Schedule of Events | Symposia

A Communication Subspace Relays Behaviorally-relevant Information from Human Prefrontal to Motor Cortex

Poster Session A - Saturday, March 29, 2025, 3:00 – 5:00 pm EDT, Back Bay Ballroom/Republic Ballroom

Neha Binish¹ (neha.binish@uni-tuebingen.de), Robert T. Knight², Randolph F. Helfrich¹; ¹Hertie Institute for Clinical Brain Research, University Medical Center Tübingen, Tübingen, Germany, ²Helen Wills Neuroscience Institute, UC Berkeley

Flexible routing of context-dependent neural information between cortical areas is essential for adaptive human behavior. The mechanisms underlying this selective routing at the large-scale network level remain poorly understood. Here, we assessed how contextual information is encoded in the prefrontal cortex (PFC) and transmitted to the motor cortex (M1) to enable efficient visual target detection. We recorded intracranial electroencephalography (iEEG) from subdural grid electrodes in 12 pharmacoresistant epilepsy patients. All participants exhibited a clear behavioral advantage of the predictive context, responding faster to predicted than random targets (P = 0.0333). Despite a clear behavioral difference, univariate analyses of neural recordings from PFC and M1 did not reveal context-specific electrophysiological signatures predictive of the behavioral benefit. However, both regions, PFC and M1, displayed high-dimensional neural dynamics. We hypothesized that context-relevant information might be confined to a subset of PFC population activity. Using multivariate analyses, we identified a low-dimensional communication subspace between PFC and M1 in humans. Critically, PFC activity projected onto this subspace correlated with behavioral outcome (rho = 0.79, P = 0.0037). Hence, latent subspace dynamics tracked context-dependent behavior. These findings reveal a population-level mechanism for the selective routing of contextual information between PFC and M1, offering new insights into how neural circuits dynamically adapt to varying task demands. Our results highlight the importance of multivariate analysis approaches to uncover behaviorally-relevant neural dynamics that are often hidden in high-dimensional neural activity patterns. In sum, our findings highlight the role of communication subspaces in mediating flexible, context-dependent behavior.

Topic Area: EXECUTIVE PROCESSES: Other