The Capacity of Short-Term Memory

Information a person focuses on and chooses to retain enters into short-term store, at least for a brief period of time.

Psychologists have devised two main strategies to estimate the capacity of short-term memory: span measures and the recency effect in free recall.

1. Span Measures

In 1887, Joseph Jacobs used memory span as a measure of how information can be stored in short-term memory. Jacobs presented his participants with a random sequence of digits or letters, and then asked them to repeat the items back in the same order.

Memory span was the longest sequence of items recalled accurately at least 50% of the time. The average number of items recalled was between five and nine, and digits were recalled better (9.3 items) than letters (7.3 items).

Jacobs’ approach was limited. First, his research lacked mundane realism—the use of an artificial situation that closely resembles a natural situation, because his span tasks were not representative of everyday memory demands.

Second, if we could only remember a few letters, we would be unable to remember the following sequence of 10 letters:

In fact, of course, we can remember that sequence because it’s easy to organize the information in memory.

He claimed we should focus on chunks (integrated pieces or units of information). About seven chunks of information can be held in short-term memory at any time.

The question of what constitutes a chunk depends on your personal experience. For example, “IBM” would be one chunk if you know about International Business Machines, but it would be three if you didn’t.

Simon (1974) tested Miller’s ideas on chunking Opens in new window by studying memory span for words, two-word phrases and eight-word phrases.

At the level of chunks, Simon argued that an entire phrase forms a single chunk.

The number of chunks recalled varied less over conditions than the number of words—it fell from six or seven with unrelated words to four with two-word phrases and three with eight-word phrases.

As with Jacobs’ earlier research, the study by Simon (1974) lacks mundane realism in that the demands on the participants were very different from those of our everyday lives.

Cowan et al. (2005) argued that the capacity of short-term memory is often exaggerated. For example, people may recall some of the information from long-term memory Opens in new window rather than short-term memory Opens in new window and may also rehearse the information while it is being presented.

Cowan et al. (2005) reduced the involvement of long-term memory and rehearsed by using the running memory task—a series of digits ended at an unpredictable point and participants then recalled the items from the end of the list.

The average numbers of items recalled was 3.87, suggesting the true capacity of short-term memory is about four items. Subsequent research has confirmed that the capacity of short-term memory is about three or four chunks (Mathy & Feldman, 2012).

2. The Recency Effect

A familiar example of the recency effect is the observation that a pop group is only as good as their last hit song.

People generally have a good memory for most-recent things (e.g. the movie they saw last).

In relation to short-term memory Opens in new window, the recency effect Opens in new window can be measured using free recall—participants see a list of words or syllables, and immediately recall them in any order.

The recency effect is demonstrated by the fact that the last two or three items in a list are usually much better remembered than items from the middle of the list.

These last few items are well remembered because they are in the short-term store when the list presentation ends.

Glanzer and Cunitz (1966) introduced an interference task involving counting backwards by threes for 10 seconds between the end of the list and the start of recall. This eliminated the recency effect but had no effect on recall of the rest of the list.

The two or three words at the end of the list were in a fragile state in short-term memory, and so were easily wiped out by counting backwards. In contrast, the other list items were in the long-term store and so were unaffected.