You are in: Home » History of the Unit
Summary of Research
Cognitive Neuropsychology
This is an area in which the Unit is particularly strong and active, suggesting that it should be covered in this overview. We have, however, opted to discuss the neuropsychological work of the Unit under the various topic areas to which it contributes, rather than as a separate topic. Anyone wishing to pick out this line of work should note the work on attention done by Burgess, Duncan, Marcel and Shallice, the research on language, reading and writing by Marcel, Marslen-Wilson, Patterson and Shallice, and the research on memory by Baddeley,
Gathercole and Shallice. Neuropsychological research on motor skills has been carried out by Wann and Wing, and in the general area of vision by Marcel and by Wilkins.
Attention (Projects 57-59)
The study of attention has been a major theme of the Unit's work for many years, and it comprises three current project areas which form a good example of the points just discussed. Project 57 is concerned with visual selective attention, and is represented principally by the work of Duncan, who has developed and elaborated the theory that was first mentioned in the last Progress Report, and which is now about to appear as a major theoretical paper [13]. The model, developed in collaboration with Humphreys (Birmingham), has been extensively explored using the task of visual search, which has supported the importance of two concepts, namely a template reflecting potential targets, and the characteristic features of the range of non-targets. The model gives a good account of existing neuropsychological and neurophysiological data and Duncan will spend next year carrying out collaborative single-cell recording work on monkeys, with Desimone in NIH Washington, developing the neurophysiological aspects of the model. The possibility of linking neuropsychological and neurophysiological theory represents an exciting development in cognitive psychology, and one which I would expect to become more marked in future years.
A major focus of much of the Unit's work over the last five years has been on Project 58 which is concerned with executive processes in cognition. These are so central and ubiquitous in cognitive psychology that it is tempting to place very large amounts of the Unit's work in this category. For example, the concept of working memory which is concerned with the temporary storage and manipulation of information as part of more complex processing tasks, is clearly very much concerned with executive processes, and could equally well be categorised as research on attention as on memory.
The study of the process or processes responsible for the attentional control of action plays a central role in the Unit's work. Shallice has worked on this problem for a number of years, and his joint model with Norman represents one of the few available conceptualisations of the attentional control of action. This was incorporated by Baddeley into his general model of working memory. In the case of neuropsychological studies, Baddeley and Shallice and Duncan all share an interest in patients with frontal lobe damage, who appear to offer a possible clue to the operation of this system. Using such patients, Shallice and Baddeley have both observed an association between amnesia in frontal patients and confabulation, and both have drawn conclusions about the implications of this for the processes of normal retrieval.
Duncan's interest in the attentional control of action stems in part from his work on driving, leading him to an approach using dual-task performance, a route that has also been used by Baddeley in studying the role of working memory in complex skills such as chess playing and the acquisition of complex computer games. In future, Duncan and Baddeley plan to collaborate in using psychometric techniques to explore the relationship between the functioning of the controlling Central Executive component of working memory and older concepts of general intelligence. They plan to use individual difference measures and secondary task procedures, in a collaborative study with a U.S. Air Force group concerned with fundamental research into individual differences.
In addition to the above-mentioned work, Marcel's demonstrations of the role of intention (in projects 68.3 and 68.4) are clearly aspects of executive function. In the first case the nature of an intended action seems to determine parameters of spatial representation, in the second case the nature of an intention seems to determine accessibility and fluency of motor control.
Given the parallels and similarities that are emerging from work in this area, it must be tempting to ask why the four groups have not worked together more extensively. The answer is that they were initially working on quite separate projects, Shallice on frontal lobe patients, Duncan on driving, Baddeley on short-term memory and Marcel on perceptual-motor control, all with adequate theories that appeared to be working quite well. The fact that these somewhat independent theories are now converging on broadly common and compatible conclusions is a point in their favour. Cognitive psychology is at present simply not in a state where detailed agreement on theory is a sensible prerequisite of a research programme. I would suggest however that the APU provides good evidence that cognitive psychologists with broadly compatible assumptions can work in a fruitful and cumulative way.
Further evidence for this comes from Project 59 which is concerned with the analysis of the structure and functions of consciousness. This is clearly one of the most important but most difficult questions of psychology. Relevant work at the APU includes Marcel's demonstration that information can be available to a subject and can be used without the subject being consciously aware of that information. He has continued to explore this area empirically, using the phenomenon of blindsight, whereby patients with cortical blindness are able to make visual judgements about objects of which they report no phenomenological awareness. There has been a recent increase in the amount of activity concerned with the issues of consciousness, a development in which Marcel has played an important role. He plans in future to turn to the issue of awareness of bodily sensation, a topic that has been curiously neglected by psychologists, despite their considerable interest in the associated problem of pain.
