You are in: Home » History of the Unit
8. COGNITIVE ERGONOMICS
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.
8.1 Design of information (Barnard, Threlfall, Wright)
This is an area in which the Unit has been active for a number of years. Our recent concern has been with devising procedures that will allow the research findings from our own and other work to be applied to practical design problems (290). What is required here is an "applications theory"; such a theory should summarise existing knowledge and direct further research, while offering the practical designer a set of procedures for evaluating existing designs, and methods for improving them (286). An adequate theory will inevitably be complex, requiring designers to have access to expertise in a range of disciplines including behavioural research (294). However, we have already shown that such an approach (a) has more power and generality than such empirical design tools as readability formulae (292) and (b) provides useful guidelines for materials as varied as manuals, forms, tables and narrative prose, as well as the range of new technologies that are currently developing (285; 288; 289).
Research into the psychological factors influencing ease of understanding written materials has shown that people's expectations are crucial. Inappropriate expectations lead to errors in use of tables (300) and the inaccurate completion of forms (30). It is therefore critical that the writer should bear in mind the knowledge and expectations of the reader, a fact that applies to materials as diverse as medical trial forms (299) and legal language (291). Nevertheless writers have difficulty incorporating such advice into their own writing style (297 U). Hence the importance of understanding the process of writing (287). We are therefore developing a research programme which has begun by examining people's editing skills (297 U). The programme will go on to study the contribution of the writer's memory and knowledge structures to the writer's organisation of ideas, and will examine the effect of the communicative task on the expression of these ideas.
8.2 Man-Machine communication
8.2.1 Pictographic instructions for equipment (Barnard, Hull, Marcel)
This project has examined how people understand and follow pictographic instructions for using equipment. In the first phase experimental work concentrated on the structural segmentation of sequences of pictograms. Although structural factors proved to have significant influences on performance, other factors emerged as important contributors to understanding and equipment use. They included (1) people's conception of both the equipment itself and the task in hand; (2) their strategies of combining equipment operation and instruction usage and (3) their assumptions about graphic representation and communication (135; 29). These factors are being investigated in a second phase of the project, which has received further support from British Telecom.
This work is part of the Unit's broad interest in Man-Machine Interaction, but also has theoretical relevance to the relation between tacit and conscious knowledge as well as to the processes underlying the planning of actions.
8.2.2 Conceptual models and cognitive skills (Young)
Young has used a Production Systems approach to investigate users' conceptual models of interactive devices, such as computer text-editing programmes or pocket calculators (304). By analysing several different designs of calculator it has been possible to make explicit the models implied by the calculators' behaviour, and from these to derive predictions about the users' performance (305). Complexities in the model, for example, give rise to corresponding difficulties in the use of the device. The models are of at least two different sorts, and this, together with their concreteness and tractabi1ity, has served to elucidate the rather confused notion of "conceptual model" itself (306 U).
These ideas have application in a project being carried out for British Telecom on accessing structures in viewdata. Two aspects of the Prestel system are being examined. Work is now in progress examining the difficulties in terms of the mismatch between the user's conceptual model of the topic area (or of Prestel itself) and the model assumed by the information designer. This investigation will be extended to study the difficulties encountered when browsing through large networks of information, in particular to examine the factors that determine whether the user remains oriented or becomes "lost".
To help with both the calculator and the viewdata studies, we have designed and built a force-sensitive platform for recording keypresses, which we hope will be of some general utility. Advice is currently being sought on the possibility of patenting and/or marketing this device.
8.2.3 Man-computer interaction: the IBM project (Barnard, Hammond, Long, MacLean, Morton)
A substantial part of the Unit's recent work on human-computer interaction has been carried out under a collaborative research project with IBM (1977-1980). The first research project concentrated on the use of interactive systems by "casual" users - people such as town planners, engineers, or accountants, who have little or no formal training in computing. The initial emphasis of the work was to explore a methodological framework for research in this new ares. The proposed framework make's use of a variety of conceptual tools and empirical methods (159). In an attempt to explore the problems, a study investigated the effects of introducing a new interactive computer system into the working environment of a large local authority. This study revealed many illustrative problems as well i some of the beneficial effects of computerisation (94 95; 122). A more, detailed observational study was also carried out to document, identify and classify the types of difficulties encountered by users attempting to solve problems interactively (96; 97; 98). These observational studies served to help identify underlying factors which could usefully he studied in the more controlled environment of a laboratory which permits the systematic variation of interactive dialogues. So far we have investigated the structure of command sequences, focussing on the consistency of this structure and its compatibility with natural English. We have also studied effects of the design of information displayed on the terminal (28), The confusability of the names for commands and the conceptual difficulty of the user's task have also been shown to influence ease of use (93). Other studies in the series have examined relationships between instructional information and the structure of the dialogue, the influence of the phrasing of "questions" to be solved and the extent to which users have a constrained or free choice for ordering elements within a computer command sequence. The research has also explored the factors which influence the design decisions made by professional programmers when developing interactive languages for "casual" users.
