The Role of Emotion in Economic Behavior

Home , Affective forecasting, Gratification

CHAPTER 9

IMMEDIATE AND whether to purchase a stock, she might imagine EXPECTED EMOTIONS the disappointment she would feel if she bought it and it declined in price, the elation Consequentialist Models of Decision Making she would experience if it increased in price, and possibly emotions such as regret and relief Economic models of decision making are that she might experience if she did not pur- consequentialist in nature; they assume that de- chase the stock and its price either rose or fell. cision makers choose between alternative The key feature of expected emotions is that courses of action by assessing the desirability they are experienced when the outcomes of a and likelihood of their consequences, and inte- decision materialize, but not at the moment of grating this information through some type of choice, at the moment of choice they are only expectation-based calculus. Economists refer cognitions about future emotions. to the desirability of an outcome as its “utility,” Immediate emotions, by contrast, are experi- and decision making is depicted as a matter of enced at the moment of choice and fall into one maximizing utility. of two categories. “Integral” emotions, like ex- This does not, however, imply that con- pected emotions, arise from thinking about the sequentialist decision makers are devoid of consequences of one’s decision, but integral emotion or immune to its influence. To see emotions, unlike expected emotions, are expe- why, it is useful to draw a distinction between rienced at the moment of choice. For example, “expected” and “immediate” emotions in the process of deciding whether to purchase (Loewenstein, Weber, Hsee, & Welch, 2001; the stock, Laura might experience immediate Loewenstein & Lerner, 2003). Expected emo- fear at the thought of the stock’s losing value. tions are those that are anticipated to occur as “Incidental” emotions are also experienced at a result of the outcomes associated with differ- the moment of choice, but arise from ent possible courses of action. For example, if dispositional or situational sources objectively Laura, a potential investor, were deciding unrelated to the task at hand (e.g., the TV pro-

138 9. The Role of Emotion in Economic Behavior 139 gram playing in the background as Laura called her brokerage house).1 The notion of expected emotions is perfectly consistent with the consequentialist perspective of economics. Nothing in the notion of utility maximization rules out the idea that the utility an individual associates with an outcome might arise from a prediction of emotions; for example, one might assign higher utility to an Italian restaurant dinner than a French restaurant dinner because one anticipates being happier at the former. While not explicitly denying the FIGURE 9.1. Consequentialist model of decision making. idea that utilities might depend on expected emotions, however, most economists until recently viewed detailed accounts of such emotions as outside the purview of their discipline. opportunity. Marketers also attempt to cap- Integral immediate emotions can also be in- italize on immediate emotions—for example, corporated into a consequentialist framework, charitable organizations that make potential although it takes one farther afield from con- donors feel guilty about what they squander ventional economics. Integral emotions, it can their money on while less fortunate people and in fact has been argued, might provide de- starve. cision makers with information about their The food industry is particularly motivated own tastes—for instance, to help inform Laura to capitalize on immediate emotions. Mrs. of how she would actually feel if she purchased Field’s Cookies, for example, has been known the stock and it rose or declined in value. How- to pump enticing cookie smells into the atmo- ever, this assumes, contrary to the usual as- sphere of shopping malls to stimulate hunger sumption in economics, that people have an (Hoch & Loewenstein, 1991). A company imperfect understanding of their own tastes. named “ScentAir” sells similar odors (e.g., An influence of incidental immediate emo- “Glazed Donut,” “Iced Cinnamon Pretzel,” tions on decision making would pose a much “Blue Cotton Candy”) to businesses looking to more fundamental challenge to the consequen- stimulate hunger.2 By contrast, the dieting in- tialist perspective, because such emotions, by dustry often attempts to market its services by definition, are irrelevant to the decision at focusing people on the positive emotions they hand. Any influence of incidental emotions can anticipate experiencing once they are fi- would suggest that decisions are influenced by nally able to fit into the perfect pair of jeans. factors unrelated to the utility of their consequences. Enter Behavioral Economics Figure 9.1 presents a schematic representation of the traditional perspective of econom- Fortunately, many economists would view the ics. Although immediate emotions are repre- snapshot of their discipline presented above as sented in the figure, they would not be part of outdated. This is largely attributable to the any traditional economist’s representation of advent of “behavioral economics,” a subdis- their framework, because they play no role in cipline of economics that incorporates more decision making; they are “epiphenomenal” psychologically realistic assumptions to in- by-products of, but not determinants of, deci- crease the explanatory and predictive power of sions. economic theory. The field first achieved prom- However, a great deal of market activity can inence in the 1980s and has been gaining influ- be understood in terms of both expected and ence since then. And much of the thrust of immediate emotions. Much advertising at- behavioral economics has involved, or at least tempts to inform consumers, whether accu- could be construed as involving, an enhanced rately or not, about emotions that they can ex- understanding of emotions. pect to feel if they do or do not buy a particular The first, and less controversial, interaction good. “One-day-only” sales, for example, are of behavioral economics with emotions was to probably effective because they make consum- question the neglect of the topic and to begin to ers think that they will regret not seizing the examine exactly how utility depended on out- 140 I. INTERDISCIPLINARY FOUNDATIONS comes. For example, whereas conventional EU assumes that people choose between alter- economics assumes that the utility of an out- native courses of action by assessing the come depends only on the outcome itself, some desirability or “utility” of each action’s possi- economists showed how counterfactual emo- ble outcomes and linearly weighting those utili- tions (e.g., regret), which arise from consider- ties by their probability of occurring. The nor- ing alternative outcomes that could have oc- mative status of the EU model was enhanced by curred, can influence decision making. Note von Neumann and Morgenstern’s (1944) dem- that these analyses focus on expected emotions onstration that it could be derived from a prim- and hence help to elaborate the connection itive, intuitively appealing set of axioms—for among outcomes, emotions, and utility, but do example, that preferences are transitive (if A is not challenge the consequentialist perspective. preferred to B, and B is preferred to C, then A More recently, economists as well as psy- should be preferred to C). In addition to its chologists who are specifically interested in de- normative appeal, this model’s assumption that cision making have begun to take greater ac- decisions are based on EU, rather than ex- count of immediate emotions. Some of the pected value, gives it descriptive appeal as well. research has shown that immediate integral For instance, it assumes that the difference in emotions play a critical role in decision mak- happiness (i.e., utility) between winning $1 and ing. However, other research has shown that winning $2 is not necessarily equal to the dif- immediate emotions, and especially but not ex- ference in happiness between winning $101 clusively incidental emotions, often propel de- and winning $102 (though the difference in cisions in different directions from expected value is equal). emotions—that is, in directions that run con- However, empirical research has docu- trary to the predictions of a consequentialist mented many behavioral phenomena that are perspective. The new research thus suggests inconsistent with the basic axioms, and thus in- that the consequentialist perspective is much consistent with the predictions of the EU too simple to be a descriptively valid account of model, and many of these anomalies can be at- actual behavior. tributed to unrealistic assumptions about the In this chapter, we review some of the critical determinants of expected emotions and the in- (consequentialist) assumptions and predictions fluence of immediate emotions. Several models of the dominant economic models of risky deci- have accounted for some of these anomalies by sion making, intertemporal choice, and social making more realistic assumptions about the preferences. For each of these areas, we first determinants of expected emotions. We next discuss behavioral phenomena that are anoma- review some of these theoretical innovations. lous from the consequentialist perspective, but We then discuss anomalies that can potentially that are rectified once the role of expected emo- be explained by taking account of the influence tions is taken into account. Next, we discuss of immediate emotions. phenomena that can potentially be illuminated by taking account of immediate emotions, both Innovations to the EU Model integral and incidental. We conclude by pro- Involving Expected Emotions posing directions for future research on the role of emotion in decision making. Relaxing the Asset Integration Assumption

