Vulnerability of long-term memory to temporal delays in amphibians.

Two experiments with toads were designed to test the memory-decay hypothesis that extinction (i.e., nonreinforced) performance is a function of time since the last reinforcement. In Experiment 1, toads (Rhinella arenarum, formerly Bufo arenarum) received 15 daily acquisition trials each reinforced with access to water during 300 s in a runway and were then randomly assigned to one of 6 retention intervals (RIs): 1, 4, 8, 16, 32, and 64 days. Extinction started after the RI and lasted 8 additional daily trials. Overall extinction performance was a logarithmic function of the RI. Although 4 extinction trials produced similar performance than 4 days of RI (consistent with memory decay), 8 extinction trials produced lower performance than 8 days of RI (consistent with a decremental effect of nonreinforcement). In Experiment 2, two groups of toads received 15 daily acquisition trials each reinforced with access to water for either 30 or 600 s, thus producing two reward magnitudes. After an 8-day RI, extinction performance was weaker after training with the small, than with the large reward magnitude. These results suggest that, at least in early extinction, the instrumental performance of toads is strongly influenced by the time since the last reinforcement.