Memory processes in familiar voice recognition

Research suggests that there are cross-modal differences in the retrieval of person specific information from faces and voices. Specifically, Hanley, Smith and Hadfield (1998) have shown that we are much more likely to recall the occupation and name from a celebrity’s face that we deem familiar than from their voice. In person recognition terms, a failure to ‘place’ a person is known as the familiar-only error. The aim of this thesis was to elucidate the cognitive factors that may be contributing to the large number of familiar-only responses that occur in voice recognition. Chapter 3 extended Hanley et al’s (1998) initial observation of a familiar-only bias in voice recognition with a new selection of stimuli (Experiments la and Id). The hypothesis that participants were poorly calibrated in their familiarity judgments with faces and voices was tested using a within-subjects design (Experiments lb, lc and le). Even under these testing conditions participants found the retrieval of person specific information to voices deemed familiar particularly problematic. However, Chapter 4 failed to produce a modality difference in occupational and name recall to familiar faces and voices when participants were required to make confidence judgments. Chapter 4 also used confidence judgments to generate ROCs and z-ROCs. This method raised the possibility that face recognition may entail a memory component that relies on recall whereas voice recognition may entail a familiarity process. However, methodological aspects of the study were not sensitive enough to provide strong support for this conclusion. Chapter 5 showed that applying a blur filter to faces can successfully reduce overall recognition performance to a comparable level with voices. This new blurred face condition was used as a matched control with voices in Chapter 6. The critical finding was that even though blurred faces and voices did not differ in overall performance, they differed significantly in terms of the memory processes that they rely on. Specifically, blurred faces are associated more strongly with remembering, whereas voices are more strongly associated with knowing (Experiment 4a and 4c). This difference can not be attributed to differences in subjective confidence (Experiment 4b). The results suggest that the cross-modal difference in person-identification can be understood in terms of the qualitatively different recognition memory processes that each modality relies on.