NeuroVentures Funded Grant

Surya Ganguli, Applied Physics
Kwabena Boahen, Bioengineering
Tirin Moore, Neurobiology

We have made substantial progress in systems neuroscience towards understanding how sensory stimuli are represented within purely sensory areas, and how simple motor behaviors are encoded within motor networks. However, we currently lack even the beginnings of a unified theoretical framework for thinking about how the interaction of sensory and motor signals give rise to cognitive phenomena deep within the brain, far from either the sensory or motor periphery.

We propose that one fundamental goal of top-down and bottom-up cortical circuitry is to predict sensory inputs based on past motor commands and the temporal statistics of the world, and use these predictions to adaptively guide future motor and cognitive functions. Such predictions occur for example a few times every second when we rapidly move our eyes to examine new locations in visual space. The sensorimotor circuits underlying these eye movements make predictions about the locations of interesting objects in order to both guide eye movements as well as to adaptively deploy our attention. We will combine theory, modeling, physiology and behavior to unravel the nature of the predictions occurring in macaque sensorimotor circuits for eye movements, and their relationship to adaptive attentional deployment. Moreover, we will expose such circuits to new situations in which well-established relationships between motor commands and sensory statistics are fundamentally altered. We will explore how these alterations force these circuits to learn new predictions, and how they affect the ability of these circuits to allocate attention. In short, we will explore how changing the structure of sensorimotor experience changes one of the most fundamental aspects of cognition, namely our ability to pay attention to stimuli within our environment.