Learning with misattribution of reference dependence (Q2155240)

This paper discusses errors in learning that arise when an agent neglects how the sensation of elation or disappointment relative to expectations contributes to the overall utility, and wrongly attributes this component of the utility to the intrinsic value of an outcome. In a sequential-learning environment, the effect of this form of misattribution is modeled. A recency bias is derived: the misattributor's beliefs over-weight recent experiences and under-weight earlier ones. It is also derived, in the long-run, convergence to steady-state beliefs with, for a loss-averse misattributor, a pessimistic bias and a tendency to undervalue prospects in proportion to their variability.