ABSTRACT
This chapter reviews the model of implicit memory effects in perceptual identification tasks called retrieving effectively from memory implicit (REMI). It uses Simon Dennis's critique of optimization modeling to explain why a Bayesian analysis provided guidance for the development of the REMI model of forced-choice tasks. In the REM model, memory traces are represented as vectors that include both content and context features. Simon argued that the conclusions that economists draw from their optimizing models seldom depend critically upon the optimizing assumptions, but they do depend critically on the auxiliary assumptions of departures from rationalitySimon's critique explains why the Bayesian approach worked for the forced-choice model: the goal was clear, the alternatives were fixed, and the empirical assumptions about similarity undoubtedly held. The assumption of including context in the memory probe is an essential component of the model; whether the assumption is construed as a departure from rationality depends on the framing.
