You are in: Home » History of the Unit
6. MEMORY
General Notes. This material has been scanned from the original typescript, while we have done our best to remove errors, some may well remain. You can access other parts of this particular Progress Report either from the menu at the bottom of this entry or by navigating back to the Unit history timeline. Reference for this report are indexed by number and these can be found in a dedicated section also accessible from the menu at the bottom of this entry.
Short-term and working memory (Baddeley, Hitch, Hull, Lewis, Morton, Ottley)
The concept of a short-term working memory has continued to be a -teful one. Recent work has presented the case for the visuo-spatial scratch pad, a system responsible for visual imagery (21) and has explored in further detail the articulatory loop, a concept devised to reflect the role of subvocal speech in memory. Recent experiments iave shown the need to distinguish between the process of articulation iCMlft which is responsible for word length and articulatory suppression effects, and the subsystem which feeds the process of articulation, which is responsible for the influence of phonemic •iailarity on memory. He hope in future to explore the question of .-ether this subsystem is one involving a "deeper" articulatory code which is not subject to disruption by articulatory suppression, or one representing some form of auditory memory (20 U).
The component of working memory most difficult to study is the central executive. We continue to use a strategy of attempting to split off from this central core subsystems which can be analysed separately. In addition to a possible auditory imagery system, it seems likely that we shall be able to identify a verbal memory component which is capable of holding speech-like material without subvocalisacion, and with relatively few attentional demands. It appears to be the case that only when this system becomes overloaded does it make heavy demands on the central attentional system (13). Further research in this area will pay more attention to the microstructure of performance when subjects are required to carry out two tasks simultaneously.
Experiments have been carried out on the characteristics of a stimulus suffix which make it effective as a disrupter of immediate oemory. First it has been shown that the set of suffixes used in an experiment can affect the size of Che suffix effect (166 U). Secondly it has been shown that certain acoustic manipulations of the suffix can lead to a reduction in its effect. This has led to the idea that an early analysis may allow sounds to be classified as 'speech' or 'non-speech' sounds. Precise periodic regularity, for example, is more characteristic of non-speech sounds (158).
A series of experiments (17) has used words of varying length as suffixes and prefixes. Results suggest two effects, a suffix effect of the type studied by Morton et al which is maximal with short suffix words, and affects che last memory item, and a separate effect which is maximal with long suffix words. This effect, influences the retention of earlier items and appears to reflect the temporal decay of a short-term memory trace.
6.2 Psycholinguistics and memory (Barnard, Dennett, Long)
One of the aspects of our recent work on language and memory includes the development of a model of question-answering. This attempts to account for variations in answering behaviour to logically equivalent questions. Initial ambiguities in interpreting questions are resolved by relationships between the structure and content of an incoming question and its referring memory representation (26 U). A new approach to short-term memory has also been developed which seeks to link classic memory phenomena to the operation of an understanding mechanism. This is modelled in the form of a multiprocessor organisation for a set of functionally independent processes for understanding, representing and producing language (27 U).
Work initiated some years back on the analysis of telephone conversations has resulted in the availability of a large corpus of natural telephone conversations. Work based on this corpus (31) examined the temporal structure of natural conversations and indicated that the management of conversational turns is not necessarily impaired by the absence of visual clues as has often been inferred on the basis of more artificial laboratory studies.
A new project has also been started to extend the work on both comprehension and question answering. The former component focuses on text comprehension and the part played by pronouns in the integration of information in texts. The question answering component, like the work cited above, is focussing on the relationship between the structure and content of incoming questions and the structure of the information about the text as it is represented in memory. The work is carried out using computer-presented sentences. The work will also examine issues pertinent to future developments of the use of computer-based systems in the automated office of the future.
6.3 Memory for faces (Baddeley. Simmonds, Woodhead)
We have continued to look for ways in which face recognition might be improved. By analogy with verbal memory, we predicted that a deeper and more elaborate encoding of faces would lead to better subsequent recognition. Unfortunately, accompanying faces with further details of the person's character and lifestyle had no effect on performance, a result that was disappointing from an applied viewpoint (25 U; 274) but theoretically interesting (11 11). A feature which did prove to influence subsequent recognition was the pose of the original photograph. Three-quarter views are consistently better than frontal or profile poses (25 U; 275). Since this is inconsistent with current practices (e.g. passport photographs, police Wanted notices), it is proposed to investigate it further using a more realistic situation involving presentation through the local press, using the response of the general public as a measure. Finally, we have found consistent individual differences in ability to remember faces, indicating the feasibility of selection tests for good recognisers (276). People who are good at remembering faces were also good at remembering paintings, but did not differ in verbal memory. This result suggests separate visual and verbal memory systems, but no separate component within the visual system specialised for remembering faces (277).
