The difficulty of investigating what is seen in a brief visual presentation is captured in the following quotation: "The apparently simple question: ‘What did you see?’ requires the observer to report both what he remembers and what he has forgotten" (p. 1). Sperling devised an elegant procedure to address this dilemma. In his procedure, subjects were shown in a brief presentation (5-500 ms) an array of letters arranged in a matrix. One of two types of recall instructions followed. In full-report, subjects were simply told to report as many letters of the entire array as they could remember. Sperling found that subjects could usually recall an average of about 4.5 items, regardless of display size or stimulus duration. He referred to this quantity as the "span of apprehension," or "immediate memory-span."
Sperling contrasted the above findings with results obtained using a partial-report recall technique in which following stimulus presentation, the subject was cued, at random, to recall a subset (e.g., a single line) of the items in the display. Sperling found that subjects could usually recall about 75-90% of items from any given, randomly cued line. The logical extension of this finding was that, immediately following stimulus presentation, much more information is available to the subjects than would be indicated by the full-report technique. That is, if subjects can recall 75% of the items from any given randomly cued row, then 75% of the entire display must be available immediately following stimulus display, as opposed to the 33% indicated by full report. This finding is called partial-report superiority.
Sperling also manipulated the duration of the interval separating stimulus offset and presentation of the cue. As is well known, Sperling found that partial report superiority diminished as cue interval increased. At a one-second cue interval, partial-report and full-report performance were equal. Sperling also found evidence for strategy shifts with increasing cue interval. Whereas at brief cue intervals subjects waited for the cue to indicate which portion of the stimulus display to recall, they often tried to anticipate where the cue would appear at longer cue delays. This latter strategy was evidenced by better recall of particular stimuli and poorer recall for other stimuli.
Sperling made many important conclusions based on his results, which were the basis for much future work. For example, he argued convincingly that the sensation created by a physical stimulus outlasts its physical duration, and that this phenomenon must imply some sort of memory storage. In his words, "stimulus information is . . . ‘stored’ for a fraction of a second as a persisting image of the objective stimulus" (p. 20). The concept of sensory storage was subsequently integrated into nearly all models of information processing. Moreover, he recognized some of the important properties of this storage mechanism, later called iconic memory. For example, a briefly stored visual image is highly susceptible to interference from subsequent stimuli. That is, "the available information is sensitive to interference by noninformational visual stimuli which follow the exposure" (p. 21). Finally, Sperling pointed to the precategorical nature of information stored in iconic memory. Thus, "The units of a visual image . . . are always those of an equivalent ‘objective image,’ the physical stimulus" (p. 21).
The influence of behaviorism is evident in Sperling’s paper. For example, attentional orienting is referred to as an "unobservable response." Nevertheless, the notion of storage reflects the shift to the then emergent information processing paradigm.