Abstract
It is commonly accepted that face perception relies on middle spatial frequencies around 8 cycles per face-width. However, it has recently been shown that the spatial frequencies critical for face recognition is not fixed but rather depends on size (Oruc and Barton 2010). This effect is especially pronounced at small sizes, where recognition is mainly based on low spatial frequencies (i.e. gross forms in the image), compared to larger sizes where middle frequencies, which include finer details, are used. This scale-dependent recognition strategy has been shown in the context of just-visible face images. However, whether this would hold for supra-threshold perception is not clear. We address this question using an adaptation paradigm. We examine face identity aftereffects using blurry and intact adaptors of two different sizes: small (2 deg) and large (10 deg). The blurry adapting faces included the low spatial frequencies hypothesized to be used at the small size, but not the higher portion of the spectrum used at the larger size. Intact adaptors induced significant aftereffects regardless of size. The blurry adaptor similarly produced aftereffects that were as large as the intact adaptors, but only at the small size. In contrast, the large blurry adaptor failed to generate any aftereffects. Since the large blurry adaptor was physically identical to the small one and therefore contained the same identity information, the lack of obtaining any identity aftereffects suggests that the observers were not able to access this low-frequency information for the purposes of perceiving identity specifically at this large size. Overall, these results confirm that size is a determining factor in processes of face perception. Importantly, our results suggest that degraded blurry faces are easier to recognize at smaller sizes (i.e. at farther viewing distances).
Meeting abstract presented at VSS 2013