Five experiments tested implications of the idea that instrumental
(operant) extinction involves learning to inhibit the learned response. All
experiments used a discriminated operant procedure in which rats were reinforced
for lever pressing or chain pulling in the presence of a discriminative stimulus
(S), but not in its absence. In Experiment 1, extinction of the response (R) in
the presence of S weakened responding in S, but equivalent nonreinforced
exposure to S (without the opportunity to make R) did not. Experiment 2
replicated that result and found that extinction of R had no effect on a
different R that had also been reinforced in the stimulus. In Experiments 3 and
4, rats first learned to perform several different stimulus and response
combinations (S1R1, S2R1, S3R2, and S4R2). Extinction of a response in one
stimulus (i.e., S1R1) transferred and weakened the same response, but not a
different response, when it was tested in another stimulus (i.e., S2R1 but not
S3R2). In Experiment 5, extinction still transferred between S1 and S2 when the
stimuli set the occasion for R's association with different types of food
pellets. The results confirm the importance of response inhibition in
instrumental extinction: Nonreinforcement of the response in S causes the most
effective suppression of responding, and response suppression is specific to the
response but transfers and influences performance of the same response when it
is occasioned by other stimuli. Theoretical and practical implications are
discussed.