Abstract
Cells in V1 show increased spiking when texture elements within the neuron's receptive field belong to a figure compared to when those same elements belong to the background (Lamme, 1995). We examined whether figure-ground signal enhancement could be measured in a population of neurons in human V1 with fMRI. Previous fMRI work found no evidence of V1 signal enhancement for figures defined by 2nd-order texture contours, but a demanding fixation task may have prevented figures from being perceived (Schira et al., 2004). In the present study subjects viewed oriented line-segment displays (19° x 19°) while BOLD fMRI data were acquired using a rapid event-related design. Ground stimuli were composed of parallel lines at one of two orientations. Figure stimuli were identical to ground but also contained two figure regions (6° x 6°, opposite sides of fixation with inner edges at 2° eccentricity) in which line orientation was orthogonal to the background. To control attention across conditions subjects detected infrequent oddball stimuli in which the line segments were of slightly higher density. Separate localizer scans were used to find voxels in V1 responsive to figure but not background regions (V1 ROIs). We averaged data within the V1 ROIs and computed conditional hemodynamic timecourses using standard linear methods. Data from four subjects revealed a 13% increase in BOLD response in the V1 ROIs to figure compared to ground stimuli. This result provides evidence that figure-ground segregation results in increased neuronal activity in human V1 that is detectable with fMRI.
Support Contributed By: NIH EY11862