Keynotes | K&Ts | GACs | Talks | Posters | Search

Poster E in Poster Session E: Thursday, August 6, 10:30 am – 12:15 pm, Kimmel Center, Shorin & Rosenthal Rooms

Cortical gradient of signal-noise subspace alignment

Tomoya Nakamura¹, Jungwoo Kim¹, Ali Moharramipour², Thomas Naselaris³, Hakwan Lau¹, Kendrick Kay³, Seng Bum Michael Yoo⁴; ¹Institute for Basic Science, ²RIKEN, ³University of Minnesota, ⁴Sung Kyun Kwan University

Presenter: Tomoya Nakamura

Neural responses to repeated presentations of the same stimulus exhibit stochastic variability, or neural noise, which is often correlated across neuronal populations. The relationship between the structures of noise and signal critically constrains information encoding, yet how this relationship may vary across the human cortex remains unclear. Here, we quantified the geometric alignment between signal and noise subspaces across cortical regions using fMRI data collected during a working memory task. Applying Generative modeling of Signal and Noise (GSN) to multivoxel activity patterns, we estimated signal and noise covariance matrices per brain region. We found that signal–noise alignment systematically increased along the cortical processing hierarchy, with transmodal regions exhibiting higher alignment than unimodal regions. The findings suggest that the structure of neural variability follows intrinsic functional gradients.

Topic Area: Methods, Tools, Theory & Neural Coding