Categorical perception of face identity in noise isolates configural processing.

Abstract

Neuropsychological evidence suggests that face recognition based on configural (holistic) information can occur in isolation from recognition based on local feature cues. The present study shows that configural processing can be isolated experimentally in normal subjects. A phenomenon is reported that exists only for upright whole faces, namely categorical perception (CP) of face identity in noise. Three discrimination tasks (ABX, better likeness, and similarity ratings) were used to test for perceptual distortion across the category boundary predicted from binary classification of face morphs. Noise was added such that any single local region provided unreliable cues to identity. Under these conditions, CP was found for upright faces but not for inverted faces or single features, even with more than 10,000 trials. The CP-in-noise signature phenomenon was then used to show that configural processing survives image plane rotations of 45 degrees-90 degrees.

Dissociations of Face and Object Recognition in Developmental Prosopagnosia

“…Complementing these cases of prosopagnosia are case studies demonstrating that the converse dissociation, spared face recognition with severe object agnosia, can also exist (McMullen, Fisk, & Phillips, 2000;Humphreys & Rumiati, 1998;Moscovitch, Winocur, & Behrmann, 1997). In addition to case studies, neuroimaging (Kanwisher, 2000;McCarthy, Puce, Gore, & Allison, 1996), neurophysiology (Kreiman, Koch, & Fried, 2000;Gross, Rocha-Miranda, & Bender, 1972), and psychophysics (McKone et al, 2001;Tanaka & Farah, 1993;Young et al, 1987) all provide evidence that faces and objects are processed by different mechanisms. Many hypotheses have attempted to characterize the domain of the recognition mechanisms used to process faces (Gauthier et al, 1999;Moscovitch et al, 1997;Farah, 1991;Levine & Calvanio, 1989;Diamond & Carey, 1986), but past cases of prosopagnosia and our current study do not provide a definitive answer.…”

“…Holistic processing is characterized by the integration of facial information into a gestalt (Tanaka & Farah, 1993;Young et al, 1987), whereas configural processing usually describes sensitivity to the precise spatial layout of the facial features (Freire et al, 2000;Diamond & Carey, 1986) or features in the context of a facial image (McKone et al, 2001;Tanaka & Farah, 1993). Some evidence indicates that the mechanisms which carry out these computations are dissociable neuropsychologically (Le Grand et al, 2001;Barton et al, 2002), developmentally (Carey & Diamond, 1994;Mondloch et al, 2002), and behaviorally (Freire & Lee, 2000;Le Grand et al, 2001), though recent behavioral and neuroimaging work has questioned this conclusion (Yovel & Kanwisher, in press).…”

“…Object recognition typically involves feature processing (Biederman, 1987), but face recognition also involves holistic processing and configural processing (McKone, Martini, & Nakayama, 2001;Freire, Lee, & Symons, 2000;Tanaka & Farah, 1993;Young, Hellawell, & Hay, 1987). Holistic processing is characterized by the integration of facial information into a gestalt (Tanaka & Farah, 1993;Young et al, 1987), whereas configural processing usually describes sensitivity to the precise spatial layout of the facial features (Freire et al, 2000;Diamond & Carey, 1986) or features in the context of a facial image (McKone et al, 2001;Tanaka & Farah, 1993).…”

Higher-level mechanisms detect facial symmetry

“…The advantage for facial symmetry detection was also narrowly tuned to vertical, disappearing with a 458 tilt. In contrast, other face-coding mechanisms continue to be engaged by misoriented faces up to 908 (Moscovitch et al 1997;Murray et al 2000;Sturzel & Spillman 2000;McKone et al 2001), again suggesting distinct mechanisms for detecting facial symmetry and other aspects of face perception.…”

Cross-species differences in color categorization

“…Rather than teach monkeys the range of colors to be labeled green or blue, we simply asked whether there might be a natural boundary within that range. Thus, in the present study, we examined blue and green categories in the simplest procedure inclined to produce categorical responding (McKone, Martini, & Nakayama, 2001); we then modeled the data to verify the underlying category boundaries.…”