hide
Free keywords:
-
Abstract:
Learning the contingencies of a new experiment is not an easy task for animals. Individuals learn in an idiosyncratic manner, revising their approaches multiple times as they are shaped, or shape themselves, and potentially ending up with different asymptotic strategies. Their long-run learning curves are therefore a tantalizing target for the sort of quantitatively individualized characterization that modelling can provide. However, any such model requires a flexible and extensible structure which can capture radically new behaviours as well as slow changes in existing ones. To this end, we suggest a dynamic input-output infinite hidden semi-Markov model whose latent states are associated with specific components of behaviour. This model includes a countably infinite number of potential states and so has the capacity to describe substantially new behaviours by introducing extra states, while dynamics in the model allow it to capture more modest adaptations to existing behaviours. We fit the model to data collected from more than 100 mice as they learn a contrast detection task over multiple stages and around ten thousand trials each. The resulting fits offer comprehensive insight into animal learning on the given task, which we extract by studying a number of different properties of the collection of fits. Despite large individual differences, we find three major stages of learning, the transitions between which are marked by distinct additions to task understanding, and which most animals progress through as they gain expertise.