar but with different us. In some cases, the object may be in-
This content downloaded from 200.12.184.4 on Tue, 13 Nov 2018 01:50:33 UTC
All use subject to https://about.jstor.org/terms
50 Journal of Economic Literature, Vol. XXXVI (March 1998)
This content downloaded from 200.12.184.4 on Tue, 13 Nov 2018 01:50:33 UTC
All use subject to https://about.jstor.org/terms
Elster: Emotions and Economic Theory 51
control of the will. Nobody can blush at urges or impulses." The action tendency
will, but some people can make them- of shame is to hide or disappear; that of
selves cry. It is possible to imitate facial guilt, to make atonements or to confess;
expressions and bodily postures that that of envy and malice, to destroy; that
signal fear, anger, sadness, and joy, of love, to approach and touch the other
although most people can't do it (Ek- person; that of anger, to hurt the person
man 1992). In cultures where words are who has hurt oneself. Fear has two ac-
usually accompanied by vivid move- tion tendencies: fight or flight. Pride,
ments of the hands, lying can be de- regret, and some other emotions seem
tected by the absence of these gestures, to have none.
because the concentration required for An action tendency may lead to im-
a convincing lie interferes with the mediate action or be modified by one of
spontaneous gesturing. Most people are several regulatory systems. In the for-
not capable of such detection, however, mer case, by assumption, no choice is
at least not without special training (Ek- involved. In the latter case, too, the
man 1992). modification of the action tendency may
occur without choice, through inhibi-
2.7 Valence
tory mechanisms of the autonomic nerv-
This is the term psychologists use to ous system (Frijda 1986, pp. 158-61,
denote the fact that emotions can be lo- 405). From the choice-theoretic point
cated on a pleasure-pain scale, with a of view, social regulation of action ten-
neutral zero point of emotional indiffer- dencies is the more important phe-
ence. Although emotions that are high nomenon. Social norms can inhibit
on arousal also tend to be high on posi- spontaneous action tendencies. A fear-
tive or negative valence, there are ex- induced tendency to flee may be kept in
ceptions. First, because arousal itself check by norms against cowardice (Nor-
has many components that co-vary only man Dixon 1976, p. 197). The destruc-
imperfectly with each other (Frijda tive urge in envy may be inhibited by
1986, pp. 168-71), the very idea of a the strong condemnation of envy that is
correlation between valence and arousal found in virtually all societies. Yet so-
is ambiguous. Second, an emotion may cial norms can also amplify action ten-
be high on arousal and low in valence dencies, as in the case of revenge (see
(embarrassment), and vice versa (bore- 5.5).
dom). Third, mixed emotions such as
the bittersweet feeling of nostalgia may 3. Choosing Emotions
be high on arousal but have zero net va-
lence. This being said, for most of the Emotions can be valuable or useful.
emotions that matter for behavior high Joy and love are intrinsically pleasant,
arousal goes together with high valence. and anger can be useful to deter an en-
Both contribute to the urgency of ac- emy. They can also be undesirable or
tion tendencies and to the short-circuit- harmful. Shame and grief are intrinsi-
ing of cognition. cally unpleasant, and fear may reduce
one's capacity to deal with danger and
2.8 Action Tendencies
also expose one to the disapproval of
As defined by Frijda (1986, pp. 70, others. Hence it would seem natural to
78), these are "states of readiness to ask whether emotions-having them or
execute a given kind of action. . . . Ac- avoiding them-could be the object of
tion tendencies have the character of rational choice. To the extent that emo-
This content downloaded from 200.12.184.4 on Tue, 13 Nov 2018 01:50:33 UTC
All use subject to https://about.jstor.org/terms
52 Journal of Econonmic Literature, Vol. XXXVI (March 1998)
tions are valued for their impact on oth- which allows non-cooperators to sur-
ers, we may also ask whether people can vive.3
choose the expressions on the basis of People may also simulate emotional
which others impute emotions to one- expressions for purposes of deterrence.
self. I shall first discuss some economic Richard Nixon deliberately cultivated
models that address the last issue, and an appearance of erratic and emotional
then consider some that rely on the idea behavior, in order to persuade the Sovi-
that the emotions themselves can be ets that he could not be counted on to
chosen. react rationally to a first strike (Walter
Isaacson 1992, pp. 163-64, 181-82).
3.1 Choosing Emotional Expressions
Even though others may suspect that
In a model of cooperative behavior, the person who behaves emotionally is
Frank (1988) relies on the fact that doing so merely to cultivate an appear-
emotional expressions are under the ance of irrationality, the possibility that
partial control of the agent. His argu- he might be truly irrational can be suffi-
ment has three premises. First, emo- cient to modify their behavior. Thus
tional dispositions are imperfectly in the finitely iterated prisoner's di-
indicated to others by the outward lemma, cooperation can be sustained al-
expressions of occurrent emotions. most up to the last game if there is
Second-and that is why the indica- common knowledge that one of the
tion is only imperfect-even those players may be irrational (David Kreps
who do not have the dispositions can et al. 1982).
simulate the expressions. Third, simula-
tors can be detected, at some (fixed) 3.2 Choosing Emotions
cost to the detectors. Applying these
premises to a world in which agents To discuss the question whether emo-
engage in one-shot prisoner's di- tions can be chosen, I shall make two
lemma interactions with others, Frank distinctions. First, there is a distinction
shows that there will be some specific between occurrent emotions and emo-
proportion of honest and dishonest tional dispositions. In theory, a person
individuals in the equilibrium state might be irascible but never angry, if
(frequency-dependent polymorphism). 3The model is elegant and not implausible. I
In that state, the honest individuals would like to question, though, its mixed evolu-
are exactly indifferent between scruti- tionary-cum-intentional basis. To explain the
emergence of honesty, Frank appeals to natural
nizing and not scrutinizing their inter- selection. If we start off with a share of coopera-
action partners, because the sum of tors less than the equilibrium share, "the coopera-
the direct and opportunity costs of scru- tors will get a higher average payoff, which means
that their share in the population will grow"
tinizing a potential partner are ex- (Frank 1988, p. 62). To explain the emergence of
actly equal to the expected opportunity scrutiny, however, he appeals to "the decision fac-
ing a cooperator who is trying to decide whether
cost of not scrutinizing him. The latter
to pay the cost of scrutiny" (Frank 1988, p. 61).
cost is the probability that a partner Rather than assuming that there is a gene for scru-
chosen at random will be dishonest tinizing behavior as well as one for honest behav-
multiplied by the difference between ior he argues that scrutiny is a matter of rational
decision. I find it hard to imagine the early homi-
the cooperative payoff and the "sucker" nids-to whom this argument is presumably meant
payoff in the prisoner's dilemma. When to apply-making that kind of decision. The alter-
the proportion of cooperators exceeds native view-that there is a gene for scrutinizing-
is also implausible, because on Frank's theory it
the equilibrium share, the probability would have to be switched off and on according to
falls so that scrutiny no longer pays, the share of cooperators in the population.
