A memory and decision model for eyewitness identification

A computer simulation model of eyewitness identification called WITNESS is proposed and fit to data from three experiments: Juslin et al. (1996), Tunnicliff and Clark (2000), and Wells et al. (1993). These three experiments directly compared two procedures for selecting foils for lineups: by selecting foils that either match the photograph of the suspect or match the description of the perpetrator given by the witness. The model assumes that (a) memory of a perpetrator is incomplete and error-prone, (b) lineup alternatives are matched to this error-prone memory trace, (c) identification decisions are based on a combination of relative and absolute match information, and (d) lineup rejections are based solely on absolute match information. The model provided good fits to data, although with important deviations between the model and data. Implications and limitations of the model, as well as future development of the model, are discussed. Copyright (C) 2003 John Wiley Sons, Ltd.