Abstract
Holistic processing for faces (and objects of expertise) can be measured in tasks where subjects are instructed to selectively attend to a part (e.g., the mouth) of two objects to compare them and told to ignore the other parts. Holistic processing is evidenced by a congruency effect between the correct response on the target and irrelevant part. In addition, configural processing can be operationally defined as a reduction of this congruency effect when the two parts are misaligned. Most accounts of holistic and configural effects postulate a representational locus as the basis for these phenomena (e.g., Farah et al., 1998) but recent work using formal modeling argues for a decisional locus (Wenger & Ingvalson, 2002). In several experiments using a sequential matching paradigm, we investigate the importance of the configuration at encoding. On each trial, a first stimulus (S1) is encoded, followed by a cue indicating which part is the target (top/bottom half), followed by a second stimulus (S2) on which a same/different judgment is made. Different groups were tested with aligned or misaligned S1 stimuli, with either faces or objects (asymmetrical Greebles). For objects, we find a congruency effect for both aligned and misaligned S2s when the S1 is misaligned, but not when the S1 is aligned. This suggests that for novel objects, requiring subjects to attend to misaligned S1 parts can reduce the subjects' ability to ignore any irrelevant part on S2 stimuli. For faces, we find strong congruency and configural effects at S2 for both aligned and misaligned S1 faces, indicating that encoding configuration had little effect on holistic processing. These results are consistent with the idea that the locus of holistic and configural effects is not representational and could result from attentional biases. Alternatively, to account for these results the notion of holistic encoding could include the capacity to construct a holistic template from disjunct parts.