This content downloaded from 200.12.184.4 on Tue, 13 Nov 2018 01:50:33 UTC
All use subject to https://about.jstor.org/terms
Elster: Emotions and Economic Theory 53
CHOSEN BY
Self Otlhers
Direct
Positive Negative
(self-control)
Occurent Oratory
emiiotions "Sliaiing"
Indirect"Smig
Positive Negative
(Olympic Gamiies) (Becker)
Inducing guilt
Eimiotional (Becker; Frank)
Edispostionsl Psycliotlherapy
dispositions Psychotherapy Inducing guilt or slhamiie
(Kandel anid Lazear)
Table 1.
This content downloaded from 200.12.184.4 on Tue, 13 Nov 2018 01:50:33 UTC
All use subject to https://about.jstor.org/terms
54 Journal of Economic Literature, Vol. XXXVI (March 1998)
This content downloaded from 200.12.184.4 on Tue, 13 Nov 2018 01:50:33 UTC
All use subject to https://about.jstor.org/terms
Elster: Emotions and Economic Theory 55
sheet. It seems plausible that these good face on things" (Montaigne 1991,
emotional gains by themselves were pp. 731-32). Yet as is clear from these
large enough to justify the huge con- examples, the avoidance method has its
struction expenses. Yet because this ex- costs. People probably make different
perience could not have been planned, tradeoffs between the costs and bene-
the expenses may not have been justi- fits of staying away from situations on
fied ex ante. The point is not that no- which they might suffer an emotional
body could count on the Norwegians reaction they want to avoid. In Mon-
being so successful. It is that if their taigne's case, it is clear that to him
victories had been predictable, they peace of mind was more important than
would have generated much less excite- material losses.
ment. If the actual emotional satisfac- Gary Becker (1996, ch. 12) has re-
tion from Norwegian success in the cently offered a model of such avoid-
Games was a decreasing functionf(p) of ance behavior with respect to guilt and
the ex ante probability p of Norwegian love. Except for those who enjoy meet-
success, the expected emotional satis- ing beggars because it make them "feel
faction p f(p) may have been too small superior or lucky" (Becker 1996, p.
to justify the investment, for the actual 232), people avoid beggars because they
p or even for any p. want to avoid the feelings of guilt trig-
gered by the encounters. It is not clear
3.6 Avoiding Unfavorable Occasions
from Becker's discussion whether the
Because Montaigne was aware of the root cause is the negative valence of the
difficulties of direct self-control, he rec- emotion or the monetary loss associated
ommended the indirect method. To with the giving which is the action ten-
prevent undesirable emotions, we may dency of guilt. When he asserts that
either ensure that the events that might "people do not want to encounter beg-
trigger them do not occur or that, if gars, even though they may contribute
they do, we do not come to know about handsomely after an encounter"
them. Montaigne (1991, p. 1075) (Becker 1996, p. 233), the phrase that I
adopted both strategies: "I shun all oc- have italicized suggests the first read-
casions for annoyance and keep myself ing. On the second reading, we would
from learning about things going expect "because" rather than "even
wrong." Concerning the first strategy, though." If the case of guilt is supposed
he refers to the example of King Cotys: to be analogous to that of love, the sec-
"He paid handsomely when some beau- ond reading must be chosen. Here,
tiful and ornate tableware was offered Becker's argument is that high-income
to him, but since it was unusually frag- individuals stay away from low-income
ile he immediately smashed the lot, rid- individuals of the opposite sex, because
ding himself in time of an easy occasion they know that they might fall in love
for anger against his servants." Con- with them and that if they do so they
cerning the second, he said about him- will want to share their income with
self that "I prefer people to hide my them. In this case the emotion itself has
losses and my troubles from me. . . . I positive valence, but is avoided because
prefer not to know about my estate-ac- of the loss of income that will be pre-
counts so as to feel my losses less ex- dictably induced by the associated ac-
actly. Whenever those who live with me tion tendency. Because guilt as well as
lack affection and its duties I beg them love may induce generous feeling and
to deceive me, paying me by putting a because generosity is costly, individuals
This content downloaded from 200.12.184.4 on Tue, 13 Nov 2018 01:50:33 UTC
All use subject to https://about.jstor.org/terms
56 Journal of Economic Literature, Vol. XXXVI (March 1998)
who care about their income stay away ning. Second, the course of external
from occasions that might trigger these events should be reasonably predict-
emotions. Yet in the case of love, this able. Occurrent emotions are jointly
avoidance behavior might have the cost caused by emotional dispositions and
that if the set of partners is very re- events that are outside the agent's con-
stricted, one might not meet anyone trol, and if the latter cannot be foreseen
who triggers the emotion. (The plot of there may not be much point in modify-
Pride and Prejudice would not have got- ing the former. A stoic attitude may be
ten off the ground if Darcy had acted in useful if one goes bankrupt, but might
accordance with Becker's analysis.) otherwise be a killjoy. And third, the
Hence the rational individual looking cost of using the technology should not
for a spouse might want to restrict the exceed the benefits. In the case of psy-
search to a set of optimal size, which chotherapy, there is no need to discuss
would also include some individuals the satisfaction of the second and third
with lower earning potential. And for conditions, as the first is manifestly not
some people, finding the right person satisfied. An important study by Robyn
might be more important than any asso- Dawes (1994) shows that with the ex-
ciated costs. ception of some forms of behavioral
therapy, psychotherapy essentially does
3.7 Emotional Character Planning
not offer any benefits over and above
There is a time-honored tradition in those that would be provided by talking
philosophical and religious thought that with any warm and interested person.
the wise man (women were rarely The efficacy of therapy may be en-
thought capable of the same feats) hanced by the therapist's belief in the
should be able to shape his emotional efficacy of his theory, but because that
dispositions. For Seneca and Plutarch, is true of therapists in all schools of
for instance, the strategy of emotional therapy this does not prove that effi-
self-management discussed above was cient technologies are available, only
only second-best. The man fully in con- that the self-confidence of the therapist
trol of himself would have no need to can be an important factor.
avoid occasions for anger, because he
3.8 Inducing Emnotions in Others
would have trained himself never to
feel anger. The contemporary version of If emotions are predictably triggered
this stoicist argument is psychotherapy. by beliefs about the external environ-
Many people seek therapy because they ment, other people could try to induce
are worried about their emotional reac- emotions by manipulating either the be-
tion patterns-their tendency to feel in- liefs or the environment. The art of
appropriately angry or guilty, their in- rhetoric, as practiced by orators and as
ability to feel love or to grieve, and so analyzed in Aristotle's -Rhetoric, is
on. They want to become a different largely about manipulating other peo-
kind of person, someone who would ple's emotional reactions by shaping
neither have excessively strong nor ex- their beliefs. The recent practice of
cessively weak emotional reactions to "shaming," which is intended to be a
events and people they encounter. more humane supplement to incarcera-
For this strategy to be rational ex tion, is supposed to work by altering the
ante, three conditions must be fulfilled. environment in which the offenders
First, there must exist an efficient and find themselves. Both ideas, however,
reliable technology of emotional plan- have serious difficulties.
