Wiersma Visiting Professor Lecture: Dr. Stefano Fusi

Please join us on Tuesday, February 24 at 4:00PM for the Wiersma Visiting Professor Lecture* in Chen 100.

Refreshments will follow in the Chen breezeway

Speaker: Dr. Stefano Fusi, Professor of Neuroscience, Columbia University

Title: The geometry of abstraction: from cognitive to spatial maps

Abstract: Neurons in the mammalian brain often exhibit complex, non-linear responses to multiple task variables (mixed selectivity). Despite the diversity of these responses, which appear disorganized, it is often possible to observe an interesting structure in the representational geometry: task-relevant variables are encoded in approximately orthogonal subspaces within the neural activity space. This encoding is a signature of low-dimensional disentangled representations, it is typically the result of a process of abstraction and allows linear readouts to readily generalize to novel situations. We show that these representations are observed across multiple brain areas in non-human primates (collaboration with the Salzman group). By studying the dynamics of the geometry in five brain areas (hippocampus, dorsolateral prefrontal cortex, anterior cingulate cortex, orbitofrontal cortex, and amygdala), we could formulate precise hypotheses about the neural mechanisms underlying decision-making. I will then present new results from the collaboration with the Axel and Abbott groups (data collected by Rajyashree Sen): we observed a spatial representation in the dorso-medial prefrontal cortex (dmPFC) that extrapolates from explored to unexplored regions of space. Using GCaMP6f imaging in miniscope-implanted mice, we recorded excitatory neurons in the dorso-medial prefrontal cortex (dmPFC), hippocampal dorsal CA1 (dCA1), as mice freely explored a square arena. Position could be decoded with high accuracy from both dmPFC and dCA1 populations, with the hippocampus showing the highest overall decoding accuracy. In dmPFC, a linear regressor trained to predict the animal's x-position in one half of the arena accurately generalized to the opposite half. Similarly, for the y-position. This indicates a linear, disentangled representation of space that might support generalization. In contrast, the dCA1 showed markedly reduced generalization, consistent with a localized place-cell code.

*The lecture and visiting professorship are named for Cornelius Wiersma who came to the Biology Department at Caltech in 1934, representing the field of neurophysiology, a precursor to the field of neurobiology. His scientific career focused first on the neuromuscular system, then on the central nervous system, and finally on the visual system.