The term memory implies the capacity to encode, store, and retrieve information. The possibility that memory might not be a unitary system was proposed by William JAMES (1898) who suggested two systems which he named primary and secondary memory. Donald HEBB (1949) also proposed a dichotomy, suggesting that the brain might use two separate neural mechanisms with primary or short-term storage being based on electrical activation, while long-term memory reflected the growth of relatively permanent neuronal links between assemblies of cells.

Empirical support for a two-component view began to emerge in the late 1950s, when Brown (1958) and Peterson and Peterson (1959) observed that even small amounts of information would show rapid forgetting, provided the subject was prevented from maintaining it by active rehearsal. The characteristic forgetting pattern appeared to differ from that observed in standard long-term memory experiments, leading to the suggestion that performance depended on a separate short-term store. Such a view was vigorously opposed by Melton (1963), leading to a period of intense activity during the early 1960s that was concerned with the question of whether memory should be regarded as a unitary or dichotomous system.

By the late 1960s, the evidence seemed to strongly favor the dichotomous view. A particularly influential source of evidence was provided by a small number of neuropsychological patients who appeared to have a specific deficit of either the short-term or the long-term system. The clearest evidence of preserved short-term (STM) and impaired long-term memory (LTM) comes in the classic amnesiac  syndrome. Particularly influential was case H.M. who underwent bilateral excision of the HIPPOCAMPUS in an attempt to treat intractable epilepsy. H.M. was left with a profound amnesia, unable to commit new material to memory, whether visual or verbal, and showing no capacity to learn his way around a new environment, to recognize people who worked with him regularly, or to remember the content of anything he read or saw. His STM, on the other hand, as evidenced by the capacity to hear and repeat back a string of digits such as a telephone number, was quite normal (Milner 1966).

The opposite pattern of memory deficit was demonstrated by Shallice and Warrington (1970) in a patient, K.F., who was unable to repeat back more than two digits, but whose long-term learning capacity and everyday memory were well within the normal range. His lesion was in the left hemisphere in an area known to be associated with language. Subsequent studies have shown that language and short-term phonological memory are often impaired in the same patient, but that the two areas are separable and the symptoms dissociable. When tested on the Petersons' short-term forgetting task, patients like K.F. proved to show very rapid forgetting, whereas densely amnesiac patients show normal performance, provided their amnesia is pure and unaffected by more general intellectual deficits (Baddeley and Warrington 1970).

Evidence from normal subjects paralleled the neuropsychological research in suggesting the need for at least two separate memory systems. Many memory tests appeared to show two separate components, one that was durable and long-term while the other showed rapid dissipation. For example, if a subject hears a list of twenty unrelated words and is asked to recall as many as possible in any order, there will be a tendency for the last few words to be well recalled, the so-called recency effect. However a delay of only a few seconds is sufficient for the effect to disappear, while recall of earlier items remains stable. When this paradigm was applied to neuropsychological patients, those with STM deficits showed preservation of the long-term component, but little or no recency, while amnesiac patients showed the opposite pattern.

Finally the learning characteristics of the two systems appeared to differ. The short-term system has a limited capacity, but appears to be relatively insensitive to speed of presentation, and in the case of verbal material to be sensitive to the sound or phonological characteristics of the material presented. The long-term system, on the other hand, has a huge capacity but a relatively slow rate of acquisition of new material, and a tendency to encode verbal material in terms of its meaning rather than sound (Baddeley 1966a, 1966b; Waugh and Norman 1965).

The 1960s saw a growing interest in developing mathematical models of learning and memory, with the most influential of these being that of Atkinson and Shiffrin (1968) which became known as the modal model. However, problems with a simple dichotomy rapidly emerged, leading to the wide-scale abandonment of the field by many of its investigators.

One problem stemmed from Atkinson and Shiffrin's assumption that the probability of an item being stored in LTM was a simple function of how long it was maintained in the short-term system. A number of studies demonstrated that active and vigorous verbal rehearsal might link to very little durable LTM (Craik and Watkins 1973; Bjork and Whitten 1974). This prompted Craik and Lockhart (1972) to propose their levels of processing theory of memory. This proposed that an item to be remembered, such as a word, could be processed at a series of encoding levels, beginning with the visual appearance of the word on the page, moving on to the sound of the word when pronounced, and, given further and deeper processing, to the meaning of that word and its relationship to other experiences of the subject. Craik and Lockhart suggested that the deeper the level of encoding, the more durable the memory trace. There is no doubt that this simple formulation does capture an important characteristic of long-term learning, namely, that encoding material richly and elaborately in terms of prior experience will lead to a comparatively durable and readily retrievable memory trace.

Note however that levels of processing is not an alternative to a dichotomous view; indeed Craik and Lockhart themselves postulate a primary memory system as part of their model, although this aspect of their work receives very much less attention than the concept of encoding levels.