This content downloaded from 200.12.184.4 on Tue, 13 Nov 2018 01:50:33 UTC
All use subject to https://about.jstor.org/terms
Elster: Emotions and Econonmic Theory 57
In the case of oratory, the basic prob- tion behind your action. Such motiva-
lem is that the cool-headedness that is tion-dependent emotions are very com-
required for rational manipulation of mon. A gift may be met with gratitude
the audience may interfere with the ef- or with resentment, depending on the
ficacy of the manipulation. On the one motivation that the recipient imputes to
hand, Chaim Perelman and Lucie 01- the donor (William Miller 1993, ch.1).
brechts-Tyteca (1969, p. 24) argue "An that income distribution that could be
"the man swayed by passion argues tolerable as an accidental or random
without taking sufficiently into account event ... might lead to violent revolt if
the audience he is addressing" and seen to be the result of conscious
hence is less effective in swaying them. choice on the part of another economic
On the other hand, if Ekman (1992) is agent" (Jack Hirshleifer 1987, p. 317).
right in his claim that it is difficult to Similarly, an expression of contempt
simulate the full range of emotional may induce shame if seen as spontane-
expressions, and Elaine Hatfield, John ous, but cause anger if seen as intended
Cacioppo, and Richard Rapson (1994) to induce shame, which is why the pol-
are right in their argument that emo- icy of punishing criminals by "shaming"
tional contagion is based on (i) mimicry them risks being counterproductive
of the emotional expression of others (June Tangney, interviewed in the New
and (ii) feedback from the expressions York Timnes, January 16, 1997).
that are mimicked to the emotions
3.9 Inculcating Emnotional Dispositions
themselves, we would expect this pro-
in Others
cess to be less effective if the expres-
sions are only partly realized. Depend- A similar problem can arise with re-
ing on the net effect of these two spect to the last technique for "choos-
opposing mechanisms, rational provoca- ing emotions," the case in which one
tion of the emotions of others through agent tries to create an emotional dispo-
oratory could be either more or less ef- sition in another agent. Thus both
fective than speeches which are them- Becker (1996, pp. 152-55) and Frank
selves grounded in passion. Although (1988, p. 93) argue that parents try to
this does not amount to an intrinsic inculcate guilt in their children-not
limitation on the purposive inducement the occurrent emotion (although some
of emotions in others, it could make the parents no doubt try to induce that
task quite difficult. too), but the disposition to have it.
The strategy of shaming runs into the Whereas I find Frank's argument some-
problem that emotions may be trig- what plausible, I believe that Becker's
gered by beliefs about the motivations argument fails. In Frank's analysis, par-
of others (Rabin 1993). If you buy a car ents try to inculcate guilt in the chil-
that is fancier than mine, I may be envi- dren for the benefit of the children. Es-
ous. If I also believe that you enjoy my sentially, guilt acts as a current proxy
envy, it may turn into resentment. If I for long-term rewards that might other-
believe that you bought the car in order wise fail to motivate the agent. Rational
to make me envious, it may become parents will allocate some resources in
murderous. In the first case, the emo- reducing the rate of time discounting in
tion is triggered by (my belief about) their children and some resources in in-
your action; in the second case, by (my culcating guilt, in amounts determined
belief about) your emotion; in the third by the equalization of marginal produc-
case, by (my belief about) the motiva- tivity.
This content downloaded from 200.12.184.4 on Tue, 13 Nov 2018 01:50:33 UTC
All use subject to https://about.jstor.org/terms
58 Journal of Economic Literature, Vol. XXXVI (March 1998)
This content downloaded from 200.12.184.4 on Tue, 13 Nov 2018 01:50:33 UTC
All use subject to https://about.jstor.org/terms
Elster: Emotions and Economic Theory 59
This content downloaded from 200.12.184.4 on Tue, 13 Nov 2018 01:50:33 UTC
All use subject to https://about.jstor.org/terms
60 Journal of Economic Literature, Vol. XXXVI (March 1998)
organism will take into account, the in- portunity costs.4 In many cases, the
ferences actually drawn from a potential organism might cope perfectly well
infinity, and the set of live options from by adopting and following mechanical
which it will choose." Along similar decision rules, such as "when you hear a
lines, the cognitive psychologists Philip sound you cannot identify, stand still"
Johnson-Laird and Keith Oatley (1992) or "when food tastes bitter, spit it out."
argue that because the ideal of "impec- In reality, of course, that's not how we
cable rationality" assumes that "there cope with novelty or bitter-tasting
are no surprises, no misunderstandings, food-not because the program is
no irresolvable conflicts," it cannot unfeasible but because natural selec-
guide action in situations that are char- tion has wired us differently. It is some-
acterized by these features. Instead, what misleading, therefore, to assert
"emotions enable social species to co- that emotions are a "supplemental"
ordinate their behaviour, to respond to principle that "fills the gap" between
emergencies, to prioritise goals, to pre- reflex-like behavior and fully rational
pare for appropriate actions, and to action.5
make progress towards . . . even though We can take this argument one step
individuals have only limited abilities to further. If we do not and cannot re-
cogitate." According to LeDoux (1996, spond to emergencies by following a
p. 176), if you were a small animal mechanical decision rule, it may be be-
faced with a bobcat and "had to make a cause our cognitive faculties are tempo-
deliberate decision about what to do, rarily clouded by the emotional arousal
you would have to consider the likeli- caused by the emergency. The emotion
hood of each possible choice succeeding serves as a functional equivalent for the
or failing and could get so bogged down rational faculties it suspends, by induc-
in decision making that you would be ing the very behavior that is rationally
eaten before you made the choice." Be- required and that reason, if left undis-
low, I cite a statement from Damasio turbed, could have come up with by it-
(1994) to the same effect. self. The emotions do solve problems-
These authors argue that emotional but problems that are to some extent of
responses enhance our capacity to make their own making. The capacity for the
good decisions, not by guiding us to the emotions to supplement and enhance
best possible decision, but by ensuring rationality would not exist if they did
that we make some decision in situ- not also undermine it.
ations where procrastination is likely to
be disastrous. The implicit premise of Richard Thaler (1980) argues that neglect of
their reasoning, however, is that ratio- opportunity costs and excessive focus on out-of-
pocket expenses is a frequent source of cognitive
nality amounts to what I have called else- irrationality. The neglect of the opportunity costs
where an addiction to reason (Elster that are created by the fact that decision making
1989c, p. 117). Some people do indeed takes time is also an important and pervasive
source of irrationality.
have a craving to make all decisions on 5 For such claims, see de Sousa (1987, p. 194),
the basis of "just" or sufficient reasons. Johnson-Laird and Oatley (1992, p. 206). The
That, however, makes them irrational point I am making here is well stated by LeDoux
(1996, p. 175): "In responding first with its most-
rather than rational. A rational person likely-to-succeed behavior, the brain buys time.
would know that under certain condi- This is not to say that the brain responds automat-
tions it is better to follow a simple me- ically for the purpose of buying time. The auto-
matic responses came first, in the evolutionary
chanical decision rule than to use more sense, and cannot exist for the purpose of serving
elaborate procedures with higher op- responses that came later."