The picture which is emerging from the results is a complex one. The usability" of interactive systems is clearly determined by a multiplicity of variables whose specific influences require careful conceptual and empirical evaluation. Adequate, but simple guidelines for system designers are likely to be difficult to establish and validate, and we may have to explore more sophisticated routes to designing truly usable systems.
The progress made on this first collaborative project was sufficient to encourage IBM to extend and expand their support for the Unit's work in the form of a new three-year project. The aims of this new project are to build on the body of empirical findings obtained in the earlier research, both by examining a broader range of interactive systems in use in industry and by continuing the laboratory studies of specific variables underlying interactive dialogue. In addition, we hope to develop further theoretical techniques for analysing human understanding in this kind of complex task environment.
8.3 Perception and decision in transport systems (Brown, Copeman)
During the past three years, effort devoted to this project and to the T.R.R.L. contract on "Drivers' Attitudes and Behaviour in Relation to Road Traffic Offences", has been largely confined to the modelling of road accident causation, using the wide variety of data available including those collected during earlier T.R.R.L. contract research on experiential and exposure factors in accident liability (e.g. 64). Given that the vast majority of road accidents result from mismatches at the interface between the individual and other components of the traffic system, attention has been concentrated on the risk incurred when certain types of individual, or certain groups, encounter specific traffic conditions. This approach keeps open the remedial options of improving driver behaviour or reducing traffic system demands.
A model of the differential development of perceptual-motor and decision skills has been used to explain the experiential effects observed in road accident statistics (48; 51). Decision theory has been used to model traffic offending (49) and to demonstrate the individual's problem in balancing costs, which arise mainly from factors intrinsic to the traffic system, against payoffs, which relate largely to extrinsic factors. Evidence for the effects of speed stress on individual drivers and on traffic flow has been reviewed (45). Psychological understanding of driving fatigue and its prevention have been reviewed in some detail (42) and earlier views on the compatibility of driver and pedestrian behaviour are questioned (47).
The application of psychological theory and ergonomics research to the design of accident countermeasures has also been reviewed (44; 46). The concept of 'danger compensation' is seen to be particularly useful in explaining why reductions in objective risk arising from road safety measures often produce disappointing changes in accident rates.
Much of the theory and practice recently reviewed under this project has been generalised to provide the initiative for current work on "individual Differences in Accident Liability".
We now plan to test hypotheses derived from recent modelling of accident liability. This will initially involve the collection of data on experiential and exposure influences in drivers' levels of confidence in their perceptual-motor and decision skills, and the implications for self-imposed demand within the traffic system. Depending on the outcome of this phase, further investigations will explore the potential for training naive drivers in hazard perception.
8.4 Individual differences in accident liability (Brown, Duncan, McKenna)
Following Brown's submission of a comprehensive research proposal to a Working Party of the Environmental Medicine Committee in 1977 the Committee accepted the Working Party's recommendation that: "A high priority should be given to psychological factors in the causation and prevention of accidents, particularly those which lead to remedial and preventive measures, or test the effectiveness of these". The Neurosciences Board subsequently approved a five-year programme of work in this field and two supernumerary members of staff, Duncan and McKenna, were recruited in 1978 to work on selected aspects of the programme.
Two earlier applied lines of research, conducted in other laboratories, are being extended and integrated at the theoretical level. The first is the demonstration of an association between selective attention and road accident liability among truck and bus drivers. The second is the association found between 'cognitive style' and performance on a wide range of perceptual and perceptual-motor tasks. Of particular interest is the correlation established between an individual's 'field-dependence' and his performance in real-world situations.