6.4 Memory in everyday life (Baddeley, Bekerian, Lewis, Harris, Wilkins)
Although the last 20 years has seen an enormous amount of research on the psychology of human memory, this has been largely confined to laboratory studies, typically of verbal material. It is important to assess the extent to which these results can be generalised outside the laboratory. Investigations have included work on the role of memory in retrospective surveys (5; 19), work on the ability to recall emotional events (35 U), memory for conversations (36 U) and an evaluation of a saturation advertising campaign (37). This involved an attempt by the BBC to acquaint the public with changes that were to occur in the assignment of radio programmes to particular wavelengths. Despite the fact Chat our subjects would have heard this information over a thousand times, their ability to recall it, or to reproduce it on a display analogous to a radio dial, was appallingly bad. This result reinforces laboratory studies which indicate that mere repetition does not guarantee learning.
6.5 Memory aids and remembering to do things (Baddeley, Harris, Lewis, Nimmo-Smith, Wilkins)
Previous psychological work on memory aids has concentrated on the effectiveness of classical mnemonics in improving performance on laboratory tasks. Two interview studies (99; 100) revealed thac memory aids reported to be used most often are not mnemonics but external aids such as diaries. Also these aids were more often used as prompts to remember to do things than for mere information storage. Consideration of the different types of prompting aid used, with their various advantages, points to a number of effectiveness criteria against which such aids can be judged and which help in constructing new ones (99; 101).
We began the investigation of remembering co do chings by re-quescing people to accivate a miniature event recorder at a succession of prearranged times over the course of the week (245), a task similar to the regular taking of medication. We compared a group of subjects who performed well on a verbal memory task with a group who performed poorly. Those with good verbal memory performed less well, indicating that remembering to do something is very different from memory as measured by conventional verbal tests.
These provocative results have been followed up in two ways. First a detailed study of the way in which people observe a clock before responding at prearranged times has revealed that the forgetting of a planned action may he quite rapid (103 U). Secondly, a collaborative study has been set up with local geriatricians to monitor the ability of elderly patients to cake medication as prescribed. Two aids are being compared, one a compartmentalised pill container, and the other a booklet.
6.6 Closed head injury and memory (Baddeley, Harris, Sunderland)
Memory difficulties are among the most common complaints from patients who have suffered severe closed head injury. Previous research has also shown closed head injured patients to be impaired on certain objective memory tests. In our first study we have been investigating Che relationship becween the everyday memory problems reported by patients (or observed by their relatives) and patients' performance on conventional memory tests. Data collection from over 100 subjects is now complete and analysis is under way. We are finding interesting inter-group differences in both objective and subjective measures of performance and interesting relationships among the various measures.
Our future plans include studying whether objective tests given just before discharge from hospital can predict later memory problems (a year post-injury), as assessed by a postal questionnaire that we are currently developing.
6.7 Amnesia (Baddeley)
Research in this area has been limited by a lack of suitable patients. The last few years have however seen a growing consensus of results in this area which has allowed the formulation of a "minimal model" (7) together with a reformulation of the theoretical problem. At present progress in this area seems to be constrained primarily by our inadequate theories of retrieval in normal memory. An extension of the scope of such theories in ways which should bear on our understanding of amnesia has been suggested (11 U).
6.8 Management of memory disorders (Harris, Sunderland)
We have conducted an informal survey of British rehabilitation units and other medical departments thought to be involved in the management of memory problems following brain damage (102 U). This revealed (a) considerable interest, coupled with a lack of expertise or training and (b) a desire for more information, contacts and research on this topic. As a result of the survey we are planning a meeting of interested researchers and clinical staff to exchange information, discuss future research, and decide what else should be done to help clinical staff.
6.9 Theoretical work on long-term memory (Baddeley, Bekerian, Hammersley, Morton)
An attempt was made by Baddeley to fit the results from a series of recent applied memory studies into the framework provided by the currently dominant views of human memory, namely Craik and Lockhart's Levels of Processing Hypothesis, and Tulving's Encoding Specificity Principle. Data were taken from the Unit's work on context-dependent memory in divers (91), memory for faces (274), amnesia (7) and word finding in aphasic patients. A paper pointing out the shortcomings of the Levels of Processing Hypothesis has already been published (1), while a later paper suggesting modifications to this and to the Encoding Specificity Principle is in preparation (11 U).
The concept of headed records has been developed by Morton, Bekerian and Hammersley (following on from work by Norman and Bobrow) to provide a framework for discussing natural memory and its limitations. The record is considered to be the basic unit of memory. The framework is unusual in postulating that individual records have no direct interconnections, hence departing from the currently dominant view that memory is based on an associative network. It is further postulated that records cannot be directly addressed, a major difference from other theories of memory, and that they are unmodifiable and essentially free from decay. Records are accessed from headings which may include a subset of information from the record but could also include other information, such as the emotional state of the individual at the time of laying down the record. The headings themselves are addressed from a description which is constructed from information currently available to the appropriate mechanism (which we call the characterised (157 U).
Morton et al use this framework to account for a variety of natural memory phenomena and to treat a number of standard experimental paradigms, in particular work relevant to the effects of the form of questions on memory for events. Experiments have been carried out on memory for films and stories (92 U), conversation (36 U) and emotional events in an individual's history (35 U).