In its original form, the EU model assumes that DECISION MAKING UNDER RISK people do not narrowly focus on potential outcomes when making a decision, but rather on Most decisions, including decisions of eco- how those outcomes affect their overall wealth. nomic importance, entail an element of risk, Thus the utility of a particular outcome is not because the consequences of alternative courses simply based on that outcome, but instead on of action are rarely known with certainty. Thus the integration of that outcome with all assets decision making under risk is a central topic in accumulated to that point. However, as origi- economics. nally noted by Markowitz (1952) and devel- Since first proposed by Daniel Bernoulli oped more fully by Kahneman and Tversky (1738/1954), the “expected utility” (EU) (1979), people typically make decisions with a model has served as the normative benchmark narrower focus. When evaluating the potential for decision making under risk in economics. outcomes of a decision, people tend to think in 9. The Role of Emotion in Economic Behavior 141 terms of incremental gains and losses, rather than they do about two smaller ones, as as- than in terms of changes in overall welfare. sumed in Loomes and Sugden (1982), then Suppose, for example, that Bob must decide Gamble A will be preferred to Gamble B. Simi- whether to accept or reject a gamble that offers larly, B is likely to be preferable to C. Since A is a 50% chance of winning $20 and a 50% preferred to B, and B is preferred to C, then chance of losing $10. If Bob currently possesses transitivity requires that A is preferred to C. $1 million in wealth, then the EU model as- However, in fact C is preferred to A, since sumes that he views the gamble as offering a choosing A over C exposes one to the risk of 50% chance of experiencing the utility of one large regret instead of two small ones. $1,000,020 and a 50% chance of experiencing the utility of $999,990. Markowitz (1952) ar- State 1 State 2 State 3 gued, however, that most people would instead Gamble A $10 $20 $30 process the gamble as it was presented, namely Gamble B $20 $30 $10 as offering a 50% chance of experiencing the Gamble C $30 $10 $20 utility of winning $20 and a 50% chance of experiencing the disutility of losing $10.3 Disappointment aversion and regret aversion theories have only met with modest empirical support. One problem with the predictive Relaxing the Assumption That Utility Is Strictly validity of regret aversion theories may be that Defined over Realized Outcomes anticipated regret only influences decision Another problematic assumption of the EU making when the possibility of regret is salient model is that unrealized outcomes do not influ- (Zeelenberg & Beattie, 1997; Zeelenberg, Beat- ence how we feel about realized outcomes. For tie, van der Plight, & De Vries, 1996). Con- example, suppose you anticipate a pay raise of sider, for example, the following gambles, in $10,000 and subsequently receive a $5,000 which one of four colors can be drawn with raise. Although the raise is a gain relative to the varying probability: status quo, you will likely code it as a loss, since it fails to meet expectations. Indeed, Gamble A Gamble B Koszegi and Rabin (2006) have recently pro- 90% chance of White, 90% chance of White, posed a model assuming that gains and losses which pays $0 which pays $0 are defined relative to expectations, rather than 6% chance of Red, 7% chance of Red, which pays $45 which pays $45 the status quo. 1% chance of Green, 1% chance of Green, Additionally, several modifications of the EU which pays $30 which pays –$10 model incorporate the tendency to compare 3% chance of Yellow, 2% chance of Yellow, what happens to what was expected to happen which pays -$15 which pays –$15 (e.g., Loomes & Sugden, 1986; Mellers, Schwartz, Ho, & Ritov, 1997). Other theories Since Green wins $30 in Gamble A and loses attempt to account for regret, a counterfactual $10 in Gamble B, choosing B could produce re- emotion that arises from a comparison be- gret if Green is drawn. This very salient poten- tween the outcome one experiences as a conse- tial for regret could lead to a preference for A quence of one’s decision and the outcome one over B, even though such a preference violates could have experienced as a consequence of monotonicity. However, the gambles can be re- making a different choice. Early versions of re- written to make the possibility of regret less sa- gret theory (e.g., Loomes & Sugden, 1982) pre- lient: dicted that regret aversion could lead to violations of fundamental axioms of the EU model, Gamble A Gamble B such as monotonicity (i.e., stochastically domi- 90% chance of White, 90% chance of White, nating gambles are preferred to the gambles which pays $0 which pays $0 they dominate). 6% chance of Red, 6% chance of Red, Regret can also lead to violations of transi- which pays $45 which pays $45 tivity. Consider, for example, the three gambles 1% chance of Green, 1% chance of Green, which pays $30 which pays $45 below. Assume that there are three equally 1% chance of Blue, 1% chance of Blue, likely states of nature; the table lists what each which pays –$15 which pays –$10 gamble pays if a particular state of nature is re- 2% chance of Yellow, 2% chance of Yellow, alized. If people care more about one big regret which pays –$15 which pays –$15 142 I. INTERDISCIPLINARY FOUNDATIONS

Note that Gambles A and B′ are equivalent to series of questions. In the control treatment, Gambles A and B, respectively; A and A′ both participants answered the questions while in have an expected value of $2.55, and B and B′ their natural (presumably not highly aroused) both have an expected value of $2.75. How- state. In the arousal treatment, participants ever, the potential for regret is no longer sa- were first asked to self-stimulate (masturbate) lient. Rather, B′ pays at least as much as A′ for while viewing erotic photographs, and were each possible color. Thus, even though A and presented with the same questions only after A′ are equivalent, A′ is likely to be less attrac- they had achieved a high but suborgasmic level tive than A, only because the way A′ and B′ are of arousal. When asked about their intention to framed obfuscates the potential for regret.4 use birth control in the future, aroused partici- However, note that regret is often more pants were less likely to report intending to use salient in prospect than in retrospect.5 a condom. Although arousal affected partici- Consider, for example, a study by Gilbert, pants’ risk attitudes, it did not affect their risk Morewedge, Risen, and Wilson (2004) that ex- perception. For example, aroused participants amined the extent to which subway passengers were no less likely to endorse this statement: regretted missing their train. Passengers who “If you pull out before you ejaculate, a woman entered a subway station within 6 minutes of can still get pregnant.” Although the authors missing the train (experiencers) were told that did not ask questions that would permit they missed their train by either 1 minute or 5 mediational analyses, the preliminary results minutes. They were then asked to report how suggest that immediate emotions had a direct much regret they felt. These ratings were com- effect on (predicted) behavior. pared to the ratings of passengers leaving the When experienced at more moderate levels, station (forecasters), who were asked to imag- however, affect can mediate the relationship ine how much regret they would feel if they between cognition and behavior. Antonio missed their train by 1 or 5 minutes. Fore- Damasio and his colleagues (Damasio, 1994; casters anticipated feeling greater regret if they Bechara, Damasio, Tranel, & Damasio, 1997) missed their train by 1 minute than by 5 min- have argued that decision makers encode the utes, though actual regret did not depend on consequences of alternative courses of action how close experiencers came to catching the affectively, and that such “somatic markers” train. A subsequent study suggested that the ef- critically influence decision making. Damasio fect was driven by forecasters’ inability to real- and colleagues have further argued that the ize how quickly they would absolve themselves ventromedial prefrontal cortex (VMPFC) plays of responsibility for the disappointing out- a critical role in this affective encoding process. come. Bechara et al. (1997) investigated the proposed Although work remains to be done to incor- role of the VMPFC in an experiment in which porate more determinants of expected emo- patients suffering damage to the VMPFC and tions into consequentialist models of decision non-brain-damaged individuals played a game making under risk, great progress has been in which the objective was to win as much made. We now discuss risky choice phenomena money as possible. Players earned hypothetical driven by immediate emotions. money by turning over cards that yielded either monetary gains or losses. On any given turn, players could draw from one of four decks, two Innovations to the EU Model Involving Immediate Emotions of which included $100 gains and two of which contained $50 gains. The high-paying decks Integral Emotions Influence Risky Decision Making also included a small number of substantial losses, resulting in a net negative expected When sufficiently strong, immediate emotions value for these decks. Bechara et al. (1997) can directly influence behavior, completely found that both nonpatients and those with precluding cognitive decision making VMPFC damage avoided the high-paying (Loewenstein, 1996). Ariely and Loewenstein decks immediately after incurring substantial (2005) experimentally examined the influence losses. However, individuals with VMPFC of sexual arousal on (hypothetical) risky deci- damage resumed sampling from the high- sion making (see also Loewenstein, Nagin, & paying decks more quickly than nonpatients Paternoster, 1997). Male participants were did after encountering a substantial loss. Thus, given a laptop computer and asked to answer a even though patients understood the game and 9. The Role of Emotion in Economic Behavior 143 wanted to win, they often went “bankrupt.” Bechara et al. (1997) reasoned that patients “knew” the high-paying decks were risky, but that their failure to experience fear when contemplating sampling from these decks made risky draws more palatable.6 While the Bechara et al. (1997) study has not been immune to crit- icism (see Maia & McClelland, 2004, for a particularly compelling critique, and Dunn, Dalgleish, and Lawrence, 2005, for a review of several critiques), the somatic marker hypothesis remains intuitively appealing. Other evidence suggesting that integral emotion influences decision making comes from studies of consumers’ willingness to insure against a variety of risks. Johnson, Hershey, Meszaros, and Kunreuther (1993), for example, asked participants how much they would be willing to pay for flight insurance that pro- FIGURE 9.2. Kahneman and Tversky’s prob- tected against death due to “any act of terror- ability-weighting function. From Kahneman and ism” or “any reason.” Since terrorism is only Tversky (1979). Copyright 1979 by the Econo- one of many reasons why a plane might crash, metric Society. Reprinted by permission. consequentialist models of decision making predict that participants will pay more for insurance