Abstract

The sequence-in/sequence-out cerebellar machinery is considered from the computational point of view. We outline a learning framework which discriminates short-term from long-term learning and is able to explain single-trial adaptation to unexpected loads