Abstract
In the face processing literature, holistic/configural (HP) has classically been dissociated from featural processing (FP) (Sergent, 1984). HP, the interactivity of feature processing, promotes the generally observed efficiency in recognizing/discriminating individual faces and forms the basis of phenomena such as the whole-part advantage (Tanaka & Farah, 1993) and the composite-face effect (Young et al., 1987). FP, characterized by a lack of such interactivity, renders a local, serial processing style suboptimal for face processing (as seen in acquired prosopagnosia). Past investigations have assessed HP/FP behaviorally, by e.g. discrimination/recognition of features embedded in the facial context, or presented in isolation. Here we suggest that the extent to which HP/FP is engaged in varies depending on the information that can be derived from full-face stimuli, and furthermore can be assessed by means of fixation patterns. Participants' eye-movements were recorded during a delayed face-matching task. The stimuli were morphs created from personally familiar/unfamiliar faces that were either easily discriminable (50% difference), or more difficult (20%, 50% blurred faces). The rationale was that greater similarity (20%) would decrease HP and increase reliance on individual features (FP). Contrariwise, HP should increase with dissimilarity, or when featural information is unavailable, as given for blurred faces (Collishaw & Hole, 2000). For easily discriminable faces, participants fixated the face-center, below the eyes (Hsiao & Cottrell, 2008). However, the individual features (eyes, mouth) were fixated more with increased similarity. The distributed nature of fixation patterns, along with more fixations, cannot be attributed to lower performance, as this pattern was not found for blurred faces. Here, despite decreased performance, fixations remained even more centrally located, as if seeing the whole face from the central point was the optimal strategy. Our results are the first to demonstrate that stimulus quality and similarity can determine processing style, which is directly linked to the observed pattern of eye-gaze fixations.