Department of Brain and Cognitive Sciences
McGovern Institute for Brain Research
Bayesian computation through cortical dynamics
Past experiences impress upon neural circuits information about statistical regularities of the environment, which help us in all manners of behavior, from reaching for one’s back pocket to tracking a friend’s voice in a crowd and making inferences about others’ mental states. The effect of statistical regularities on behavior is often described in terms of Bayesian theory, which offers a powerful and principled framework for understanding the combined effect of prior beliefs and sensory evidence in perception, cognition, and sensorimotor function. The effects of experience on neural activity, on the other hand, is typically described in terms of cellular mechanisms that govern the response properties of neurons. For example, natural statistics are thought to shape tuning properties of neurons through adjustments of synaptic connections in early sensory areas. Therefore, there is a fundamental gap between our understanding of how behavior exploits statistical regularities and how the nervous system represents past experiences.
Recent studies have provided a deeper understanding of how neural circuits perform behaviorally relevant computations though an analysis of the geometry and structure of population cortical activity in trained animals as well as in-silico activity in artificial neural networks. Using this emerging multidisciplinary approach within the context of a Bayesian timing task in monkeys, we investigated how neural circuits in frontal cortical areas might encode prior statistics and how the dynamic patterns of activity they generate could support Bayesian integration. Our results indicate that prior statistics establish curved manifolds of neural activity that warp the underlying representations and create biases in accordance with Bayes-optimal behavior. This finding uncovers a simple and general principle for how prior beliefs may be embedded in the nervous system and how they might exert their influence on behavior.
A pizza lunch will be served.