SIGNALING THE OMISSION OF A RESPONSE-CONTINGENT OUTCOME REDUCES DISCRIMINATIVE CONTROL

The effect of training a positive discriminative stimulus (S+) as a signal for the nonreinforcement of an instrumental response (S-) on the ability of that stimulus to evoke its original instrumental response was examined in three experiments using rats. In all three experiments, two different stimuli were established as S+s for different response-outcome relations. In Experiment 1, an S+ was less effective in controlling its original response after it had undergone training as an S- for a new response that earned the same outcome than it was after training as an S- for a response that earned a different outcome. Experiment 2 established that this effect was not mediated by Pavlovian inhibitory conditioning produced by the negative correlation between the S+ and the outcome during S- training. Simply arranging a negative correlation between S+ and the outcome whose occurrence it had previously signaled did not impair the ability of that S+ to elicit its original response. In Experiment 3, the response-evoking properties of an S+ were found to be undermined by using the S+ as a signal for the simple extinction of a new response trained with the same outcome, but not with a different outcome. These results suggest that positive discriminative stimuli use their associations with the outcomes earned in their presence to control the responses that earned those outcomes.