Whilst Engineering is concerned with improving equipment from the point of view of mechanical and electrical design and Psychology is concerned with the study of the mind and behaviour, Engineering Psychology is concerned with adapting the equipment and environment to people, based upon their psychological capacities and limitations (Blum, 1952) with the objective of improving overall system performance (involving human and machine elements). As Sanders & McCormick (1987) put it, "... it is easier to bend metal than twist arms", by which they mean that the design of the device to prevent errors is likely to be more successful than telling people not to make errors. According to Wickens (1992) the role of Engineering Psychology is distinct from both Psychology and Engineering in that it arises from the intersection of the two domains. He also distinguishes Engineering Psychology from Ergonomics (see note 1) to suggest that "the aim of engineering psychology is not simply to compare two possible designs for a piece of equipment ... but to specify the capacities and limitations of the human ... from which the choice for a better design should be directly deductible" (pp. 3-4, Wickens, 1992 cites Poulton, 1966).

Ergonomics is distinct from Engineering Psychology in that it is mutidisciplinary (incorporating Psychology, Engineering, Physiology, Environmental and Computer Science), but the boundaries are fuzzy and Ergonomics shares the overall goals of Engineering Psychology. The objectives of Ergonomics (cf. Human Factors) are shared by Engineering Psychology, which are to optimise the effectiveness and efficiency with which human activities are conducted as well as to improve the general quality of life through "increased safety, reduced fatigue and stress, increased comfort [and] ... satisfaction." (Sanders & McCormick, 1992, p. 4).

It is difficult to delineate the genesis of both Engineering Psychology and Ergonomics, but both can be traced back to a general interest in problems at munitions factories during the First World War (Oborne, 1982). Machines that were designed to be operated by men seemed to have production-related problems when operated by women. These difficulties were resolved when it was realised that the problems were related to equipment design rather than the people operating them, i.e. they were designed to be operated by men and not women. The mis-reading of altimeters by pilots in the Second World War stimulated further interest in Engineering Psychology. A study by Grether (1949) illustrated that the traditional three needle altimeter (where the three pointers read 10,000s, 1,000s and 100s of feet respectively) not only took pilots over 7 seconds to interpret but nearly 12 percent of the readings contained errors of a 1000 feet or more. Grether showed conclusively that superior designs could dramatically reduce both reading time and error rates. This study, perhaps more than any other, indicates the importance of Psychology in the design of devices. Despite this evidence, it is sometimes difficult to gain acceptance from the Engineering community, and to change design, as the following quote from an accident report in 1958 (some 9 years after Grether's original study) shows:

"The subsequent investigation ... showed that the captain had misread his altitude by 10,000 feet and had perpetuated his misreading error until the aircraft struck the ground and crashed."  Rolfe (1969) p.16

The Need for a Psychology of Engineering

We are all familiar with the frustrations that accompany one's use of technology in the home and at work. Norman (1988) provides an abundance of examples on this subject. The Information Technology revolution has led to computers pervading almost every aspect of our lives from programming Video Cassette Recorders (VCRs) and Microwave Ovens, to withdrawing cash from Automatic Teller Machines, to purchasing rail tickets, to performing most aspects of our work. Yet why do these devices, which are supposed to make our lives easier, seem to thwart our best intentions? One reason is that users of these devices perceive the problem to be with themselves rather than with the technology. People often blame themselves when failing to comprehend the manufacturer's instructions or when errors occur (Reason, 1990). Also, the problems are usually of a small, relatively trivial and individual nature, and do not affect other people. These problems are often only minor hassles compared to major events, such as incidents in the aviation and nuclear industries. On the face of it there is little comparison between errors with VCRs and errors on the flightdeck of an aircraft. However, Reason (1990) argues that at the basic level of interfacing human thought processes with technology there are many similarities. Despite the obvious differences in training, level of skill and knowledge in operating VCRs and aircraft, basic error types such as 'mode error' (i.e. errors that occur when devices have different modes of operation and the action appropriate for one mode has different consequences in other modes: Norman, 1986) have been found to occur in both environments.

