The Neuronal Underpinnings of Non-invasive Laminar fMRI

The six layers of cortex form distinct computational units that together govern the information flow and processing required for complex behavior. Hence, unravelling the brain's computational strategies requires understanding the layer-specific organization of the neocortex. Until recently, layer-resolved recordings have been confined to animal models, ignoring specific properties of the human brain and limiting our ability to study uniquely human functions such as language. The unprecedented opportunity to combine laminar electrophysiology recordings in humans with High-field fMRI in the same human subjects could close this gap; however, inherent vascular artifacts prevent a straightforward depth-resolved interpretation of the BOLD signal. To disentangle vascular from neuronal processes, we propose to develop a neurovascular coupling model by combining the laminar electrophysiology data with laminar fMRI in the same human subjects. To this end, we will exploit a unique opportunity to record ground truth, layer-resolved neuronal activity in epilepsy patients. We hypothesize that such a model - validated using data from patients and controls and multiple tasks - will resolve layer-specific neuronal activation from fMRI responses and will be applicable to the general population, across tasks, and cortical areas ? without the need for additional electrophysiological data. 1