Mechanisms for Internal Models in a Cerebellum-like Circuit

Project Summary/Abstract Predicting the consequences of action is a vital function of the nervous system. The hypothesized neural substrate are so-called internal models that transform information about outgoing motor commands and the current sensory state into predictions of sensory input. Such internal models are likely critical for a wide range of sensory, motor, and cognitive functions and their disruption has been implicated in neurological disorders such as autism and schizophrenia. Nevertheless, it has proven challenging to understand how internal models are implemented in neural circuits in the mammalian brain. Our prior studies were successful in developing a detailed mechanistic understanding of how neurons in the electrosensory lobe (ELL) of mormyrid fish predict and cancel out the sensory consequences of a simple behavior--the electric organ discharge (EOD) pulse. However, because these studies were performed in immobilized animals, the nature of the predictions studied was limited in scope and complexity. This renewal uses novel methods for neural recording and high-resolution behavior monitoring in freely swimming fish to study the more complex internal models underlying the remarkable active electrolocation abilities of electric fish. Computational modeling approaches will be used both to rigorously define the problem facing the active electrosensory system and to generate and test realistic circuit-level models of how they may be solved. The key components of such models, including synaptic plasticity, recurrent and feedforward connectivity, and biophysical compartmentalization of axonal and dendritic spikes, are common to many neural systems including the cerebellum, hippocampus, and neocortex. Hence insights from these studies are expected to be widely relevant to understanding how internal models are implemented in neural systems.