Repetition priming between parts and wholes: Tests of a computational model of familiar face recognition

Andrew W. Ellis*, A. Mike Burton, Andy Young, Brenda M. Flude

*Corresponding author for this work

Research output: Contribution to journalArticleResearchpeer-review

18 Citations (Scopus)

Abstract

Four experiments are reported which assess repetition priming between parts of famous faces and whole faces. Their results are compared with the predictions of a computational model based on principles of interactive activation with competition (IAC). Experiment 1 showed that familiarity decisions to whole-face photographs were primed more by prior recognition of the whole face than by recognition of part of the face (the internal features). Experiments 2 and 3 found that priming from recognizing one part of a face to recognizing a different, non-overlapping part was less than the priming observed from one part to itself. These results are as predicted by the IAC model. That model also predicts that the recognition of part of a face will be primed more by prior recognition of the whole face than by recognition of the same part, but Expt 4 found more priming from a part to itself than from the whole to the part. This result can be accommodated within a revised IAC model if at least some missing features are treated as adding noise to the recognition process rather than simply a lack of input. We suggest that this treatment of missing features is compatible with the notion of global features suggested by the concepts of wholistic coding. The ability of alternative, episode-based theories of repetition priming to explain the obtained results is discussed.

Original languageEnglish
Pages (from-to)579-608
Number of pages30
JournalBritish Journal of Psychology
Volume88
Issue number4
DOIs
Publication statusPublished - Nov 1997
Externally publishedYes

Fingerprint

Dive into the research topics of 'Repetition priming between parts and wholes: Tests of a computational model of familiar face recognition'. Together they form a unique fingerprint.

Cite this