This dissertation examines visual processing of contextually modulated artificial and natural stimuli in primary visual cortex on a local scale. Understanding how local features are integrated into a global structure or ignored as irrelevant background is a critical step in comprehending human vision. To investigate these mechanisms, first it was necessary to measure the relationship between inferred neural responses, such as those obtained with blood oxygenation level-dependent (BOLD) fMRI, and local stimuli. From this point, orientation-dependent contextual modulation was analyzed locally or with a contour. While focusing on primary visual cortex, these experiments with stimuli of increasing complexity provide a foundation for how local features are grouped into global structures.
BOLD fMRI provides a non-invasive method to measure the inferred neural response in humans. Because BOLD fMRI is a result of interaction between neural activity, blood flow, and deoxyhemoglobin concentration, it is not obvious that there is a linear relationship between these mechanisms as well as established functions, like the contrast response function (CRF). Chapter 2 measures the BOLD response to single Gabor patches of increasing contrast with two pulse sequences: Gradient Echo (GE) and Spin Echo (SE). GE measurements include signals from large and small veins while SE measurements eliminate the signal from large veins. Comparing these signals, at ultra-high field strength (7 Tesla) found the relationship between the CRF and BOLD fMRI for local stimuli is not linear with GE measurements.
Chapters 3 and 4 focus on orientation-dependent contextual modulation of a single Gabor patch or of a vertical line of Gabor patches. In the periphery, surrounds of parallel orientation suppress the center stimulus while surrounds of orthogonal orientation facilitate the center stimulus. The relationship between the BOLD response and these suppressive or facilitative mechanisms was measured on a local scale (Chapter 3). Then, to compare the mechanisms for orientation-dependent contextual modulation and contour integration, performance in a contour detection task was measured over an extensive parameter space (Chapter 4). These data show that the BOLD response to suppressive stimuli do not behave as predicted by psychophysical results and that orientation-dependent contextual modulation and contour integration operate over different spatial scales, and likely different neural mechanisms.
This dissertation provides data on the relationship between the BOLD response and local stimuli as well as data on the neural mechanisms behind orientation-dependent contextual modulation, contour integration, and texture classification. An over-arching theme is that inferred neural responses, such as those measured with BOLD fMRI, behave differently on a local scale than a global scale. However, other non-invasive measures provide details for how local stimuli are processed and further integrated into a global structure. Future work can incorporate computational models of neural activity and the BOLD response to clarify why measured responses differ on a local scale compared to a global scale.
University of Minnesota Ph.D. dissertation. May 2010. Major: Neuroscience. Advisors: Dr. Cheryl A. Olman and Dr.Daniel J. Kersten. 1 computer file (PDF); vi, 76 pages. Ill. (some col.)
Schumacher, Jennifer Frances.
Human neurophysiological mechanisms of contextual modulation in primary visual cortex..
Retrieved from the University of Minnesota Digital Conservancy,
Content distributed via the University of Minnesota's Digital Conservancy may be subject to additional license and use restrictions applied by the depositor.