The role of conscious awareness in short-term or working memory is also an important but under-explored topic; the work by Teasdale on intrusive stimulus-independent thoughts, and by Watts and Levey on the role of such thoughts in insomnia, indicate that this area can be tackled empirically by using the very simple expedient of asking subjects at irregular intervals just what they are thinking about. Such simple but potentially useful thought-reporting methods will be used to extend existing models to explore the role of conscious awareness in working memory.
Audition (Projects 60-61)
This is an area in which a single very successful integrated group is now establishing much closer links with other aspects of the Unit's work. During the past five years, the work on auditory warnings has continued to bear fruit, leading to a number of patents and British and International Standard specifications. Sets of auditory warnings have been delivered to the Civil Aviation Authority for civilian fixed-wing aircraft, and have been developed for three separate military helicopters, as well as being fitted in North Sea oilfield helicopters. Other applications include the development of an auditory warning system for hospitals (intensive care wards and operating theatres), and railway track-side crews.
Although this is an example of a very successful application of basic research in psychoacoustics, it also illustrates the potential danger of successful applied work, since our psychoacoustics section could spend its whole time developing alarms for the very many situations in which they are needed. Ideally, these projects would be taken on by industry; however our industrial partners typically do not have the necessary psychological expertise. An alternative is to delegate the running of the project to a junior scientist at the APU. In principle this is feasible, but unfortunately suitable psychoacousticians are few and far between, and once trained tend to leave for more lucrative jobs elsewhere.
Meanwhile Patterson has been actively developing a model of auditory sensation and extending it to the problem of speech recognition. This has led to his being the coordinator of a European funded ESPRIT Basic Research Action which involves collaboration between the APU and speech scientists from a range of disciplines and countries. Within the APU, the ESPRIT project will involve collaborative work on speech between a number of previously separate groups including that of Cutler and Norris on speech perception, Houghton and Shanks on connectionist modelling and Baddeley on phonological memory.
Language and Speech (Projects 62-64)
Although language consists of separable words, spoken language in fact comprises a virtually continuous stream of sound. This raises the problem of how to segment the incoming signal into separate words. Cutler and Norris in collaboration with Mehler and Segui in Paris have shown that this process differs between French, where the discrimination is based on syllabic units and English, which appears to rely on the tendency for English words to have their stress on the initial syllable.
Cutler and Norris propose to test a number of models of this process, which is of course very important for attempts to develop devices for automatic speech recognition. As mentioned earlier, this will involve the detailed development of connectionist models of speech recognition, as part of the ESPRIT BRA project.
The Unit has continued to be active in the study of reading and writing and their breakdown in dyslexia and dysgraphia. Detailed analysis of the reading deficit in patients who have become dyslexic following brain damage has been continued by Patterson, Marcel and Shallice. This again is an area that has been strongly influenced in recent years by the development of ambitious connectionist models of normal reading. Patterson has been collaborating with McClelland in Pittsburgh and Seidenberg in Montreal to develop such a model, her concern being particularly with the exploration of the extent to which "lesioning" such a model will produce errors that are analogous to those produced by certain types of dyslexic patient. A similar approach is being pursued by Shallice in collaboration with Hinton in Toronto. This work has so far proved extremely promising and will be actively pursued over the coming years. Other aspects of the Unit's programme in reading research involve (a) relationships between impaired reading and other language skills such as speech segmentation, (b) models for skilled and disordered reading in various orthographies and (c) plans for more sophisticated neurological localisation of components of the reading process via PET-scanning activation studies.
Learning and Memory (Projects 65-66)
This continues to be an active area, with the Unit involved in research on both working memory and long-term memory. One of the major developments in this area over the last five years has been the attempt to explore the real-world validity of one subsystem of working memory, the Articulatory or Phonological Loop, a system that is concerned with the phonological storage and rehearsal of verbal material. Earlier work was largely confined to laboratory studies, which have left open the question of whether the system is of any practical importance. However, recent studies by Baddeley in collaboration with Vallar and Papagno in Milan, and Wilson in Southampton involving neurological patients with short-term memory deficits suggest that the system is of real importance, both in understanding language, and in learning new phonological sequences, such as new words in one's native or a foreign language. Further research on normal children by Gathercole and Baddeley indicates that the capacity of this system has a major influence on the initial acquisition of vocabulary and is important for the early stages of reading. Future work aims to continue longitudinal studies to investigate further the ramifications of the system. We also plan to develop a more specific model of the phonological store, as part of the previously mentioned ESPRIT BRA speech project, using parallel distributed processing or connectionist techniques to implement models of memory storage and retrieval in ways that appear to be plausible both psychologically and physiologically.
