Schedule of Events | Search Abstracts | Invited Symposia | Symposia | Poster Sessions | Data Blitz

Sketchpad Series 

SuperEEG: A Gaussian Process Model of Human Electrocorticographic Data Revisited and Optimized

Poster Session F - Tuesday, March 10, 2026, 8:00 – 10:00 am PDT, Fairview/Kitsilano Ballroom

Jose Carmona-Sanchez¹ (), Lucy L W Owen; ¹University of Montana

Understanding how cognition emerges from distributed neural systems requires methods that capture the brain’s fast, large‑scale coordination during tasks. Direct measurement at this scale is limited in humans because ECoG offers millisecond precision but sparse spatial coverage. Our method, SuperEEG, addresses this by learning a population‑level covariance model that relates activity across cortical locations and then using Gaussian process regression to reconstruct full‑brain field potentials from a patient’s limited electrodes. Here we reposition SuperEEG as a general tool for studying cognition—spanning attention, decision‑making, language, and motor control—and describe extensions that make the model more neurobiologically grounded and task‑general. First, we (i) systematically tune the radial‑basis function length‑scale (λ) to balance spatial smoothness and local specificity and (ii) incorporate neuroanatomically and functionally informed priors into the covariance (e.g., volume conductance, inter‑regional distance, structural connectivity from diffusion imaging, and functional connectivity constraints). We benchmark reconstructions with cross‑validation at held‑out electrodes and quantify how training within versus across patients and tasks affects accuracy and interpretability. We further evaluate whether anatomically informed priors improve identification of transient, task‑evoked network states on tens‑of‑milliseconds timescales using the multi‑task Kai Miller ECoG dataset. Together, these advances provide a practical pipeline for estimating fast, whole‑brain dynamics from sparse recordings, enabling individualized tests of how distributed coordination supports cognition.

Topic Area: METHODS: Electrophysiology