This content downloaded from 200.12.184.4 on Tue, 13 Nov 2018 01:50:33 UTC
All use subject to https://about.jstor.org/terms
Elster: Emotions and Econonmic Theory 61
This content downloaded from 200.12.184.4 on Tue, 13 Nov 2018 01:50:33 UTC
All use subject to https://about.jstor.org/terms
62 Journal of Economic Literature, Vol. XXXVI (March 1998)
received a large sum of play money, and bring valuable business but also hap-
a smaller sum when they draw from pens to be the archenemy of your best
decks C and D. When taking a card friend, and proceeding or not with a
from A and B, they also sometimes had particular deal. The brain of a normal,
to pay back a very large amount of intelligent, and educated adult reacts to
money. As they learn about the struc- the situation by rapidly creating scenar-
ture of the game, normal subjects ios of possible response options and re-
mostly take cards from C and D and lated outcomes" (Damasio 1994, p.
usually end up ahead. Brain-damaged 170). He then argues that (what he
subjects, by contrast, stick to decks A takes to be) the rational-choice ap-
and B even though they regularly go proach to this problem would involve
bankrupt halfway through the game. impossibly complex calculations. The
Damasio's explanation is that these pa- decision would take an "inordinately
tients suffered from an inability to be long time" or might never be made at
motivated by mental representations of all. Because we are, as a matter of fact,
future states. Although they had normal able to make such decisions quite rap-
skin conductance reactions to monetary idly and efficiently, something else
loss, they differed from normal subjects must be going on:
in having no anticipatory responses in
Consider again the scenarios that I outlined.
the period immediately preceding their The key components unfold in our minds in-
selection of a card from a bad deck.6 stantly, sketchily, and virtually simultane-
Brain-damaged patients, then, tend ously, too fast for the details to be clearly de-
to be emotionally flat and to have de- fined. But now imagine that before you apply
any kind of cost/benefit analysis to the prob-
fective decision making capacities. It
lem, something quite important happens:
remains to characterize the relation When the bad outcome connected with a
between these two features. A strong given response option comes into mind, how-
claim is that "Reduction in emotion may ever fleetingly, you experience an unpleasant
constitute an . . . imnportant source of gut feeling. Because the feeling is about the
body, I gave the phenomenon the technical
irrational behavior" (Damasio 1994, p.
term somatic state . . .; and because it marks
53: italics in original). A weaker claim, an image, I called it a marker.
asserting correlation but not causation,
is that "The powers of reason and the [The] somatic marker . . . forces attention on
experience of emotion decline to- the negative outcome to which a given action
may lead, and functions as an automated
gether" (Damasio 1994, p. 54).
alarm signal which says: Beware of danger
To support the strong claim he first ahead if you choose the option which leads to
describes a decision-making problem this outcome. The signal may lead you to re-
that the owner of a business might con- ject, immediately, the negative course of ac-
front, "faced with the prospect of meet- tion and thus make you choose among other
alternatives. The automated signal protects
incr nr nnt with a nn.ihlP O]liPnt whobn t-n
___0 __ - -- ' -- _ ' - you against future losses, without further ado,
6 In a recent experiment (Antoine Bechara et al. and then allows you to choose from among
1997) it is also shown that normal subjects "began fewer alternatives. There is still room for us-
to generate anticipatory skin conductance re- ing a cost/benefit analysis and proper deduc-
sponses (SCRs) whenever they pondered a choice tive competence, but only after the auto-
that turned out to be risky, be ore they knew ex- mated step drastically reduces the number of
plicitly that it was risky, whereas patients [with options .7
prefrontal damage] never developed anticipatory
SCRs, although some eventually realized which 7 Damasio (1994, p. 173); italics in original.
choices were risky." In other words, conscious Elsewhere (pp. 175, 187) he also acknowledges
awareness of costs and benefits is neither a neces- positive somatic markers that lead to the pursuit of
sary nor a sufficient condition for rational choice. specific options rather than their elimination.
This content downloaded from 200.12.184.4 on Tue, 13 Nov 2018 01:50:33 UTC
All use subject to https://about.jstor.org/terms
Elster: Emotions and Economic Theory 63
Once again, however, going by one's [which] have been connected, by learn-
gut feelings is not the only way to cut ing, to predicted future outcomes of
through the maze'of a complex deci- certain scenario" (Damasio 1994, p.
sion-problem. One can also, for in- -174; italics deleted). The basic princi-
stance, flip a coin. Damasio might ples of reinforcement theory, however,
counter that this procedure is inferior make it implausible that this mecha-
to going by gut feelings, which enable nism could guide the decision whether
one not only to make swifter decisions to deal with a businessman who is the
but also better ones. But the coin-toss- enemy of one's best friend. For rein-
ing heuristic is only the most simple of forcement to establish behavior, it
many rules of thumb that are used in should ideally occur soon after the be-'
complex decision making problems. The havior in question, occur invariably
general idea of satisficing (Simon 1955) when the behavior is chosen, and the
can be spelled out in many ways (see for behavior itself should be one that is
instance Richard Nelson and Winter chosenfrequently. None of these condi-
1982). Medical diagnosis and prognoses tions is even approximately satisfied in
can be done very efficiently by me- the example. Concerning (ii), Damasio
chanical point systems that rely on a (1994, p. 185) acknowledges that some
small number of variables. In fact, such somatic markers may operate uncon-
methods almost invariably tend to per- sciously, as when "worker bumblebees
form better than intuition based on "gut 'decide' on which flowers they should
feeling" (Dawes, David Faust, and Paul land." In this case, no emotions are in-
Meehl 1989). In opposing gut feelings volved. He gives no evidence or argu-
to hyperrational cost-benefit calculation ment to show that unemotional hunches
Damasio is setting up a strawman.8 could not be the rule in other cases too.
