Continuing the Cognitive Science Seminal Papers series with the study that used a simple auditory cue to show that the visual system holds far more information than we can report—and that this information fades within a fraction of a second.
A Simple Introduction
Sperling’s 1960 monograph begins with a puzzle that anyone who has looked at a briefly flashed display will recognize. When observers are shown an array of letters for a fraction of a second and asked to report everything they saw, they typically manage only about four or five letters. Yet they insist that they saw more. According to Sperling, “The apparently simple question: ‘What did you see?’ requires the observer to report both what he remembers and what he has forgotten” .
The problem, as Sperling framed it, is that a whole report confounds two things: what is available to vision and what survives in memory long enough to be written down. To separate them, he borrowed a technique from statistics—sampling—and adapted it to perception. The observer would be shown an array, and after the display disappeared, a tone would indicate which row to report. Since the observer did not know which row would be called, the report could not be prepared in advance. If performance on the cued row was better than performance on the whole array, then more information must have been available than the whole report revealed.
The paper does not use the term “iconic memory.” That name came later, from Neisser. According to the Citation Classic commentary, Sperling’s method demonstrated “that the subject has a very short-term visual memory of the array” and measured “the decay of this memory during the half-second following the exposure” .
10 Key Takeaways for Cognitive Science and Learning
- Whole report underestimates what is seen. According to Sperling, observers could report only about 4.5 letters from arrays of up to twelve, regardless of exposure duration from 15 to 500 milliseconds .
- Partial report reveals more. When cued to report a single row immediately after the display, observers could report 75–90% of the letters in that row. According to Sperling, this implies that far more than four or five letters were available immediately after the exposure .
- The cue must come quickly. According to Sperling, the advantage of partial report diminished as the cue was delayed, and at about one second the partial report was no better than the whole report .
- The information is stored as a visual image. Sperling argued that the available information is held in a rapidly fading visual image of the stimulus. According to him, “stimulus information is . . . ‘stored’ for a fraction of a second as a persisting image of the objective stimulus” .
- The store is precategorical. According to Sperling, the units of this visual image are those of the physical stimulus, not of categorized letters. This is why selection by location works but selection by category (letters versus numbers) does not produce the same advantage .
- The store is vulnerable to interference. According to Sperling, a bright post-exposure field drastically reduced accuracy, showing that the available information is sensitive to subsequent visual stimulation .
- The span of immediate memory is an individual constant. According to Sperling, the number of letters reported in a whole report is relatively invariant across stimulus arrangements and exposure durations. It is a characteristic of the observer, not of the stimulus .
- Order of report is not the explanation. According to Sperling, the partial-report advantage does not depend on the order in which letters are reported or on their position on the stimulus. The advantage comes from being able to read a persisting visual image before it fades .
- The finding has been widely replicated. According to Sperling, the main results were reproduced with different equipment, and similar findings were reported by Klemmer and Loftus and by Averbach using different techniques .
- The paper changed how memory was studied. According to the Citation Classic commentary, Sperling’s ambition was to use the inferential methods of physics to make inferences about internal mental states. “Like the atom, the span had been split” .
According to Sperling, “The high accuracy of partial report observed in the experiments does not depend on the order of report or on the position of letters on the stimulus, but rather it is shown to depend on the ability of the observer to read a visual image that persists for a fraction of a second after the stimulus has been turned off” .
The Sampling Problem: According to Sperling, the whole report is a test that is too short to reveal what is available. The partial report is a way of sampling the available information without exceeding the observer’s memory span.
Sensory Storage: According to Sperling, the visual system holds information in a form that outlasts the physical stimulus. This storage is brief, large in capacity, and precategorical.
The Paper’s Core Ideas
The partial report method. Sperling presented arrays of letters for 50 milliseconds. In the whole report condition, observers reported as many letters as they could. In the partial report condition, a tone after the display indicated which row to report. According to Sperling, the partial report was consistently more accurate than the whole report when the cue came immediately .
The decay function. According to Sperling, the accuracy of partial report declined sharply as the cue was delayed. At zero delay, observers had about nine letters available from a twelve-letter display. At one second, the partial report was no better than the whole report .
The visual image hypothesis. According to Sperling, the available information is stored as a visual image that persists after the stimulus ends. The evidence included observers’ reports that the display seemed still readable when the tone arrived, the effect of a bright post-exposure field, and the characteristics of errors .
Selection by location versus selection by category. According to Sperling, partial reports by location (which row) were more accurate than whole reports, but partial reports by category (letters versus numbers) were not. This suggested that the store is precategorical: selection works on physical features, not on meaning .
The Seven Experiments
Sperling’s monograph reports seven experiments, each addressing a distinct question about the availability and decay of visual information.
