Type of Document Dissertation Author Bennett Day, L. Brooke URN etd-07052007-153108 Title The Role of Processing Strategy in the Cross-race Effect Degree Doctor of Philosophy Department Psychology, Department of Advisory Committee
Advisor Name Title E. Ashby Plant Committee Chair Albert Stiegman Committee Member Colleen Kelley Committee Member Dianne Tice Committee Member Jon Maner Committee Member Keywords
- Processing Strategy
- Cross-Race Effect
- Eyewitness Memory
- Face Recognition
Date of Defense 2007-06-27 Availability unrestricted AbstractThe cross-race effect (CRE) states that people are better able to accurately recognize faces of their own race as opposed to faces of other races. Although this finding has been studied extensively over the past 35 years, there still exists some disagreement over the primary cause of the effect. One of the suggested mechanisms involves processing strategy as it relates to face recognition. A configural processing strategy takes into account the whole face, including the spatial relationships between features, while a featural processing strategy simply focuses on the separate features. Objects with which people have more expertise (such as faces) tend to benefit from a more configural processing strategy. Thus, CRE researchers have suggested that the effect may exist due to a differential use of processing strategy, with a more configural strategy being used for same-race faces when compared to cross-race faces.
The present research examined the involvement of configural and featural processing in the CRE for Black and White participants by using Navon stimuli (large letters composed of smaller letters) to either manipulate or measure processing strategy. Both studies included the signal detection measures of hits, false alarms, A' (a measure of discrimination accuracy), and B"D (a measure of response bias). Experiment 1 examined the efficacy of a processing strategy training task in reducing the CRE. Participants (140 White) completed one of three tasks (configural processing, featural processing, or control) either prior to or directly after encoding a series of Black and White faces. Participants were later asked to recognize those faces from a larger series of old and new faces. It was predicted that training with configural stimuli would increase cross-race face recognition, whereas training with featural stimuli would decrease same-race face recognition. This predicted result did not occur, but participants in each training condition did show significant improvements in their cross-race face recognition with regard to hit rate. Participants who completed any of the tasks prior to the encoding portion of the study demonstrated more cross-race accuracy. There were also significant correlations between response criterion and accuracy rates that suggested participants who reported more cross-race experience were using a more liberal criterion for recognition.
Experiment 2 was designed to examine the type of processing strategy used by participants when viewing same and cross-race faces, with a secondary goal of comparing processing strategy to the extent of the appearance of the CRE. Participants (42 Black, 51 White) were presented with a series of Black and White faces, each of which was followed by a Navon stimulus. Response latencies to configural and featural stimuli were examined as a function of the race of the face preceding the stimulus. A recognition test for these faces was included to provide a measure of participants' CRE. It was predicted that participants would show quicker responses to featural stimuli that followed cross-race faces than same-race faces, and configural stimuli that followed same-race faces than cross-race faces. Although participants did not demonstrate the predicted effect, they did show a general configural bias. An examination of the relationship between participants' CRE and outcome measures suggested that participants who were quicker to respond to processing stimuli were also likely to have a more liberal response criterion, perhaps relying more on familiarity than recollection for their recognition judgments. Explanations for these findings were discussed, along with suggestions for future research.
Filename Size Approximate Download Time (Hours:Minutes:Seconds)
28.8 Modem 56K Modem ISDN (64 Kb) ISDN (128 Kb) Higher-speed Access Bennett DayLDissertation.pdf 868.05 Kb 00:04:01 00:02:04 00:01:48 00:00:54 00:00:04