Pavlovian conditioning is the consequence of more than just the number of CS-US pairings

Three Pavlovian conditioning experiments with rats were performed to test predictions from models of associative learning. The procedure in all 3 experiments involved assessing conditioned responding to a Light that was compounded with either perfect or imperfect auditory signals for food. In Experiment 1 a partially reinforced Light which signaled 100% of the food pellets acquired a greater ability to elicit a conditioned response when the auditory cues with which it was paired were relatively poor predictors of the food. In Experiment 2 the Light was still a moderate positive predictor for food, but food was presented both in the presence and absence of the Light. Responding to the Light again was reliably higher when the auditory cues were poor predictors of food. Experiment 3 replicated the results from Experiment 2 and included two control conditions to assess sensitivity to the absolute validity of the Light. In both control conditions the Light was uncorrelated with the US. In one condition the number of food presentations was the same as in the experimental condition, while in the second the number of Light-food pairings was maintained. Results suggested that regardless of the absolute contingency of the Light responding was higher when the auditory cues were uncorrelated with food. These findings suggest that associative models of cue competition make accurate predictions of conditioned behavior following exposure to multiple predictors of reinforcement