chapter  8
61 Pages

Face Perception: An Information Processing Perspective

The chapters in the first half of this book have focused almost exclusively on the representations used in facial perception. This research has resulted in the creation of computational models that specify the process by which pattern spaces are constructed from sensory input. These representational models have been shown to account for a wide variety of experimental data including distinctiveness (Valentine, chap. 3, this volume), similarity, typicality (Busey, chap. 5, this volume; Steyvers & Busey, chap. 4, this volume), and generalization across viewpoint and lighting (Edelman & O’Toole, chap. 10, this volume). The purpose of this chapter, in contrast, is to discuss mathematical models for processing the psychological evidence resulting from these pattern spaces (O’Toole, Wenger, & Townsend, chap. 1, this volume; Townsend, Solomon, & Smith, chap. 2, this volume). Our approach is aimed at tackling problems associated with the hypothesized rules and processes that operate on psychological evidence spaces (O’Toole et al., chap. 1, this

volume). Whereas representational models are mainly concerned with information, this work focuses on information processing. Information processing models of facial perception specify the operations or procedures by which facial representations produce behavioral decisions. Our assumptions concerning the nature of information processing have been formalized in a mathematical model called the fuzzy logical model of perception (FLMP).