Abstract
Objects in the real world are rarely seen in isolation. Generally they are seen embedded in complex backgrounds. This reality necessitates the ability of recognition mechanisms to disregard background information and selectively respond to objects of interest. Our goal in this study is to examine how putative object-specific neural responses are influenced by the presence of background clutter. We focus specifically on the domain of faces since a rich body of work has accumulated regarding neural correlates of face recognition. The recording technique we use is magneto-encephalography (MEG). The M170 response of the occipito-temporal region of the human brain has previously been implicated in face recognition. However it is unknown how this response changes when faces are viewed against complex natural backgrounds rather than in isolation. We measured the amplitude and latency of the M170 for color faces embedded either in natural scenes or in homogeneous backgrounds with luminance values representative of the natural scenes. The same face stimuli were also presented with thin white halos separating the face from the background to probe the effect of segmentation cues. Our MEG data indicate that the latency of the M170 response is invariant to natural backgrounds. In contrast, the amplitude of the M170 response is decreased by the presence of a natural background but then restored by the presence a halo. Taken together, these results tentatively suggest that the M170 response might precede segmentation of a face from background, since a post-segmentation response would be expected to exhibit a greater amount of invariance to clutter than what we have observed in our data.