Proceedings of the Annual Meeting of the Cognitive Science Society

In two experiments we manipulated the prior probabilityof occurrence for two alternatives. After a first learningsession, in a second session the cue to bias the decisionwas reversed. Our investigation shows that subjects areable to learn the reverse bias only when the bias of the firstsession is in line with their expected outcome. When, dur-ing the first session, the actual outcome of the bias is notin line with the expected outcome, there is an inhibitionfor the reversal bias learning in the second session. Weinvestigate this phenomenon with computational modelsof choice showing that the inhibition of reversal is due toan increase in the rate at which subjects accumulate evi-dence for repeated, unexpected stimuli. We discuss a pos-sible theoretical explanation that links this phenomenonto similar results found in the literature on reversal learn-ing and to the effect of novelty on learning.