One of the major features of recent years in long-term memory research has been the growth of interest in autobiographical memory, the capacity to recall incidents of one's earlier life. This area of memory research has until recently been curiously neglected, probably because of methodological problems in deriving quantitative measures. It is however of considerable practical importance in areas such as eyewitness testimony, and in assessments of the effects of case presentation features on the memory and decisions of juries, both areas that have been studied by Bekerian.
Clinical disturbance of autobiographical memory has also evoked considerable interest; this includes work by Shallice and by Baddeley on confabulation in frontal lobe patients mentioned earlier, and research by M. Williams on disturbances in the precision of autobiographical memory in patients who have attempted suicide. The observation of varying patterns of deficit in autobiographical memory in amnesic patients has led to the development of a clinical test of autobiographical memory by Baddeley in collaboration with Wilson in Southampton and Kopelman at Charing Cross. This is part of an ongoing programme concerned with the development of clinical memory tests which has so far produced a test of everyday memory in collaboration with Wilson and Cockburn of Rivermead Rehabilitation Centre in Oxford. We plan to standardise and publish clinical tests of verbal and visual long-term memory, semantic memory and of the various components of working memory.
Connectionist and other network models have also recently had a major impact on theories of learning and memory. Such parallel models appear to capture the characteristics of the processes of storage and retrieval much more naturally than does the more conventional computer analogy, in which information is stored perfectly in a range of discrete memory locations. We believe that the full exploitation of parallel network models for psychology will require a combination of the understanding of their strengths and weaknesses with a willingness to confront them with carefully collected empirical data. The
Unit has a strong experimental tradition, and is currently developing its skills in modelling. Houghton is actively exploring connectionist architectures for both long- and short-term learning, while Shanks is concerned with models of concept acquisition and decision. Norris is concentrating on the use of connectionist models in speech perception, a process that almost certainly involves a component of phonological storage. Finally, McLaren, who will be joining the Unit as part of his King's College Research Centre Fellowship, will be working on a connectionist model of spatial attention, probably in collaboration with Duncan.
Perceptual-motor Skills (Projects 67-68)
In the past, Wing has been concerned with developing a theoretical account of noise in the timing of movement. With his theoretical account of timing now widely accepted, Wing is currently evaluating its applicability to new tasks including the coordination of multiple effectors. Recently, in collaboration with Ulrich in Germany, he has started to extend the account of motor variability into the force domain.
Time and force are rather elemental aspects of voluntary movement. A more usual focus of attention in research into the control of action is position in space, or, in the case of skills such as writing or drawing, the trajectory of position changes as reflected in the trace left by the pen on paper. Wann and Nimmo-Smith have been investigating strategic constraints on the way such voluntary positioning movements are executed. Mechanical equipment such as a computer pen-plotter is often driven in a very jerky style. This contrasts strongly with the smooth changes in velocity typically observed in human handwriting and which has led people to suggest that movement trajectories are organised to minimise the rate of change of acceleration. APU research has challenged some common simplifying assumptions within this perspective. Reformulated equations of motion are supported by behavioural data but, more importantly, they suggest a way in which the motor system could be informed of the consequences of jerky movement. This therefore constitutes a model of feedback that has implications for the refinement of control in motor development or rehabilitation of neurologically impaired patients. It is expected that this will develop into a useful theoretical basis for current research projects at the APU in motor skills training (even though, regretfully, these will proceed less rapidly with Wann's lectureship appointment to Edinburgh).
Although variability in output means that our movements are not perfectly accurate, the motor system appears to compensate for errors automatically so that our actions usually achieve a successful outcome. The coordination between opening and positioning of the hand when reaching for an object in the environment is an example that has been studied by Wing in collaboration with Fraser and Turton. If normal use of visual information to guide the hand toward the object is restricted, a compensatory increase in hand opening is observed that improves the chances of encompassing the object. Current research by Haggard seeks to assay the separability of these strategically coordinated elements by imposing additional demands at various phases of reaching. Comparison of the effects of cognitive interference by a secondary task and the effect of mechanical perturbation to movement seem particularly promising.
