Changing Myself, Changing My Fate: How anticipating negative outcomes prompts self-relevant change

People often spontaneously apply “recency beliefs” to form predictions of future outcomes based on prior outcomes. Different recency beliefs in the same context (e.g. that future outcomes are either positively or negatively correlated with recent outcomes) can lead to starkly different anticipations for the same future outcome. We propose that these anticipations of future outcomes can have a dramatic impact on people’s preference for self-relevant change. We show that in both hypothetical and real settings, when recency beliefs and prior outcomes lead people to pessimism about future outcomes, they make self -relevant changes, as if they were “thwarting fate” by shifting the salient aspect of their identity. However, when circumstances instead lead to optimism about future outcomes, they make self-consistent choices, as if “embracing fate,” by emphasizing the currently salient identity. (130 words) In many aspects of life, the future is uncertain. People often spontaneously form predictions for future outcomes by applying some naive “recency” heuristic, assuming that future outcomes will tend to either positively or negatively relate to recent outcomes. These beliefs, in turn, can influence the appeal of choices that represent change. Choosing change can be beneficial, particularly when a person is pessimistic about controllable future outcomes (e.g., a baseball player practicing harder for a game). However, such choices are often irrelevant to the outcome (e.g., a player wearing his cap inside-out and backwards when falling behind in a game, Gmelch 2003). We propose that people will prefer changes involving self-representation when they are pessimistic about future outcomes, even in the absence of a causal link, and will avoid such changes, preferring consistency in self-representation, when they are optimistic. First, we validate a framework in which the attribution of outcomes to either skill or chance impacts “recency” beliefs about how anticipated future outcomes can be predicted by prior outcomes. We then show that the anticipation of future outcomes has an important and novel impact on unrelated but self-relevant choices, as if the choice could serve as a “switch” between fates. When the combination of prior outcomes and recency beliefs yield pessimism about future outcomes, people prefer options representing a shift in the salient aspect of self-identity, as if to “thwart fate”. However, when the combination of prior outcomes and recency beliefs instead yield optimism for future outcomes, people prefer self-consistency, as if to “embrace fate”. Recency beliefs and predictions of the future People often use prior outcomes to inform their predictions for the future, based on a recency belief. The positive recency belief (hot-hand, Gilovich, 1985; hot-outcome, Edwards, 1961) implies that past outcomes are more likely to repeat in the future. The negative recency belief (gambler’s fallacy, Tune, 1964; stock-of-luck, Leopard, 1978) implies that future outcomes are more likely to differ from past outcomes. Many factors have been shown to contribute to differences in people’s recency beliefs, including perceived intentionality (Caruso, Waytz & Epley 2010), randomness and causation (see Oskarsson, Van Boven, McClelland & Hastie, 2009 for a review). These factors can generally be understood as representing a broader distinction between attributing outcomes primarily to skill or chance. In many common activities, this distinction is important but difficult to judge, even requiring empirical study (e.g. playing poker, Croson, Fisherman & Pope 2008, Levitt & Miles 2011; financial fund management, Fama & French 2010). Different people may, in fact, reach different conclusions about whether skill or chance determines a specific future outcome. As we will confirm in a pilot study, differences in people’s attributions of a given outcome to skill or chance will affect that person’s recency beliefs. When skill is emphasized, people may infer that, since recent outcomes can be informative of the current level of skill, future outcomes will reflect recent outcomes, yielding positive recency. However, when chance is emphasized instead, people may anticipate that outcomes will balance out, even in a short sequence (per the “law of small numbers”, Tversky & Kahneman 1971), and thus predict future outcomes that differ from recent outcomes, yielding negative recency. Although this framework is intuitive and is suggested by prior findings, a thorough literature review failed to reveal any research relating beliefs in skill vs. chance in a single context to the adoption of different recency beliefs. Therefore, in the pilot study we will manipulate recency beliefs by merely framing the mechanism of a specific task in a fixed context as either skill or chance. Confirming the effect of this manipulation on recency beliefs