Search Papers | Poster Sessions | All Posters
Poster B155 in Poster Session B - Thursday, August 8, 2024, 1:30 – 3:30 pm, Johnson Ice Rink
Do vision and imagery share common principal signal components?
Tiasha Saha Roy1 (), Jesse Breedlove1, Ghislain St-Yves1, Kendrick Kay1, Thomas Naselaris1; 1University of Minnesota
Most people experience mental imagery as an approximation to seeing. However, brain activity during acts of imagery typically exhibits lower signal-to-noise ratio (SNR) compared to vision, particularly in the early visual areas. In a previous work visual signal compression was suggested to be a plausible explanation for this apparent reduction of signal variance, and hence SNR, during imagery. In this work we explore the specific dimensions of visual coding that are preserved during imagery. We used an autoencoding voxel-to-voxel framework on data from a 7T fMRI imagery experiment to estimate the principal components of the visual and imagery signal spaces. The first principal imagery component separates imagery activity patterns in a similar way as the first principal visual component that separates visual activity patterns, implying that the same type of visual features are represented. However, visual variance is substantially lower along the imagery components, which is consistent with the compression hypothesis. Our results so far suggest that while the principal visual and imagery signal components exhibit apparent similarities in their coding for certain features, they do not seem to be completely identical.
Keywords: vision mental imagery fMRI vox-to-vox models