covering all types of crashes than for comes. They speculate that affect-rich prizes insurance just covering terrorism. However, elicit greater degrees of hope and fear, and thus Johnson et al. (1993) found that participants an extreme overweighting of small probabili- were willing to pay slightly more for insurance ties and an extreme underweighting of large protecting against terrorism.7 probabilities. Indeed, Rottenstreich and Hsee Additional evidence of integral emotions’ (2001) found that participants’ willingness to impact on risky decision making comes from pay to avoid an electric shock was insensitive studies of probability weighting. The EU model to the probability of the shock, whereas will- assumes that the weight an outcome’s probabil- ingness to pay to avoid losing $20 was ex- ity receives in decision making is independent tremely sensitive to the probability of the loss.8 of the outcome; in fact, the model assumes lin- ear probability weighting (i.e., that outcomes Incidental Emotion Influences are weighted in exact proportion to their likeli- Risky Decision Making hood of occurring). However, more recent models of decision making under risk have In a study of market index returns across 26 challenged this assumption, suggesting instead countries from 1982 to 1997, Hirshleifer and that probabilities are weighted nonlinearly, as Shumway (2003) found that the amount of in Figure 9.2 (Kahneman & Tversky, 1979). sunshine (relative to expected amount of sun- Kahneman and Tversky’s (1979) proposed shine for a given time of year) was positively “probability-weighting function” suggests that and significantly correlated with market re- small probabilities are overweighted and large turns. The authors speculate that the phenome- probabilities are underweighted. non may be driven by incorrect attributions of Despite the innovation, models such as good mood to positive economic prospects Kahneman and Tversky’s (1979) still assume rather than correct attributions to the sunshine that probability weights are independent of (cf. Schwarz & Clore, 1983). Similarly, outcomes. This suggests, for example, that a Edmans, García, and Norli (2007) have found 1% chance of losing $1 has the same psycho- that stock market returns plummet when a logical impact as a 1% chance of losing your country’s soccer team is eliminated from an im- life. Rottenstreich and Hsee (2001) suggest that portant tournament (e.g., the World Cup). the probability-weighting function is flatter for They also document a dip in market returns affect-rich outcomes than for affect-poor out- following important losses in other sports (e.g., 144 I. INTERDISCIPLINARY FOUNDATIONS cricket, rugby, and hockey) in countries where or 10 years than to avoid a shock experi- those sports are popular.9 enced within the next 3 days. These anomalies can be reconciled with the DU model if one takes account of the observa- INTERTEMPORAL CHOICE tion that utility is not strictly a function of realized outcomes, but also of emotions experi- Models of intertemporal choice address how enced while waiting for those outcomes to decision makers choose between alternatives occur. Loewenstein (1987) proposes that peo- involving costs and benefits that are distributed ple derive utility from “savoring” future good over time. The “discounted utility” (DU) outcomes and disutility from dreading bad out- model is the dominant model of intertemporal comes.10 Indeed, in a brain imaging study in choice in economics (Samuelson, 1937). Struc- which participants were confronted with the turally, this model is closely parallel to the EU prospect of a real impending shock, Berns et al. model—and, like the EU model, has been de- (2006) found that components of the brain’s rived from a series of intuitively compelling ax- “pain matrix” (a cluster of regions that are ac- ioms (Koopmans, 1960). However, a number tivated during the experience of pain) are also of anomalies have been identified that call into active in anticipation of shock. Furthermore, question the descriptive validity of these axi- providing support for the idea that utility from oms, and thus the predictions of the DU model anticipation plays a causal role in the desire to (Loewenstein & Prelec, 1992). We next review expedite negative outcomes, individual differ- anomalies that can be reconciled with this ences in activation in response to anticipatory model once more realistic assumptions are pain predict individual tendencies to expedite made about the determinants of expected emo- shocks.11 tions; we then discuss anomalies that can be explained by taking account of immediate emo- Incorporating Affective Forecasting Errors tions. For the DU model to be descriptively valid, people must be able to forecast accurately how they Innovations to the DU Model will react emotionally to future outcomes. Involving Expected Emotions However, there is by now substantial evidence Relaxing the Assumption That Utility that people have difficulty making such fore- Is Strictly Defined over Realized Outcomes casts. Consider, for instance, a study by Brickman, Coates, and Janoff-Bulman (1978) in Like the EU model, the DU model assumes which lottery winners, persons with paraplegia that utility (and thus expected emotion) is or quadriplegia, and a control group were asked only a function of realized outcomes. If peo- to report their current happiness on a 5-point ple devalue future emotions, they should scale. The lottery group (n = 22) consisted of want to experience pleasurable outcomes im- people who had recently won at least $50,000 in mediately and postpone painful outcomes the Illinois state lottery. The paraplegic and whenever possible. However, contrary to this quadriplegic participants (n = 29) had become basic assumption, in many situations people paralyzed within the past year. Lottery winners prefer to get unpleasant outcomes over with reported a mean level of happiness virtually quickly, or to “leave the best for last.” In an identical to that of the control group (4.00 vs. early study documenting this phenomenon, 3.82), whose happiness was significantly differ- Loewenstein (1987) asked 30 undergraduates ent from, but surprisingly close to, the mean how much they would be willing to pay im- happiness level among paraplegic and quadri- mediately to obtain a kiss from the movie plegic participants (2.96). Although the lottery star of their choice and to avoid receiving a winners and the paraplegic and quadriplegic (nonlethal) 110-volt shock, after several time participants were not prospectively asked to delays. Contrary to the predictions of the DU predict their future happiness (since they could model, respondents were willing to pay more not be identified beforehand), it seems likely to experience a kiss delayed by 3 days than that both groups would have overestimated the an immediate kiss or one delayed by 3 hours hedonic impact of their future circumstances.12 or 1 day, and were also willing to pay more Loewenstein and Adler (1995) examined to avoid a shock that was delayed for 1 year whether people could predict falling subject to 9. The Role of Emotion in Economic Behavior 145 the “endowment effect” (Thaler, 1980), which mediately after the coin toss. The experimenter refers to the tendency for people to value an then flipped a coin and paid participants ac- object more highly if they possess it than they cordingly. Participants then rated how they felt would value the same object if they did not. In at that moment. Some participants were also the typical demonstration of the effect (see, asked to report what they would think after the e.g., Kahneman, Knetsch, & Thaler, 1990), coin toss, and once the coin had actually been some participants (sellers) are endowed with an tossed, they were asked to report their actual object and given the option of trading it for thoughts. Kermer et al. (2006) found that peo- various amounts of cash; other participants ple expected losing $3 to diminish their happi- (choosers) are not given the object, but are ness (relative to happiness reported at the given a series of choices between receiving the beginning of the experiment) more than it actu- object and receiving various amounts of cash. ally did.13 This suggests that the predictors in Although the objective wealth position of the Loewenstein and Adler (1995) may have accu- two groups is identical, as are the choices they rately based their predicted selling prices on face, endowed participants hold out for signifi- how they would actually feel after losing an ob- cantly more money than those who are not en- ject. Sellers, by contrast, may have based their dowed. Loewenstein and Adler (1995) in- selling prices on unrealistically negative fore- formed some participants that they would be casts of how they would feel after losing an ob- endowed with an object (a mug engraved with ject.14 their school logo) and asked them to predict In a behavioral economic model of inter- the price at which they would sell the object temporal choice that incorporates affective back to the experimenter once they were en- forecasting errors, Loewenstein, O’Donoghue, dowed. These participants, and others who did and Rabin (2003) propose that people exagger- not make a prediction, were then endowed ate the degree to which their future tastes with the object and given the opportunity to will resemble their current tastes. Conlin, sell it back to the experimenter. Participants O’Donoghue, and Vogelsang (2007) find evi- who were not yet endowed substantially dence of such “projection bias” in catalog or- underpredicted their own postendowment sell- ders of cold-weather-related clothing items and ing prices. In a second study, selling prices were sports equipment. People are overinfluenced by elicited from participants who were actually the weather at the time they make decisions, as endowed with an object and from others who measured by their likelihood of returning the were told they had a 50% chance of getting the item: A decline of 30° F on the date an item is object. Selling prices were substantially higher ordered increases the probability of a return by for the former group, and the valuations of 3.95%. participants who were not sure of getting the Economists have incorporated more realistic object were indistinguishable from the buying assumptions about expected emotions into prices of participants who did not have the ob- models of intertemporal choice. However, some ject. phenomena, driven by immediate emotions, re- Loewenstein and Adler’s (1995) results sug- main anomalous from the perspective of such gest that participants who were not endowed models. We now turn to these phenomena. with an