LOSS OF CONTROL BY A LESS VALID OR BY A LESS SALIENT STIMULUS COMPOUNDED WITH A BETTER PREDICTOR OF REINFORCEMENT

An unsalient stimulus, or one imperfectly correlated with reinforcement, may acquire significant control over responding, provided that it is the only available signal for reinforcement, but may fail to acquire control if it is reinforced only in conjunction with a second, more salient or more validstimulus. Astimulus imperfectly correlated with reinforcement may also losecontrol over responding if having initially been reinforced in isolation, it is subsequently reinforced only in conjunction with another, more validstimulus. If the effects of relative salience are to be explained in exactly the same way as those of relative validity, we should expect a similar loss of control by an unsalient stimulus, A, if, after initial consistently reinforced trials to A alone, subjects subsequently receive reinforcement only in the presence of a compound stimulus, A + B. Two experiments on discrete-trial discrimination learning in pigeons and one on conditioned suppression in rats confirm this expectation. The results have implications for theories of selective association in conditioning and discrimination learning.