Abstract
Psychophysical research has established that there are channels selective for the spatial frequency (SF) components of visual stimuli (e.g. Blakemore & Campbell, 1969, J. Physiol., 203, 237–260). Building on our previous work (Sowden, Myers & Payne, 2004, Perception, 33, 173), we sought to link such psychophysically defined SF channels to the underlying neural substrate using the fMRI-adaptation method (Grill-Spector & Malach, 2001, Acta Psychologica, 107, 293–321). We first identified area V1 using retinotopic mapping (Sereno et al., 1995, Science, 268, 889–893). Then observers viewed runs of 6 blocks of counter-phasing sinusoidal gratings (8Hz temporal frequency) interspersed with fixation blocks, each block lasting for 60 seconds. Of these blocks, one contained grating patterns of the same SF (1.27 c/deg) while the other blocks were composed of this SF and two others symmetrically chosen to be (+/−) 0.25, 0.5, 1, 2 or 3 octaves different. Given that typical V1 SF channel bandwidths are a little over 1 octave, we would expect that multiple SF selective populations of neurons would be stimulated in the blocks spanning the greatest SF ranges. Further, previous work shows that a 25msec exposure to a grating - a fifth of the duration here - is enough time for adaptation to occur (Georgeson & Georgeson, 1987, Vis. Res., 27, 369–379) and that recovery from adaptation is unlikely to have occurred during the 250 msec gap between each repetition of the same SF (Greenlee et al, 1991, Vis. Res., 31, 223–236). Thus, at the present repetition frequency, we should predict greater adaptation in the blocks spanning the wider range of SF's and therefore adapting multiple cell populations. In line with this, the BOLD response in V1 showed a greater drop for the blocks with wider ranges of stimulus SF's; the +/− 3 octave block showed the biggest drop in signal, hence greatest adaptation. Our findings indicate that fMRI can be used to isolate the activity of individual SF channels.