A second difficulty for the modal model lay in the neuropsychological evidence. It may be recalled that patients with an STM deficit performed poorly on tasks such as immediate memory span and recency, but were normal in their LTM performance. The modal model suggested, however, that the short-term system acts as a crucial antechamber to long-term learning, hence predicting that such patients should have impaired learning capacity, and indeed should show poor performance on a wide range of tasks that were assumed to be dependent on the limited-capacity short-term system. They showed no evidence of this, with one such patient being an efficient secretary, while another ran a shop and raised a family.

This problem formed the focus of work by Baddeley and Hitch (1974), who attempted to simulate the neuropsychological STM deficit by means of a dual task technique. Subjects were required to hold and rehearse sequences of digits varying in length while at the same time performing a range of other tasks that were assumed to depend upon the limited-capacity store. It was assumed that longer sequences of digits would absorb more of the store, until eventually capacity was reached, leaving the main tasks to be performed without the help of the short-term system. A range of tasks were studied including long-term learning, reasoning, and comprehension. A clear pattern emerged suggesting that concurrent digits did impair performance systematically, but by no means obliterated it. This led to a reformulation of the STM hypothesis and the postulation of a multicomponent system which was termed working memory. It was suggested that this comprised a limited capacity attentional control system, the central executive, together with at least two slave systems, one concerned with maintaining visual-spatial information, the sketchpad, while the other was responsible for holding and manipulating speech-based information, the phonological loop.

The concept of working memory has proved extremely fruitful, not only in accounting for the initial neuropsychological evidence but also in being applicable to a wide range of tasks and subject groups, and more recently, providing a very fruitful basis for a range of neuroradiological studies concerned with the neuroanatomical basis of working memory (see Smith and Jonides 1995).

As in the case of STM, the concept of LTM has also undergone a detailed analysis in the last twenty years, again resulting in a degree of fractionation. One of the strongest cases for a basic distinction is that between implicit and explicit memory (see IMPLICIT VS. EXPLICIT MEMORY). Once again this distinction was heavily influenced by neuropsychological evidence, when it was observed that even densely amnesiac patients could nevertheless show comparatively normal learning on certain tasks, including the acquisition of motor skills, classical conditioning, and a whole range of procedures that come under the general term of priming. The classic demonstration within this area was that of Warrington and Weiskrantz (1968), who showed that amnesiac patients who were shown a list of words were totally unable to recall or recognize the words, but were able to demonstrate learning by perceiving the words more rapidly when they were presented in fragmented form. Subsequent work showed that learning was also preserved when tested by cueing with the first few letters of the word (e.g., present CROCODILE, test with CRO -- --), or with a fragment of the word, (C -- O -- O -- I -- E). Equivalent phenomena have been demonstrated in other modalities, and have shown to be widely demonstrable in normal subjects (see Roediger 1990 for a review).

Over the last decade there has been substantial controversy as to how best to explain this pattern of results. There is still some support for attempts to account for the data within a unitary system, but my own view (Baddeley 1998) is that this is no longer a tenable position. In particular, the neuropsychological evidence seems to argue for a distinction between an episodic LTM system (depending on a circuit linking the temporal lobes, the frontal lobes, and parahippocampal regions), and a whole range of implicit learning systems, each tending to reflect a different brain region.

While these systems are of considerable interest in their own right, and as ways of analyzing perceptual and motor processing, it can be questioned as to whether they should be referred to as memory systems, as they typically involve relatively automatic retrieval processes that are often not under the direct control of the subject. In contrast, episodic memory is the system that typifies our experience of recollecting the past. Indeed, Tulving (1985) suggests that its crucial and defining feature is the recollective process, accompanied by the feeling of familiarity, a process he refers to as ecphory. There have in recent years been a growing number of studies concerned with the phenomenological aspect of memory, often with considerable success (see Gardiner 1988).

A second proposed distinction within LTM is that between semantic and episodic memory (see EPISODIC VS. SEMANTIC MEMORY). Semantic memory refers to the stored knowledge of the world that underlies not only our capacity to understand language but also our ability to take advantage of prior knowledge in perceiving and organizing both the physical and social world around us. The need for such a store of information was initially made obvious by attempts to develop computer-based systems for comprehending text, such as that of Quillian (1969). These stimulated attempts to understand semantic memory in human subjects, and prompted Tulving (1972) to propose that semantic and episodic memory are distinct systems. At first sight, the evidence appeared persuasive. Densely amnesiac patients may perform normally on semantic memory tests while showing no evidence of new episodic learning (Wilson and Baddeley 1988). However, semantic memory tests typically involve accessing old memories, whereas episodic tests are principally concerned with the laying down of new memory traces. When amnesiac patients are required to extend their existing semantic memory systems, for example, by learning about the developing political system within their country, or learning new routes within their town, learning appears to be catastrophically bad. An alternative way of conceptualizing semantic memory is to suggest that it represents the residue of many episodic memories, with access being based on generic commonalities, rather than the retrieval of a specific episode. The nature of semantic memory and its neuroanatomical basis continues to be a very active research area, with neuropsychological evidence again being particularly cogent (see Patterson and Hodges 1996).

