Generative effects of extinction learning emerging from an embodied associative network

The adaptation to changes in the environment has a great importance for survival of the individual. A process which actively drives behavioral adaptation is extinction learning - the process of ceasing a previously learned behavior in response to altered reinforcement contingencies. An often neglected aspect of extinction learning is the generation of new, previously non-reinforced behaviors in parallel to the gradual elimination of the learned response. However, the mechanism that generates these behaviors is loosely understood. I analyzed the behavior of 16 rats executing a T-maze context-dependent extinction task. The analysis confirmed the presence of extinction learning on a sample level and uncovered a variety of individual behaviors observed during extinction. Then by adapting the model proposed in Donoso et al.. In review, I demonstrated that an associativemodel can produce extinction learning as well as behavioral variability similar to the one observed in rats.