object failed to predict how painful it would be to part with the object once they pos- Innovations to the DU Model sessed it. That is, non-endowed participants Involving Immediate Emotions made “affective forecasting” errors when predicting their future attachment to the object. Relaxing the Assumption of Exponential Discounting However, a recent study by Kermer, Driver- Linn, Wilson, and Gilbert (2006) suggests that The DU model assumes that people discount it may be the sellers who are making the affec- future flows of utility at a fixed discount rate tive forecasting error (see also Galanter, 1992). based on when the utility will be experienced. Kermer et al. (2006) first asked participants to Discounting at a fixed rate (i.e., “exponential” report their baseline affect. Participants then discounting) means that a given time delay received a $5 show-up fee and were told that a leads to the same amount of discounting re- coin would be flipped to determine whether gardless of when it occurs. According to the they would win an additional $3 or lose $2. DU model, delaying the delivery of a good by 1 Next, they predicted how they would feel im- day leads to the same degree of time discount- 146 I. INTERDISCIPLINARY FOUNDATIONS ing whether that delay makes the difference be- Thus impulsivity may reflect factors other tween consuming the good tomorrow rather than a devaluation of expected emotions. Im- than today or in 101 days rather than 100 mediate emotions may also produce non- days. However, an overwhelming amount of exponential discounting. To examine the influ- empirical work suggests that people (as well as ence of immediate emotions on impulsivity, animals) do not discount the future exponen- McClure, Laibson, Loewenstein, and Cohen tially (Kirby & Herrnstein, 1995; Rachlin & (2004) measured the brain activity of partici- Raineri, 1992). Rather, people care more about pants with functional magnetic resonance im- the same time delay if it is proximal rather than aging (fMRI) while they made a series of distal—a general pattern that has been refereed intertemporal choices between small proximal to as “hyperbolic time discounting” (Ainslie, rewards ($R available at delay d) and larger de- 1975). For example, delaying consumption of a layed rewards ($R′ available at delay d′), where pleasurable good from today to tomorrow is $R <$R′ and d < d′. Rewards ranged from $5 more distressing than delaying consumption to $40 Amazon.com gift certificates, and the from 100 days from now to 101 days from delay ranged from the day of the experiment to now. Hyperbolic time discounting predicts that 6 weeks later. McClure et al. (2004) investi- people will behave farsightedly when the con- gated whether there were brain regions that sequences of their decision are delayed. In such showed elevated activation (relative to a resting situations, decision makers will place great state benchmark) only when immediacy was an weight on long-term costs and benefits. How- option (i.e., activation when d = 0, but no acti- ever, when consequences are immediate, hyper- vation when d > 0), and whether there were re- bolic time discounting will produce behavior gions that showed elevated activation when that appears impulsive.15 participants were making any intertemporal Consider, for example, an experiment by decision irrespective of delay. McClure et al. Read, Loewenstein, and Kalyanaraman (1999) (2004) found that time discounting is associ- in which participants were asked to select 1 of ated with the engagement of two neural sys- 24 movies to watch. Some of the movies were tems. Limbic and paralimbic cortical struc- “highbrow” (e.g., Schindler’s List), and some tures, which are known to be rich in were “lowbrow” (e.g., The Mask). Some par- dopaminergic innervation, were preferentially ticipants were asked to choose a movie to recruited for choices involving immediately watch that night, whereas others were asked to available rewards. In contrast, fronto-parietal choose a movie to watch in the future. Consis- regions, which support higher cognitive func- tent with hyperbolic discounting, “lowbrow” tions, were recruited for all intertemporal movies (ones that are high in short-run bene- choices. Moreover, the authors found that fits, but low in long-run benefits) were most when choices involved an opportunity for im- popular among participants selecting a movie mediate reward, thus engaging both systems, for immediate viewing.16 greater activity in fronto-parietal regions than Behavioral economists have made great in limbic regions was associated with choosing progress in modeling hyperbolic discounting larger delayed rewards.17 These results suggest (e.g., Laibson, 1997). Such models implicitly that the experience of immediate emotion assume that discounting leads to impulsive rather than the devaluation of expected emo- behavior by diminishing the importance of ex- tion may, at least in some situations, drive pected emotions. However, when the timing of impulsivity.18 consumption is held constant, various other situational factors can also lead to impulsivity. Integral Emotions Influence Intertemporal Choice Walter Mischel (1974) and colleagues, for example, have extensively studied the impact of Suppose you are deciding whether or not to physical proximity of rewards on the impulsivi- buy a CD for $10. The DU model predicts that ty of children. Children faced with the choice you will buy the CD if the anticipated pleasure between a small immediate reward (e.g., one of listening to it exceeds its “opportunity cost” marshmallow immediately or two marshmal- (i.e., the forgone pleasure that could have been lows in 15 minutes) and a larger delayed re- purchased with the $10). However, Frederick, ward (two marshmallows) tend to behave more Novemsky, Wang, Dhar, and Nowlis (2006) impatiently when the immediate reward is visi- suggest that people do not spontaneously con- ble. sider opportunity costs when deciding whether 9. The Role of Emotion in Economic Behavior 147 or not to purchase goods. Frederick et al. chasing decisions. Insula activation has (2006) asked participants whether they would previously been observed in connection with (hypothetically) be willing to purchase a desir- aversive stimuli such as disgusting odors able video for $14.99. They simply varied (Wicker et al., 2003), unfairness (Sanfey, whether the decision not to buy it was framed Rilling, Aronson, Nystrom, & Cohen, 2003), as “not buy this entertaining video” or “keep and social exclusion (Eisenberger, Lieberman, the $14.99 for other purchases.” Although the & Williams, 2003). Thus when the delayed two phrases described objectively equivalent costs of immediage indulgence are not explic- actions, the latter highlighted the pleasure that itly represented (as in, e.g., McClure et al., would be forgone by purchasing the video. 2004), but rather implicitly captured by prices, Frederick et al. (2006) found that drawing at- participants appear to rely on an anticipatory tention to opportunity costs significantly re- “pain of paying” (Prelec & Loewenstein, 1998) duced the proportion of participants willing to to curtail their spending. purchase the video, suggesting that many participants were not spontaneously considering Incidental Emotions Influence Consumer Choice opportunity costs (cf. Jones, Frisch, Yurak, & Kim, 1998). In other research conducted wiht Cynthia If many people do not take opportunity costs Cryder (Rick, Cryder, & Loewenstein, 2008) into account when deciding whether or not to investigated whether individuals chronically purchase goods, then how do they make such differed in their tendency to experience antici- decisions? In a project wiht Brian Knutson, patory pain when making purchasing deci- Elliott Wimmer, and Drazen Prelec (Knutson, sions. We hypothesized that individuals who Rick, Wimmer, Prelec, & Loewenstein, 2007), typically experience an intense pain of paying we investigated this question in an experiment may generally spend less than they would ide- in which participants chose whether or not to ally like to spend, whereas individuals who ex- purchase a series of discounted consumer perience minimal pain of paying may typically goods while having their brains scanned with spend more than they would ideally like to fMRI. The goods ranged in retail price from spend. We developed a “Spendthrift–Tight- $10 to $80, and were offered at a 75% dis- wad” scale to measure individual differences in count to encourage spending. Participants were the pain of paying and found that tightwads given $20 to spend and were told that one of outnumbered spendthrifts by a 3:2 ratio in a their decisions would be randomly selected to sample of more than 13,000 people. Rick count for real. At the conclusion of the experi- (2007) hypothesized that incidental sadness ment, participants indicated how much they could help both tightwads and spendthrifts liked each product and how much they would overcome their prepotent affective responses to be willing to pay for it. spending. The hypothesis was based on previ- We found that the extent to which partici- ous experimental work suggesting that sadness pants reported liking the products correlated deepens deliberation (e.g., Tiedens & Linton, positively with activation in nucleus 2001) and motivates people to change their cir- accumbens, a target of dopaminergic projec- cumstances (e.g., Lerner et al., 2004). Rick tions that has previously been associated with (2007) tested the hypothesis in an experiment anticipation of gains and self-reported happi- in which tightwads and spendthrifts decided ness (Knutson, Adams, Fong, & Hommer, whether or not to purchase a variety of goods 2001). Moreover, consumer surplus (i.e., the while listening to neutral or sad music. As pre- difference between self-reported willingness to dicted, tightwads spent more when sad than pay for the good and its price) correlated posi- when in a neutral state, and spendthrifts spent tively with activation in medial prefrontal cor- less when sad than when in a neutral state. tex, a region previously associated with the re- ceipt of unexpectedly large gains (e.g., Knutson, Fong, Bennett, Adams, & Hommer, SOCIAL PREFERENCES 2003). We also found that activation in both regions correlated positively with purchasing Although there are widely accepted normative decisions. However, we found that activation benchmarks for risky decision making and in insula during the period when subjects first intertemporal choice, no such benchmarks exist saw the price correlated negatively with pur- for how people should behave toward others. 148 I. INTERDISCIPLINARY FOUNDATIONS

