Two kinds of fMRI repetition suppression?

Abstract

In many cortical regions, the fMRI response to a repeated stimulus is reduced compared to the response to a novel stimulus. This phenomenon, known as fMRI repetition suppression, or fMRI adaptation, has been used as a tool to study neural representations at the subvoxel level. However, many aspects of fMRI adaptation are not understood. For example, it is unclear whether reductions caused by immediate repetition of a stimulus index the same representations as reductions caused by repetition after a longer interval. We examined this issue by measuring the effects of both kinds of repetition within the same experimental session. In the first half of the experiment, immediate (within-trial) repetition effects were examined by presenting two visual scenes in each trial, which could either be identical images (no change), different views of the same place (viewpoint change), or different places (place change). In the second half of the experiment, cross-trial repetition effects were examined by presenting subjects with single scene images which could either be identical to those presented in the first half of the experiment (old view), previously-unseen views of the places shown in the first half of the experiment (new view), or previously-unseen places (new place). Consistent with previous results, within-trial repetition effects in scene-responsive regions were entirely viewpoint specific: response reduction was only observed when a place was repeated from the same view (no-change) but not when a place was repeated from a different view (viewpoint-change). Surprisingly, however, cross-trial repetition effects were primarily viewpoint invariant: strong response reduction was observed when places were repeated from different views (new view). These results raise the possibility that within-trial and across-trial fMRI repetition effects may be generated by different neural mechanisms, as suggested by recent neurophysiological studies (Sawamura et al., 2006). Alternative interpretations of these results will also be discussed.

Footnotes

 This research was supported by the following grants: NIH R01 EY-016464, Whitehall Foundation 2004-05-99-APL.