Some effects of discriminable goal-box conditions on the learning of a successive discrimination

In an attempt to confirm and extend a previous reslt, rats were trained on two tasks where a signal delivered at the start of each trial indicated which of two paths through a maze would be rewarded. In Experiment I both paths led to the same goal-box, and it was found that performance was better when the state of the goal-box was different on trials with each of the two signals. In Experiment II the two paths led to spatially separated goal-boxes. It was found that when the states of the two goal-boxes were discriminably different but the state of each of them remained the same from trial to trial, performance was better than when their states varied irregularly. It is suggested that these results have interesting implications for theories of behaviour. Now at Department of Psychology, University of Hull.