Abstract
Studies examining spatial maps of where information is represented in ventral temporal cortex (VTC) have revealed critical organizational principles of the neural underpinnings of visual recognition. However, less is known about the fine spatiotemporal dynamics of information processing across VTC, which can also reveal critical details about the organizational and computational principles underpinning recognition. To fill this gap, we examined data recorded from 383 electrodes (electrocorticography, ECoG) in or on category-sensitive VTC of 38 participants. Temporally specific, multivariate information theoretic measures were used to assess how category-sensitive information evolved over time and space. At the group level and within individual subjects, the onset latency of this information was positively correlated with posterior to anterior VTC coordinate, consistent with fast-feedforward frameworks of the ventral visual stream. However, electrodes also demonstrated a decrease in the rate of information accumulation along the posterior to anterior VTC axis, with information accumulating more slowly and peaking later along a continuous gradient down VTC. We hypothesized that slower rates of information accumulation may be caused by increased diversity of sources that information is being integrated over in more anterior VTC regions. To test this, we examined the functional connectivity between each category-sensitive VTC electrode and all other electrodes implanted in the patients’ brains. The results support our hypothesis, showing that the rate of information accumulation was negatively correlated with functional connectivity to non-visually responsive brain regions, but not connectivity to visually responsive brain regions. This study suggests a novel organizational and computational principle in VTC: the rate of category-sensitive information accumulation is shaped by the diversity of non-visual sources that a region integrates over, which correlates with a posterior-to-anterior gradient of VTC organization.
Acknowledgement: National Institute of Mental Health Award NIH R01MH107797 (to A.S.G), National Science Foundation Award 1734907 (to A.S.G.).