Effect of picture-word transfer on accuracy and latency of recognition memory.

Abstract

Temporal parameters of recognition memory for pictures and words were investigated in a reaction time experiment. Dark-adapted 5s self-initiated 250msec. exposures of arrays containing eight different stimuli-either pictorial silhouettes or corresponding block-lettered verbal labels-in the peripheral cells of a 3X3 grid. Following every array and a subsequent 3-sec. delay interval, a test item--again, either a four-letter word or a picture-appeared in the center grid square for 250 msec. By quickly pressing a yes or no button, 5s indicated whether the test item had been present or absent in the preceding array. Each 5 served under all of the following conditions: word array-word test, picture-picture, picture-word, and word-picture. Order of presentation was completely randomized for each 5. Analyses of both accuracy and reaction time data revealed interactions between mode of presentation, mode of test, and decision strategy. Such results may be partially dependent on ability to form a visual memory image of the array, though further active encoding processes may be responsible for subsequent disruption of that image.Recent accounts of verbal memory have often included as the first of several discrete stages of stimulus processing a peripheral image, or icon, which is directly related to the stimulus input but is extremely short-term. Sperling (1960) demonstrated that 5s have rather large-capacity visual memories that rapidly decay over intervals up to 1.0 sec. He proposed a model that attributes the persisting memory for visual materials which occurs in spite of rapid icon decay to the operation of verbal encoding processes (Sperling, 1967).Similarly, Posner and Mitchell (1967) hypothesized that peripheral visual images may be displaced by nonvisual codes. They found that same response latencies were longer for different-case letters, i.e., uppervs. lowercase, than for identical-case letters at interstimulus intervals (ISIs) below 1.5 sec., but at ISIs of 1.5-2.0 sec., the advantage of similar-case decisions disappeared. Posner, Boies, Eichelman, & Taylor (1969) demonstrated that an interpolated mental addition task eliminated 1 This article is taken in part from a thesis submitted by the first auther in partial fulfillment of the requirements for the master's degree to the University of Arizona.2 Requests for reprints should be sent to Terry