There has been some concern in recent years about safety (Stanton, 1996). The incidents at Three Mile Island (in the USA) and Chernobyl (in the former USSR) are often cited in the press and technical literature. A recent near-incident at a nuclear utility in the UK has seemingly reinforced this concern. Whilst these nuclear power plants employ different technologies there is one common factor to these, and other, incidents: namely human beings. Reason (1990) reports that 92% of all significant events in nuclear utilities between 1983-1984 were caused by people and of these only 8% were initiated by the control room operator.

Thus, the scope of Engineering Psychology needs to consider all aspects of the human-technology system. Consideration of the human element of the system has been taken very seriously since the publication of the President's commissions report on Three Mile Island (Kemeny, 1979) which brought serious problems to the forefront. The summary of the main findings of the report highlights a series of "human, institutional and mechanical failures." It was concluded that the basic problems were people-related, i.e. the human aspects of the systems that design, build, operate and regulate nuclear power. Some reports have suggested 'operator error' as the prime cause of the event. However, the failings at Three Mile Island included:

• deficient training which left operators unprepared to handle serious accidents;
• inadequate and confusing operating procedures that could have led the operators to incorrect actions;
• design deficiencies in the control room, for example in the way that information was presented and controls were laid out;
• serious managerial problems within the Nuclear Regulatory Commission.

None of the deficiencies explain the root cause of the incident in terms of 'operator error', which is an all too familiar explanation in incidents involving human-technology systems. Reason (1987), in an analysis of the Chernobyl incident, suggested two main factors of concern. The first factor relates to the cognitive difficulties of managing complex systems: people have difficulties in understanding the full effects of their actions on the whole of the system. The second factor relates to a syndrome called 'groupthink': small, cohesive and elite groups can become unswerving in their pursuit of an unsuitable course of action. Reason cautions against the rhetoric of "it couldn't happen here" because, as he argues, one of the basic system elements (i.e. people) is common to all nuclear power systems.

Demand-Resource Theory in Engineering Psychology

Solutions to the problems raised in people interacting with technology come in two main forms; either to reduce demand or to increase resources in situations of work overload or vice versa in situations of work underload. The dual concepts of demands and resources are prevalent in Engineering Psychology and particularly pertinent when considering the capacities and limitations of people in technological environments. Wickens (1992) proposes a theory of multiple pools of attentional resources in relation to different information processing demands - speech and text utilise a verbal information processing code and draw upon a different pool of attentional resources to tones and pictures which utilise a spatial processing code. Wickens argues that when the attentional resources assigned to the verbal processing code are exhausted, workload demands may be increased further by using the alternative spatial information processing code through the presentation of tones or pictures (although these pools are not wholly mutually exclusive).

The concept of demands and resources provides a conceptual framework for Engineering Psychology. Demands and resources could come from the task, the device and the user. For example, user resources (e.g. knowledge, experience and expertise) and demands (e.g. user goals and standards) interact with task demands (e.g. task goals and standards) and task resources (e.g. instruction manuals and training). This interaction is mediated by demands (e.g. device complexity) and resources (e.g. clarity of the user-interface, which could reduce demands) of the device being operated.

This is a familiar concept in discussions of task workload, and it is implied that demand resource imbalance can occur as both task underload and task overload, both of which are detrimental to task performance. An illustration of the relationship between demands and resources is provided by the Tale of Procrustes (Oborne, 1982). In Greek mythology, Procrustes was an ingenious robber who conned travellers into parting with their gold. His trick was very simple. He offered weary travellers all the food and wine they wanted and they could either pay for what they had consumed or accept his hospitality without payment and take a bed for the night. Most travellers opted for latter, at which point Procrustes added one more clause: that the traveller had to fit one of his two spare beds exactly. Most accepted without question and ate and drank their fill. When it came time for them to bed down for the night Procrustes showed them the two beds, one was very long and the other very short. At this point Procrustes threatened to make them fit the bed by either cutting off their legs to fit the short bed or stretching them to fit the long bed. Most traveller opted to pay the exhorbitant bill instead! Oborne (1982) suggests that the Procrustean approach often appears to be taken by designers, who design tasks that either stretch people beyond their physical and/or mental capacities or tasks that are physically and/or mentally constrictive. Both ends of the spectrum result in a dissatisfactory outcome for the individual, as well as poor performance of the system. So we end up paying for poor design in terms of discomfort, errors, dissatisfaction and poor performance. Some times the price can be counted in terms of human life.