will facilitate testing our main hypothesis, that the anticipation of future outcomes impacts preference for options representing self-relevant change. “Managing” the future – from predictions to self-relevant change Given that the self is central in causal inferences (Langer & Roth 1975; Ross & Sicoly, 1979), the change (vs. consistency) of the actor’s identity may be highly relevant to recency inferences. While the past and future outcomes of a given person will usually be seen as constituting the same sequence, we propose that a change in the person can disrupt the coherence of the perceived sequence, impeding recency inferences. As a result, when one player literally replaces another in a game, we predict that recency beliefs will be suppressed, which we test in the pilot study. Further, we propose a parallel between inter-personal and intra-personal change, such that selfrelevant changes which signify a shift in personal identity may similarly suppress recency inferences. As a result, when people are pessimistic about future outcomes due to the combination of prior outcomes and their recency beliefs, self-relevant change will seem more appealing. This proposition is consistent with current views of personal identity which describe a multi-faceted and malleable conception of the self (Markus & Wurf 1987), such that people shift their identity in response to external cues (Brewer & Gardner 1996; Ramírez-Esparzaa et al. 2006) and anticipated changes in identity can, in turn, affect choices for the future (Bartels & Urminsky 2011). Our proposed hypothesis implies that people will select change in the self when they are pessimistic about a future outcome, but avoid change in the self when optimistic. To test this, we manipulate recency beliefs to induce pessimism or optimism after a given outcome and compare people’s choices for self-relevant options that represent change or consistency. Choices constitute a common means of self-representation (Ariely & Levav 2000; Kim & Drolet 2003), with different choices reflecting different aspects of one’s identity (Stephens, Markus & Townsend, 2007). We test our hypothesis with consumption choices in Study 1, and with choices between consistency and change in self-description in Study 2. We find that options representing a change in self-representations are chosen differently in pessimistic vs. optimistic situations. When a pessimistic future outcome is anticipated, people prefer options reflecting self-relevant change, as if to “thwart fate”, but when an optimistic future outcome is anticipated, people prefer options reflecting self-consistency, as if to “embrace fate”. Thus, our proposed framework (see Figure 1) incorporates insights from two literatures that have previously been treated as largely unrelated: personal identity and belief formation in sequences of outcomes, enabling us to better understanding how people predict and “manage” future outcomes. Pilot Study: Recency effects disrupted by identity-change In this pilot study, we employed a hypothetical ball-throwing game, manipulating the mechanism framing (skill vs. chance), recent outcomes (success vs. failure), and identity consistency (same vs. different) in a 2x2x2 between-subjects design. The purpose is to confirm that, per our framework, the combination of mechanism beliefs and recent outcomes determines people’s anticipations of uninterrupted future outcomes but, a shift in the player’s identity suppresses these recency effects. Method Students (N=170) from a large public university in China filled out a brief questionnaire in exchange for a candy snack. Participants were asked to imagine visiting an amusement park with a friend and playing a game in which they could win a prize. The goal of the game was to throw a ping-pong ball into a basket that was moving back and forth at a random speed (stimuli for all studies are in the Appendix). All participants were told that the probability of scoring a basket each time was roughly 50%, so that the perceived game difficulty would be held constant. We framed the game as either determined by chance or skill, based on the name of the game (“Lucky Shooting” vs. “Master Shooting”) and the accompanying promotional slogan. Participants were told to imagine that they and the friend decided to play six rounds together, that either they had initially started to play or the friend had started, and that the first player got either two hits or two misses. In the same-player conditions (N=87), participants were asked to predict the outcome of the next shot (hit or miss), performed by the same player. In the different-player conditions (N=83), they were told that they had switched places with their friend and were asked to predict the next shot performed by the other person (i.e., the participant taking over for the friend or the reverse). Since, we found no effect of who the initial player was (self vs. other), we collapsed across the initial player’s identity.