Duration and Capacity of Short-Term Memory Explained
Working memory is also sometimes called short-term memory. Some researchers sometimes use these terms interchangeably, as will we throughout this literature.
As the term short-term memory suggests, information is briefly stored there. However, the duration of items residing in working memory is, as we will see, much longer than that of items residing in sensory memory Opens in new window.
Working memory retains less information; its capacity is unlike the unlimited capacity of the visual icon. Working memory is limited to storing just a small number of items.
Whereas information in the different sensory stores is specific to modality, coding in working memory can be acoustic, semantic, or visual. How the different types of codes are put to use is discussed at length in Alan Baddeley’s model of working memory.
For now it is helpful to use the analogy of a workbench when thinking about working memory. A workbench is a space where one can construct something using parts and tools.
For example, if one wanted to build a spice rack, s/he would need several pieces of wood, some nails, and varnish. In addition to this, s/he would need saws to cut the wood to size, a hammer, and a paintbrush. With these items in hand one could then set about to create the final product.
Working memory is the mental equivalent of this workbench. It is a place where data (the parts) can be temporarily stored so that they can be operated on by cognitive processes (the tools).
In this sense working memory is the site where conscious thinking takes place. It is here that you remember a phone number, figure out how to navigate your way around a new city, or solve an arithmetic problem.
How long does information last in working memory?
This is a crucial question. If information is lost rapidly from short-term memory, this must limit our ability to think about several things at once.
In a classic study, Peterson and Peterson (1959) demonstrated the duration of items that reside in short-term memory. They presented participants with items to be remembered. A tone was sounded following the presentation of the items.
In separate trials the tone was sounded at varying time intervals. Sometimes it would go off immediately; during other trials there would be a substantial delay. The participants were instructed to recall the items upon the sounding of the tone. In this condition, all participants were able to remember the items correctly, no matter the length of the delay.
In a second condition, presentation of a three-digit number followed presentation of the items. Participants were instructed to count backward in threes from this number. If the number was “796,” they would have to count out “793,” “790,” “787,” and so on. A tone was sounded, again at varying intervals, cueing recall of the item.
At this point the results were quite different. The ability to remember the items deteriorated rapidly as the period of delay increased. After an 18-second recall interval, the accuracy of recall had dropped to just 5%, meaning participants could only recall 5% of the items correctly. Figure 1 shows idealized results for both conditions of such a study.
You may have guessed what’s happening here. In the first condition, participants were able to rehearse the trigrams. Rehearsal, by the way, is the mental repetition or mental “practicing” of some to-be-learned material. Rehearsal refreshes items in short-term memory; it keeps them active and prevents them from decaying.
In the second condition, participants were prevented from rehearsing because of the backward counting. Here the items underwent decay. Using this paradigm, Brown (1958) and Peterson and Peterson (1959) were able to determine that the duration of short-term memory was about 18 seconds. Our next discussion involves capacity.
How much can working memory hold?
It is easy to devise an experiment that tests this question. It would involve presenting participants with a series of lists, each containing a different number of items.
For example, we could present a list of four digits, then a five-digit list, a six-digit list, and so on. If there were a limit to how many digits could beheld in working memory, we would see a dramatic drop off in retention as we arrived at that limit.
Many early experiments of this sort demonstrated that this was indeed the case. On average, individuals can retain about seven items, give or take two, commonly denoted as 7+2. This limit has come to be called the “magical number seven” (Miller, 1956).