However,many economic models make the sim- trials, responders received the same offer, but plifying, but unrealistic, assumption that people this time from a computer. As in Blount (1995), are strictly self-interested. Below we review participants were more willing to accept low behavioral economic models of social prefer- offers from computer proposers than from hu- ences that have incorporated more realistic as- man proposers. Moreover, activation in ante- sumptions about the determinants of expected rior insula, a region commonly implicated in emotions in social interactions. We then review the experience of pain (e.g., Knutson et al., anomalies driven by immediate emotions. 2007), was greater for unfair offers from human proposers than for fair offers from human Expected Emotion: Relaxing proposers. Insula activation was also signifi- the Pure Self-Interest Assumption cantly greater in response to unfair offers from human proposers than in response to unfair of- Economists frequently study social preferences fers from computer proposers. In fact, whether in the context of the “ultimatum game” (Guth, players rejected unfair offers from human pro- Schmittberger, & Schwarze, 1982). In the typi- posers could be predicted reliably by the level cal ultimatum game, a “proposer” offers some of their insula activity. Thus it appears that in- portion of an endowment to a “responder,” tegral emotions influence responders’ behav- who can either accept the offer or reject it. If ior in the ultimatum game (cf. Pillutla & the responder accepts the offer, the money is di- Murninghan, 1996). vided according to the proposed split. If the re- Behavioral economists have created more sponder rejects the offer, both players leave descriptively valid models of social prefer- with nothing. Since purely self-interested re- ences by relaxing the assumption of pure self- sponders should accept any positive offer, self- interest. However, some phenomena driven by interested proposers should offer no more than immediate emotion cannot be explained by the smallest positive amount possible. How- such models. We review such anomalies below. ever, average offers typically exceed 30% of the pie, and offers of less than 20% are frequently Integral Emotions Influence rejected (see Camerer, 2003). Social Preferences Several behavioral economic models have emerged that incorporate a taste for fairness.19 Recent work on the “identifiable-victim effect” Rabin (1993) proposes a model in which peo- (Small & Loewenstein, 2003), which refers to ple derive utility from reciprocating intentional the tendency to give more to identifiable vic- (un)kindness with (un)kindness (see also tims than to statistical victims, suggests that in- Dufwenberg & Kirchsteiger, 2004). Blount tegral emotions play a role in generosity to- (1995) conducted an interesting variant of the ward others (see also Schelling, 1968; Kogut & ultimatum game to investigate the role of inten- Ritov, 2005). Subsequent research has demon- tions in social behavior. Some responders were strated that people are also more punitive to- told that the proposer with whom they were ward identifiable wrongdoers than toward paired would make an offer, as in the standard equivalent but unidentified wrongdoers, and ultimatum game. Other responders were told that anger mediates the effect of identifiability that the offer would be randomly generated. on punishing behavior (Small & Loewenstein, Blount (1995) found that responders were will- 2005). To capture these phenomena, as well as ing to accept significantly less when the offer a variety of experimental findings, Loewenstein was generated randomly than when it came and Small (2007) have proposed a dual-process from the proposer. model of helping behavior in which a sympa- Sanfey, Rilling, Aronson, Nystrom, and Co- thetic but highly immature emotional system hen (2003) conducted a similar study in which interacts with a more mature but uncaring de- participants played the ultimatum game while liberative system. having their brains scanned with fMRI. Partici- pants, all responders, were told they would Incidental Emotions Influence play the ultimatum game with 10 different hu- Social Preferences man proposers (though offers were actually de- termined by the experimenters). Responders re- Andrade and Ariely (2006) investigated the im- ceived five “fair” offers ($5 for proposer, $5 for pact of incidental emotions on behavior in the respondent) and five unfair offers. In 10 other ultimatum game. They induced either inciden- 9. The Role of Emotion in Economic Behavior 149 tal happiness or anger, and then had partici- Second, substantial research supports the pants play the role of responder in an ultima- idea that immediate emotions also play an im- tum game in which they were offered $4 of a portant role in decision making. Integral imme- $20 endowment. After deciding whether to ac- diate emotions arise from contemplating the cept or reject the offer, participants then played potential outcomes of a decision. In some the role of proposer in a second ultimatum cases, these emotions seem to play a beneficial game, with a presumably different partner than role in decision making, informing decision in the first game. Andrade and Ariely (2006) makers about their own values. But in other found that happy responders were less likely cases, such as the disproportionate fear com- than angry responders to reject unfair offers in monly associated with flying as opposed to the initial ultimatum game. Surprisingly, how- driving, integral emotions may cause people to ever, proposers who were initially induced to act contrary to their own material interests. In feel happy made more selfish proposals in the contrast to the mixed role played by integral second ultimatum game. The authors reasoned emotions, it is much more difficult to justify the that angry individuals, who were more likely to well-documented role of incidental emotions, reject unfair offers than happy individuals in which by definition are unrelated to the deci- the initial ultimatum game, misattributed their sion at hand. behavior to stable preferences rather than to in- In general, research on expected emotions is cidental affect. Later, due to a “false-consensus far more advanced than that on immediate effect” (Ross, Greene, & House, 1977; but see emotions. As a result, there is a pressing need Dawes & Mulford, 1996), the previously angry for more research to examine the causes and individuals inferred that others would also be consequences of immediate emotions, and to likely to reject unfair offers and therefore, as understand the complex interplay of immediate proposers, made very generous offers. By con- and expected emotions in the production of trast, the authors reasoned that happy individ- behavior. In some cases, immediate and ex- uals, who were less likely to reject unfair offers pected emotions seem to complement one an- than angry individuals in the initial ultimatum other. This is true, for example, when immedi- game, also misattributed their behavior to sta- ate emotions provide decision makers with a ble preferences. Accordingly, previously happy better understanding of their own values—an individuals inferred that others would also be understanding that may help them to better unlikely to reject unfair offers and therefore, as predict their own future feelings. For instance, proposers, made very selfish offers.20 the experience of anticipatory guilt may help students who are contemplating cheating on an exam to appreciate the guilt they would experi- CONCLUSION ence after doing so. In other cases, however, immediate and expected emotions come into As the foregoing review indicates, emotions conflict. For example, the immediate effect of a influence economic behavior in two distinct positive mood may be to make decision makers ways. First, people anticipate, and take into ac- more inclined to take risks—but, by a different, count, how they are likely to feel about the po- consequentialist path, a positive mood might tential consequences of alternative courses of also make decision makers more risk-avoidant, action. As discussed, such a role for expected with the goal of not risking a disturbance to the emotions is entirely consistent with consequen- positive feelings (Isen, Nygren, & Ashby, 1988; tialist economic accounts of decision making. Kahn & Isen, 1993). Research on the role of expected emotions in The clash between immediate and expected decision making has taken a variety of direc- emotions is also a major cause of self-control tions. It has assessed the types of emotions that problems. For example, people are often driven people actually experience when different out- by immediate emotions to eat, drink, and make comes are realized, with a special focus on merry, but in some of these situations, contem- counterfactual emotions. It has examined peo- plation of expected emotional consequences ple’s predictions of what emotions they will ex- may discourage indulgence. Psychologists have perience, and the accuracy of such predictions. for decades been developing “dual-process” And, it has sought to determine the degree to models that can be interpreted in such terms which decisions are in fact guided by predicted (see Evans, 2008, for a review), and in recent emotions. years economists have begun to follow their 150 I. INTERDISCIPLINARY FOUNDATIONS lead. Thaler and Shefrin (1981) were the first studying the impact of weather on the stock economists to do so; their model adopts a market have generally taken a rather simplistic principal–agent framework, in which a far- view—that bad weather should lead to nega- sighted planner (the principal) attempts to rec- tive emotions, which should in turn lead to oncile the competing demands of a series of negative price movements. But psychologists myopic doers (the agents). More recently, many studying the impact of emotions on risk taking dual-process models have focused on the prob- find that different specific negative emotions lem of self-control (Brocas & Carrillo, 2006; can have very different effects. More relevant Fudenberg & Levine, 2006; Benhabib & Bisin, to the central theme of this chapter, that they 2005; Loewenstein & O’Donoghue, 2004; have found negative emotions can exert con- Bernheim & Rangel, 2004). flicting effects on risk taking, depending on Although most of the dual-process models whether the mechanism is consequentialist or proposed by economists have sought to adhere more reflexive. to the standard consequentialist perspective, Economists’ understanding of the role of introducing a role for immediate emotions emotions in economic behavior has made enor- should raise questions about whether such a mous strides in recent decades. However, there perspective is “up to the job” of providing a is still a long distance to go. useful account of human behavior. Behavior under the control of immediate emotions bears little resemblance to the reflective weighing of NOTES costs and benefits that is the prototype of rational economic decision making. Instead, it is a 1. Note that the distinction between expected and im- much more reflexive process that often drives mediate emotions closely maps onto other com- behavior in exactly the opposite direction from monly discussed distinctions in economics and psy- that suggested by a weighing of costs and bene- chology, such as the broad distinction between fits. Whether behavior driven by immediate cognition and emotion, or Adam Smith’s (1759/ emotions even warrants the label of “decision 1981) distinction between the “impartial spectator” making” seems questionable. and the “passions.” 