A Note on Case-Based Optimization with a Non-Degenerate Similarity Function

The paper applies the ��realistic-ambitious�� rule for adaptation of the aspiration level suggested by Gilboa and Schmeidler (1996) to a situation in which the similarity between the available acts is represented by a non-degenerate function. The paper shows that the optimality result obtained by Gilboa and Schmeidler (1996) in general fails. With a concave similarity function, the best corner act is chosen in the limit. Introducing convex regions into the similarity function improves the limit choice. A sufficiently fine similarity function allows to approximate optimal behavior with an arbitrary degree of precision.