Purchase this article with an account.
Olivia S. Cheung, Jennifer J. Richler, Thomas J. Palmeri, Isabel Gauthier; Revisiting the role of spatial frequencies in the holistic processing of faces. Journal of Vision 2007;7(9):506. doi: https://doi.org/10.1167/7.9.506.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Goffaux and Rossion (2006; G&R) reported that 1) holistic processing (HP) is supported by low- but not high-spatial frequencies (LSF, HSF) and 2) HP has a perceptual locus. We addressed each of these claims in two experiments using sequential matching tasks with face composites. In Experiment 1, observers judged whether target parts of two faces (e.g., tops) were the same while ignoring irrelevant parts (e.g., bottoms). In G&R, a ‘partial’ design was used, with irrelevant parts always being different. In our study, a ‘complete’ design was used where task-relevant and task-irrelevant parts could be same or different, enabling the measures of both sensitivity and response bias. This is important because previous studies using the complete design have reported differential response biases based on stimulus characteristics such as alignment and orientation (e.g., Wenger & Ingvalson, 2002). We replicated the G&R results using the partial design trials, finding greater evidence for HP with LSF compared with HSF. However, results using the complete design revealed that there was a significant difference in response bias for HSF compared with LSF, indicating that the effects in G&R arose from differential response biases. In Experiment 2, we used a version of the complete design where participants responded to both parts of the face, and again found significant differences in response bias for HSF and LSF. Critically, analyzing these results based on General Recognition Theory (Ashby & Townsend, 1986) revealed strong evidence for decisional holism, with less consistent perceptual contributions. In summary, our results show that 1) differences in HP between LSF and HSF reported by G&R can be accounted for by differences in response bias and 2) HP has a strong decisional component.
This PDF is available to Subscribers Only