Abstract
The aim of this study was to determine whether reinforcer-specific conditioned responding would occur in a situation in which responding was not thought to be mediated by a representation that encodes information about the specific properties of the reinforcer. The force of the pigeon's keypeck was monitored during first and second-order conditioning with either food or water as the unconditioned stimulus (US). Each pigeon was trained with four different stimuli: a first-order cue predicting that responding would be reinforced with grain (S1f), a first-order cue predicting water as the reinforcer (S1w), a second-order stimulus predicting the S1f (S2f), and a second-order cue predicting the S1w (S2w). Following conditioning, the pigeons were selectively satiated with one of the two reinforcers and presented with the first and second-order cues in an extinction test. At the end of training, the pigeon's keypecks were less forceful to the S1w than to the S1f. There was not, however, a reliable difference between the force of the pecks to the S2f and the S2w. These force differences are consistent with the conclusion that the topography of the keypecks was systematically related to the nature of the primary reinforcer during first but not during second-order conditioning. The results from the selective satiation test are difficult to interpret. There was no evidence to indicate that second-order responding was mediated by a detailed representation of the primary US, but a detailed representation of the reinforcer may have been mediating first-order responding. Taken together, these findings are consistent with the view that a representation of the reinforcer is an important determinant of the topography of conditioned responding.