Abstract
Perception of sensory inputs is modulated by shifts in endogenous, ongoing brain states. Specifically, previous studies have tied endogenous states measured by the pre-stimulus neural activity to behavior in visual tasks. However, it remains unclear whether the endogenous shifts modulate neural coding and category tuning in the ventral stream, which could provide a neural pathway for behavioral modulation. To address these questions, we collected intracranial electroencephalography (iEEG) data from a large cohort of 32 patients while viewing visual images. We analyzed the iEEG data recorded from 230 channels showing category-selectivity for 5 different categories of visual stimuli: faces, human bodies, words, places, and tools. A generalized linear model was trained to classify the category of the stimuli, and the trained model was used to extract the linear subspace that maximized the category-selectivity in post-stimulus neural activity. We used the projection of neural activity in this linear subspace as the neural metric of category-selectivity of single trials, and evaluated the dependency between the pre-stimulus oscillatory activity and the post-stimulus category-selective activity. We found that the pre-stimulus oscillatory activity predicted the magnitude of the post-stimulus category selectivity on a single-trial basis. Specifically, different patterns of pre-stimulus activity led to different degrees of category tuning in the category-selective areas. These results demonstrate that endogenous activity modulates category tuning in ventral temporal cortex, providing a potential neural basis for perceptual modulation by endogenous activity.
Meeting abstract presented at VSS 2018