Abstract
Functional MRI data are typically interpreted as measurements of average, local neural population activity. However, when the local neural population encodes multiple aspects of a stimulus or behavioral state, quantitative inference becomes more difficult. For example, interactions between neighboring visual stimuli may confound interpretation of the local BOLD signal. To better understand the contributions of task, timing, and stimulus geometry to the fMRI signal, we conducted four experiments measuring the fMRI response to small sinusoidal grating patches with either parallel or orthogonal surrounding gratings. Experiments were conducted in a 7 Tesla scanner with 1.2mm isotropic resolution. Targets were presented at 8%, 16% and 32% contrast while manipulating: (1) spatial extent of parallel (strongly suppressive) or orthogonal (weakly suppressive) surrounds, (2) stimulus onset asynchrony between the target and surround, (3) temporal structure of stimulus presentation (block vs. event-related design), and (4) locus of spatial attention. Previous work has shown that the localized fMRI response in primary visual cortex (V1) during iso-orientation suppression does not reliably increase with greater luminance contrast for small targets (Schumacher & Olman, 2010), and can be predicted by long-range patterns too large to be detected by V1 receptive fields (Joo, Boynton & Murray, 2012). Consistent with these findings, we observed that the localized V1 response to sinusoidal grating patches reliably indicated target contrast only when attention was directed away from the stimulus, or when a blocked stimulus presentation was used. For attended event-related stimuli, the V1 fMRI response to luminance contrast was conflated with higher-order pattern responses, reflecting second-order contrast between the target stimuli and flanking context. This work highlights the important role of both attention and pattern perception during early visual processing, as well as our limited ability to make inferences about diverse local neural activity from the fMRI signal.
Meeting abstract presented at VSS 2014