Learning gaps: Decision under Described and Experienced signals

In classical studies of Bayesian updating (Grether, 1992; Holt and Smith, 2009) this has been mostly studied in situations of risk, individuals are given the signal distribution and its relationship with events that determine outcomes. Although useful this characterization assumes that the true relationship between signal and state is always available to individuals. For example, there might be diseases that share symptoms, but doctors learn from their own experience which might be more common in the population they are treating. 

In this project I investigate how information processing differs in described signals vs experienced signals. I do this by adapting a typical Decision from experience-decision from description methodology to noisy signals and using the hedging method of Baillon et al. (2021) to derive uncertainty attitudes and beliefs. I compare the individuals updating and uncertainty attitudes in the experience information case vs the described information case, and what is the effect on the attitudes of different correlations between signal and state space change.  Preliminary results from a pilot study suggest than in the Experience condition individuals tend to under react to unfavorable signals more, while there is suggestive evidence of a higher insensitivity to probabilities in the experience condition, a common finding in the literature Experience-Description gap  (Cubitt et. Al, 2022; Abdellaoui et al., 2011)