Psychology World — home
Blog

Short-term memory vs working memory: two names, one argument

Short-term memory holds information briefly; working memory holds and works with it. Where the terms came from, why they overlap, and the four-not-seven limit.

Cognition12 min read
By Psychology World

Short-term memory is the system that holds a small amount of information, in a very accessible state, for a short time. Working memory is memory as it is used to plan and carry out behaviour. In the words of Nelson Cowan's 2008 review, it "includes short-term memory and other processing mechanisms that help to make use of short-term memory". The exam version is shorter: short-term memory stores, and working memory stores and manipulates.

That split is a good place to start, but it hides an argument. Researchers have not settled whether the two are separate systems or one system described in two ways, and even journal articles use the terms loosely. The number everyone remembers, seven, was not quite the claim its author made. Below: where each name came from, what the evidence says about capacity and duration, and what to write when an exam asks you to tell the two apart.

  • Short-term memory holds information for a short time. Working memory holds it and uses it, for example in mental arithmetic.
  • "Working memory" was used by Miller, Galanter and Pribram in 1960. Baddeley and Hitch built it into a multi-component model in 1974.
  • Miller's 1956 paper counted memory span in chunks, not items, and called the recurring seven "a pernicious, Pythagorean coincidence". Cowan puts the limit more typically at three or four chunks.
  • Peterson and Peterson's 1959 result is often quoted as a duration. Later work found interference behind much of it, and Cowan calls decay "still pretty much open to debate".
  • A 2012 review found that the difference between the two is "far from clear", so check how your course defines the terms.

The difference in one table

Short-term memory and working memory compared, as the terms are usually taught.
Short-term memoryWorking memory
What it namesA limited amount of information held in a very accessible state for a short timeMemory as it is used to plan and carry out behaviour, including short-term memory and the processing that makes use of it
What happens to the informationIt is maintainedIt is maintained and manipulated
Where the term comes fromUsed by Broadbent (1958) and Atkinson and Shiffrin (1968); related to the primary memory of James (1890)Used by Miller, Galanter and Pribram (1960); developed by Baddeley and Hitch (1974)
ModelA single holding place, as in the store modelsA multi-component system: central executive, phonological loop, visuospatial sketchpad, episodic buffer
Typical taskRepeating digits backJudging sentences or checking sums while remembering a list
Usual measureSimple span taskComplex span task

The table is the textbook split, not a boundary that research has fixed. The sections below show where it blurs.

What each term describes

Cowan defines short-term memory as the faculties of the mind that can hold a limited amount of information in a very accessible state temporarily. Hearing a seven-digit number and repeating it straight back is a short-term memory task: it asks you only to keep the digits. Researchers call this a simple span task, and Aben, Stapert and Blokland list digit span, word span and letter span among its examples. People often keep such items alive by repeating them, which our article on maintenance and elaborative rehearsal covers.

Working memory is the broader idea. Cowan notes that it has been defined in three slightly different ways: as short-term memory applied to cognitive tasks, as a multi-component system that holds and manipulates information in short-term memory, and as the use of attention to manage short-term memory. In all three, holding information is a means, not the goal. Cowan's own examples are keeping the partial results while you solve an arithmetic problem without paper, combining the premises of a long argument, and baking a cake without adding the same ingredient twice.

In the laboratory, a working-memory measure adds a second job to the holding job. Aben and colleagues describe complex span tasks as simple span tasks with a demanding secondary task added, such as solving sums (operation span) or judging whether a sentence is correct (reading span).

Where the two names came from

Key dates in the history of the two terms.
YearWhoWhat changed
1956MillerMemory span described as constant in chunks, not bits
1958BroadbentShort-term store limited by decay, in a model of memory drawn as a series of boxes
1959Peterson and PetersonRecall of a three-consonant string falls to below 10% after 18 seconds of counting backwards
1960Miller, Galanter and Pribram"Working memory" used for memory as it is used to plan and carry out behaviour
1968Atkinson and Shiffrin"Short-term memory" used, in a slightly different way, for a limited, temporary holding place
1974Baddeley and HitchEvidence that one module cannot account for all temporary memory; working memory as a multi-component system
2000BaddeleyEpisodic buffer added as a fourth component
2001CowanCapacity reconsidered in "The magical number 4 in short-term memory"

The two names started in different places. Miller, Galanter and Pribram used "working memory" in 1960 for temporary memory seen from a functional standpoint, so from their point of view, Cowan notes, there is no clear distinction between short-term and working memory.

