Psychological Science http://pss.sagepub.com/ Symmetry in Cold-to-Hot and Hot-to-Cold Valuation Gaps Geoffrey Fisher and Antonio Rangel Psychological Science 2014 25: 120 originally published online 12 November 2013 DOI: 10.1177/0956797613502362 The online version of this article can be found at: http://pss.sagepub.com/content/25/1/120 Published by: http://www.sagepublications.com On behalf of: Association for Psychological Science Additional services and information for Psychological Science can be found at: Email Alerts: http://pss.sagepub.com/cgi/alerts Subscriptions: http://pss.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav >> Version of Record - Jan 10, 2014 OnlineFirst Version of Record - Nov 12, 2013 What is This? Downloaded from pss.sagepub.com by Cendri Hutcherson on January 10, 2014 PSSXXX10.1177/0956797613502362Fisher, RangelSymmetry in Empathy Gaps 502362research-article2013 Research Article Psychological Science 2014, Vol. 25(1) 120 –127 Symmetry in Cold-to-Hot and Hot-to-Cold © The Author(s) 2013 Reprints and permissions: sagepub.com/journalsPermissions.nav Valuation Gaps DOI: 10.1177/0956797613502362 pss.sagepub.com Geoffrey Fisher1 and Antonio Rangel1,2 1Division of Humanities and Social Sciences and 2Computation and Neural Systems, California Institute of Technology Abstract Individuals commonly mispredict their future preferences when they make decisions in a visceral state different from their anticipated state at consumption. In the research reported here, we asked subjects to bid on different foods while exogenously varying their hunger levels at the time of decision and at the time of consumption. This procedure allowed us to test whether cold-to-hot and hot-to-cold gaps are symmetric in size and driven by similar mechanisms. We found that the effect size was symmetric: Hungry subjects overbid 20¢ for a snack they would eat later when they were satiated, and satiated subjects underbid 19¢ for a snack they would eat later when they were hungry. Furthermore, we found evidence that these gaps were driven by symmetric mechanisms that operate on the evaluation of visceral features of food, such as taste, as opposed to more cognitive features, such as healthiness. Keywords valuation, misprediction, mistakes, empathy gaps, projection bias, symmetry, decision making, preferences Received 5/14/13; Revision accepted 7/10/13 It is well known that the utility, or value, derived from in a hot state (e.g., eating a hamburger when they are consumption is modulated by emotional and physiological hungry) while they are in a cold state at the time of deci- states at the time of consumption. For example, the plea- sion (e.g., after just having eaten). Hot-to-cold gaps refer sure of drinking water is larger when one is thirsty than to the opposite situation (e.g., forecasting the value of when one is quenched. A basic question is whether indi- eating dessert at the end of the meal in a satiated state viduals anticipate the effect of these “visceral” states on while being hungry at the time of decision). An impor- their utility when making decisions about future consump- tant open question is whether both types of gaps are tion. For instance, can a hungry grocery shopper buy the symmetric in the following two ways. First, do individuals correct amount of food to consume throughout the week? underestimate their change in preferences to the same A sizable body of evidence has shown that individuals in degree when going from cold-to-hot states as when a “cold” state (e.g., satiated) systematically underestimate going from hot-to-cold states? Second, are symmetric the increase in consumption value that they would experi- mechanisms at work in generating both types of empathy ence in a “hot” state (e.g., hungry; Badger et al., 2007; gaps? Gilbert, Gill, & Wilson, 2002; Loewenstein, Nagin, & The answer to these questions matters for several rea- Paternoster, 1997; Nisbett & Kanouse, 1968; Sayette, sons. First, they inform beliefs about the likelihood that Loewenstein, Griffin, & Black, 2008; Van Boven & individuals make mistakes of similar magnitude in both Loewenstein, 2003). This phenomena is known as a cold- types of situations, as well as the extent to which both to-hot empathy gap in psychology (Loewenstein, 1996), types of mistakes can be addressed with similar policy and as a projection bias in behavioral economics (Loewenstein, O’Donoghue, & Rabin, 2003). Corresponding Author: Empathy gaps can arise in two different scenarios. Antonio Rangel, California Institute of Technology, Division of Cold-to-hot gaps refer to situations in which people fore- Humanities and Social Sciences, MC 228-77, Pasadena, CA 91125 cast the value of an event that will occur when they are E-mail: [email