Theoretical interest lies initially in the relationships between field dependence and selective attention, and also between these factors and other individual characteristics, especially those which influence the attribution of causality in behavioural events. A battery of laboratory tests has been constructed to explore some of these issues.
Performance on this test battery is currently being correlated with task performance among three populations exposed to risk of accident. By the kind cooperation of the Transport and Road Research Laboratory, data are being collected from accident-involved and non-involved motorists. The Ministry of Defence (R.A.F.) has generously agreed to the collection of comparable test data among pilots undergoing selection and training. In addition, London Transport has kindly permitted data to be collected among bus drivers at their Training School. These studies will considerably broaden our knowledge of the specific implications of selective attention and cognitive style in the training and practice of perceptual and perceptual-motor skills in dynamic environments.
Once we have completed the collection of data on field dependence and selective attention among trainee drivers, trainee pilots and accident involved motorists, these data will be used to test alternative theories on the causal bases for the two characteristics in question. A planned development of this work is the initiation of prospective studies among naive driver populations, to investigate the potential for influencing skilled performance and accident liability via special training in attention and perception.
High and sustained levels of attentional and perceptual skill are fundamental to the error-free performance of all sorts of everyday tasks. The consequences of accidents in performance in many of these tasks can be tragic for the individual and expensive for society. It is therefore hoped that this programme of research will continue to receive support beyond its initial five-year term, in order to pursue to successful conclusions the promising leads which have been established to date. In the longer term, a pool of A.P.U. expertise in this field could provide an invaluable source of information on a whole range of accident provoking situations, where individuals and groups within the workforce are exposed to occupational risk and hazard.
8.5 Automated testing (Baddeley, Watts, Williams)
With the development of microprocessors, it is becoming increasingly easy to carry out routine psychometric assessment automatically, without requiring the presence of an experienced tester. It is important that this developing technology be used wisely. The project is therefore concerned with evaluating certain key features of automated testing. Our first study took a series of standard neurological tests and presented them automatically to a range of patients. Results showed that patients were able to respond appropriately to the machine, which provided scores that were broadly consistent with the pattern expected from the location of the patient's lesion. Patients were independently assessed by a neuropsychologist and we are at present relating his evaluation to an evaluation based purely on automated test scores. A second study was concerned with "tailored testing" whereby a standard test is presented automatically, using a programme which selects questions on the basis of the subject's response. This allows the test to be shortened substantially. Subjects were tested on normal and automated versions of the Mill Hill Vocabulary Test and Raven's Progressive Matrices Test of Intelligence. Results showed that the machine-presented version took half as long to administer and in both cases scores correlated very highly with the same subject's score on the conventional test. However while the absolute scores on the vocabulary test were equivalent in the two forms, subjects tended to score less highly on the automated version of the Matrices test. This suggests that automation is feasible, but that population norms based on the conventional test may not be applicable. We are at present attempting to devise a suitable automated memory test. We plan to use samples of elderly patients and attempt to relate performance on the test to subjective reports of memory problems. If a satisfactory test can be developed, it may provide a useful screening device for detecting the early stages of senile dementia.
8.6 Perceptual and control problems in postal and telecommunications systems (Baddeley, Barnard, Brown, Dennett, Hull, Marcel, Poulton, Wright)
This project covers all the research and advisory activities of the A.P.U. under the human factors consultancy agreed annually with the Post Office (now with British Telecom).
Research activities during the past three years have included:
1. Investigations of pictographic instructions, by Marcel and Barnard
2. Studies of the temporal structure of directory enquiry conversations, by Barnard.
3. Experiments by Barnard and Dennett on the comprehension of text, with relevance for viewdata and word processing systems.
4. Investigations by Young and Hull of data accessing structures, with relevance for viewdata in general and Prestel in particular.
Among the advisory activities undertaken within the consultancy agreement, mainly by Brown, but also occasionally by Baddeley, Barnard Marcel, Poulton and Wright, are included the following:
(a) Ergonomic aspects of data-logging records used by telephone switchboard operators.
(b) Trials of a computerised Directory Enquiry system.
(c) Workspace and environmental difficulties encountered by designers of new telephone exchanges.
(d) Format of the data-logging dockets used by engineers within the repair service.
(e) Design of user instructions for postage label vending machines.
(f) Format of artificial speech in recorded announcements.