Abstract
A wealth of research has addressed the nature and location of retinotopic (eye-centered) maps (in occipital, temporal, parietal and prefrontal cortex), which show egocentric coding of a hemifield or quarterfield of visual space (for review, see Wandell et al., 2007). However, few studies have examined allocentric, non-lateralised reference frames, such as those posited for object-based coordinates, which can be affected by hemispatial neglect (Behrmann and Tipper, 1999). We ran an fMRI study to compare eye-centered (retinotopic) and object-centered (objectotopic) reference frames during an attentional experiment, where participants monitored the motion of a variable number (1 to 4, out of 6) of target gratings (defined by colour) and detected transient changes in their velocity. Three of the gratings were located inside each of two discs at either side of fixation, providing object-based grouping. Multi-Voxel Pattern Analysis (MVPA) dissociated retinotopic and objectotopic spatial coding in eye-centered and object-centered coordinates independently. This revealed extensive hemifield-specific retinotopy in occipital and parietal cortex, but no sign of objectotopy. Furthermore, univariate analyses revealed substantial independence in the load-scaled activity in each hemisphere, consistent with previous behavioural reports (Alvarez and Cavanagh, 2005). The current dissociation suggests that retinotopic reference frames are automatically engaged during attentional monitoring, but remains equivocal regarding objectotopic reference frames. We discuss several possibilities. Perhaps object-based reference frames do not exist and are unnecessary to explain clinical findings (Mozer, 1999). Alternatively, object-based reference frames are only activated by specific task requirements.
Medical Research Council.