Extensive research spanning theory, psychophysics, and physiology has investigated how we rely on statistical regularities in the environment to improve our sensorimotor behavior: (1) Bayesian theory has provided an understanding of how one should take advantage of statistical regularities, (2) psychophysical experiments have documented the impact of such regularities on behavior, and (3) electrophysiology experiments have identified neural signals that reflect those regularities. An important consideration is that statistical properties of the environment are rarely stable. Therefore, a most pressing and unresolved question at the frontier of this interdisciplinary body of work is how malleable brain signals, through experience, gradually acquire information about new environmental statistics. Here, we will tackle this problem by developing a sensorimotor behavioral paradigm in the non-human primate model that demands adaptive statistical learning (Aim 1). We will use this paradigm to test specific computationally-motivated hypotheses regarding how the structure and dynamics of neural activity in candidate regions of the frontal cortex change throughout learning (Aim 2). Finally, we will use a dynamical systems approach to analyze the laminar organization of learning signals in the frontal cortex to tease apart functional sub-circuits with distinct input-output properties that support sensorimotor learning (Aim 3).
Understanding the neural circuits and mechanisms of sensorimotor adaptation and learning is essential for addressing behavioral deficits associated with sensorimotor dysfunction and for identifying effective strategies for intervention. Our research brings together electrophysiological studies in nonhuman primates with computational modeling to provide an understanding of how cortical circuits support sensorimotor learning, and help lay a foundation for developing new strategies for treatment.