Experiment 1: Immediate memory. According to Sperling, this experiment asked whether the span of immediate memory is independent of the number and arrangement of letters. Observers gave whole reports of arrays ranging from three to twelve letters. The result was that each observer reported a roughly constant number of letters—about 4.3 on average—regardless of array size or arrangement. The span, according to Sperling, is an individual characteristic .
Experiment 2: Exposure duration. According to Sperling, this experiment asked whether the span depends on how long the display is visible. Exposure durations were varied from 15 to 500 milliseconds. The result was that exposure duration had little effect on the number of letters reported. The span, according to Sperling, is not an artifact of brief presentation .
Experiment 3: Partial report. According to Sperling, this experiment asked whether more information is available than the whole report reveals. Observers were cued by a tone, after the display, to report one row. The result was that partial reports were consistently more accurate than whole reports. Immediately after a twelve-symbol display, observers had about 9.1 letters available—more than twice the whole-report span .
Experiment 4: Decay of available information. According to Sperling, this experiment asked how quickly the available information fades. The cue was delayed by intervals from 0 to 1 second. The result was that accuracy declined sharply with delay. At delays beyond about a quarter of a second, most of the extra information was gone; at one second, partial report was no better than whole report .
Experiment 5: Exposure parameters. According to Sperling, this experiment asked how the available information depends on the visual conditions before and after the display. A bright post-exposure field was introduced, following Baxt’s earlier method. The result was that the bright field reduced accuracy to about half. This, according to Sperling, supports the view that the available information is a visual image sensitive to subsequent stimulation .
Experiment 6: Letters and numbers. According to Sperling, this experiment asked whether partial report by category (letters versus numbers) works as well as partial report by location. Observers were cued to report only the letters or only the numbers from a mixed array. The result was that category cues gave little or no advantage over whole report. According to Sperling, this shows that the store is precategorical: selection by physical location works, but selection by meaning does not .
Experiment 7: Order of report. According to Sperling, this experiment asked whether the partial-report advantage depends on the order in which items are reported. Observers were cued to report one row first and then the other. The result was that order of report had only a slight effect on overall accuracy, and for some observers no effect at all. According to Sperling, the partial-report advantage cannot be explained by order of report .
From Method to Cognitive Science
The paper was not written as cognitive science, but according to many later accounts it became foundational for several reasons.
The concept of iconic memory. According to later accounts, Neisser named the visual buffer “iconic storage” or “iconic memory,” and paired it with echoic storage for audition. The concept became part of the standard architecture of information processing.
The information-processing framework. According to the summary, the notion of storage in Sperling’s paper reflects the shift to the information-processing paradigm, even though the influence of behaviorism is still visible in his language (for example, calling attentional orienting an “unobservable response”) .
The debate it provoked. According to later reviews, Holding argued that the partial-report advantage could be explained by guessing and reduced response interference rather than by a persisting visual image. Coltheart defended the sensory storage interpretation. The debate over what iconic memory is, and whether it is truly precategorical, has continued for decades.
The citation record. According to Google Scholar, the paper has been cited more than 6,500 times. The Citation Classic commentary recorded over 455 citations by 1979, and the number has grown steadily since .
Sperling and Learning Science
The distinction between availability and accessibility. According to Sperling, more information is available to the visual system than can be reported. This distinction—between what is present and what can be accessed—has implications for how we think about attention, memory, and the limits of report.
The brevity of sensory storage. According to Sperling, the visual image fades within about a second. This means that any task that requires reporting more than a few items from a brief display must either use a partial report or accept that much of the information will be lost. The design of displays and instructions must take this limit into account.
The role of cues. According to Sperling, a cue that arrives in time allows the observer to read from the persisting image. A cue that arrives too late finds nothing but what has already been encoded into a more durable but limited store. This is a lesson about the timing of prompts and feedback in learning.
The limits of the analogy. Sperling’s store is visual and precategorical. Whether other modalities have analogous stores, and how those stores interact with meaning and categorization, are questions that later research has pursued.
Conclusion
Sperling’s monograph is not cognitive science. It is an experimental psychology paper about what can be seen in a brief exposure. But according to many of its readers, it made sensory memory an object of precise measurement—and in doing so, helped establish the idea that the mind has a short-term storage system with a specific capacity and a specific decay rate.
The paper’s legacy is in the concept of iconic memory, in the partial report method, and in the distinction between what is available and what can be reported. The method has been adapted and criticized, the concept has been refined and debated, but the question Sperling asked—how much can be seen in a single brief exposure?—remains a starting point for research on visual memory.
Sperling himself, according to his Citation Classic commentary, hoped that the style of his research rather than its specific content would have the greatest influence. “Like the atom, the span had been split” . The paper is a reminder that a simple method, applied to a simple question, can change how a field thinks about the mind.
This post is part of the Cognitive Science Seminal Papers series. The previous entry covered Chomsky’s review of Skinner’s “Verbal Behavior” (1959).
About the Author
