Lecturer's Précis - Baddeley (2000)

"The Episodic Buffer: A New Component of WorkingMemory"

Copyright Notice: This material was written and published in Wales by Derek J. Smith (Chartered Engineer). It forms part of a multifile e-learning resource, and subject only to acknowledging Derek J. Smith's rights under international copyright law to be identified as author may be freely downloaded and printed off in single complete copies solely for the purposes of private study and/or review. Commercial exploitation rights are reserved. The remote hyperlinks have been selected for the academic appropriacy of their contents; they were free of offensive and litigious content when selected, and will be periodically checked to have remained so. Copyright © 2004-2018, Derek J.Smith.


First published online 07:40 GMT 2nd November 2004, Copyright Derek J. Smith (Chartered Engineer). This version [2.0 - copyright] 09:00 BST 4th July 2018.


TECHNICAL ASIDE - VARIETIES OF BUFFER: [The term "buffer" came originally from computer science. This is an abbreviated version of how we explained the notion and practice of buffering in Sections 1.3 and 2.3 of our e-paper on "Short-Term Memory Subtypes in Computing and Artificial Intelligence" (Part 4).] "Buffers" in general are simply conveniently sized areas of temporary computer storage, and they are important because they allow "buffering", that is to say, "the temporary holding of information in transit" (Purser, 1987, p312). Bit String Buffering: Computer designers in the early 1950s added small free-standing storage units between the relatively fast data bus and the relatively slow peripheral device controllers. As data arrived in the form of a slow stream of bits from a peripheral device, it was loaded into an "input buffer", and only when that buffer was full was the accumulated stream released - either at a higher serial rate, or in parallel - onto the bus to complete its journey to its destination in CPU or Main Memory. This short delay left the CPU free to be getting on with something else. A similarly arranged "output buffer" could buffer data en route from the output record area to an output device. Main Memory Buffering: It is also possible to use Main Memory for buffering. This can be done by doubling (etc.) the number of Main Memory record blocks allocated to each file, and by then routinely loading these one record ahead of the ongoing processing. This sort of "double buffering" introduced a degree of record-level overlap into the system, such that the input delay for record <n+1> coincided with the processing time for record <n>, which, in turn, coincided with the output delay for record <n-1>. For very fast processes, "treble buffering" (etc.) could be resorted to, thus creating even more overlap. Instruction Buffering (or "Pipelining"): Pipelining is a technique "whereby multiple instructions are overlapped in execution" in the computer's fundamental FETCH-and-EXECUTE processing cycle (Patterson and Hennessy, 1996, p125). This is accomplished by providing an "Instruction Buffer", to contain instruction <n + 1> in FETCH mode, while the Instruction Register [reminder] contained instruction <n> in EXECUTE mode. "Store and Forward" Buffering: This is a form of buffering which occurs (in addition to all the others) at the subordinate nodes in a communications network. It is needed whenever three or more computers have been interconnected, and it allows transmissions to be stored temporarily at node #2 en route for node #3, should the line to node #3 be down or otherwise engaged, or should the CPU at node #3 be too busy to accept the input immediately. Modern telecommunications - including the technology currently bringing you this page - devotes much of its attention to such problems, and network designers need to carry out a complex traffic queuing analysis before deciding how much buffering store to provide at each intermediate node.

1 - Introduction

Baddeley begins the present paper by listing the main difficulties with the original Baddeley and Hitch (1974) Working Memory Model of Memory (henceforth "WMM") as it stood at the end of the millennium, thus .....

1 - The Phonological Loop: The phonological loop concept accounts for a number of empirical observations, namely the phonological similarity effect [glossary], the word-length effect [glossary], and the phonological recoding effect [glossary]. The idea of a sound-processing loop also fits in well with a body of clinical evidence, because this particular faculty is generally more inhibited in dyspraxics [glossary] than dysarthrics [glossary]. However, it has recently been less successful at explaining the articulatory suppression effect [glossary], where the effect size is significantly less than the model would predict.

2 - Prose Recall: The WMM also has little to say about the phenomenon of "chunking" [glossary] in prose recall. Certainly, the capacity limit of the phonological loop seems to be more a limitation on the number of meaningful phrases than on the number of words per se, but what is not known is how different input streams are integrated and whether the chunks of meaning are held in the phonological loop, LTM, or somewhere else altogether. Baddeley sees value, indeed, in Ericsson and Kintsch's (1995) notion of a "long-term working memory" (LTWM) [glossary] capable of using LTM itself as STM [mediated, in our view, by some sort of second messenger sensitisation process - see Section 6 of our e-paper on "The Neurobiology of Memory" and the entry for "protein kinase studies" in our Memory Glossary.]

