Counteracting memories acquired in parallel by limited pairings in eyeblink conditioning

Animals adapt to environments by learning even when experience is limited, which is critical for their survival in nature. However, the relationship between experience richness and learning efficiency remains elusive. Here we investigated this issue using mice eyeblink conditioning with three distinct training paradigms. Surprisingly, 1/9th of sensory pairings than conventional ones were not only sufficient for acquisition of conditioned responses, relying on the activity of deep cerebellar nuclei as usual, but also resulted in more efficient learning. On the other hand, the learning established by limited training was more susceptible for extinction. Modeling of establishment and extinction processes, coupled with experimental validation of the model prediction, unveiled that memories for execution of eyeblink and its suppression are independently formed irrespective of their temporal order.