Relative validity of contextual and discrete cues

Contextual conditioning during relative validity training was explored in 3 experiments that used an appetitive Pavlovian conditioning preparation with rats. Magazine entries were the conditioned response. In Experiment 1, true-discrimination (TD: AX+, BX-) training generated weaker conditioning of X than did pseudodiscrimination (PD: AX+/-, BX+/-) training. The context showed a similar relative validity effect. Also, both PD training and simple partial reinforcement (X+/-) reduced contextual conditioning more than did unsignaled food, a demonstration of relative validity using partial reinforcement. Experiments 2 and 3 used within-subject and between-subjects designs, respectively, and showed that relative validity was determined by the summation of differences in conditioning to both the common element (X) and the context. Our results are consistent with an attentional model or with a computational comparator model but not with the Rescorla-Wagner (R. A. Rescorla & A. R. Wagner, 1972) model.