Hungry rats were trained to press a lever for food pellets prior to an assessment of the effect of a shift in their motivational state on instrumental performance in extinction. The first study replicated the finding that a reduction in the level of food deprivation has no detectable effect on extinction performance unless the animals receive prior experience with the food pellets in the nondeprived state (Balleine, 1992;. When tested in the nondeprived state, only animals that were reexposed to the food pellets in this state between training and testing showed a reduction in the level of pressing during the extinction test relative to animals tested in the deprived state. The magnitude of this reexposure effect depended, however, on the amount of instrumental training. Following more extended instrumental training, extinction performance was unaffected by reexposure to the food pellets in the nondeprived state whether or not the animals were food deprived at the time of testing. A second study demonstrated that the resistance to the reexposure treatment engendered by overtraining was due to the animals' increased experience of the food pellets in the deprived state during training rather than to the more extensive exposure to the instrumental contingency. In contrast to the results of the first two experiments, however, a reliable reexposure effect was detected after overtraining in a fmal study, in which the animals were given greater reexposure to the food pellets in the nondeprived state.

No abstract

Rats were conditioned across 2 consecutive days where a single unsignaled footshock was presented in the presence of specific contextual cues. Rats were tested with contexts that had additional stimulus components either added or subtracted. Using freezing as a measure of conditioning, removal but not addition of a cue from the training context produced significant generalization decrement. The results are discussed in relation to the R. A. Rescorla and A. R. Wagner (1972), J. M. Pearce (1994), and A. R. Wagner and S. E. Brandon (2001) accounts of generalization. Although the present data are most consistent with elemental models such as Rescorla and Wagner, a slight modification of the Wagner-Brandon replaced-elements model that can account for differences in the pattern of generalization obtained with contexts and discrete conditional stimuli is proposed.

Forward and backward blocking of taste preference learning was compared in rats. In the forward condition, thirsty rats were exposed to a flavor (A) in sucrose solution (+) or in water (-), after which they were exposed to A in compound with another flavor (B) in sucrose solution (i.e., AB+). In the backward condition, these phases were reversed. Consumption of B alone was assessed when rats were food deprived. In the forward condition, rats given A+ consumed less B than rats given A-, providing evidence of forward blocking, whereas in the backward condition, rats given A+ drank more of B than those given A-. Subsequent experiments found that alternating but not blocked preexposure to A and B, when given prior to training, produced blocking of B whether A+ was given before or after AB+, suggesting that prior failures to observe backward blocking reflect failures of discrimination.

No abstract