"Short-term memory" belongs to the store models. Cowan notes that Broadbent (1958) and Atkinson and Shiffrin used it in slightly different ways, and he follows Atkinson and Shiffrin's 1968 chapter. Cowan describes Baddeley and Hitch as first thinking of short-term memory as the unitary holding place of such models. Then they realised that the evidence fitted a multi-component system that could not be reduced to a single store, and they used "working memory" for that whole system. After their 1974 chapter, Cowan writes, the term became much more dominant in the field. Aben and colleagues call one finding an important argument for the model: a concurrent problem-solving task disrupts the memorising of a list of items only to a minor extent, which contradicts a single system for both storage and processing.

The first version of the model had three parts. The central executive is not a memory system itself but coordinates the other two. The phonological loop and the visuospatial sketchpad hold verbal-phonological and visual-spatial material. In a 2000 paper, Baddeley pointed to "a number of phenomena that are not readily captured by the original model" and proposed a fourth component, the episodic buffer. These are components of a model, not brain structures.

Why the estimate is now about four, not seven

You will still see "7 ± 2 items", Miller's "magical number seven", given as the capacity of short-term memory. That is not quite what Miller's 1956 paper said. Its main point was a distinction between bits and chunks: the span of immediate memory is constant in chunks, and almost independent of how many bits each chunk holds, over the range examined. His example comes from Hayes, who found a span of five words when each word was drawn at random from a set of 1,000 English monosyllables. That could just as well have been called a span of 15 phonemes, Miller wrote, but people were recalling five words, not 15 phonemes.

Miller was also cautious about the seven. His paper ends with a list of sevens, from the seven wonders of the world to the seven-point rating scale and the seven digits in the span of immediate memory. He proposed to withhold judgment, and wrote that he suspected it was "only a pernicious, Pythagorean coincidence". Cowan adds that Miller's later autobiographical essay indicates he was never very serious about the number seven, and that it was a rhetorical device.

Cowan accepts that memory span is approximately seven items in adults, but notes that there is no guarantee that each item is a separate entity. Later studies tried to take account of strategies that make memory more efficient, such as grouping. They suggested that the limit is more typically three or four units. Cowan's 2001 article puts the number in its title.

For an exam, the safe version is this. Miller's "about seven" is a memory span, and each item in it can be a chunk that holds several things. Cowan's "three or four" is the estimate when such strategies are taken into account.

Why there is no single answer for how long it lasts

Textbooks and revision pages often give a fixed duration. The open textbook Psychology 2e from OpenStax, for example, says that short-term memory storage "lasts 15 to 30 seconds". The sources behind this article do not support one number. The best-known experiment shows why.

In 1959, Peterson and Peterson gave people a three-consonant string, read aloud, and had them count backwards from a three-digit number straight afterwards, to prevent rehearsal. They tested recall after 3 to 18 seconds, in steps of 3. According to MacLeod's chapter on interference theory, recall showed an exponential decline as the interval lengthened, to below 10% at the longest. It became the classic short-term forgetting function. Peterson and Peterson read it as rapid decay, and it is often summarised as a duration of about 18 seconds.

Keppel and Underwood questioned the claim that interference played no role. They looked only at the first three trials and found "a severe proactive effect produced by a single prior item" (p. 156). Interference appeared to build up from trial to trial, which could not be seen when all trials were averaged together. So the 1959 curve is better read as forgetting under particular conditions, with earlier trials interfering, than as a fixed lifespan. Our article on proactive and retroactive interference explains the mechanism. Whether items in short-term memory also decay with time, Cowan writes, "is still pretty much open to debate".

Why the line between the two stays blurry

In 2012, Aben and colleagues reviewed how the terms are actually used and found a mess:

  • The two terms are often used interchangeably.
  • Some authors define short-term memory as the ability to maintain information for seconds, while others use the same definition for working memory.
  • Many studies treat the two as coexisting, while others claim that working memory has replaced the older term.

Their conclusion is that the differentiation between the two is "far from clear". Correlational studies have not consistently separated them, and there is evidence for a large or even complete overlap. The tasks do not settle it, because there are no standard short-term or working-memory tasks, and each one probably measures a mix of processes.

Cowan puts the confusion down largely to definitions. He says it is "largely the result of different investigators using different definitions", and that the distinction "depends on the definition that one accepts". His own definition puts short-term memory inside working memory. Aben and colleagues write that it is hard to imagine a model in which working memory is a part of short-term memory, while there are claims that working memory consists of short-term memory plus additional processes.

Some researchers go narrower: Engle and colleagues have sometimes used "working memory" only for the processes that control attention, which Cowan calls "a slightly confusing discrepancy in the usage of terms". The practical result is that the same word can mean different things in two textbooks.

