I can't recognize your face but I can recognize its movement

Abstract

Idiosyncratic facial movements can provide a route to facial identity (review in Roark, Barrett, Spence, Abdi, & O'Toole, 2003). However, it is unclear whether recognizing a face in this way involves the same cognitive or neural mechanisms that are involved in recognizing a static face. Three studies on a developmental prosopagnosic (C.S.) showed that although he is impaired at recognizing static faces, he can discriminate between dynamic identities (Experiments 1a and 1b) and can learn to name individuals on the basis of their idiosyncratic facial movements (Experiment 2), at levels that are comparable to those of matched and undergraduate control groups. These results suggest a possible cognitive dissociation between mechanisms involved in dynamic compared to static face recognition. However, future work is needed to fully understand this dissociation.

Notes

¹ Because we were only interested in the extent to which C.S. was using facial information to recognize famous faces, we did not test a group of controls. Several studies have already shown that normals can accurately recognize and match famous faces when they are limited to internal features (Clutterbuck & Johnston, 2005; Ellis, Shepherd, & Davies, 1979; Young, Hay, McWeeny, Flude, & Ellis, 1985).

² Inverted faces are not believed to engage the same mechanisms involved in face processing as do upright faces (Diamond & Carey, 1986; Scapinello & Yarmey, 1970). Normal face processing (i.e., upright faces) depends on both configural and featural processing. Face inversion can interfere with both types of processing (Maurer et al., 2002; Riesenhuber, Jarudi, Gilad, & Sinha, 2004; Yovel & Kanwisher, 2004). For normal participants, inversion decreases performance in face recognition by 15–20% (Diamond & Carey, 1986; Scapinello & Yarmey, 1970; Yin, 1969). For many prosopagnosics, inversion does not affect face-processing performance (Boutsen & Humphreys, 2002; Gauthier, Behrmann, & Tarr, 1999; Marotta, McKeeff, & Behrmann, 2002) because they are impaired at configural and/or featural processing.

³ Blurring a face removes high spatial frequency information that conveys information about individual facial features and forces participants to use configural or relational information for face processing (see Collishaw & Hole, 2000).

⁴ We are very grateful to Harold Hill and Alan Johnston for allowing us to use this stimulus set.

⁵ Given that participants were required to give one of three responses on each trial (1, 2, or 3), and each response appeared equally often (7 times for each response), chance level performance corresponded to 33%.

⁶ In 3dMeNow, the developer uploads images of a face (front and/or profile views), and application tools are used to place markers around the main features of each facial image (outer contour of the head, hairline, eyes, nose, and mouth). Upon placing these markers, the software computes the structure of the head and produces a 3D model of the face.