protected] Downloaded from pss.sagepub.com by Cendri Hutcherson on January 10, 2014 Symmetry in Empathy Gaps 121 instruments. Second, theories in behavioral economics 50 snacks at the end of that session. Ratings were made and psychology have posited that cold-to-hot and hot-to- on an integer scale from −2 to 2 in response to the ques- cold gaps are symmetric and driven by similar mecha- tion, “How much would you enjoy that particular food at nisms, but this has not been previously tested. the end of TODAY’s experiment?” The purpose of these Previous work has provided strong evidence for the ratings was to familiarize the subjects with the entire set existence of empathy gaps, but it has not provided a defi- of foods prior to the main bidding task. nite answer to either of the two symmetry questions. In Second, on each day, subjects entered bids for the fact, the vast majority of experimental studies have right to eat each of the foods at the end of the second day focused on the cold-to-hot case (Badger et al., 2007; of data collection. They were explicitly told whether they Gilbert et al., 2002; Loewenstein et al., 1997; Nisbett & would be hungry or satiated at that time. Bids were made Kanouse, 1968; Sayette et al., 2008; Van Boven & in integers from $0 to $4 by pressing a button. At the Loewenstein, 2003). One important exception is Read beginning of the first day, they were informed that at the and van Leeuwen (1998), who compared hungry-satiated end of the second session they would need to remain in and satiated-hungry empathy gaps in real food choice, as the lab for 20 min, and the only thing that they would be we did in the present experiment. However, unlike our able to eat was whatever they purchased from us through methodology, theirs did not permit a direct comparison their bids. At the end of Day 2, one of the foods on which of the extent to which changes in utility were underesti- the subject had bid (from either date) was selected, and mated in both cases. With respect to the second question, their bid was implemented using the rules of a Becker- as far as we know, no previous experiments have inves- DeGroot-Marschak (BDM) auction (Becker, DeGroot, & tigated the mechanisms at work in projection bias, nor Marschak, 1964).1 Subjects were given $4 in bidding cash the extent to which they are symmetric. and were allowed to keep whatever they did not spend. The bids provided a measure of the perceived value at Method the time of decision of eating the food at the end of Session 2. Subjects Third, for each of the foods, separate taste and healthi- ness ratings were made on integer scales from −2 to 2. One hundred one students from the California Institute Subjects provided taste ratings in response to the ques- of Technology took part in two behavioral sessions. To tion “how tasty [do] you believe that food to be, indepen- encourage subjects to return for the second session, we dent of any health considerations.” Healthiness ratings paid them $10 after the first session and $40 after the were made in response to the question “how healthy [do] second session. The experiment was approved by the you believe that food to be, independent of any taste California Institute of Technology Institutional Review considerations.” The ratings were collected in blocks, Board. with the order randomized across subjects. These ratings provided a measure of the perceived attributes of each Stimuli food at the time of decision. Subjects completed the four tasks separately, but the Each subject saw two different food sets, each containing same procedure was followed for each one. Each trial 50 different snacks. These snacks consisted of a variety of began with a fixation cross (500 ms), after which one of candies, fruits, chips, and energy bars. Subjects saw one the foods was presented inside a white rectangle. The set of foods in the first session and a different set in the white rectangle disappeared after 500 ms, and then sub- second session. The order of the sets was determined ran- jects were free to take as long as they liked to give their domly for each subject. We used two different sets to avoid rating or place their bid. Response feedback (either the consistency biases. During pretesting, all of the foods were amount of their bid or their rating) appeared on screen rated on average as being neutral or appetitive. for 1 s prior to the start of the next trial. As shown in Figure 1c, the experiment had a 2 × 2 Tasks factorial design, with conditions varying across subjects. For each date, we exogenously manipulated subjects’ The experiment consisted of two sessions that occurred at hunger by asking them either to fast for 4 hr prior to the the same time of day but were separated by 3 to 5 days.