Abstract
Faces are processed holistically, which is often measured using the composite paradigm, a matching task requiring selective attention to part of a face. One popular index of holistic processing (HP)–the alignment effect used in the partial design–is confounded with response biases, whereas a different measure–the congruency × alignment effect used in the complete design–produces a more valid measure of HP. Because the two measures can yield different conclusions, we re-visit the role of HP in two phenomena where the complete design has not yet been used: the face-inversion effect (FIE) and the other-race effect (ORE).
Recognition of inverted faces (Yin, 1969) or upright faces of an unfamiliar race (Meissner & Brigham, 2001) is often impaired, with a reduction in HP posited as the basis of reduced performance (Rhodes et al., 1989; Hole, 1994). However, support for this claim has been mixed (Sekuler et al., 2004; Stokes et al., VSS 2010) and composite studies of these effects have only used the partial design. Here we obtain categorically different conclusions regarding the contributions of HP to the FIE and ORE depending on how HP is measured.
When investigating the FIE via the composite paradigm using the partial design, HP was only observed for upright but not inverted faces. With the complete design, however, inverted faces were also processed holistically at longer exposure durations. Similarly, when Caucasian and Asian participants were tested with same- and other-race faces, the partial design failed to capture an ORE in HP. In contrast, a significant ORE was observed using the complete design. Additionally, in both experiments, only partial design measures correlated with response bias.
HP was reduced but not abolished for other-race faces and delayed for inverted faces, which is consistent with reduced processing efficiency when objects of expertise depart from familiar conditions.
Supported by a grant to the Perceptual Expertise Network from the James S. McDonnell Foundation, the Temporal Dynamics of Learning Center (SBE-0542013), an NSF Science of Learning Center, an NEI award 2 R01 EY013441 and a grant from the Vanderbilt Vision Research Center (P30-EY008126) to IG, and a grant from the Hong Kong Research Council (HKU 744209) to WH.