Selection from visual persistence by perceptual groups and category membership.

Following Sperling, the nature of the representation of visual information during visual persistence has been investigated by comparing partial-report (PR) and whole-report (WR) estimates of available information. A PR superiority is considered evidence for the representation of the cued stimulus dimension in visual persistence. In general, PR cues based on a physical characteristic produce a PR superiority, but PR cues based on a category distinction give no higher estimates of available information than is obtained with WR. These findings have been used to support an interpretation of visual persistence based upon a storage system metaphor (e.g., iconic memory), whereby a critical characteristic of the stored information is its "literal" precategorical nature. The present experiments explored whether there are reasonable alternative explanations for the fact that only physical PR cues typical produced a PR superiority. Experiments 1 and 2 demonstrate that the effectiveness of physical PR cues depends upon the "goodness' of the perceptual groups defined by the cued dimension. Perceptual grouping within multi-letter displays was varied according to the principles of proximity (Exp. 1) and similarity (Exp. 2), and the results showed greater PR superiorities when the demand characteristics of the cues were compatible with the implied perceptual groups in the displays. Experiments 3 and 4 establish that PR cues based upon a category distinction (letter-digit) produce a PR superiority when both cue onset latency and cue uncertainty are equated across PR and WR conditions. Circular alphanumeric displays were used, and category PR cues and WR cues were either presented in separate trial blocks (Exp. 3) or intermixed at two possible cue delays relative to display onset (-1000 msec or 0 msec). A PR superiority was found in all conditions. In addition, Experiment 5 shows that the magnitude of this category PR superiority decreased systematically with increases in cue delay (-900 msec, -300 msec, +300 msec, and +900 msec), and in Experiment 6, it was found that the PR superiorities for both physical and category cues decrease at comparable rates with increased cue delay. Since perceptual grouping influences the effectiveness of physical PR cues and category PR cues produce a PR superiority under appropriate conditions, the results question the validity of interpretions of visual persistence that imply the existence of a literal, precategorical storage system. It is suggested that a multichannel view of the visual system provides a more adquate theoretical conceptualization of visual persistence.