This objection does not, however, af- Damasio's correlational claim is ar-
fect Damasio's claim that in most com- resting. His causal claim, however,
plex decisions people do, as a matter of seems unsupported. The mechanism he
fact, consult their gut feelings. When suggests-reinforcement learning-can-
confronted with a novel challenge for not perform the work it is supposed to
which no rule of thumb is available, do. There seems to be a cluster of abili-
some people procrastinate more or less ties-to react emotionally, to delay
indefinitely, while others, for better or gratification, and to avoid procrastina-
for worse, make a snap decision based tion-that are linked in the brain, but it
on some salient feature of the situation. seems too early to tell whether the abil-
Damasio claims (i) that more often than ity to have emotional reactions plays a
not this feature has great predictive direct causal role in sustaining the other
value for making a good choice or at abilities.
least avoiding a bad one, and (ii) that its
salience is signaled by an occurrent 5. Emnotions and Interest
emotion. Concerning (i), he appeals to a
regular reinforcement process, although From the point of view of economic
he does not use that term. "Somatic theory, the most interesting issue con-
markers are . . . emotions and feelings cerning the emotions may be the nature
of the interaction between emotion and
s Moreover, as I argue in Section 5 below, there
other motivations. Among the latter, I
is no reason to believe that the emotionally in-
duced curtailment of the perceived feasible set al-
shall only consider material self-interest
ways has beneficial effects. ("interest" for short), although similar
This content downloaded from 200.12.184.4 on Tue, 13 Nov 2018 01:50:33 UTC
All use subject to https://about.jstor.org/terms
64 Journal of Economic Literature, Vol. XXXVI (March 1998)
problems arise for the relation between that the agent is willing to invest re-
emotions and impartial motivations. For sources in making the rival worse off up
instance, a wealthy liberal might on im- to the point where he derives more util-
partial grounds prefer to send his chil- ity from making himself better -off
dren to a public school, but his emo- (Hirshleifer 1987). In modeling altru-
tional attachment to the children might ism, we can make a similar assumption
induce a preference for a better-quality (Becker 1976, chs. 12 and 13). Eco-
private education. nomic analyses of regret (David Bell
1982; Graham Loomes and Robert Sug-
5.1 The Cost-benefit Model of Emotions
den 1982) also assume that agents
By far the most common way of mod- weigh satisfaction from actual outcomes
eling the interaction between emotions and emotions generated by counterfac-
and interests is to view the former as tual beliefs. Strictly speaking, none of
psychic costs or benefits that enter into these analyses need to rely on valence,
the utility function on a par with satis- in the sense of subjective feelings of
factions derived from material rewards. pleasure and pain. All that is needed is
In this perspective, the only relevant as- that we can draw indifference curves
pect of the emotions is their valence. that reflect the tradeoffs involved. We
We may use Becker's analysis of beg- may think of emotional valence as the
gar-induced guilt to illustrate two ways underlying mechanism behind these
in which the pleasure and pain associ- tradeoffs, but it need not be directly re-
ated with the emotions may enter into flected in the formal analysis. In a mod-
the utility function. First, the guilt itself eling perspective, "emotional altruism"
is a cost. Even if I do not have any is indistinguishable from "reason-based
money with me, I may cross the street altruism." In the following I assess the
to avoid coming face to face with a beg- usefulness of this approach with respect
gar whose visible misery would induce to guilt, shame, envy, indignation, love,
the unpleasant feeling of guilt. Second, vindictiveness, hatred, and contempt.
the guilt may induce behavior that is
5.2 Guilt
costly in the material sense. If I do have
money with me, I know that if I come Let us assume that a person is
face to face with the beggar I would tempted to steal a book from the li-
give him something to alleviate my brary. If he feels guilty about doing it,
guilt. More accurately, I would give up he may abstain. If he steals the book
to the point where the marginal utility and then feels guilt, he may return the
of money in alleviating my guilt equals book to the library. On the Becker-
its marginal utility for other purposes. Frank assumption that guilt is to be
(If crossing the street is costly, this modeled as a cost, both the abstention
would also have to be taken into ac- from stealing and the return of the book
count.) would be explained by a simple cost-
In other analyses, the encounters that benefit analysis. This approach has the
trigger the emotion is taken for given great advantage that it allows us to ac-
and not subject to choice. The question count for the undeniable existence of a
of choice arises only because the agent tradeoff between moral emotions and
has to weigh emotional satisfaction self-interest. The world is not made up
against other satisfactions, as in the of two exclusive and exhaustive catego-
choice of the amount to give to the beg- ries, those who would steal a book
gar. In modeling envy, we may assume whenever there was no risk of detection
This content downloaded from 200.12.184.4 on Tue, 13 Nov 2018 01:50:33 UTC
All use subject to https://about.jstor.org/terms
Elster: Emotions and Economic Theory 65
and those who would never do so. Many account for the tradeoff between guilt
people would go ahead and steal the and interest and yet does not imply that
book if but only if its value to them was a reluctant agent would buy the guilt-
sufficiently high or its value to others erasing pill. I conjecture that the model
sufficiently small. To model such behav- would involve some kind of non-inten-
ior, we could talk "as if' guilt and inter- tional psychic causality rather than de-
est add up to an inclusive utility, with liberate choice. To illustrate what I
the marginal disutility from guilt being have in mind, I shall sketclh a model
an increasing function of (say) the num- drawn from catastrophe theory, along
ber of people on the waiting list for the the lines of a model of the relation be-
book and the marginal utility from tween personal opinion and conformism
interest a decreasing function of (say) proposed by Abraham Tesser and John
the time the agent expects to use the Achee (1994). In a catastrophe model,
book. the surface describing the behavior of a
Whether or not this model of the in- dependent variable as a function of two
teraction between emotion and interest independent variables folds in on itself
is predictively adequate, I submit that it in a cusp. Within a certain range, a
is basically flawed. If guilt were nothing given constellation of the independent
but an anticipated or experienced cost, variables is thus consistent with several
an agent whose guilt deters him from values of the dependent variable.
stealing or retaining the book should be Suppose that the agent is initially un-
willing to buy a guilt-erasing pill if it willing to steal the book, but that as its
was sufficiently cheap. I submit that no value to him increases he finally decides
person who is capable of being deterred to do so.9 Suppose conversely that the
by guilt would buy the pill. In fact, he agent has stolen the book, but that as its
would feel guilty about buying it. For value to others increases he finally re-
him, taking the pill in order to escape turns it to the library. In the first case,
guilt and be able to steal the book suppose that its value to others is 10
would be as morally bad as just stealing and that he decides to steal it just when
it. He would not see any moral rele- its value to him reaches 15. In the sec-
vance between stealing the book in a ond case, suppose that its initial value
two-step operation (taking the pill to to him is 15 and the initial value to oth-
steal the book) and stealing it in a one- ers is 6. On the cost-benefit model, he
step operation. There is a strict analogy would return it when its value to others
between this argument and a point reaches 10. On the catastrophe model,
that I have made elsewhere (Elster he might not do so until its value to oth-
1997), viz. that a person who discounts ers reached 15. The reason for this
the future very highly would not be mo- asymmetry is found in the mechanism
tivated to buy a pill that would reduce of dissonance reduction (Leon Festin-
his rate of time discounting. To want to
be motivated by remote consequences
9 Technically, the dependent value has to be
of present behavior is to be motivated continuous rather than the dichotomous choice
by remote consequences of present be- between stealing the book and not stealing or re-
turning it. We can assume, therefore, that the de-
havior. Similarly, to want to be immoral
pendent variable is a propensity to steal the book
is to be immoral. A person willing to and that the agent proceeds to steal it once the
take the guilt-erasing pill would not propensity reaches a certain level. Alternatively,
we could use an example in which the choice vari-
need it.
able is continuous, such as the amount to give to
We need, therefore, a model that can the beggar in the street.
