Estimating cortical reorganization and neural fill-in using fMRI population receptive field (pRF) mapping.
33.4037, Sunday, 17-May, 8:30 am - 12:30 pm, Pavilion
Jessica Thomas1, Erik Runeson1, Ione Fine1, Geoffrey Boynton1; 1Department of Psychology, University of Washington, Seattle, WA
Introduction: There is substantial interest in the use of population receptive field (pRF) 1 modeling to examine cortical reorganization after vision loss. Our lab has been developing methods for quantifying bottom-up changes in population receptive fields that are independent from the effects of spatio-temporal BOLD nonlinearities and various top-down influences including neural ‘fill-in’. Previously we demonstrated how estimates of cortical reorganization can be biased by either not modeling scotomas appropriately 2, or by using a predictable stimulus 3. Here we examine how these phenomena interact with the fixation instability typical of individuals with low vision. Methods: In normally sighted individuals, we measured fMRI responses in cortical areas V1-V3 with a 2x2 stimulus design: Drifting bar vs. multifocal: stimuli were either predictable drifting bar sequences or unpredictable multifocal sequences; and Full-field vs. scotoma: stimuli were either full-field or contained a central mask simulating a foveal scotoma. Throughout each condition the fixation spot moved in a periodic pattern designed to simulate a typical form of nystagmus. Data were analyzed in a 2x2 design: Scotoma modeled or not modeled: stimuli containing scotomas were modeled either with or without the inclusion of the scotoma. Eye-movements modeled or not modeled: the moving fixation point was either modeled according to eye-tracker data or assumed to be static. Results: We found that unbiased pRF estimates can be obtained as long as an unpredictable stimulus is used, scotomas are modeled, and eye-movements are compensated for. Failing to compensate for any or all of these parameters will result in biased pRF estimates resembling cortical reorganization; including increased receptive field size and eccentricity values, particularly for pRFs near the scotoma. Finally, we show that unbiased pRFs can accurately estimate the neural “fill-in” elicited by predictable stimuli.