Thinking (Projects 69-70)
Johnson-Laird, Anderson and Byrne have been further testing the proposal that much thinking involves the setting up and manipulation of mental models. They have systematically compared this approach with an alternative view of thinking based on the operation of sets of logical rules, an approach that dominated the area prior to Johnson-Laird's theory. The work has been concerned with two criticisms, that the model is too vague to be specified, and that it is not demanded by the empirical evidence. Areas of thinking that have been considered include the comprehension of discourse, everyday informal inferences, and deductions based on prepositional reasoning, relational reasoning and quantification. In each case a formalisation of the proposed mental model has been developed and programmed, and empirical evidence has been collected. The evidence consistently favours the mental model approach and in a number of cases produces data that are very difficult to explain in terms of the rival rule-based hypothesis. This work will continue, but unfortunately not in Cambridge, since Johnson-Laird has accepted a Chair at Princeton University.
Until very recently, studies on animal learning and cognition have proceeded quite independently of equivalent work on human subjects. There are now welcome signs of the two areas beginning to come together, partly under the influence of the development of connectionist learning models, and partly as a result of the development of cognitive theories of conditioning in animals, which appear to have implications for human cognition. Within this latter category, Shanks has been testing models of the judgement of causality, and finding that an associative model based on principles of conditioning offers a better account of the data than one based on normative rules. Similarly, animal conditioning models have proved highly successful in accounting for the results of a categorisation task in which, for example, subjects have to decide on the basis of symptoms whether a hypothetical patient is suffering from one disease or another. The conditioning model has been simulated using a connectionist network; future work will study its capacity to cope with more complex categorisation tasks, including that of diagnosticians making decisions on the basis of actual disease patterns.
Human-Computer Interaction (Projects 71-72)
This is an area in which the Unit has worked for many years, starting at a time when it was virtually unexplored, and reaching a point now at which it is one of the most intensively studied areas of applied psychology. One of the difficulties in working in this area is that the technology changes so rapidly that particular systems become obsolete before they have been adequately studied. This means that much of the applied work in the area has to be based on rule-of-thumb techniques, which ideally should be based on earlier evidence. Since the user of such information is typically not a psychologist, a major problem is how the necessary information can be delivered, and it is this that has concerned both Barnard and Young in recent years. Both have chosen to use AI techniques, with Young and Green opting to develop a cognitive model of the user that allows the software designer to program in the system and task characteristics, while Barnard has developed an expert system in which the rules are based upon Interacting Cognitive Subsystems, his very ambitious model of cognition. Working with these ideas, Young and Barnard are currently collaborating on an
ESPRIT Basic Research Action, aimed at bringing modelling techniques from cognitive and computer science closer to the practical needs of software designers.
We remain active in the related area of exploring the cognitive demands of technical communications. Wright has continued her detailed analysis of the representation of linguistic information, exploring the conflict that sometimes occurs between the linguistically graceful and the logically desirable. She continues to provide practical help for those concerned with designing forms, as well as other varieties of communication, for example via the maps and signs that attempt to guide the visitor around a large and complex hospital. Investigation of the problems people have in communicating technical information suggests that they are often insufficiently aware of the alternatives available, whether to use words or pictures, and whether to provide an enquirer with the quickest or the easiest way. The development of desktop publishing offers an increasing range of potential resources for the communicator; if they are to be used effectively, we need a more detailed and sophisticated understanding of the problems facing both the communicator and the recipient of the information.
Transport System Users (Project 73)
Research on the behaviour of transport users has been a prominent feature of the Unit's work for many years, and continues to be an area of considerable activity. A previous project involving Brown, Duncan, Groeger and P. Williams, concerned with comparisons amongst novice drivers, normal experienced drivers and experts revealed, among other things, that the normal experienced drivers were poorer than either of the other two groups on certain components of driving such as mirror-checking. The tasks concerned seemed to have the characteristic that failure to perform them does not normally lead to immediate feedback, although in the long run failure can be extremely serious. This has clear implications for road safety.
Detailed analysis of the skill of driving suggests that it comprises many different and independent subskills, making prediction of success in trainees very difficult. Executive functions appear to be less important than might have seemed likely, a conclusion that is consistent with a small study of head-injured patients. This showed that even patients with clear evidence of executive dysfunction show little impairment in driving skills, and suggests that medical criteria for return to driving should be re-examined.
A series of new externally-funded projects involving Groeger, Brown, Chapman and Grande has just begun. One collaboration (with Paris and University College London) concerns estimating time to coincidence or collision. It attempts to model the decisions drivers make at junctions. This should provide assistance to traffic engineers devising countermeasures to the problems of risk-taking at junctions. A second related study is concerned more directly with perception of risk, using video tapes of specific situations, and requiring drivers to assess the danger and relate it to their driving behaviour (this is funded by a contract from General Accident PLC).
