Abstract
Recent behavioral studies have shown that in addition to three-dimensional (3D) shape, two-dimensional (2D) surface reflectance information (color and texture) is important for perception of facial identity (e.g., O'Toole et al., 1999; Lee & Perrett, 2000; Jiang et al., 2006; Russell et al., 2006, 2007). However, it is unclear how shape and surface reflectance of faces are processed in the human brain.
In the present study, we used a long-term event-related fMRI adaptation paradigm to identify brain regions that encode these two types of facial information. Following 3-second adaptation to an adapting face, participants were asked to match the identity of a briefly presented (300 ms) test face and the adapting face. Using a 3D morphable model (Blanz & Vetter, 1999), we manipulated the shape and surface reflectance properties of the test stimulus with respect to the adapting stimulus. Four test conditions were included: (1) repetition of the same adapting face; (2) variation in 3D shape only; (3) variation in 2D surface reflectance only; (4) variation in both 3D shape and 2D surface reflectance. We also implemented a 15-degree viewpoint change between the adapting and test faces to reduce the contribution from low-level features.
We found that changes in either 3D shape or 2D reflectance affected participants' matching performance. Neurally, despite the change in viewpoint, repetition of the same face identity induced significant activation suppression in the functionally localized right and less significantly in the left fusiform face areas (“FFA”). Release from adaptation occurred when changes in either 3D shape or 2D surface reflectance were introduced to the test stimulus, but changes in the two properties induced the strongest adaptation release. Combined, our results indicate that both face shape and reflectance information are coded for face identification in face-sensitive areas. Moreover, these two types of facial information may share overlapping neural representation.