Boring, Matthew J
(2022)
Bottom-up and top-down contributions to ventral temporal cortex organization and dynamics.
Doctoral Dissertation, University of Pittsburgh.
(Unpublished)
Abstract
Ventral visual cortex is a hierarchical recursive network that facilitates object recognition. Many questions regarding the balance between bottom-up and top-down constraints on the functional organization and response dynamics of ventral temporal cortex (VTC) remain unanswered. Here, we map spatial and temporal properties of category-selective neural populations in VTC and investigate long-range functional connectivity patterns that are associated with key neural dynamics in these populations. These maps demonstrate systematic changes in neural dynamics and functional connectivity patterns at successive stages of the ventral visual hierarchy, which are not well characterized by fast feedforward models. Regions within VTC that were highly selective for specific object categories demonstrated a complex organization, with multiple adjoining patches selective for words and faces, each with distinct dynamics.
To understand how bottom-up and top-down interactions influence local neural representations, we specifically examined the responses of one key region involved in reading, word-selective ventral occipitotemporal cortex (wVOT), sometimes referred to as the Visual Word Form Area. We replicate the finding that activity in this region demonstrates a dynamic shift in its representation 250 ms after viewing real words. Early activity from wVOT was sufficient to disambiguate visually dissimilar words, whereas later activity could disambiguate words sharing all but one letter. This transition was strongest for real words compared to pseudowords, consonant strings, and false fonts and was associated with increased functional connectivity with anterior VTC and early visual cortices. This suggests that bottom-up information, potentially across multiple eye movements, and top-down information, like phonology and semantics, encourage dynamic shifts in wVOT representations.
Finally, we review recent and seminal findings of wVOT’s development and response properties. In doing so, we arrive at a model wherein wVOT’s localization is constrained by preexisting white matter pathways specialized for evolutionarily older functions, including visual to phonological transformations necessary for lip/speech reading and visual to semantic transformations necessary for object naming. This model help explains why wVOT responds to non-visual linguistic stimuli and why it has a consistent localization across individuals. Together, this work provides a systematic investigation into how bottom-up and top-down influences shape the organization and dynamics of category-selective VTC.
Share
| Citation/Export: |
|
| Social Networking: |
|
Details
| Item Type: |
University of Pittsburgh ETD
|
| Status: |
Unpublished |
| Creators/Authors: |
|
| ETD Committee: |
|
| Date: |
26 May 2022 |
| Date Type: |
Publication |
| Defense Date: |
15 March 2022 |
| Approval Date: |
26 May 2022 |
| Submission Date: |
1 April 2022 |
| Access Restriction: |
No restriction; Release the ETD for access worldwide immediately. |
| Number of Pages: |
188 |
| Institution: |
University of Pittsburgh |
| Schools and Programs: |
School of Medicine > Neurobiology |
| Degree: |
PhD - Doctor of Philosophy |
| Thesis Type: |
Doctoral Dissertation |
| Refereed: |
Yes |
| Uncontrolled Keywords: |
vision, object recognition, reading, words, faces, cognitive neuroscience, neural development, intracranial electroencephalography, magnetoencephalography, functional magnetic resonance imaging, ventral stream, fusiform gyrus, neural dynamics, timescales, functional connectivity |
| Date Deposited: |
26 May 2022 04:24 |
| Last Modified: |
26 May 2022 04:24 |
| URI: |
http://d-scholarship.pitt.edu/id/eprint/42453 |
Metrics
Monthly Views for the past 3 years
Plum Analytics
Actions (login required)
 |
View Item |
![[feed]](/images/feed-icon-32x32.png){kind=icon}