The dynamic relationship between neural category discrimination and perceptual behavior measured with iEEG across tasks

Regions of ventral temporal cortex (VTC) demonstrate category selective responses that when disrupted lead to perceptual deficits affecting the category that area is selective for. However, the relationship between dynamic category selective activity in VTC and perceptual behavior is unclear. To investigate this relationship, intracranial EEG data was collected from 32 participants while they completed a 1-back category localizer task containing pictures of faces, words, bodies, hammers, houses, and phase-scrambled objects. Participants were asked to press a button if any image was repeated twice in a row (exemplar level repeat detection). Multivariate classifiers were trained on non-repeat image trials to predict which category the participant viewed during repeat trials. Then the relationship between the confidence of these category classifiers and behavior was examined. Around peak category selectivity (50-250 ms post-stimulus presentation) lower classifier confidence was associated with faster reaction times (RT) on repeat trials. This suggests that greater repetition suppression of peak category selective activity facilitates perceptual behavior during the 1-back task. Conversely, higher classifier confidence was associated with faster RT around the time of response (500-750 ms), suggesting that greater category information around this time facilitates perceptual behavior. We then applied these category classifiers to data from a gender discrimination task to determine if these relationships generalized. Greater face category confidence from 500-750 ms was associated with faster RT like during the 1-back task; however, there was no relationship between classifier confidence and behavior around peak face selectivity (50-250 ms). These results suggest that perceptual behaviors have a task general relationship with category selective VTC activity around the time of response and a task specific relationship around the time of peak VTC selectivity. Future work is necessary to determine the local and network-level neural features that give rise to dynamic category selective responses that are related to perceptual behavior.