Where long-term memory fits

Long-term memory is the third term in the usual trio. Cowan describes it as a vast store of knowledge and a record of prior events, and says it exists according to all theoretical views. In his chapter, short-term and long-term memory could differ in two ways: only short-term memory would show temporal decay and a chunk capacity limit. He calls both properties still controversial, but adds that the current literature is rather encouraging regarding the existence of both decay and capacity limits.

For more on long-term memory, see our articles on forgetting and the forgetting curve and on the kinds of long-term memory in episodic versus semantic memory.

Which term to use in an exam

Use the split your course teaches and say which one you are using, with a task for each. If the question asks whether the two are the same, a careful answer is that they are conceptually different but overlap heavily, and studies have not consistently separated them.

Keeping the terms apart

Short-term memory, working memory, long-term memory, chunk, rehearsal and episodic buffer are neighbours that are easy to mix up, so it is worth putting them in one place. In Psychology World you can save them to one list and drill them as flashcards, as multiple-choice questions once the list has four or more words, or by typing the answer out. What each study mode does is on the home page. The drills help you keep the vocabulary straight. For the methods with the best evidence, read our guide to memorising psychology terminology.

Frequently asked questions

Is working memory the same as short-term memory?

Not exactly, but they overlap heavily, and researchers disagree about how they relate. By the usual textbook split, short-term memory maintains information and working memory maintains and manipulates it. Cowan treats working memory as including short-term memory. A 2012 review concluded that the differentiation is "far from clear".

What is an example of working memory?

Cowan gives keeping the partial results while you solve an arithmetic problem without paper. In the laboratory, a reading span task is one example: you judge whether sentences are correct while you remember a list of items. Repeating a list of digits back, by contrast, is a short-term memory task.

How many items can short-term memory hold?

Memory span in adults is about seven items, but an item can be a chunk that contains several smaller units. Cowan reports that when strategies such as grouping are taken into account, the limit is more typically three or four units. Individual adults range from about two to six.

How long does short-term memory last?

There is no single figure. Peterson and Peterson found recall below 10% after 18 seconds of counting backwards, but Keppel and Underwood found that proactive interference appeared to build up across trials. Cowan calls the question of decay "still pretty much open to debate".

What are the parts of Baddeley's working memory model?

There are four. The central executive is not a memory system itself: it coordinates the processes of the two stores. The phonological loop holds verbal-phonological material, and the visuospatial sketchpad holds visual and spatial material. The episodic buffer was added in 2000, and Baddeley describes it as a limited-capacity system that provides temporary storage and binds information from the other components and from long-term memory into a unitary episodic representation. They are components of a model, not brain regions.

Sources

  1. 1.Cowan, N. (2008). What are the differences between long-term, short-term, and working memory? Progress in Brain Research, 169, 323–338.
  2. 2.Aben, B., Stapert, S., & Blokland, A. (2012). About the distinction between working memory and short-term memory. Frontiers in Psychology, 3, 301.
  3. 3.Miller, G. A. (1956). The magical number seven, plus or minus two: Some limits on our capacity for processing information. Psychological Review, 63, 81–97.
  4. 4.Peterson, L., & Peterson, M. J. (1959). Short-term retention of individual verbal items. Journal of Experimental Psychology, 58(3), 193–198.
  5. 5.Keppel, G., & Underwood, B. J. (1962). Proactive inhibition in short-term retention of single items. Journal of Verbal Learning and Verbal Behavior, 1(3), 153–161.
  6. 6.Atkinson, R. C., & Shiffrin, R. M. (1968). Human memory: A proposed system and its control processes. Psychology of Learning and Motivation, 2, 89–195.
  7. 7.Baddeley, A. D., & Hitch, G. (1974). Working memory. Psychology of Learning and Motivation, 8, 47–89.
  8. 8.Baddeley, A. (2000). The episodic buffer: A new component of working memory? Trends in Cognitive Sciences, 4(11), 417–423.
  9. 9.Cowan, N. (2001). The magical number 4 in short-term memory: A reconsideration of mental storage capacity. Behavioral and Brain Sciences, 24(1), 87–114.
  10. 10.MacLeod, C. M. Interference theory: History and current status (chapter preprint).
  11. 11.Spielman, R. M., Jenkins, W. J., & Lovett, M. D. (2020). Psychology 2e, 8.1 How memory functions. OpenStax.

Look the terms up yourself

18,000+ psychology and psychotherapy terms in English and German, on iPhone, iPad, Mac and Apple Vision.

See how Psychology World works
All posts