2. On the surface, it seems somewhat unethical to arti- In closing, we note two potential (and, we ficially induce visceral states in order to sell prod- believe, fruitful) directions for future research ucts. However, food companies that failed to prey on the role of emotion in decision making. The on the affective vulnerability of consumers would first is the need to study stronger emotions than probably be driven out of business by other compa- have generally been examined in the empirical nies that did. Hence one could argue that food com- literature. Many vitally important decisions are panies that pump artificial smells into the atmo- made “in the heat of the moment,” and indeed sphere to stimulate hunger are not evil, but rather important economic decisions such as major are doing what they must to stay afloat. purchases often evoke powerful emotions. But 3. Note that narrowly focusing on gains and losses studying the impact of such emotions is rather than on changes in overall welfare suggests difficult—in part because it is difficult if not that all people, regardless of their current wealth position, view gambles the same way. Indeed, such a impossible to manipulate such strong emo- narrow focus may explain why some extraordi- tional states experimentally, and in part be- narily wealthy individuals take big risks to achieve cause people generally do not like to be studied small gains and avoid small losses (e.g., Martha when they are in heightened emotional states. Stewart, worth hundreds of millions of dollars, en- Gaining a better understanding of the role gaged in insider trading to avoid a loss of less than of immediate emotions in economic decision $50,000). making, therefore, is going to require research- 4. As Sugden (1986) notes, another problem with ers who are willing to extend themselves into regret-aversion models may be that it is recrimina- “hot” situations and creative enough to find tion—regret accompanied by the feeling that one natural experiments in which people are natu- should have behaved differently—rather than regret rally assigned to different emotional states be- that one cares about and attempts to avoid. Sup- pose, for example, you take your car to your regular fore they make important decisions. mechanic, Sue, for an oil change. You have never The second pressing need is for economic re- had a problem with this mechanic’s work, but this search that takes fuller account of the range of time she uses the wrong type of oil, which causes the insights that psychologists are developing into car to break down. In this situation, you surely re- emotions. Thus, for example, economists gret that the mistake was made, but you probably 9. The Role of Emotion in Economic Behavior 151

do not blame yourself for taking it to Sue, since you (2001), Ditto et al. (2006) found that participants’ had no reason to anticipate such a mistake based on willingness to play the game was insensitive to the her past performance. Now suppose that you in- probability of winning cookies when the cookies stead had decided to change your own oil. You have were baked in the lab, whereas willingness to play never done so before, but you decide it is worth try- was very sensitive to the probability of winning ing to save the money. Your inexperience leads you when the cookies were merely described. to use the wrong type of oil, causing the car to break 9. Also, Lerner and Keltner (2001) find that down. As in the previous scenario, you regret that dispositional (i.e., incidental) anger and fear have the mistake was made. However, now there is likely opposing effects on risk preferences. Specifically, an- to be recrimination as well: You think that you gry people tend to prefer risk (see also Fessler, should have known better than to try to change Pillsworth, & Flamson, 2004), whereas fearful peo- your own oil. ple tend to avoid it. The authors explain their results 5. Interestingly, however, Kivetz and Keinan (2006) in terms of the cognitive appraisals generated by the show that regret from choosing virtues over vices in- emotions (Smith & Ellsworth, 1985). Anger is gen- creases over time, whereas regret from choosing erally associated with appraisals of certainty and in- vices over virtues diminishes over time. dividual control, whereas fear is generally associ- 6. Note that the extent to which emotional deficits lead ated with appraisals of uncertainty and situational to poor decision making depends largely on situa- control. These incidental emotions, through their tional factors. In a similar study in which risky associated appraisals, appear to influence partici- choices had a higher expected value than riskless pants’ cognitive evaluations of the problem, thus in- choices, Shiv, Loewenstein, Bechara, Damasio, and fluencing their subsequent decisions. Damasio (2005) found that patients with damage to 10. Loewenstein’s model applies only to deterministic brain regions associated with processing emotion outcomes (e.g., a guaranteed kiss from a movie star earned more than control participants. in the future). Caplin and Leahy (2001) note that 7. One natural explanation for these results is that many anticipatory emotions (e.g., suspense) are “unpacking” vivid subsets of a larger set provides a driven by uncertainty about the future. They pro- more effective retrieval cue when people are recall- pose a model that modifies the EU model to incor- ing past causes of plane crashes (e.g., Tversky & porate such anticipatory emotions, and then show Koehler, 1994). Such an account would be consis- that it can explain a variety of phenomena (e.g., the tent with a consequentialist model of decision mak- overwhelming preference for riskless bonds over ing that allows for errors in judging probabilities. stocks). However, other work suggests that this result 11. In addition to savoring and dread, an entirely differ- should not be interpreted in purely cognitive terms. ent type of anticipation may also drive inter- Slovic, Fischhoff, and Lichtenstein (1980), for ex- temporal choice: the anticipation of memories ample, speculated that people’s willingness to insure (Elster & Loewenstein, 1992). For example, people themselves against unlikely losses may be related to may perform challenging but unpleasant activities how much these potential losses cause worry or (e.g., mountain climbing) partly because they savor concern. Consistent with this view, a number of the pleasant memories of conquering the challenge studies have shown that knowing someone who has (see also Keinan, 2006). been in a flood or earthquake, or having been in one 12. Addressing an important limitation of the Brickman oneself, greatly increases the likelihood of purchas- et al. (1978) study, Gilbert et al. (1998) conducted a ing insurance (Browne & Hoyt, 2000). Although study in which affective forecasts could be elicited this finding, like that of Johnson et al. (1993), could prior to an important life event. Specifically, Gilbert be explained in consequentialist terms as resulting et al. (1998) studied assistant professors’ forecasts from an increase in individuals’ expectations of ex- of how they would feel after their tenure decisions; periencing a flood or earthquake in the future, the the investigators compared these forecasts to the effect remains significant even after subjective ex- self-reported well-being of others whose tenure de- pectations are controlled for (Kunreuther et al., cisions had been made in the past. The sample con- 1978). sisted of all assistant professors who were consid- 8. Similarly, Ditto, Pizarro, Epstein, Jacobson, and ered for tenure in the liberal arts college of a major MacDonald (2006) conducted an experiment in university over a 10-year period, and it was divided which participants were given the opportunity to into three categories: current assistant professors, play a game that would either result in winning those whose decisions were made less than 5 years chocolate chip cookies or being required to work on earlier, and those whose decisions were made more a boring task for an extra 30 minutes. Half of the than 5 years earlier. Current assistant professors participants were only told about the cookies, predicted that they would be much happier during whereas for the other half the cookies were freshly the first 5 years after a positive decision, but that baked in the lab and placed in front of the partici- this difference would dissipate during the subse- pants as they decided whether or not to play the quent 5 years. Thus they expected to adapt much game. Consistent with Rottenstreich and Hsee more slowly than others actually did: There was no 152 I. INTERDISCIPLINARY FOUNDATIONS

