Self priming in face recognition

Abstract

Recently Burton, Bruce and Johnston (1990) have presented an interactive activation and competition model of face recognition. They have shown that this IAC model presents a parsimonious account of semantic and repetition priming effects with faces. In addition, a number of new predictions are evident from the model's structure. One such prediction is highlighted by Burton et al. themselves - that for short prime-target stimulus onset asynchronies (SOAs) a face should prime the recognition of a target name (or vice versa), 'self priming'. This thesis examined this prediction and found that it held for a design in which items were repeated across prime type conditions (same, associated, neutral and unrelated). Further, cross (face prime/name target) and within-domain (name prime/name target) designs were found to produce equivalent degrees of self and semantic priming (Experiments 1 and 2). Closer examination of the Burton et al. model suggested that the effect of domain equivalence for self priming should not hold for a design in which the stimulus items are not repeated across prime type conditions (i.e. subjects are presented with each item only once). This prediction was confirmed in Experiments 3, 4, 5 and 6.The time courses of self and semantic priming were investigated in two experiments where the interstimulus interval (ISI) between prime and target, and prime presentation times were varied. The results proved difficult to accommodate within the Burton et al. model, but it is argued that they did not provide a sufficient basis on which to reject the model. Finally, the self priming paradigm was applied to the study of distinctiveness effects. Faces judged to be distinctive in appearance were found to produce more facilitation than faces judged to be typical in appearance. Similarly, caricatured representation of faces were found to produce more facilitation than veridical or anticaricatured representations. The results of the distinctiveness studies are discussed in terms of the Valentine's (1991a; 1991b) exemplar-based coding model and Burton, Bruce and Johnston's (1990) IAC implementation. It is concluded that the results of these experiments lend support to the Burton et al. model.