No survey of memory would be complete without comment on one aspect of memory that has been both active and controversial in recent years, namely, the attempt to apply the lessons learned in the laboratory to everyday functioning. Although the link between the laboratory and the field has occasionally appeared to be excessively confrontational (e.g., see Neisser 1978; Banaji and Crowder 1989), the interaction has on the whole been a fruitful one. This is particularly true of clinical applications of the psychology of memory, where, as we have seen, the study of memory deficits in patients has been enormously influential in changing our views of the normal functioning of human memory.

See also

-- Alan Baddeley


Atkinson, R. C., and R. M. Shiffrin. (1968). Human memory: A proposed system and its control processes. In K. W. Spence, Ed., The Psychology of Learning and Motivation: Advances in Research and Theory. New York: Academic Press, pp. 89-195.

Baddeley, A. D. (1966a). Short-term memory for word sequences as a function of acoustic, semantic and formal similarity. Quarterly Journal of Experimental Psychology 18:362-365.

Baddeley, A. D. (1966b). The influence of acoustic and semantic similarity on long-term memory for word sequences. Quarterly Journal of Experimental Psychology 18:302-309.

Baddeley, A. D. (1998). Human Memory: Theory and Practice, Revised ed. Needham Heights, MA: Allyn and Bacon.

Baddeley, A. D., and G. Hitch. (1974). Working memory. In G. A. Bower, Ed., The Psychology of Learning and Motivation. New York: Academic Press, pp. 47-89.

Baddeley, A. D., and E. K. Warrington. (1970). Amnesia and the distinction between long- and short-term memory. Journal of Verbal Learning and Verbal Behavior 9:176-189.

Banaji, M. R., and R. G. Crowder. (1989). The bankruptcy of everyday memory. American Psychologist 44:1185-1193.

Bjork, R. A., and W. B. Whitten. (1974). Recency-sensitive retrieval processes. Cognitive Psychology 6:173-189.

Brown, J. (1958). Some tests of the decay theory of immediate memory. Quarterly Journal of Experimental Psychology 10:12-21.

Craik, F. I. M., and R. S. Lockhart. (1972). Levels of processing: A framework for memory research. Journal of Verbal Learning and Verbal Behavior 11:671-684.

Craik, F. I. M., and M. J. Watkins. (1973). The role of rehearsal in short-term memory. Journal of Verbal Learning and Verbal Behavior 12:599-607.

Gardiner, J. M. (1988). Functional aspects of recollective experience. Memory and Cognition 16:309-313.

Hebb, D. O. (1949). Organization of Behavior. New York: Wiley.

James, W. (1890). The Principles of Psychology. New York: Holt, Rinehart and Winston.

Melton, A. W. (1963). Implications of short-term memory for a general theory of memory. Journal of Verbal Learning and Verbal Behavior 2:1-21.

Milner, B. (1966). Amnesia following operation on the temporal lobes. In C. W. M. Whitty and O. L. Zangwill, Eds., Amnesia. London: Butterworths, pp. 109-133.

Neisser, U. (1978). Memory: What are the important questions? In M. M. Gruneberg, P. E. Morris, and R. N. Sykes, Eds., Practical Aspects of Memory. London: Academic Press.

Patterson, K. E., and J. R. Hodges. (1996). Disorders of semantic memory. In A. D. Baddeley, B. A. Wilson, and F. N. Watts, Eds., Handbook of Memory Disorders. Chichester, England: Wiley, pp. 167-186.

Peterson, L. R., and M. J. Peterson. (1959). Short-term retention of individual verbal items. Journal of Experimental Psychology 58:193-198.

Quillian, M. R. (1969). The teachable language comprehender: A simulation program and theory of language. Communication of the ACM 12:459-476.

Roediger, H. L. (1990). Implicit memory: Retention without remembering. American Psychologist 45:1043-1056.

Shallice, T., and E. K. Warrington. (1970). Independent functioning of verbal memory stores: A neuropsychological study. Quarterly Journal of Experimental Psychology 22:261-273.

Smith, E. E., and J. Jonides. (1995). Working memory in humans: Neuropsychological evidence. In M. Gazzaniga, Ed., The Cognitive Neurosciences. Cambridge, MA: MIT Press, pp. 1009-1020.

Tulving, E. (1972). Episodic and semantic memory. In E. Tulving and W. Donaldson, Eds., Organization of Memory. New York: Academic Press, pp. 381-403.

Tulving, E. (1985). How many memory systems are there? American Psychologist 40:385-398.

Warrington, E. K., and L. Weiskrantz. (1968). New methods of testing long-term retention with special reference to amnesic patients. Nature 217:972-974.

Waugh, N. C., and D. A. Norman. (1965). Primary memory. Psychological Review 72:89-104.

Wilson, B. A., and A. D. Baddeley. (1988). Semantic, episodic and autobiographical memory in a post-meningitic amnesic patient. Brain and Cognition 8:31-46.

Further Readings

Baddeley, A. D. (1999). Essentials of Human Memory. Hove: Psychology Press.

Parkin, A. (1987). Memory and Amnesia: An Introduction. Oxford: Blackwell.