Proceedings of the Annual Meeting of the Cognitive Science Society

We propose an interaction in expectation formation between returns of an investment and the favorability of new information. Such an interaction can have consequences in trading behavior like leading to the Disposition Effect or differences in the profitability of selling and buying decisions. We introduce a context sensitive Reinforcement Learning model to capture this effect and validate it in a pre-registered investment experiment. Using a Bayesian Hierarchical model fitting approach we find the interaction to stem mainly from participants incorporating unfavorable information more strongly when in a gain position and less so when in a loss. By providing increasing levels of additional information about the price movements we are able to mitigate these effects in a second phase of the experiment. Speaking for a strong effect of belief formation, very clear information is needed to mitigate the observed averse belief and investment patterns.