Item

Abstract

Re: Freeman J, Brouwer GJ, Heeger DJ, Merriam EP (2011) Orientation Decoding Depends on Maps, Not Columns. J Neurosci 31(13): 4792- 4804.

Freeman et al. observed a coarse-scale topographic map of orientation preference in human V1. The map was tightly colocalized with the angular-position component of the retinotopic map - a bias towards radial orientations. They found that orientation decoding accuracy was degraded by removing the angular-position map from the responses to different orientations, which was interpreted as "proving that the coarse- scale orientation map was necessary for orientation decoding".

Radial bias can be avoided by the use of stimuli matched everywhere for their radial components, such as spirals of opposite sense. Previous studies have reported significant accuracy in discriminating anti- clockwise and clockwise spirals from patterns of activity in human V1 (Mannion et al., 2009; Seymour et al., 2010), leading to the conclusion that "the influence of a radial bias is not necessary for the discrimination of stimulus orientation on the basis of fMRI activity patterns" (Mannion et al., 2009).

How can these apparently disparate conclusions be reconciled? One possibility is that the brief (1.5 second) presentation time of each orientation in the study by Freeman et al. coupled with the presence of a coarse radial bias in the resulting voxel time courses might have obscured the subtle variations observed by the previous studies that used spiral patterns as stimuli and presented them for longer periods of time (12-15 s). While we share the concern of Freeman et al. regarding the potential impact of coarse scale biases on orientation classification, we suggest that their failure to find significant orientation information in human V1 when radial bias is controlled for should be interpreted with caution.