Abstract
Although holistic processing is thought to underlie normal face recognition ability, widely discrepant reports have emerged about this link in an individual differences context (Konar et al., 2010; Richler et al., 2011). Progress in this domain may have been impeded by the widespread use of subtraction scores, which typically lack validity due to their contamination with control condition variance. A more valid alternative is to regress, rather than subtract, control conditions from primary conditions (Peter et al., 1993). Using 43 participants, we measured the relationships amongst the Cambridge Face Memory Test (CFMT; Duchaine & Nakayama, 2006) and two holistic processing measures, the composite task (CT; Young et al., 1986) and the part-whole task (PW; Tanaka et al., 1993). For the CT and the PW, we contrasted the results for regressing versus subtracting out the control conditions (parts for the PW; misalignment effect for the CT) from the primary conditions (wholes for the PW; alignment effect for the CT). Both regression based measures of holistic processing correlated with the CFMT and with each other. Interestingly, the "complete design" subtraction measure of the CT used by Richler and colleagues (2011) correlated almost as highly with the CFMT as did our regression based measure, suggesting that its contamination with control task variance has a relatively modest end result. However, the subtraction based PW measure correlated neither with the CFMT nor with the CT, suggesting that the PW takes a major hit in validity when computed with a subtraction based approach. In sum, using a regression based approach, we confirm the hypothesized link between holistic processing and face recognition and demonstrate a theoretically crucial link amongst holistic processing measures. The present results illustrate the utility of a regression based approach for associating and dissociating aspects of human cognition via individual differences.
Meeting abstract presented at VSS 2012