要旨
It is widely established that a reward prediction error signal can be observed in the human brain. When subjects believe there is a 50% chance of earning a reward and then receive it, areas like the ventral striatum and the ventromedial prefrontal cortex show BOLD activations correlated with a positive prediction error. This finding is based on the belief of a 50% reward chance, a reward expectation. This study aims to determine where and how these reward expectations are updated. Specifically, it asks whether BOLD signals in the ventral striatum and ventromedial PFC, known to encode reward prediction errors (Schultz, Dayan, and Montague, 1997; Caplin et al., 2010; Rutledge et al., 2010; Niv et al., 2012), also reflect updates to a subject’s reward expectations, even without actual rewards. In an fMRI experiment, participants faced binary lotteries represented as partially occluded pie charts. While in the scanner we sequentially revealed more and more of the pie chart, dynamically resolving ambiguity about the probability of winning $20 or $100 as a trial progressed. The pie chart’s composition was revealed only gradually to participants, so that they could sequentially update their belief about the probability of winning the prize. At a random time during this gradual resolution, as early as after the first view of the occluded pie chart or as late as after the full pie chart had been revealed, we elicited participants’ valuation of the current pie chart-lottery using an incentive- compatible procedure. Behavioral data show that, in line with theoretical predictions, participants’ valuations closely track the lottery’s winning probability when comparing two lotteries with identical prizes (which amounts to pure belief elicitation). Our procedure results in a gradually updated reward expectation that is under clean experimental control. In the presence of ambiguity, participants’ revealed beliefs are slightly more pessimistic, indicating ambiguity aversion. When the lotteries’ prizes are not identical, participants’ elicited probability equivalents reveal their degree of risk aversion. The neuroimaging portion tests whether updates to reward expectations produced BOLD activations in reward prediction areas. This study seeks to see if reward prediction-error encoding areas represent updates to reward expectations, mirroring punishment expectation errors seen in similar locations (Seymour et al., 2003). This approach tests neuroeconomic theories making joint predictions about beliefs, behavior, and neural data, offering new empirical constraints on models of belief evolution.