This content downloaded from 200.12.184.4 on Tue, 13 Nov 2018 01:50:33 UTC
All use subject to https://about.jstor.org/terms
66 Journal of Economic Literature, Vol. XXXVI (March 1998)
This content downloaded from 200.12.184.4 on Tue, 13 Nov 2018 01:50:33 UTC
All use subject to https://about.jstor.org/terms
Elster: Emotions and Economic Theorn 67
lof 1976; Robert Axelrod 1986; Didier opportunity for a mutually profitable
Abreu 1988; James Coleman 1990) business transaction, this need not be
have argued that social norms work the case; and even when it is the case,
through material sanctions, involving my claim is that the costs to the sanc-
higher-order sanctions of those who fail tioner are what makes the sanction re-
to sanction norm-violators or non-sanc- ally painful to the target. It tells him
tioners. I shall not repeat the argu- that others see him as so bad that they
ments I have offered against this view are willing to forego valuable opportu-
elsewhere (Elster 1989b, 1989d), except nities rather than have to deal with
to note that they rest on an empirical him.
claim that the willingness to impose The second question concerns the
sanctions goes quickly to zero as we go mode of operation of shame in shaping
upwards in the hierarchy of sanction- behavior. According to the cost model,
ings.10 present and future shame enters into
I would like to go beyond my earlier the utility function on a par with mate-
arguments, however, to assert that the rial costs and benefits. A person who
material sanctions themselves are best has been publicly exposed to contempt
understood as vehicles of the emotion might compare three options: suicide,
of contempt, which is the direct trigger moving elsewhere to take up a new
of shame. When a person refuses to profession with a new name, and
deal with someone who has violated a sticking it out in the expectation that
social norm, the latter may suffer a fi- the contempt of others and the feeling
nancial loss. More important, he will of shame will fade after a while. If the
see the sanction as a vehicle for the immediate feeling of shame is immense,
emotions of contempt or disgust, and suicide might well be preferable to the
suffer shame as a result. The material discounted present value of the other
aspect of the sanction that matters is options. It would be hard to refute this
how much it costs the sanctioner to pe- account, as the disutility of shame can
nalize the target, not how much it costs always be stipulated to be arbitrarily
the target to be penalized. (Thus the high. I believe it is at least equally
phrase "This hurts me more than it plausible, however, to assume that
hurts you" may be intended to add to shame induces a temporary heightening
the punishment, not to soften it.) The of the discounting rate. With respect
more it costs the sanctioner to refuse to to drugs, Becker (1996, p. 329) argues
deal with the target person, the that "A habit may be raised into an
stronger will the latter feel the con- addiction by exposure to the habit
tempt behind the refusal and the more itself. Certain habits, like drug use
acute will be his shame. Although high and heavy drinking, may reduce the at-
costs to the sanctioner often go to- tention to future consequences-there
gether with high costs for the target, as is no reason to assume discount rates on
when the sanctioner renounces on the the future are just given and fixed."
The argument seems equally applic-
10 Eugene Kandel and Edward Lazear (1992, p. strong emotions. Alternatively,
able to
813) try to deal with this problem by arguing that
"the firm can be thought of as a circle. As long as a
we might follow Loewenstein (1996)
worker is told only that he is to punish the neigh- and argue that shame, like other vis-
bor on his right or suffer punishment from the one ceral factors, undermines our ability to
on his left, he will carry out the punishment." I fail
to see what this means or how it solves the diffi-
predict future subjective states. When
culty. one is in intense pain or suffering
This content downloaded from 200.12.184.4 on Tue, 13 Nov 2018 01:50:33 UTC
All use subject to https://about.jstor.org/terms
68 Journal of Economic Literature, Vol. XXXVI (March 1998)
This content downloaded from 200.12.184.4 on Tue, 13 Nov 2018 01:50:33 UTC
All use subject to https://about.jstor.org/terms
Elster: Emotions and Economic Theory 69
rather than accept a small amount. We rival is made worse off, but not as the
can imagine two emotional reactions result of my action
that explain why the second player (C) The state of the world in which
would react in this way: envy and indig- my rival is made worse off, as the re-
nation. If the second player is moti- sult of my action
vated by envy, it can be modeled as a
cost which has to be subtracted from The key to my account is the assump-
what the first player offers him. If the tion that most envious people prefer (B)
net result is negative, he will reject the over (A) and (A) over (C). We would
offer. If he is motivated by indignation, like our rival to be made worse off, but
his rejection of a bad offer will give him because of the shame attached to de-
the pleasure of revenge. If this pleasure struction from envy we do not want to
is large enough, it may offset the mate- serve as the agent of his destruction.
rial loss involved in rejection. (See Hirshleifer, 1987, for a similar dis-
I want to make two remarks on these tinction between action-dependent and
interpretations of the Ultimatum Game action-independent emotions.) This
behavior. First, I shall argue that envy doesn't mean that we don't lhave the
is more plausibly interpreted as an ac- urge to destroy, only that we don't want
tion tendency than as a cost. Second, I to act on it. When people act on the
shall suggest that envy and indignation urge, it is because in their keen desire
are not necessarily incompatible moti- for the other's misfortune they confuse
vations. Rather, a subjective feeling of (B) and (C). They fail to see, that is,
indignation may be a form of envy in that they cannot intentionally bring
disguise. Both remarks are somewhat about a state that is defined by not be-
speculative, but I believe they capture ing brought about intentionally by
some of the dynamics of the emotions themselves (Elster 1983b, ch. 11).12
that is not reflected in the simple cost- Once the deed is done, they discover
benefit model. that they are worse off rather than bet-
Pre-scientific and literary writings on ter. Again, emotion has a dual role.
envy are in unanimous agreement on While inducing action to alleviate the
one point: the envy man is not made negative emotion of envy, it prevents us
happy by carrying out his urge to de- from thinking clearly about the conse-
stroy the envied object or its possessor quences of such action.