A third project, this time funded by the CEC DRIVE Research Initiative, is concerned with the role of feedback from driving instructors during the process of learning to drive. It will attempt to relate this to the type of accidents encountered subsequently, using this information to identify the type of instructional support that might be given to the learner by an "intelligent" vehicle. This project is carried out in collaboration with the Traffic Research
Centre at Groningen. Brown is due to retire in three years' time, and the future of this area is one that is of considerable concern. We would hope to continue to work in this area, provided we can recruit senior staff who are capable of blending the many practical problems this area offers with the prospect of genuine theoretical development.
Vision (Projects 74-76)
A major component of the Unit's work in this area was the development by Watt of a theory of human vision and visual attention that combines information from psychophysics, computer science and cognitive psychology and is expressed in a recent book [561]. Unfortunately, Watt has just left to take up a Personal Chair in Psychology and Computer Science at Stirling. We hope to replace him by a senior appointment in a similar area.
Meanwhile, Wilkins' work on visual discomfort and environmental design continues to be extremely productive. He has shown that the characteristic flicker and stripe patterns that trigger epileptic attacks in photosensitive patients are also a major source of complaints of eyestrain and headache in non-epileptic subjects. This has important practical implications for a wide range of environmental design issues, including fluorescent lighting, computer VDUs and even the design of typeface, where certain types can be shown to produce clear stripe effects, and accompanying complaints of eyestrain and headache. The potential environmental significance of this phenomenon is very substantial. At the same time, Wilkins and Neary have been exploring possible mechanisms, elaborating and testing a theory of eyestrain described in the last progress report.
In the meantime, Wilkins has been applying the visual techniques developed to provide clinical tests useful in the diagnosis of a range of diseases including multiple sclerosis, optic neuritis, diabetes and glaucoma. Future proposals involve the continued investigation of visual discomfort in the office environment, and ways in which it can be alleviated, together with a continuation of the investigation and modelling of the visual symptoms associated with headache and reading.
Psychophysiology (Projects 77-78)
Work by Wilkinson on sleep has involved continued analysis of data collected from earlier fieldwork on individual sleep disturbance, together with the development of measures of sleep on/offset, using both behavioural and physiological measures. Future work will develop portable sleep detectors for use in industrial and field situations. Wilkinson is due to retire in a year's time, an event that colleagues in North America and Britain plan to mark by organising a meeting in Cambridge on sleep and performance.
Levey, who forms the second senior member of the psychophysiology section, is also due to retire, bringing to an end his very productive line of research on human classical conditioning, carried out in collaboration with Martin at the Institute of Psychiatry. Plans for the future of the psychophysiology section will be discussed below.
Emotion and Cognition (Projects 79-80)
The Unit has sought to develop links between experimental and clinical psychology in the belief that any adequate theory of cognition must be able to incorporate an understanding of the role of motives and emotions. Since it is difficult to manipulate emotions ethically within the laboratory, we felt that work on clinical patients could be potentially of great significance for understanding normal cognition. At our last progress report, we had established a group in this area, but the Board felt that it was still too soon to assess the success of our attempt to bridge the gap between clinical and cognitive psychology. We are now in a much stronger position, with an active and coherent group that is not only developing strong connections between these two areas of psychology but is also beginning to influence theory within each of them.
One indication of the extent to which research on cognition and emotion has become integrated into the work of the Unit is the fact that no fewer than eight cognitive psychologists who do not see emotion as their principal area of research, have been involved in research in cognition and emotion over the period since our last progress report. These include Baddeley, Barnard, Bekerian, Conway, Dritschel, Johnson-Laird, Levey and Wilkins. A second achievement of the group has been the publication by Williams and Watts in collaboration with Mathews and MacLeod (of St George's Hospital Medical School) of Cognitive Psychology and Emotional Disorders, a book that has been very favourably reviewed, and which we believe reflects the growing tendency for cognitive psychology and clinical psychology to interact fruitfully.
The work in this area falls into two sub-areas, the interaction of emotion and cognition (Project 79), and the analysis and treatment of emotional disorders (Project 80). Past work on cognition and emotion has included the communicative theory of emotions devised jointly by Johnson-Laird and Oatley of Glasgow. Bekerian and Conway have explored the structuring of emotional and conceptual knowledge in a way that links up with previously-described research on semantic and autobiographical memory, and with the work by Williams on autobiographical memory in attempted suicide patients.
