Abstract representations encoded by human hippocampal neurons support behavioral inference and can be induced through verbal instruction.
Hristos Courellis, Juri Minxha, California Institute of Technology, United States; Araceli Cardenas, Taufik Valiante, Toronto Western Hospital, Canada; Adam Mamelak, Cedars-Sinai Medical Center, United States; Ralph Adolphs, California Institute of Technology, United States; Stefano Fusi, Columbia University, United States; Ueli Rutishauser, California Institute of Technology, United States
Session:
Posters 2B Poster
Presentation Time:
Fri, 25 Aug, 13:00 - 15:00 United Kingdom Time
Abstract:
Abstract: Desirable properties of neural representations, namely high dimensionality and abstract geometry, can exist in harmony in the brain despite their presumed trade-off. However, the process by which such properties emerge at the level of the neural population, particularly in the human brain, remain unstudied. Here, we study and manipulate neural representational geometry in the human brain by conducting single-unit recordings in neurosurgical patients who are performing an inferential reasoning task. We find that the hippocampus uniquely supports inference behavior through changes in its representational geometry, and that these changes can be induced in minutes through verbal instruction.