significant difference in reported well-being be- with activation in limbic regions, making partici- tween those who had and had not received tenure in pants’ visceral attraction to the cake more either the first 5 or the next 5 years afterward. influential. 13. By contrast, participants accurately predicted how 19. But see Dana, Weber, and Kuang (2007) for evidence much winning the coin flip would increase their suggesting that some actions that appear to reflect a happiness. taste for fairness may in fact reflect a desire to ap- 14. Why are people often unable to accurately predict pear to have a taste for fairness. their affective reactions to negative events? Kermer 20. Incidental emotion also influences prosocial behav- et al. (2006) suggest that people do not realize how ior. Darlington and Macker (1966), for example, capable they are of finding “silver linings.” For ex- found that incidental guilt increased participants’ ample, participants who were asked to report their willingness to donate blood. Alice Isen and her col- thoughts before and after losing the coin flip were leagues (e.g., Isen & Levin, 1972; Isen, Horn, & significantly more likely to think about their $2 Rosenhan, 1973; Isen, Clark, & Schwartz, 1976) profit after losing the coin flip than before the coin have found in a variety of settings that incidental was flipped. Conversely, participants were more happiness (induced, e.g., by finding a dime in a likely to think they would focus on the $3 loss be- phone booth or receiving free cookies) increases fore the coin was flipped than they actually did after people’s willingness to help others (e.g., by picking losing the coin flip. Other researchers (e.g., Schkade up their dropped papers or by helping the experi- & Kahneman, 1998) attribute affective forecasting menter with a subsequent task; but see Isen & errors to “focusing illusions,” whereby people exag- Simmonds, 1978). Incidental gratitude also in- gerate the impact of specific narrow changes in their creases people’s willingness to help others (Bartlett circumstances on well-being. Both are plausible ex- & DeSteno, 2006). Although the preceding studies planations of the affective forecasting errors docu- did not deal directly with money, note that the help- mented in the studies discussed here. ing behavior they documented did involve expendi- 15. However, Kivetz and Simonson (2002) suggest that tures of costly resources (e.g., blood, effort, atten- some people have a hard time selecting luxuries tion). (items that are presumably high in short-run benefits, but low in long-run benefits) over cash when either would be received shortly after the decision. REFERENCES They demonstrate that choosing luxuries over cash is easier when the consequences of the decision are Ainslie, G. (1975). Specious reward: A behavioral the- delayed. ory of impulsiveness and impulse control. Psycholog- 16. Goldstein and Goldstein (2006) document a similar ical Bulletin, 82(4), 463–496. phenomenon among Netflix customers, who watch Andrade, E. B., & Ariely, D. (2006). Short- and long- and return low-brow movies right away, but let term consequences of emotions in decision making. high-brow movies sit around much longer before Manuscript in preparation, University of California, watching them. Berkeley. 17. However, note that since the rewards were gift cer- Ariely, D., & Loewenstein, G. (2005). The heat of the tificates, the consumption they afforded was not im- moment: The effect of sexual arousal on sexual deci- mediate in any conventional sense. To address this sion making. Journal of Behavioral Decision limitation, McClure, Ericson, Laibson, Loewen- Making, 18(1), 1–12. stein, and Cohen (2007) ran an experiment in which Bartlett, M. Y., & DeSteno, D. (2006). Gratitude and participants were asked not to drink any liquids prosocial behavior: Helping when it costs you. Psy- during the 3 hours preceding their session. While chological Science, 17(4), 319–325. having their brains scanned with fMRI, participants Benhabib, J., & Bisin, A. (2005). Modeling internal com- made a series of choices between receiving a small mitment mechanisms and self-control: A neuroeco- amount of juice or water immediately (by having it nomics approach to consumption-saving decisions. squirted into their mouths) and receiving a larger Games and Economic Behavior, 52(2), 460–492. amount of juice or water up to 20 minutes later. Bechara, A., Damasio, H., Tranel, D., & Damasio, A. R. Like McClure et al. (2004), McClure et al. (2007) (1997). Deciding advantageously before knowing the found that limbic regions were preferentially re- advantageous strategy. Science, 275, 1293–1295. cruited for choices involving immediately available Bernheim, B. D., & Rangel, A. (2004). Addiction and juice or water, whereas fronto-parietal regions were cue-triggered decision processes. American Eco- recruited for all choices. nomic Review, 94(5), 1558–1590. 18. The results are consistent with earlier behavioral re- Bernoulli, D. (1954). Specimen theoriae novae de search by Shiv and Fedorikhin (1999), who found mensura sortis [Exposition of a new theory on the that cognitive load increases the likelihood of measurement of risk]. Econometrica, 22(1), 23–36. choosing cake over fruit salad. The McClure et al. (Original work published 1738) (2004) results suggest that cognitive load interfered Berns, G. S., Chappelow, J., Cekic, M., Zink, C. F., more with activation in fronto-parietal regions than Pagnoni, G., & Martin-Skurski, M. E. (2006). Neu- 9. The Role of Emotion in Economic Behavior 153

robiological substrates of dread. Science, 312, 754– ory and anticipation. In G. Loewenstein & J. Elster 758. (Eds.), Choice over time. New York: Russell Sage Blount, S. (1995). When social outcomes aren’t fair: The Foundation. effect of causal attributions on preferences. Organi- Evans, J. St. B. T. (2008). Dual-processing accounts of zational Behavior and Human Decision Processes, reasoning, judgment, and social cognition. Annual 63(2), 131–144. Review of Psychology, 59.[pp tk] Brickman, P., Coates, D., & Janoff-Bulman, R. (1978). Fessler, D. M. T., Pillsworth, E. G., & Flamson, T. J. Lottery winners and accident victims: Is happiness (2004). Angry men and disgusted women: An evolu- relative? Journal of Personality and Social Psychol- tionary approach to the influence of emotions on risk ogy, 36(8), 917–927. taking. Organizational Behavior and Human Deci- Brocas, I., & Carrillo, J. D. (2006). The brain as a hier- sion Processes, 95, 107–123. archical organization. Manuscript in preparation, Frederick, S., Novemsky, N., Wang, J., Dhar, R., & University of Southern California. Nowlis, S. (2006). Opportunity costs and consumer Browne, M. J., & Hoyt, R. E. (2000). The demand for decisions. Manuscript in preparation, Massachusetts flood insurance: empirical evidence. Journal of Risk Institute of Technology. and Uncertainty, 20(2), 271–289. Fudenberg, D., & Levine, D. K. (2006). A dual-self Buchel, C., & Dolan, R. J. (2000). Classical fear condi- model of impulse control. American Economic Re- tioning in functional neuroimaging. Current Opinion view, 96(5), 1449–1476. in Neurobiology, 10(2), 219–233. Galanter, E. (1992). Utility functions for nonmonetary Camerer, C. (2003). Behavioral game theory: Experi- events. American Journal of Psychology, 65, 45–55. ments in strategic interaction. Princeton, NJ: Prince- Gilbert, D. T., Morewedge, C. K., Risen, J. L., & Wil- ton University Press. son, T. D. (2004). Looking forward to looking back- Caplin, A., & Leahy, J. (2001). Psychological expected ward: The misprediction of regret. PsychologicalSci- utility theory and anticipatory feelings. Quarterly ence, 15(5), 346–350. Journal of Economics, 116(1), 55–80. Gilbert, D. T., Pinel, E. C., Wilson, T. D., Blumberg, S. Conlin, M., O’Donoghue, T., & Vogelsang, T. J. (2007). J., & Wheatley, T. P. (1998). Immune neglect: A Projection bias in catalog orders. American Eco- source of durability bias in affective forecasting. nomic Review, 97(4), 1217–1249. Journal of Personality and Social Psychology, 75(3), Damasio, A. R. (1994). Descartes’ error: Emotion, rea- 617–638. son, and the human brain. New York: Putnam. Goldstein, D. G., & Goldstein, D. C. (2006). Profiting Dana, J., Weber, R. A., & Kuang, J. X. (2007). Ex- from the long tail. Harvard Business Review, 84(6), ploiting moral wiggle room: Experiments demon- 24–28. strating an illusory preference for fairness. Economic Guth, W., Schmittberger, R., & Schwarze, B. (1982). An Theory, 33(1), 67–80. experimental analysis of ultimatum bargaining. Jour- Darlington, R. B., & Macker, C. E. (1966). Displace- nal of Economic Behavior and Organization, 3(4), ment of guilt-produced altruistic behavior. Journal of 367–388. Personality and Social Psychology, 4(4), 442–443. Hirshleifer, D., & Shumway, T. (2003). Good day sun- Dawes, R. M., & Mulford, M. (1996). The false consen- shine: Stock returns and the weather. Journal of Fi- sus effect and overconfidence: Flaws in judgment or nance, 58(3), 1009–1032. flaws in how we study judgment? Organizational Hoch, S. J., & Loewenstein, G. (1991). Time- Behavior and Human Decision Processes, 65(3), inconsistent preferences and consumer self-control. 201–211. Journal of Consumer Research, 17, 492–507. Ditto, P. H., Pizarro, D. A., Epstein, E. B., Jacobson, J. Isen, A. M., Clark, M., & Schwartz, M. F. (1976). Dura- A., & MacDonald, T. K. (2006). Visceral influences tion of the effect of good mood on helping: Foot- on risk-taking behavior. Journal of Behavioral Deci- prints on the sands of time. Journal of Personality sion Making, 19(2), 99–113. and Social Psychology, 34(3) 385–393. Dufwenberg, M., & Kirchsteiger, G. (2004). A theory of Isen, A. M., Horn, N., & Rosenhan, D. L. (1973). Ef- sequential reciprocity. Games and Economic Behav- fects of success and failure on children’s generosity. ior, 47(2), 268–298. Journal of Personality and Social Psychology, 27(2), Dunn, B. D., Dalgleish, T., & Lawrence, A. D. (2005). 239–247. The somatic marker hypothesis: A critical evaluation. Isen, A. M., & Levin, P. F. (1972). The effect of feeling Neuroscience and Biobehavioral Reviews, 30(2), 1– good on helping: Cookies and kindness. Journal of 33. Personality and Social Psychology, 15(4), 294–301. Edmans, A., García, D., & Norli, Ø. (2007). Sports sen- Isen, A. M., Nygren, T. E., & Ashby, F. G. (1988). Influ- timent and stock returns. Journal of Finance, 62(4), ence of positive affect on social categorization. Moti- 1967–1998. vation and Emotion, 16(1), 65–78. Eisenberger, N. I., Lieberman, M. D., & Williams, K. D. Isen, A. M., & Simmonds, S. F. (1978). The effect of (2003). Does rejection hurt: An fMRI study of social feeling good on a helping task that is incompatible exclusion. Science, 302, 290–292. with good mood. Social Psychology, 41(4), 346–349. Elster, J., & Loewenstein, G. (1992). Utility from mem- Johnson, E. J., Hershey, J., Meszaros, J., & Kunreuther, 154 I. INTERDISCIPLINARY FOUNDATIONS