Teasdale has been concerned with the role of mood congruent memory in maintaining negative mood states, and has gone on to study methods of controlling the flow of negative thoughts, which have been shown to help maintain the depressed mood state in patients. A series of experiments have used Baddeley's working memory model to explore methods of controlling such thoughts, demonstrating the importance of the Central Executive component rather than peripheral slave systems in the process. Future plans involve adapting Barnard's model of Interacting Cognitive Subsystems to explore further the control of negative thoughts, with a view to developing therapeutic methods for encouraging and maintaining more positive affect in such patients. Barnard plans to devote more time to this and rather less to human-computer interaction throughout the coming five years.
Work on the analysis and treatment of emotional disorders has concentrated on patients suffering from depression, and from phobias. Work by Watts on depression has demonstrated genuine memory deficits which are not simply due to subjects having a bias against responding freely, and has suggested that the problem occurs at least partly as a result of the reduced organisational activity indulged in by the subject. Depressed patients tend, for example, to fail to pick out the more important features of text, and are more likely to rely on simple rote rehearsal. On the other hand, use of imagery is unimpaired, and Watts has successfully applied these findings to the treatment of work problems in students. This particular line of research is regarded as complete.
Attempted suicide presents a major practical problem which at present has not yielded readily to treatment. Williams has investigated the autobiographical memory of patients in an attempt to elucidate the nature of the hopelessness that appears to be an important determinant of whether or not the patient will attempt suicide again. Williams' work suggests that although such patients can access autobiographical memories, they report memories that are excessively general, and appear to have great difficulty in producing detailed specific recollections. Subsequent work showed that a similar phenomenon occurs in other depressed patients, and furthermore indicated that the phenomenon is also present in ex-patients who have recovered from their suicidal crisis. He proposes to examine this phenomenon in more detail, in particular exploring the relationship between autobiographical memory, hopelessness and the control of action, exploring the extent to which autobiographical memories of the past can be related to the patient's prediction of what will happen in the future. In collaboration with Scott and Ferrier at Newcastle, Williams will be studying a group of depressed patients longitudinally, with a particular view to exploring whether the generic style of recall identified in depressives will predict dysfunctional attitudes and subsequent relapse.
A series of experiments on patients' memory for phobic objects (such as spiders) by Watts and collaborators has shown a pattern of results with some similarities to the lack of specificity in autobiographical recall by depressives. Phobics tend to have both very poor recognition memory and rather undifferentiated cognitive representations of phobic stimuli. This lack of precision appears to be associated with poorer subsequent emotional habituation to the stimuli. It is as if, in both the autobiographical memories of suicide patients, and the perception of fear-inducing stimuli by phobics, the degree of associated anxiety can be reduced by avoiding detailed processing and hence "distancing" the stimulus. Unfortunately however, this distancing mechanism appears to make it harder for the patient subsequently to adapt and come to terms with the source of anxiety. This emphasis on impoverished stimulus processing differentiates Watts' approach from that of Lang who emphasises the importance of the response aspects of anxiety. Future work will explore this hypothesis further, attempting to improve phobic encoding by inducing positive mood, and by attentional instructions. If successful, this should not only improve subsequent recall, but also lead to better desensitization.
A related problem is that of dealing with emotional memories which may continue to cause acute distress in patients long after the occurrence of the event. Recent work suggests that these memories again tend to be preoccupied with the patient's emotional response, with little attention to the details of what occurred. It is suggested that this prevents effective emotional processing, and that better ways of working through emotional memories may be available. The work is closely related to that of Williams on generic autobiographical memory, and of Teasdale on intrusive thoughts and mood biasing.
Finally, a collaboration between Watts and Levey is investigating a cognitive treatment of sleep dissatisfaction. This work is conceptualised in a working memory framework, and uses articulatory suppression as a means of blocking verbal behaviour, while requiring suppression to occur at irregular intervals, a task that occupies but does not stimulate and overload the Central Executive component. Preliminary results appear to be encouraging, and a formal trial of the treatment is just beginning.
The Next Five Years
While we have attempted in the body of the Progress Report to indicate for each research area, the direction we plan to take, given the number of projects, it may be difficult for the reader to gain an overall feel for the broad research strategy of the Unit. The present section aims to fulfil that function.
In many ways the broad theme of the Unit's work has remained remarkably constant over its 45 years of life. We began with a strong commitment to theory, based on the concept developed by Craik of the model as a means of theorising, and using the electronic computer to model human cognition. We are still strongly committed to theory, still use an information-processing metaphor, and tend to rely quite heavily on computer modelling. In our last Progress Report we made a plea for better equipment, and I am delighted to say that the Council responded very positively.
