Abstract
Observers tend to detect horizontal and vertical objects or edges more frequently than tilted features, which is known as the oblique effect. Specifically, we are more sensitive to as well as better at estimating the angles of objects at cardinal orientations. Previous research indicates cardinal orientation preferences in V1 as evidence of the oblique effect (Furmanski & Engel, 2000; Li et al., 2003), while other research suggests that a radial bias with no cardinal superiority is present in the peripheral regions of V1 (Freeman et al. 2011). Here, we tested for the presence of radial bias or the oblique effect in high-resolution fMRI data by measuring the orientation preferences of voxels in foveal and para-foveal regions of interest. Six observers performed a two-alternative forced-choice orientation discrimination task while being scanned with 0.6 mm 7T fMRI. Observers maintained fixation within a 0.5-degree diameter circle and determined whether the the second of two gratings (2 cycles/degree, 8-degree radius) was tilted clockwise or counterclockwise compared to the first. The grating pairs had 8 possible (average) orientations; an adaptive staircase controlled the orientation difference ). Orientation discrimination threshold was computed by fitting a sigmoidal function to the proportion of correct responses as a function of orientation difference. Behavioral data revealed an oblique effect, in which better performance (lower thresholds) was found in trials with gratings tilted along the cardinal axis as opposed to those with gratings tilted along the oblique axis. However, in neither foveal nor peripheral ROIs was an oblique effect observed: the orientation preference of individual voxels could be predicted by the polar angle of the retinotopic location. This effect was significantly stronger in peripheral ROIs, compared to foveal ROIs. These findings indicate that our dataset provides physiological evidence for radial bias, but none for the oblique effect, even in the fovea.