摘要：Rats were runway trained with sequences of rewards that changed in 3 phases. In Phase 1 (24 days), the sequences were NP', SNP', and P'SNP' (n = 3), or NS', PNS', and S'PNS', where P and P' refer to 4 and 8 plain Noyes pellets, and Sand S' are 4 and 8 sucrose pellets. N was a 30-s confinement in the goal without reward. In Phase 2 (14 days) the animals were trained either SNP' or PNS' depending upon compatibility with the Phase 1 training series. A test for control of performance by reward-memory associations or position associations occurred in Phase 3 (2 days). The test series was a transfer NNN. Controls were not trained in Phase 1, but received training in Phase 2 (24 days) with either SNP' (n = 3) or PNS' (n = 3). In Phases 1 and 2, responding was slower on N trials than on rewarded trials. In Phase 3 controls responded slower on the second trial of NNN than on the initial or final trials. The animals trained in all 3 phases showed either the same pattern as the controls in the NNN test (n = 4) or approached all 3 NNN trials rapidly (n = 2). The findings illustrate the difficulty of blocking position learning by training with reward-memory associations.
