Abstract
We studies gaze dependence of adaptation to faces and scenes using fMRI. Previous studies indicate face adaptation is non-retinotopic, but its gaze dependency has not been systematically examined. In Experiment 1 we demonstrated robust adaptation to foveal faces and scenes in FFA and PPA, respectively, using a psychophysical paradigm known to induce identity aftereffect: after 4 sec adaptation to a face(scene) followed by 0.8 sec blank, either a different, or an identical image of the same category (probe) was presented for 0.8 sec. Observers reported if adaptor and test are the same. A different probe evoked significantly larger BOLD response in FFA(PPA) than the same image. In Experiment 2, adaptor and probe were displayed in periphery (7.2 deg, same hemifield), and we measured BOLD activity in the contralateral area. Adaptation to faces, but not scenes, was considerably weaker than in Experiment 1. Experiment 3 examined whether gaze affects the adaptation phase, or the subsequent testing (probe). A foveal adaptor resulted in significant aftereffect in periphery. Almost no adaptation occurred when adaptor was peripheral and probe was foveal. No such dependency was found for scenes. The BOLD activities evoked by peripheral and foveal faces/scenes were about the same. Thus, different activation is ruled out as a confound. We conclude that adaptation to faces but not scenes depends on overt attention during adaptation. These results are unlikely to be due to the reduced resolution in the periphery as scenes contained more high-spatial frequency information than faces.