Abstract
We used fMRI to study the modulation of retinotopic maps in human cortex by top-down and bottom-up attention. We presented point-light biological motion or scrambled control stimuli in a phase-encoded polar angle mapping paradigm. The background of the stimuli was either empty, or contained the opposite stimulus type. Subjects fixated and always ignored the peripheral stimuli as they performed a low-load (respond on red cross) or a high-load (respond on upright yellow or inverted green cross) task at fovea. Even in the absence of attention, and the presence of competing stimuli in the rest of the visual field, multiple brain areas, primarily in occipital and temporal cortex, still responded retinotopically to the stimuli. There was no effect of attentional load on fovea on these responses; i.e., the stimuli drove retinotopy regardless of how strongly attention was directed elsewhere. Consistent with Saygin and Sereno (2008), parietal and frontal maps were not strongly active in the absence of top-down attention to the stimuli. Although stimulus effects on retinotopic responses were subtle (and could easily be swamped by responses to the mere presence of stimuli, hence our background manipulation), lateral temporal retinotopic regions, including the superior temporal sulcus (STS), responded preferentially to biological motion, even when these stimuli were task-irrelevant. A similar pattern was seen in primary visual cortex. Overall, it appears that parieto-frontal retinotopic maps are highly sensitive to, possibly even dependent on top-down attention. Occipital and temporal maps on the other hand, maintain stable retinotopic representations under a variety of conditions. Lateral temporal areas including the STS show preferential responses to the biological motion stimuli even when they are task-irrelevant. Thus, retinotopic maps in different areas have distinct functional properties, likely reflecting their roles in real life vision and attention, for which both flexible and stable representations of space are needed.
Supported by the following grants: Marie Curie Intra-European Fellowship from the European Commission, University Research Fellowship from City University, London and Academic Senate grant from University of California, San Diego. We thank Marty Sereno, Marc Tibber, and the Wellcome Trust Centre for Neuroimaging, London.