Much of the Unit's theoretical work continues to be concerned with the detailed modelling of specific cognitive processes. However, the development of the techniques of Artificial Intelligence has led to the creating of unified theories of cognitive processing. Within the Unit three broad approaches to achieving such integration are represented. One approach is based on connectionist architectures (Parallel Distributed Processes). Such models emphasise the interaction among large numbers of inter-connected subunits. Another approach to parallel processing has been achieved by an architecture comprising a number of dedicated processing modules. In this approach, which is typified by the Interacting Cognitive Subsystems model, structural complexity is traded against processing simplicity. This work provides a link with the third approach which is based on models using a Production Systems architecture, with cognition being modelled as the operation of rules that have been learned. The next five years will see clarification of the scope and power of these theoretical approaches. My own view is that the viability of these theories depends upon the level of the cognitive functions being modelled, with some tasks such as pattern recognition being much more readily captured by connectionist models, while other tasks such as planning and problem solving may be more amenable to a Production System approach. It is likely that any powerful account of complex human behaviour will require an integration of these different approaches and the Unit is relatively unusual in having groups active in all these areas, concerned with a diversity of cognitive functions and with a strong commitment to linking these approaches to empirical data.
A second feature of the Unit's work from its inception has been a concern with applying psychological knowledge to the solution of real world problems. Rarely does psychological theory lend itself to simplistic application, and the solutions that the Unit has been able to provide takes many forms. In some areas (e.g. audition and speech) it is possible to deliver products which solve or contribute to the solution of practical problems. In other areas (e.g. driving and interacting with computers), the diversity of potential solutions means that ways have to be found for helping non-psychologists to understand the cognitive implications of the options they are considering. The Unit is starting to explore ways to bridge this communication gap, including the use of expert systems and simulations of cognitive functioning that designers can run in the task environments under consideration. Over the next five years we will assess the utility and power of these approaches to 'giving psychology away'. Moreover we expect the classes of communication being developed to have relevance to other problem domains where the application of psychological knowledge is far from simple or obvious (e.g. rehabilitation).
The Unit in its early days had a relatively strong biological emphasis, provoked in part by the military problems of the Second World War which generated an interest in environmental stressors such as heat and noise, as well as in factors such as fatigue and combat neurosis. During the 1970s, this aspect of the work of the Unit tended to become less prominent, but revived again with the advent of the group studying the relationship between emotion and cognition in normal and emotionally disturbed subjects.
I would see the development of research on cognition and emotion involving three phases. The first of these has been concerned with establishing that cognitive psychology does have a role to play in this area. I believe that we have already begun to show that the techniques and concepts of cognitive psychology can throw light on clinical problems, which in turn enrich the cognitive theory. The second stage is one of developing and testing theoretical models that incorporate both emotional and cognitive factors. This has already begun and is reflected in the many collaborations that have developed between the group whose primary concern is with emotion and cognition, and a range of colleagues in more mainstream cognitive psychology. I would see the third stage as involving the development and testing of methods of treatment. This has already started in the case of the proposed treatment for insomnia by Watts and Levey. However, in general it is realistic to expect the development and application of work in this area to be gradual, with theoretical insights leading first of all to experimental treatments, leading over a period of years to the development of standardised treatment packages and finally to their evaluation by multi-centre trial.
A second area in which the clinical and biological emphasis has steadily increased, is of course that of neuropsychology, an area in which the Unit has had, and continues to have considerable strength. Much of the theoretical progress in this area has come from applying cognitive models to single cases, often selected as having a very pure functional deficit. I believe that we are now developing our technical and conceptual skills to a point at which we are able to tackle patients with more than one deficit. This is an important development for two reasons, first because it increases the number of patients that can profitably be studied, and secondly because it means that we are in a much better position to tackle the complex cognitive deficits that accompany neurological diseases such as Alzheimer's Disease, Parkinson's Disease, and schizophrenia. This leads to a natural alliance between the cognitive psychologist, the clinician and the neuroscientist. This aspect of our work is likely to be further stimulated if the proposed Interdisciplinary Research Centre goes ahead. Even if it does not, the contacts made in discussing IRC proposals themselves seem likely to stimulate collaboration with our neurobiological and clinical colleagues in this area.
