hide
Free keywords:
-
Abstract:
Objective: Preferences and beliefs are key ingredients of any choice theory. A core idea of economics is that choice behavior reveals which (if any) preferences and beliefs can rationalize observed behavior. This paper seeks to extend this revealed preference methodology beyond behavior: We expand the theory’s domain to include a measurable correlate of neural activity that is well understood to depend on – and hence revealing of – reward expectations. Specifically, we leverage the fact that the BOLD signal in the ventral striatum has been shown to encode reward prediction errors (RPE; Schultz, Dayan, and Montague 1997; Niv et al, 2012), which reflect the difference between the obtained reward and the reward expected under a prior belief. Methods: We introduce a theory of RPE-revealed preference which makes precise what RPE measurements imply about preferences and/or beliefs. Observing a large RPE, for instance, reveals that the obtained reward was highly valued and/or unexpected. In an fMRI experiment designed to test this theory, participants faced binary lotteries with dynamically resolving ambiguity: A “roulette” wheel indicated the probability of winning a prize ($50 or $25) or receiving nothing. The wheel’s composition was revealed only gradually to participants, so that they could sequentially update their belief of winning the prize. At a random time during this gradual resolution, we elicited participants’ lottery valuation as a probability equivalent: the probability of winning $50 that would make them indifferent to the lottery under consideration. Results: Behavioral pilot data (N=35) show that, in line with theoretical predictions, participants’ valuations closely track the lottery’s winning probability (ß=1.08; 95% CI=[1.03,1.12]) when comparing two lotteries with identical prizes (which amounts to pure belief elicitation). 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 fMRI portion of the study is currently underway, and will allow us to test whether the preferences and beliefs inferred from measurements of ventral striatal activity are consistent with those revealed by behavior. Conclusion: If successful, our approach promises a rigorous way to test neuroeconomic theories that make joint predictions about behavior and neural data, as well as empirical constraints on models of hitherto unobservable evolving beliefs.
