Abstract
It has been claimed that identification of a face relies more on the perception of the whole face than the perception of its constituent parts (e.g., eyes, nose, mouth). The face inversion, face composite, and parts/wholes tasks are well-established paradigms demonstrating the holistic perception of facial identity. However, less is known about the cognitive mechanisms mediating recognition of facial expressions. In these experiments, we applied established paradigms to test the holistic processing of emotional expressions. The stimuli were face halves displaying the same (e.g., happy top/ happy bottom) or different expressions (e.g., angry top/happy bottom). In all tasks, participants were instructed to attend to and label the expression (i.e., ‘angry’ or ‘happy’) in the top (or the bottom) half of the face and ignore information in the irrelevant half. When labeling inconsistent expressions, evidence of holistic interference effects were found such that recognition of the expression in the attended half was affected by conflicting information in the irrelevant face half. Interference effects were reduced or abolished when the faces were presented upside-down, as spatial or temporal composites or as isolated parts. However, there was less evidence of facilitative effects in the consistent condition where the inversion, composite and parts manipulations had little effect on performance. Collectively, these results suggest that in situations where information about the facial expression is clear and unambiguous, processing is more analytic and less dependent on information from the whole face. However, in cases, where the information is conflicting or ambiguous, holistic face processes are engaged.