The major change that we are proposing over the next five years is of course the replacement of the Psychophysiology Section with a group concerned with rehabilitation. The problem of rehabilitating the brain-damaged patient is of course one of enormous practical and social importance, and one that has in the past been sadly neglected. We believe that developments in neuropsychology have now reached a point at which it can start to tackle these issues with some hope of success, and that the APU with its tradition of blending pure and applied research is in an ideal position to collaborate with the rapidly developing Cambridge Medical School in developing a centre of genuine excellence in this area. It will require the goodwill and collaboration of the regional administration, of the University Medical School, of our clinical rehabilitation colleagues, and of course the strong support of the Board. We have already begun local discussions and made initial contact with the Department of Health, and so far there appears to be considerable general enthusiasm for the proposal.
The proposed development will involve a substantial expansion of the clinical service at Addenbrooke's, with our own involvement being an integral part of such a newly-developed Rehabilitation Centre. Realistically, it will probably be four or five years before such a centre is up and running. Bearing that in mind, I would hope to see a gradual build-up of the Unit's rehabilitation commitment, initially relying on sub-optimal accommodation, but comprising by the end of the five-year period a team of some eight or ten scientists (see Section F). I would hope that by the time of our next progress report we would already be starting to demonstrate the fruitfulness of such an approach, while accepting that making real progress in this important but challenging area is likely to take substantially longer than five years. It is for that reason that an MRC Unit, with its remit for engaging in long-term research programmes is particularly well-placed for developing such an enterprise.
Categorisation of Publications
Over the period reviewed, the Unit has produced somewhere in the region of 750 publications, raising the problem of how to present them in a reasonably comprehensible and coherent way. We have opted to break them up according to the 10 broad subject areas covered by the report, and to present each set of publications after the relevant section of the report. We have tried to give further information about each publication by classifying them according to the following scheme:
A: Books. These range of course from sets of edited papers to major scholarly works that are likely to have an influence that goes far beyond that of most individual papers. I would particularly like to single out Shallice's "From Neuropsychology to Mental Structure" which is the first comprehensive analysis of the area of cognitive neuropsychology, and as such I think will become a classic. Secondly, I would like to draw attention to a book that gives an account of the attempt by a team at the APU together with colleagues at St George's Hospital Psychology Department to build an empirical bridge between clinical problems and the concepts and methods of cognitive psychology, published as "Cognitive Psychology and Emotional Disorders" by Williams, Watts, MacLeod and Mathews. Finally I would like to mention Johnson-Laird's "The Computer and the Mind" which, by giving a very scholarly but well written overview of cognitive science, is a very good example of the Unit's concern to present science to the general public.
B: Journal Articles. These represent the solid core of our work, and comprise work that has been subjected to the normal processes of peer reviewing. The bulk of it is empirical, but we also include a number of theoretical contributions in this category.
C: Invited Chapters and Commentaries. This section comprises mainly articles provided in response to an invitation, often to take part in a symposium, or to participate in an open discussion on a particular topic. In some cases, the papers represent the output of small highly-focussed workshops which can be extremely productive in bringing the available experts in a topic together to address issues of common interest. Although they are often not rigorously refereed, in most cases the invitation to attend is strictly limited, and the freedom to write in a more speculative vein can be extremely productive.
D: Conference Proceedings. These clearly overlap with C, but differ in that they tend to have larger numbers of people attending and to be consequently more variable in quality. They can however be remarkably productive, as was the case with the proceedings of the two conferences on "Practical Aspects of Memory" which brought together a very wide range of previously disparate researchers with common interests in applied problems, and which have over the last 10 years served as a focus for work in this area. Furthermore, invited keynote addresses at such conferences can be very effective in communicating views to a wide audience.
E: Technical Reports, Theses and Tests. These represent publications that communicate technical information to a limited audience. This is an extremely important mode of communication in a number of applied areas, notably in those of psycho-acoustics, human-computer interaction and in research on practical issues of road transport. In terms of their practical impact, technical reports and papers presented at specialist meetings are often much more important in influencing policy than are papers in academic journals. In this section we also include reports in response to requests from government bodies such as the Cabinet Office or the Roskill inquiry on fraud trials. We also include unpublished PhD theses in this category.
F: Dissemination. Cognitive psychology has implications for, and draws from a range of disciplines within both the cognitive and biological sciences. We regard it as part of our responsibility to try to inform our colleagues about cognitive psychology, at the same time of course as attempting to learn from them. We are also fortunate in working in an area that is often of general public interest. This leads to the opportunity to talk about our work on radio and television, but also gives opportunities of informing the general public about cognitive psychology in slightly less ephemeral ways through our contribution to books for the general public.
While we hope this categorisation will help the reader, it should be borne in mind that it is to some extent arbitrary, with a substantial number of papers being potentially classifiable under more than one topic and categorisation, while a few do not fit readily into any of the available categories.