H. (1993). Framing, probability distortions, and in- counting. Quarterly Journal of Economics, 112(2), surance decisions. Journal of Risk and Uncertainty, 443–477. 7(1), 35–51. Lerner, J. S., & Keltner, D. (2001). Fear, anger, and risk. Jones, S. K., Frisch, D., Yurak, T. J., & Kim, E. (1998). Journal of Personality and Social Psychology, 81(1), Choices and opportunities: Another effect of framing 146–159. on decisions. Journal of Behavioral Decision Lerner, J. S., Small, D. A., & Loewenstein, G. (2004). Making, 11(3), 211–226. Heart strings and purse strings: Carryover effects of Kahn, B. E., & Isen, A. M. (1993). The influence of pos- emotions on economic decisions. Psychological Sci- itive affect on variety seeking among safe, enjoyable ence, 15(5), 337–341. products. Journal of Consumer Research, 20(2), Loewenstein, G. (1987). Anticipation and the valuation 257–270. of delayed consumption. The Economic Journal, Kahneman, D., Knetsch, J. L., & Thaler, R. H. (1990). 97(387), 666–684. Experimental tests of the endowment effect and the Loewenstein, G. (1996). Out of control: Visceral influ- Coase theorem. Journal of Political Economy, 98(6), ences on behavior. Organizational Behavior and Hu- 1325–1348. man Decision Processes, 65(3), 272–292. Kahneman, D., & Tversky, A. (1979). Prospect theory: Loewenstein, G., & Adler, D. (1995). A bias in the pre- An analysis of decision under risk. Econometrica, diction of tastes. The Economic Journal, 105(431), 47(2), 263–291. 929–937. Keinan, A. (2006). Productivity mindset and the con- Loewenstein, G., & Lerner, J. (2003). The role of affect sumption of collectible experiences. Manuscript in in decision making. In R. J. Dawson, K. R. Scherer, & preparation, Columbia University. H. H. Goldsmith (Eds.), Handbook of affective sci- Kermer, D. A., Driver-Linn, E., Wilson, T. D., & Gilbert, ence (pp. 619–642). Oxford: Oxford University Pres. D. T. (2006). Loss aversion is an affective forecasting Loewenstein, G., Nagin, D., & Paternoster, R. (1997). error. Psychological Science, 17(8), 649–653. The effect of sexual arousal on predictions of sexual Kirby, K. N., & Herrnstein, R. J. (1995). Preference re- forcefulness. Journal of Crime and Delinquency, versals due to myopic discounting of delayed reward. 34(4), 443–473. Psychological Science, 6(2), 83–89. Loewenstein, G., & O’Donoghue, T. (2004). Animal Kivetz, R., & Keinan, A. (2006). Repenting hyperopia: spirits: Affective and deliberative processes in eco- An analysis of self-control regrets. Journal of Con- nomic behavior. Manuscript in preparation, Carnegie sumer Research, 33, 273–282. Mellon University. Kivetz, R., & Simonson, I. (2002). Self-control for the Loewenstein, G., O’Donoghue, T., & Rabin, M. (2003). righteous: Toward a theory of precommitment to in- Projection bias in predicting future utility. Quarterly dulgence. Journal of Consumer Research, 29, 199– Journal of Economics, 118(4), 1209–1248. 217. Loewenstein, G., & Prelec, D. (1992). Anomalies in Knutson, B., Adams, C. M., Fong, G. W., & Hommer, intertemporal choice. Quarterly Journal of Econom- D. (2001). Anticipation of increasing monetary re- ics, 107(2), 573–597. ward selectively recruits nucleus accumbens. Journal of Neuroscience, 21, RC159. Loewenstein, G., & Small, D. A. (2007). The scarecrow Knutson, B., Fong, G. W., Bennett, S. M., Adams, C. M., and the tin man: The vicissitudes of human sympathy & Hommer, D. (2003). A region of mesial prefrontal and caring. Review of General Psychology, 11, 112– cortex tracks monetarily rewarding outcomes: Char- 126. acterization with rapid event-related fMRI. Loewenstein, G., Weber, E. U., Hsee, C. K., & Welch, N. NeuroImage, 18, 263–272. (2001). Risk as feelings. Psychological Bulletin, Knutson, B., Rick, S., Wimmer, G. E., Prelec, D., & 127(2), 267–286. Loewenstein, G. (2007). Neural predictors of pur- Loomes, G., & Sugden, R. (1982). Regret theory: An al- chases. Neuron, 53, 147–156. ternative theory of rational choice under uncertainty. Kogut, T., & Ritov, I. (2005). The “identified victim” ef- The Economic Journal, 92(368), 805–824. fect: An identified group, or just a single individual? Loomes, G., & Sugden, R. (1986). Disappointment and Journal of Behavioral Decision Making, 18(3), 157– dynamic consistency in choice under uncertainty. Re- 167. view of Economic Studies, 53(2), 271–282. Koopmans, T. C. (1960). Stationary ordinal utility and Maia, T. V., & McClelland, J. L. (2004). A reexamina- impatience. Econometrica, 28(2), 287–309. tion of the evidence for the somatic marker hypothe- Köszegi, B., & Rabin, M. (2006). A model of reference- sis: What participants really know in the Iowa gam- dependent preferences. Quarterly Journal of Eco- bling task. Proceedings of the National Academy of nomics, 121(4), 1133–1165. Sciences USA, 101(45), 16075–16080. Kunreuther, H., Ginsberg, R., Miller, L., Sagi, P., Slovic, Markowitz, H. M. (1952). The utility of wealth. Journal P., Borkan, B., et al. (1978). Disaster insurance pro- of Political Economy, 60(2), 151–158. tection: Public policy lessons. New York: Wiley. McClure, S. M., Ericson, K. M., Laibson, D. I., Laibson, D. I. (1997). Golden eggs and hyperbolic dis- Loewenstein, G., & Cohen, J. D. (2007). Time dis- 9. The Role of Emotion in Economic Behavior 155

counting for primary rewards. Journal of Neurosci- Schwarz, N., & Clore, G. (1983). Moods, misattri- ence, 27(21), 5796–5804. bution, and judgments of well-being: Informative McClure, S. M., Laibson, D. I., Loewenstein, G., & Co- and directive functions of affective states. Journal of hen J. D. (2004). Separate neural systems value im- Personality and Social Psychology, 45(3), 513–523. mediate and delayed monetary rewards. Science, Shiv, B., & Fedorikhin, A. (1999). Heart and mind in 306, 503–507. conflict: The interplay of affect and cognition in con- Mellers, B. A., Schwartz, A., Ho, K., & Ritov, I. (1997). sumer decision making. Journal of Consumer Re- Decision affect theory: Emotional reactions to the search, 26, 278–292. outcomes of risky options. Psychological Science, Shiv, B., Loewenstein, G., Bechara, A., Damasio, H., & 8(6), 423–429. Damasio, A. R. (2005). Investment behavior and the Mischel, W. (1974). Processes in delay of gratification. negative side of emotion. Psychological Science, In L. Berkowitz (Ed.), Advances in experimental so- 16(6), 435–439. cial psychology (Vol. 7). New York: Academic Press. Slovic, P., Fischhoff, B., & Lichtenstein, S. (1980). Facts Pillutla, M. M., & Murnighan, J. K. (1996). Unfairness, and fears: Understanding perceived risk. In R. C. anger, and spite: Emotional rejections of ultimatum Schwing & W. A. Albers, Jr. (Eds.), Societal risk as- offers. Organizational Behavior and Human Deci- sessment: How safe is safe enough? San Francisco: sion Processes, 68(3), 208–224. Jossey-Bass. Prelec, D., & Loewenstein, G. (1998). The red and the Small, D. A., & Loewenstein, G. (2003). Helping ‘a’ vic- black: Mental accounting of savings and debt. Mar- tim or helping ‘THE’ victim: Altruism and keting Science, 17, 4–28. identifiability. Journal of Risk and Uncertainty, Rabin, M. (1993). Incorporating fairness into game the- 26(1), 5–16. ory and economics. American Economic Review, 83, Small, D. A., & Loewenstein, G. (2005). The devil you 1281–1302. know: The effects of identifiability on punishment. Rachlin, H., & Raineri, A. (1992). Irrationality, impul- Journal of Behavioral Decision Making, 18(5), 311– siveness, and selfishness as discount reversal effects. 318. In G. Loewenstein & J. Elster (Eds.), Choice over Smith, A. (1981). The theory of moral sentiments (D. D. time. New York: Russell Sage Foundation. Raphael & A. L. Macfie, Eds.). Indianapolis, IN: Lib- Read, D., Loewenstein, G., & Kalyanaraman, S. (1999). erty Fund. (Original work published 1759) Mixing virtue and vice: Combining the immediacy ef- Smith, C. A., & Ellsworth, P. C. (1985). Patterns of cog- fect and the diversification heuristic. Journal of nitive appraisal in emotion. Journal of Personality Behavioral Decision Making, 12(4), 257–273. and Social Psychology, 48(4), 813–838. Rick, S. (2007). The influence of anticipatory affect on Sugden, R. (1986). Regret, recrimination, and rational- consumer choice. Unpublished doctoral dissertation, ity. In L. Daboni, A. Montesano, & M. Lines (Eds.), Department of Social and Decision Sciences, Carne- Theory and decision library series: Vol. 47. Recent gie Mellon University. developments in the foundations of utility and risk Rick, S., Cryder, C., & Loewenstein, G. (2008). Tight- theory. Dordrecht, The Netherlands: Reidel. wads and spendthrifts. Journal of Consumer Re- Thaler, R. H. (1980). Toward a positive theory of con- search, 34. sumer choice. Journal of Economic Behavior and Or- Ross, L., Greene, D., & House, P. (1977). The false con- ganization, 1(1), 39–60. sensus effect: An egocentric bias in social perception Thaler, R. H., & Shefrin, H. M. (1981). An economic and attribution processes. Journal of Experimental theory of self-control. Journal of Political Economy, Social Psychology, 13(2), 279–301. 89(2), 392–410. Rottenstreich, Y., & Hsee, C. K. (2001). Money, kisses, Tiedens, L. Z., & Linton, S. (2001). Judgment under and electric shocks: On the affective psychology of emotional certainty and uncertainty: The effects of risk. Psychological Science, 12(3), 185–190. specific emotions on information processing. Journal Samuelson, P. (1937). A note on measurement of utility. of Personality and Social Psychology, 81(6), 973– Review of Economic Studies, 4(2), 155–161. 988. Sanfey, A. G., Rilling, J. K., Aronson, J. A., Nystrom, L. Tversky, A., & Koehler, D. J. (1994). Support theory: A E., & Cohen, J. D. (2003). The neural basis of eco- nonextensional representation of subjective probabil- nomic decision-making in the ultimatum game. Sci- ity. Psychological Review, 101(4), 547–567. ence, 300, 1755–1758. von Neumann, J., & Morgenstern, O. (1944). Theory of Schelling, T. C. (1968). The life you save may be your games and economic behavior. New York: Wiley. own. In S. B. Chase (Ed.), Problems in public expen- Wicker, B., Keysers, C., Plailly, J., Royet, J-P., Gallese, diture analysis. Washington, DC: Brookings Insti- V., & Rizzolatti, G. (2003). Both of us disgusted in tute. my insula: The common neural basis of seeing and Schkade, D., & Kahneman, D. (1998). Does living in feeling disgust. Neuron, 40, 655–664. California make people happy?: A focusing illusion Zeelenberg, M., & Beattie, J. (1997). Consequences of in judgments of life satisfaction. Psychological Sci- regret aversion: 2. Additional evidence for effects ence, 9(6), 340–346. of feedback on decision making. Organizational 156 I. INTERDISCIPLINARY FOUNDATIONS

Behavior and Human Decision Processes, 72(1), 63–78. Zeelenberg, M., Beattie, J. van der Plight, J., & De Vries, N. K. (1996). Consequences of regret aversion: Ef- fects of expected feedback on risky decision making. Organizational Behavior and Human Decision Pro- cesses, 65(2), 148–158.

Guilford Publications Copyright © 2008 The Guilford Press. All rights reserved under International Copyright 72 Spring Street Convention. No part of this text may be reproduced, transmitted, downloaded, or stored in New York, NY 10012 or introduced into any information storage or retrieval system, in any form or by any 212-431-9800 means, whether electronic or mechanical, now known or hereinafter invented, without the 800-365-7006 written permission of The Guilford Press. www.guilford.com