(for surveys, see Helmut Schoeck 1987 To feel envious is to feel inferior. No-
and Gonzalo Fernandez de la Mora body likes to feel inferior, especially
1987). Whereas implementation of the when there is nobody else to blame. To
action tendency of guilt does alleviate alleviate the feeling of inferiority,
the emotion and render the agent bet- people often rewrite the script so as
ter off, implementing the action ten- to be able to blame someone else for
dency of envy makes the agent worse their situation or to explain the supe-
off. The cost-benefit model of envy can- riority of the other by his immoral
not capture this feature of the emotion.
To suggest an explanation, let me first 12 There is an obvious similarity between this ar-
gument and the argrument I adduced above against
distinguish among three states of the
the idea of inducing or inculcating shame and
world: guilt. Yet whereas tle intention to bring about
state (B) is logically incoherent, the intention to
(A) The status quo induce shame or guilt is incoherent only because
people as a matter of fact feel anger when manipu-
(B) The state of the world in which my lated.
This content downloaded from 200.12.184.4 on Tue, 13 Nov 2018 01:50:33 UTC
All use subject to https://about.jstor.org/terms
70 Journal of Economic Literature, Vol. XXXVI (March 1998)
This content downloaded from 200.12.184.4 on Tue, 13 Nov 2018 01:50:33 UTC
All use subject to https://about.jstor.org/terms
Elster: Emotions and Economic Theory 71
day before" (Rhetoric 1380b 11-13). In (Milovan Djilas 1958; Miller 1990) con-
trials of collaborators in German-occu- firm the view that the passion for re-
pied countries after World War II, venge or "wrath" (Frijda 1994) can be a
those who were tried later generally re- lifetime concern. Unlike love or limer-
ceived milder sentences even when the ence, however, the durable thirst for re-
crimes were similar. venge is not a universal phenomenon.
Yet some emotions have a more dura- The spontaneous urge to retaliate may
ble character. In Becker's analysis of be universal, but its transformation into
love, the reason why the prudent man a lifelong passion occurs only in socie-
would take care to avoid low-income ties where it is amplified by strong so-
women is presumably that he might cial norms.
contract a lifelong disposition to share What one might call the "prejudice
his income. One might question, per- emotions," contempt and hatred, also
haps, whether the relationship between tend to be very durable and frequently
spouses typically involves emotions in all-consuming. Unlike most other emo-
the full sense of the term. Marital love tions, these are not necessarily trig-
may involve concern for the welfare of gered by a particular event or episode.
one's spouse, but not the strong arousal The members of one social group may
and action tendencies that we associate harbor these feelings toward members
with the emotion often referred to as of another group simply on the basis of
infatuation. That term is somewhat un- (what they take to be) their character
fortunate, however, to the extent that it rather than on the basis of anything
suggests an acute rather than a durable they have done. In the case of con-
state. The characteristic features of in- tempt, the others are viewed as intrinsi-
fatuation-all-consuming interest in the cally inferior; in the case of hatred, as
other person, heightened energy levels, intrinsically evil. Thus followers of
less need for sleep and food-can in Hitler thought Jews evil but Slavs infe-
fact endure for years. In a study of this rior (Daniel Goldhagen 1996, p. 469).
emotion, for which she coined the term As Aristotle noted, the action tendency
"limerence," Dorothy Tennov (1979) of hatred is to destroy the person who is
found that the typical duration of an the target of the emotion, whereas that
episode was from 18 months to three of anger is to make him suffer (Rhetoric
years, with some episodes lasting only a 1382a 2-16).14 Not all prejudices are
few weeks and others a whole lifetime. all-consuming. In fact, the phenomenon
Revenge behavior provides another of "prejudice with compunction" (Gor-
counterexample to the idea that emo- don Allport 1979, ch. 20) shows that the
tions can be modeled as momentary prejudiced individual need not endorse
preferences. In societies where blood his own attitude, and rather have a
feuds are common, revenge can be a meta-emotion of shame that prevents
lifetime obsession. In his outstanding
study of blood feuds in nineteenth cen- 14 Similarly (see note 2 above), the guilty person
tury Corsica, Stephen Wilson (1988, pp. may want to make himself suffer, whereas the per-
30, 280) refers to one case in which a son who is ashamed may want to destroy himself.
Yet the analogy is not perfect. The normal action
man killed six persons who had testified tendency of guilt is to try to undo the harm one
in the trial of his brother, and to an- has caused. Although the undoing may be costly
other in which a man killed all 14 wit- and to that extent involve suffering, it is only in
the special case of neurotic guilt (Otto Fenichel
nesses who had testified against his 1945, p. 105) that the need to suffer is the primary
brother. Other studies of feuding urge.
This content downloaded from 200.12.184.4 on Tue, 13 Nov 2018 01:50:33 UTC
All use subject to https://about.jstor.org/terms
72 Journal of Economic Literature, Vol. XXXVI (March 1998)
him from acting on it. Yet as the twenti- tional dispositions such as envy and in-
eth century has abundantly shown, dignation may have been selected be-
prejudice can become all-consuming. cause of their survival value.
These "standing" emotions shape In my opinion, these attempts are
preferences in a durable manner. When premature. I think we need a better un-
they are all-consuming, we can model derstanding of how emotions actually
them by a lexicographic preference or- influence behavior before we can begin
dering. For the person who is in a state to think about how they may have
of limerence or in pursuit of revenge, evolved. Until we know whether the net
there is no tradeoff between satisfaction effect of an emotional disposition is
of the emotion and material interest. positive or negative, it makes no sense
Ordinary economic activities are pur- to begin looking for an evolutionary ex-
sued only to the extent that they pro- planation.'5 In the case of anger, for in-
mote the emotional goal of the agent. stance, it may be true that irascible peo-
Unlike the person who is in the grip of ple often get their way, but that is only
an acute emotion of shame or anger, part of the story. Others will learn to
people who are subject to an all- recognize them as irascible, and walk
consuming durable passion are per- around them rather than have any deal-
fectly capable of acting in an instrumen- ings with them. Sometimes one has no
tally rational fashion. As Aristotle noted choice, but often one can find alterna-
(Politics 1312b 19-34), the angry man is tive and more reasonable partners. Iras-
irrational whereas the man animated by cible people will find themselves
hatred is not. The emotion may be shunned, which detracts from opportu-
grounded in an irrational belief, but nities for mutually favorable interac-
that is another matter. tions with others. They may gain more
in each interaction, but interact more
6. Conclusion rarely. They will not, moreover, be able
to learn that their emotional disposition
It is not possible in the short format works against them, and hence will have
of an article to bring out all the features no incentive to control themselves.
of emotion that might be relevant for They will get positive reinforcement
economists. Any survey has to be selec- from their encounters with others-
tive, and I have chosen to exclude some they find that being angry works!-but
topics that may be of interest to many they cannot get feedback from the en-
readers. In particular, I have said very counters they fail to have. I am not say-
little about the evolutionary models that ing that the net effect of irascibility is
have gained prominence through the negative, only that one cannot show it
writings of Hirshleifer (1987) and Frank to be positive simply by citing a positive
(1988). They argue, to simplify, that the impact in isolation from other effects.
emotionally induced disregard for con-
sequences can have good consequences.
