Abstract
Are configural and holistic face processing distinct? There is a lack of consistency in terms and definitions across the literature, and it is not clear whether and how these processing types are distinct from one another. One way to investigate this is to examine patterns of performance across face recognition tasks that are thought to tap into one or the other processing type. For instance, performance patterns on the part-whole task and the composite face task are both thought to reflect holistic processing, so performance on them would be expected to positively correlate. Similarly, the face inversion effect and the configural/featural effect are both thought to arise due to eliciting deficits in configural processing, so performance patterns in these two tasks would be expected to correlate. To examine whether this pattern of correlations exist, we compared performance within-subjects (N=70) across these four commonly-used face perception tasks: face inversion, part-whole, composite face, and configural/featural. Performance data were calculated in terms of reaction time, accuracy, and efficiency scores (the sum of normalized accuracy and reaction time measures). This was done to address the fact that the various tasks might express their effects in terms of RT, accuracy, or some mixture of the two. Results revealed that performance data from the conditions within a given task are strongly correlated with each other. However, there was no evidence of correlations between different tasks that might suggest that they tap into the same mechanisms. A preliminary PCA analysis suggested a similar result, with one factor emerging for each task, and the four conditions within each task loading strongly onto that factor only. Thus, our data suggest that each of these four tasks measures a distinct face processing capacity, rather than tapping into common holistic or configural mechanisms.
Meeting abstract presented at VSS 2013