Purpose: We aimed to delineate parietal and occipital cortical activity in humans for self-initiated saccades decoupled from concurrent visual input. In particular we wished to define the topographical distribution in reference to retinotopic and spatiotopic coordinate systems.
Methods: Prior to every scan the subjects were presented with a fixation cross that was horizontally displaced in an outline face between nose, left and right eye. During the fMRI-scans subjects (2 males and 2 females) performed voluntary saccades in total darkness to the previously trained positions in the imagined face. Eye movements were recorded with EOG. The latency and the direction of the saccades served to define the event related fMRI-model. The BOLD-responses to the same saccades were contrasted in reference to retinotopic space (saccades directed leftwards vs. rightwards to the current fixation) and egocentric space (saccades within left vs. right head/body centered space). The two spaces are identical for centrifugal but not for centripetal saccades.
Results: Retinopically defined perisaccadic activity was consistent in the contralateral (to saccade direction) parietal cortex. The activation in V1 was ipsilateral and in two locations: at the fovea and at extrafoveal representations.
Conclusions: Previous fMRI data show that V1 does not code for imagined saccades (Bodis-Wollner et al., 1997). Our study shows that V1 activity codes for processes associated with fixational and with eye movement processes when self-initiated saccades are decoupled from visual input. V1 activity is referenced to body centered coordinates, while the parietal cortex, consistent with other studies, codes the activity in retinotopic coordinates.