| Despite much progress in our understanding of the factors affecting eyewitness memory for faces, attempts to improve face memory have been unsuccessful. This dissertation presents a series of studies that were designed to develop procedures for training eyewitnesses to be more accurate in identifying criminal suspects.; Study 1 was designed to identify encoding strategies associated with recognition accuracy. Participants in this study completed the Benton Facial Recognition Test twice. The first administration was used to measure recognition accuracy and the second to measure encoding strategies, in which participants thought aloud as they did the test. Their verbal reports were coded for evidence of ten different encoding strategies. Holistic strategies such as the encoding of stereotype information, physical attractiveness, facial asymmetries and geometric shapes were related to accuracy, but feature encoding strategies were not. Emotional empathy correlated positively with accuracy whereas vividness of visual imagery and accuracy were not correlated. Although observers typically encode faces on only a few dimensions, those encoding faces along more dimensions were most accurate.; Studies 2 through 4 explored whether training in encoding facial asymmetries and geometric shapes improved face recognition. Across two tests of face recognition and two encoding tasks, results did not differ. Neither facial asymmetry encoding nor encoding of geometric shapes improved recognition accuracy. Those encoding geometric shapes performed consistently worse than the control group.; Study 5 tested the efficacy of instructing observers to encode multiple dimensions. The results showed that individuals who were instructed to encode a combination of stereotypes, physical attractiveness and facial asymmetries were significantly more accurate than a no-training control group.; Overall, the results suggest that encoding along multiple, holistic dimensions results in greater recognition accuracy than other types of encoding. The findings of this dissertation also support prototype matching theories of face recognition. |
