Isolating the interference caused by cue duration in partial report: a quantitative approach

In bar-probe partial report experiments, subjects are presented with a brief array of letters, followed by a cue that singles out a target letter. Using this procedure, V. M. Townsend (1973) reported a counterintuitive effect: As the duration of a cue was increased, target performance decreased dramatically. The aim of the present study was to isolate the locus of the cue-duration effect. To this end, several characteristics of the bar-probe display were manipulated in a single experiment: the interstimulus interval between the array and the cue, array density, the number of letters, and the number of symbols adjacent to the target. These factors were manipulated on a priori grounds so as to affect the different sources of information used in the bar-probe task--namely, durable storage, abstract identity information, and feature level information. The data were accurately fit by a simple quantitative, multinomial model that assumes that the different sources of information used in the bar-probe task make independent contributions to performance. The critical assumption of the model is that cue duration interferes with only one source of information--namely, feature level information.