Network Dynamics of Visual Object Recognition

<p>Visual object recognition is the principal mechanism by which humans and many animals interpret their surroundings. Despite the complexity of neural computation required, object recognition is achieved with such rapidity and accuracy that it appears to us almost effortless. Extensive human and non-human primate research has identified putative category-selective regions within higher-level visual cortex, which are thought to mediate object recognition. Despite decades of study, however, the functional organization and network dynamics within these regions remain poorly understood, due to a lack of appropriate animal models as well as the spatiotemporal limitations of current non-invasive human neuroimaging techniques (e.g. fMRI, scalp EEG). To better understand these issues, we leveraged the high spatiotemporal resolution of intracranial EEG (icEEG) recordings to study rapid, transient interactions between the disseminated cortical substrates within category-specific networks. Employing novel techniques for the topologically accurate and statistically robust analysis of grouped icEEG, we found that category-selective regions were spatially arranged with respect to cortical folding patterns, and relative to each other, to generate a hierarchical information structuring of visual information within higher-level visual cortex. This may facilitate rapid visual categorization by enabling the extraction of different levels of object detail across multiple spatial scales. To characterize network interactions between distributed regions sharing the same category-selectivity, we evaluated feed-forward, hierarchal and parallel, distributed models of information flow during face perception via measurements of cortical activation, functional and structural connectivity, and transient disruption through electrical stimulation. We found that input from early visual cortex (EVC) to two face-selective regions – the occipital and fusiform face areas (OFA and FFA, respectively) – occurred in a parallelized, distributed fashion: Functional connectivity between EVC and FFA began prior to the onset of subsequent re-entrant connectivity between the OFA and FFA. Furthermore, electrophysiological measures of structural connectivity revealed independent cortico- cortical connections between the EVC and both the OFA and FFA. Finally, direct disruption of the FFA, but not OFA, impaired face-perception. Given that the FFA is downstream of the OFA, these findings are incompatible with the feed-forward, hierarchical models of visual processing, and argue instead for the existence of parallel, distributed network interactions.</p>

Name thesis:degree_name

Doctor of Philosophy (PhD)

Level thesis:degree_level

Dissertation (PhD)

Year dc:date.available

2016

Author dc:creator

• Kadipasaoglu, Mehmet C

Contributors dc:contributor

• Nitin Tandon
• Steven J. Cox
• Valentin Dragoi

• intracranial EEG (icEEG)
• electrocorticography (ECoG)
• subdural electrode (SDE)
• broadband gamma activity (BGA)
• category-selectivity
• ventral temporal cortex (VTC)
• lateral occipital cortex (LOC)
• fusiform face area (FFA)
• visual object recognition
• visual naming
• Behavioral Neurobiology
• Cognitive Neuroscience
• Computational Neuroscience
• Medicine and Health Sciences
• Systems Neuroscience

Repository record dc:identifier

https://digitalcommons.library.tmc.edu/utgsbs_dissertations/645

OAI identifier oai:identifier

oai:digitalcommons.library.tmc.edu:utgsbs_dissertations-1687