Occasion Setters: Specificity to the US and the CS–US Association

Abstract Two experiments using pigeon subjects examined whether the action of a positive occasion setter was specific to a particular CS–US association. In both experiments subjects were trained on a switching procedure, in which a diffuse feature, A, signaled that keylight x would be followed by one reinforcer, US1, and that a second keylight y would be followed a different reinforcer, US2. A second feature, B, signaled the converse arrangement, that x would be followed by US2 and y by US1. The intention was to establish whether the features allowed the birds to anticipate the specific keylight/US associations that they signaled. This was achieved using a blocking technique, by examining the ability of feature A to prevent learning about two test stimuli, S and D, presented in compound with x. On same trials A signaled stimulus x, which was followed, as in training, by US1; on these trials x was accompanied by S. On different trials x was signaled by A but followed by the other reinforcer, US2, and accompanied by the second test stimulus, D. If A signaled specific CS–US associations, then the US would be predicted in the former case but surprising in the latter, so that acquisition of associative strength by stimulus S would be blocked, but acquisition by stimulus D would not. In both experiments stimulus S supported less conditioned responding than stimulus D, which is consistent with this interpretation. The implications of these findings for theories of occasion setting are discussed.