Face memory : the role of configural and holistic processing in face recognition

Abstract

The study of human face recognition requires us to understand how faces are perceived and then later remembered. Previous studies have often taken a perceptual perspective to show that faces are perceived more holistically than other objects. This dissertation investigates the role of configural and holistic processing of faces in the context of memory. Three series of behavioral experiments were conducted. In the first series of experiments, recognition memory of faces was examined. Three classical face recognition tasks (inversion, spatial frequency, viewpoint) were assessed using the remember/know procedure to examine how different transformations of the face influence recollection and familiarity in recognition memory. An inversion effect was observed with remember responses but not know responses. The poor observed performance with inverted faces is indicative of a disruption of holistic processing. I then separated different spatial frequencies from faces (high spatial frequencies, or HSF, and low spatial frequencies, or LSF) to examine how recollection and familiarity interact with configural and featural processing of faces. No differences were found in recollection of HSF and LSF faces. A type of contextual information about a face, viewpoint, was manipulated and I found higher remember responses to front view faces than three-quarter views, with profile views lower still. This series of experiments showed recollection is not particularly sensitive to configural processing of faces. In the second series of experiments, the importance of configural processing for face recognition was studied. A new methodology – a face puzzle task – is proposed to test memory of a face configuration. Participants were no more accurate at re-constructing the configuration of a previously studied face than they were at estimating the configuration for an unstudied faces. This series of experiments demonstrate that configural information is not used by people to memorize a face. In the third series of experiments, the nature of holistic processing was investigated. Some researchers have suggested that holistic processing of faces is sensitive to detecting changes but is poor at localizing changes. Common perceptual manipulations (the composite effect and the part-whole effect), which served as indices of holistic processing, were examined using both change detection and change localization tasks. Aligned face composites interfered with change detection but did not affect change localization. However, an advantage for whole faces was found in both the change detection task and the change localization task when compared with scrambled faces. This indicates that holistic processing of whole faces does not always reduce change localization ability. This also shows that the composite face and part-whole paradigms may be measuring different constructs of holistic processing. In the end of this dissertation, I also explore other information, such as face shape and texture, that are involved in holistic processing and discuss the holistic mechanism in face perception and memory.