Evidence From Children and Monkeys

In a study exploring the origins of cognitive dissonance, preschoolers and capuchins were given a choice between two equally preferred alternatives (two different stickers and two differently colored M&M’s, respectively). On the basis of previous research with adults, this choice was thought to cause dissonance because it conflicted with subjects’ belief that the two options were equally valuable. We therefore expected subjects to change their attitude toward the unchosen alternative, deeming it less valuable. We then presented subjects with a choice between the unchosen option and an option that was originally as attractive as both options in the first choice. Both groups preferred the novel over the unchosen option in this experimental condition, but not in a control condition in which they did not take part in the first decision. These results provide the first evidence of decision rationalization in children and nonhuman primates. They suggest that the mechanisms underlying cognitive-dissonance reduction in human adults may have originated both developmentally and evolutionarily earlier than previously thought. Cognitive dissonance is one of the most heavily studied phenomena in the history of psychology. The term cognitive dissonance describes a psychological state in which an individual’s cognitions—beliefs, attitudes, and behaviors—are at odds (Festinger, 1957). People experience cognitive dissonance as aversive (Elliot & Devine, 1994), and are motivated to resolve the inconsistency between their discrepant cognitions. Psychologists have long been interested in the nature of cognitive dissonance, as this phenomenon has implications for many areas of psychology, including attitudes and prejudice (e.g., Leippe & Eisenstadt, 1994), moral cognition (e.g., Tsang, 2002), decision making (e.g., Akerlof & Dickens, 1982), happiness (e.g., Lyubomirsky & Ross, 1999), and therapy (Axsom, 1989). Unfortunately, despite long-standing interest in cognitive dissonance, there is still little understanding of its origins—both developmentally over the life course and evolutionarily as the product of human phylogenetic history. Does cognitive-dissonance reduction begin to take hold only after much experience with the aversive consequences of dissonant cognitions, or does it begin earlier in development? Similarly, are humans unique in their drive to avoid dissonant cognitions, or is this process older evolutionarily, perhaps shared with nonhuman primate species? To date, little research has investigated whether children or nonhuman primates experience and strive to reduce dissonance. In one welcome exception in the developmental literature, Aronson and Carlsmith (1963) found that 4-year-old children who obeyed an experimenter’s mild warning not to play with an attractive toy later liked the toy less than did children who had obeyed an experimenter’s severe warning not to play with the toy. Aronson and Carlsmith interpreted this result in terms of cognitive dissonance: Because children seek to make their attitudes consistent with their behaviors, when they followed the warning and avoided the toy, their liking for the toy decreased. This effect, however, relied on an induced behavior—obedience to an adult’s admonition—rather than on more self-driven decisions on the part of the children. We believe that a demonstration that children shift their attitudes because of counterattitudinal selfdriven behavior would provide clearer evidence that they are motivated to resolve cognitive dissonance in their everyday lives. There is also relatively limited work on cognitive-dissonance reduction in other species, despite the fact that Festinger himself wondered about the extent to which animals experience dissonance. Indeed, Lawrence and Festinger (1962) postulated that cognitive dissonance could explain patterns of extinction Address correspondence to Louisa Egan, Department of Psychology, Yale University, 2 Hillhouse Ave., New Haven, CT 06520, e-mail: [email protected]. PSYCHOLOGICAL SCIENCE 978 Volume 18—Number 11 Copyright r 2007 Association for Psychological Science across different reward conditions. In line with this suggestion, work on cognitive dissonance in nonhuman animals has exclusively employed variants of the effort-justification paradigm (see Aronson & Mills, 1959, for a version of this paradigm with human subjects). This research has led to mixed results. Lewis (1964), for example, demonstrated that rats who pulled a weight harder to obtain a food pellet ran faster to retrieve the pellet after the work was completed. Although Lewis explained these results in terms of cognitive dissonance, other researchers have argued that they could have been due to a simple transfer-of-effort effect: A rat who has just pulled a heavy weight may run faster than a rat who has pulled a light weight because it is physiologically aroused, not necessarily because it experiences greater anticipation of the reward (see Armus, 2001). In a study supporting this view, Armus (2001) observed no differential preferences for food pellets when one food was given in response to much work and another was given in response to limited work. This negative result suggests that rats may not strive to reduce cognitive dissonance, at least in the context of effort justification. Other nonhuman species—particularly birds—have demonstrated effects similar to dissonance reduction in the context of effort-justification paradigms (Stumus vulgaris: Kacelnik & Marsh, 2002; Columba livia: Clement, Feltus, Kaiser, & Zentall, 2000; DiGian, Friedrich, & Zentall, 2004; Friedrich, Clement, & Zentall, 2004). Friedrich and Zentall (2004), for example, demonstrated that pigeons prefer to eat from a feeder that is associated with greater rather than lesser effort. The authors explained these results in terms of relative contrast effects: Pigeons who receive a piece of food after pecking many times experience a larger shift in relative hedonic status than those who simply receive a piece of food after pecking once (Friedrich & Zentall, 2004). Thus, the results of effort-justification studies of animals may be attributed to changes in the relative hedonic value of the reward, rather than changes to the animals’ attitudes per se. In the study reported here, we used a combined comparativedevelopmental approach to investigate both the developmental and the evolutionary origins of cognitive-dissonance reduction (see Hauser & Spelke, 2004). More specifically, we tested two populations—human children and nonhuman primates—on similar tasks to address the questions of how adult mechanisms for cognitive-dissonance reduction originate and when these mechanisms originated phylogenetically. This type of combined comparative-developmental approach has been used to investigate questions of origins in numerous domains of psychological inquiry, such as theory of mind (Tomasello, Call, & Hare, 2003; Tomasello, Carpenter, Call, Behne, & Moll, 2005), numerical cognition (Feigenson, Dehaene, & Spelke, 2004), and core physics knowledge (Hauser & Spelke, 2004; Spelke, 2000). In the present study, our goal was to examine whether children and nonhuman primates, like human adults, would shift their attitudes to fall in line with their decisions. We hoped to develop amethod that not only could be used with both children and monkeys, but also would provide an especially simple and direct test of cognitive-dissonance reduction—a test in which changes in behavior could clearly be attributed to attitude change per se, rather than alternative phenomena. To do this, we modified the free-choice paradigm pioneered by Brehm (1956). In the traditional free-choice paradigm, individuals rate the attractiveness of a variety of items. They are then given a choice between two items that they have rated as equally attractive. This choice is thought to induce dissonance because a decision to avoid the unchosen alternative conflicts with the many positive, preferred aspects of that alternative. After making the choice, subjects are asked to rerate all items. Typically, subjects will rerate items that they have chosen as more attractive, and items that they did not choose as less attractive, apparently changing their attitudes to fit with their choices. This rating pattern suggests that subjects change their present attitudes to be in line with their past decisions. Adapting this free-choice methodology for use with nonverbal populations, we first assessed individuals’ preferences for similar objects and determined three (A, B, and C) that were equally attractive. Next, subjects received a choice between A and B (Phase 1) and then a second choice between whatever they did not select (either A or B) and C (Phase 2). We predicted that if subjects experienced dissonance in choosing one equally preferred item over the other, then they would change their attitude toward the unchosen item, liking it less because of their decision. Therefore, in Phase 2, when they had a choice between it and another (originally equally preferred) option, they would choose the unchosen item less. Subjects also participated in a control condition in which we removed the intentional-choice phase: Rather than choose intentionally between A and B in Phase 1, subjects simply received one of the two alternatives from the experimenter. In this condition, subjects were not expected to experience dissonance—as they themselves never made a choice between the two items—and therefore were not expected to show a preference in Phase 2.