Sensory-specific associations in flavor-preference reversal learning.

In two experiments, we examined the effect of reversal learning on the status of initially learned associations. In Experiment 1, thirsty rats were first taught to associate one flavor with sucrose and another flavor with Polycose. These relations were then reversed in a subsequent phase. One of the nutrients was then devalued by being paired with lithium chloride. The results of a two-bottle flavor-choice test revealed that the most recently learned associations governed performance. In Experiment 2, we aimed to discern whether the initially learned associations in Experiment 1 were weakened or masked by reversal learning. In order to address this question, either a 1-day (Group Immediate) or a 21-day (Group Delayed) retention interval was interpolated between the reversal and devaluation phases. Subsequent flavor-choice tests revealed that Group Immediate avoided the flavor most recently associated with the devalued nutrient but that Group Delayed avoided the flavor that was initially associated with the devalued nutrient. These findings suggest that the second-learned associations do not erase, but transiently mask, the first-learned associations, which subsequently recover over a retention interval. These results suggest a parallel in the mechanisms of extinction and reversal learning.