15 Strictly speaking, this is not true. Even if a
Threats and promises that would not given emotional disposition is on the whole nega-
have been credible if made by an agent tive in its impact on reproductive fitness, we might
still be able to explain it as a suboptimal part of an
(known to be) motivated by pure mate-
optimal package solution that has been selected by
rial self-interest, can be credible if evolution. Yet whereas this would require detailed
made by an agent (known to be) willing knowledge about genetic linkages, the demonstra-
tion that the disposition has a positive net effect
to disregard his material self-interest.
would by itself go a long way towards explaining
They conjecture, therefore, that emo- why it exists.
This content downloaded from 200.12.184.4 on Tue, 13 Nov 2018 01:50:33 UTC
All use subject to https://about.jstor.org/terms
Elster: Emotions and Economnic Theory 73
The more urgent task is to under- BOYER, PETER. "Admiral Boorda's War," The New
Yorker, Sept. 16, 1996.
stand how emotions interact with other COLEMAN, JAMES S. Foundations of social theory
motivations to produce behavior. I have Cambridge, MA: Harvard U. Press., 1990.
tried to argue that the most obvious DAMASIO, ANTONIO R., Descartes' error. New
York: Putnam, 1994.
strategies-modeling emotions as psy- DAWES, ROBYN M. House of cards. New York:
chic costs and benefits or as a source of Free Press, 1994.
DAWES, ROBYN; FAUST, DAVID AND MEEHL,
preferences-miss important features
PAUL E. "Clinical Versus Actuarial Judgment,"
of the phenomenon. The role of emo- Science, Mar. 1989,243, pp. 1668-74.
tions cannot be reduced to that of shap- DIXON, NORMAN F. On the psychology of military
ing the reward parameters for rational inconmpetence. London: Futura, 1976.
DJILAS, MILOVAN. Land without justice. London:
choice. It seems very likely that they Methuen, 1958.
also affect the ability to make rational EKMAN, PAUL. Telling lies. New York: Norton,
1992.
choices within those parameters. This
EKMAN, PAUL AND DAVIDSON, RICHARD J. The
dual role of the emotions-shaping nature of emotion. Oxford U. Press, 1994.
choices as well as rewards-has ana- ELSTER, JON. Explaining technical change. Cam-
bridge: Cambridge U. Press, 1983a.
logues in pain, addictive cravings, and
_ . Sour grapes. Cambridge: Cambridge U.
other visceral factors. As in these other Press, 1983b.
cases, the claim is not that the emotions - . Nuts and bolts for the social sciences.
Cambridge: Cambridge U. Press, 1989a.
fully determine choice, or that there is
. The cenment of society. Cambridge: Cam-
no tradeoff between emotional rewards bridge U. Press, 1989b.
and other rewards. Rather, it is that the - . Solomonic judgemnents. Cambridge: Cam-
bridge U. Press. 1989c.
tradeoff itself is modified by one of the
. "Social Norms and Economic Theory," J.
rewards that is being traded off against Econ. Perspectives, Fall 1989d, 3(4), pp. 99-
the others. 117.
. Review of Becker (1996), U. of Chicago
Law Rev., 1997, 64, pp. 749-64.
REFERENCES
- . Alchenmies of the mind. Forthcoming
ABREU, DIDIER. "On the Theory of Infinitely Re- FEARON, JAMES D. AND LAITIN, DAVID D. "Ex-
peated Games with Discounting," Econo- plaining Interethnic Cooperation," Amer. Polit.
metrica, Mar. 1988, 56(2), pp. 383-96. Sci. Rev., Dec. 1996, 90(4), pp. 715-35.
AKERLOF, GEORGE. "The Economics of Caste of FENICHEL, OTTO. The psychoanalytical theory of
the Rat Race and Other Woeful Tales," Quart. neurosis. New York: Norton, 1945.
J. Econ., Nov. 1976, 90(4), pp. 599-617.
FERNANDEZ DE AL MORA, GONZALO F. Egialitar-
AKERLOF, GEORGE A. AND DICKENS, WILLIAM ian envy. New York: Paragon House Publishers,
T. "The Economic Consequences of Cognitive 1987.
Dissonance," Amer. Econ. Rev., June 1982, FESTINGER, LEON. A theory of cognitive disso-
72(3), pp. 307-19. nance. Stanford, CA: Stanford U. Press, 1957.
ALLPORT, GORDON. The nature of prejudice. FESTINGER, LEON AND BRAMEL, DANA. "The Re-
Reading, MA: Addison-Wesley, [1954] 1979. actions of Humans to Cognitive Dissonance," in
AXELROD, ROBERT. "An Evolutionary Approach Experimental foundations of clinical psychol-
to Norms," Amer. Polit. Sci. Rev., Dec. 1986, ogy. Ed.: ARTHUR J. BACHRACH. New York:
80(4), pp. 1095-1111. Basic Books, 1962, pp. 254-79.
BARON, JONATHAN. Thinking and deciding. 2nd FRANK, ROBERT H. Passions within reason. New
ed. Cambridge: Cambridge U. Press, 1994. York: Norton,1988.
BECHARA, ANTOINE ET AL. "Deciding Advanta- FRIJDA, NICO H. The emotions. Cambridge: Cam-
geously Before Knowing the Advantageous bridge U. Press, 1986.
Strategy," Science, Feb. 28, 1997, 275, pp. - . "The Lex Talionis: On Vengeance," in
1293-95. Emotions: Essays on emotion theory. Eds.:
BECKER, GARY S. The economic approach to hu- STEPHANIE H. M. VAN GOOZEN, NANNE E.
man behavior. Chicago: U. of Chicago Press, VAN DE POLL, AND JOE A. SERGEANT.
1976. Hillsdale, NJ: Lawrence Erlbaum, 1994, pp.
_ . Accounting for tastes. Cambridge, MA: 263-90.
Harvard U. Press, 1996. GOLDHAGEN, DANIEL J. Hitler's twilling execu-
BELL, DAVID. "Regret in Decision Making Under tioners. New York: Knopf, 1996.
Uncertainty," Operations Research, Sept./Oct. GORDON, ROBERT M. The structure of emotions
1982, 30(5), pp. 961-81. Cambridge: Cambridge U. Press, 1987.
This content downloaded from 200.12.184.4 on Tue, 13 Nov 2018 01:50:33 UTC
All use subject to https://about.jstor.org/terms
74 Journal of Economic Literature, Vol. XXXVI (March 1998)
This content downloaded from 200.12.184.4 on Tue, 13 Nov 2018 01:50:33 UTC
All use subject to https://about.jstor.org/terms