Purchase this article with an account.
Mitchell Meltzer, Gowtham Ganesan, Michelle Min, James Bartlett; Inversion, Configural Recognition, and Part Recognition in Long-Term Memory for Faces. Journal of Vision 2017;17(10):1026. doi: 10.1167/17.10.1026.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Investigating the role of holistic processing in face recognition memory, Meltzer et al. (VSS, 2016) presented participants with a study list followed by a test containing intact (studied) faces, conjunction faces (recombinations of studied face parts), feature faces (containing one old part and one new part), and entirely new faces. In response to each test face, participants made separate recognition judgments for the top and bottom face-halves, and for the face as a whole. Face-half judgments and whole-face judgments produced equivalent levels of intact-conjunction discrimination, suggesting that such discrimination is based on a uni-dimensional familiarity signal as opposed to a flexible recollection process. However, when holistic processing was disrupted by inverting the faces and misaligning their top and bottom halves, whole-face judgments produced higher intact-conjunction discrimination than half-face judgments, suggesting a flexible recollection process that can be controlled to meet to task demands. In support of this conclusion, the conditional probability of recognizing the parts of conjunction faces and feature faces, given their rejection as new whole faces, was low with well-formed faces, and significantly higher with inverted-misaligned faces. Meltzer et al. conjectured that holistic face processing has a "dark side," forcing participants to recognize faces based on global familiarity, preventing flexible recognition of parts as well as wholes. To further test this hypothesis, a new experiment compared performance with upright and inverted-but-well-aligned faces. Based on evidence that inversion alone disrupts holistic processing (McKone et al., 2013), we predicted that the effects of inversion would resemble the effects of inversion and misalignment combined. In fact, this was the case, but the effects were smaller and statistically weak. This might indicate that inversion disrupts but does not eliminate holistic processing, or that familiarity-based responding reflects unitization rather than holistic processing, and that inversion impairs unitization without necessarily eliminating it.
Meeting abstract presented at VSS 2017
This PDF is available to Subscribers Only