3 - Rehearsal: Since a major revision in 1986, the WMM has assumed "separable processes of storage and rehearsal" (p420). The phonological loop was seen as supporting short-term auditory memory, whilst rehearsal was seen more as an articulatory process - as "broadly equivalent to uttering the material to be recalled subvocally" (p420). Again, however, there are difficulties fitting the available data to the model, making it "more plausible to assume some form of general rehearsal" (p420) in which each separate cognitive module does what it can to remember what it most needs to remember [performing not just parallel processing, in other words, but parallel memorising into the bargain]. 

4 - Consciousness and the Binding Problem: The WMM was never intended to address the age-old problem of consciousness.  Nevertheless, the visuo-spatial sketchpad was assumed from the outset to play a role in our awareness of visual images, both at initial presentation and subsequent recall, and the phonological loop was assumed to play a similar role in our awareness of "auditory-verbal imagery" (p421). Research suggests "the existence of a store that is capable of drawing information both from the slave systems and from LTM, and holding it in some integrated form" (p421).

2 - The Proposal

Taken together, these various minor problems indicated to Baddeley "that visual and phonological information are combined in some way" (p418), and the body of the paper expands on the possibilities here. Baddeley calls his new "back-up store" the "episodic buffer", and describes it as "the fourth component" of the WMM. This is how he introduces it ..... 

"The episodic buffer is assumed to be a limited capacity temporary storage system that is capable of integrating information from a variety of sources. It is assumed to be controlled by the central executive, which is capable of retrieving information from the store in the form of conscious awareness, of reflecting on that information, and, where necessary, manipulating and modifying it. The buffer is episodic in the sense that it holds episodes whereby information is integrated across space and potentially extended across time. [.....] It can be accessed by the central executive through the medium of conscious experience. The executive can, furthermore, influence the content of the store by attending to a given source of information, whether perceptual, from other components of working memory, or from LTM." (p421.)

Baddeley gives due acknowledgement to Endel Tulving's prior claim to the term "episodic" (see, for example, Tulving, 1972), but sees a major difference between Tulving's concept of "episodic memory" [glossary] and his own "episodic buffer". For example, clinical evidence demonstrates that episodic memory is one of the first things to suffer in cases of amnesic syndrome [glossary], whereas the episodic buffer is preserved. He also saw the episodic buffer as providing "the crucial interface" (p422) between LTM as a static cognitive resource and conscious awareness as both a phenomenon and a control process. The proposed architecture is shown in Figure 1 .....  

Figure 1 - Baddeley's Episodic Buffer Model: Here we see the basic WMM prior to its present reformulation [in black], together with the enhancement proposed herein [highlighted in red]. The central executive remains the mind's principal attentional control system, and the visuospatial sketchpad and phonological loop are conceptualised as before. The new episodic buffer "is assumed to be capable of storing information in a multi-dimensional code [and] provides a temporary interface between the slave systems [.....] and LTM" (p421).

If this diagram fails to load automatically, it may be accessed separately at



Redrawn from a black-and-white original in Baddeley (2000, p421; Figure 1). This version Copyright © 2004, Derek J. Smith.

As to where the episodic buffer might be implemented biologically, Baddeley reviews a number of neuroscientific studies reflecting on the binding process. He was not looking for "a single unitary anatomical location" (p421), however, because the clear adaptive significance of the buffering resource called for the sort of resilience to injury offered by distributed resources. By and large, he suggests, the frontal lobes are the most likely seat of both the central executive and the episodic buffer.

 3 - Baddeley's Conclusions

 Here is Baddeley's principal conclusion .....

"By emphasising the importance of coordination, and confronting the need to relate WM and LTM, [the episodic buffer proposal] suggests a closer link between our earlier multi-component approach and other models that have emphasised the more complex executive aspects of WM" (p422).

4 - Evaluation

This paper is a textbook example of the successful importation into cognitive psychology of concepts born within computer science. Having already done this to excellent effect twice already by importing the concepts of working memory and the loop, Baddeley now scores a memorable hat-trick with the buffer.  

ASIDE: He was not the first to have the idea, however. Indeed, we can go back at least to Treisman (1964), who used the phrase "peripheral 'buffer' storage" (p15). Then there was Sperling (1967), who proposed a "recognition buffer" capable of converting a visual image into a "program of motor instructions" and of then storing these instructions momentarily. Sperling's approach was then taken up by both Morton (1970) and Mackay (1970), who placed "buffer systems" part-way down the speech motor hierarchy, working on a "store-and-forward" basis [we have reproduced Mackay's diagram in Section 1.2 of our e-paper on "Speech Errors, Speech Production Models, and Speech Pathology", if interested]. We then have a "rehearsal buffer" in the Atkinson and Shiffrin (1971) "Modal" Model of Memory, and no less than four of the things built into the output legs of the Ellis (1982) psycholinguistic transcoding model.

However, none of the above used the buffer as an adjunct to consciousness and attention, and so, on balance, we judge the WMM (Millennium Edition) to be a significant advance over WMM-86. Here are the key arguments put forward in this paper, in revision point format .....

5 - References

See the Master References List