Perspectives on Engineering Psychology

Three different perspectives on Engineering Psychology are offered, Engineering Psychology as:

• Ergonomics
• Human Computer Interaction
• Cognitive Engineering

Shackel (1996) starts by distinguishing Psychology from Ergonomics, to propose that Ergonomics is about fitting the device to the individual. He argues that industrialisation has exacerbated many of the problems associated with device use. First there is the problem of operating industrialised systems. Second there is the problem of tailoring mass produced devices to individual needs. Tailoring every device to everyone's needs may seem like an impossible goal, but if we know what the range of needs are we may be able to design flexibility into devices so that they meet most people's needs most of the time. For example, in a relatively simple device, like a chair, we can offer height and backrest adjustments. The challenge is either to offer the same degree of customisation for other more complex devices, like computer interfaces, or to design a standard interface that can be used by all.

Shackel argues that Ergonomics, like Psychology, suffers from being labelled a Science of Common Sense. All too often, designers seem to prefer to consult their own intuitions rather than a professional ergonomist. Device testing tends to be very informal, only involving the designers themselves, rather than being based upon a sample of the end-user population and subjected to the rigour of statistical analysis. If, indeed, good design were common sense then we would not witness the extent of disasters due to poor design in human terms (see Reason, 1990). Shackel argues that a systematic and scientific approach to the analysis and design of devices is needed. Even apparently well design devices (such as the example given by Shackel) appear to benefit from this approach, although performance problems are normally the indication of poor Ergonomics. Shackel considers the role of Ergonomics in different kinds of work and this shows the links between Engineering Psychology and Ergonomics (specifically both concerned with human-machine interaction and system performance)

Payne (1996) argues that technology and Psychology have mutually beneficial relationship, but that advances in either can exist without the other. Payne thus suggests a situation of mutual benefit but not mutual dependence. However, the one without the other may lead to a poorer outcome for both. Payne asks the question of whether advances in Psychology lead to advances in technology or vice versa? He suggests that we witness more of the latter, i.e. technological insights offer new insights for psychology. For example, the development of the Graphical User Interface (GUI: e.g. the use of Windows, Icons, Menus and Pointing devices: WIMP) owes little to psychological theory, but has enabled applied cognitive psychologists to develop greater explanations for the phenomenon of why the GUI is easier to use than character-based user interfaces (Norman & Draper, 1986). Payne argues that psychology is good at providing explanations for this kind of phenomenon but has not yet revolutionised technology. The WIMP/GUI interface might be considered to be a technological revolution, not a psychological one, whereas Psychology can offer small evolutionary improvements.

Payne cites two examples where Psychology has had modest success: in the development of the SuperBook and the application of the GOMS model. In the first example, on-line versions of books are generated automatically with additional features that enable the book to be used with enhanced functionality. This functionality was based upon psychological research on human language to design a word search facility.

In the second example, the GOMS model (based on a cognitive theory developed by Card, Moran & Newell, 1982) was used to determine the effectiveness of a new workstation. The theory-driven evaluation (i.e. "to specify the capacities and limitations of the human from which the choice for a better design should be directly deductible" Wickens, 1992) led to the company rejecting the new design.

Payne also notes the problem of coupling between Cognitive Psychology research and engineering concerns - this has led to a new, but related discipline: Human-Computer Interaction (HCI) - which is more closely aligned to engineering concerns than Cognitive Psychology. Payne indicates that HCI is rather more unifying than Cognitive Psychology. The former is largely concerned with whole tasks, such as the operation of a device, from a Video Cassette Recorder to a Nuclear Power Station, whereas the latter tend to focus on isolated processes such as perceptual categorisation, word recognition, etc.

Additionally, Payne suggests that Cognitive Psychology can benefit from advances in technology. The study of human interaction with technology (which Payne proposes is the domain of Human Computer Interaction) supplies Cognitive Psychology with phenomena which require explanation. As in the earlier example of the GUI, the success of the interface was poorly understood until Applied Cognitive Psychologists addressed this conundrum. Development of theory in this area could lead to prediction of new technology. Whereas, design in the absence of theory leads to Psychology chasing technology.

Long & Dowell (1996) argue that operational problems (such as the problems associated with the computerisation of the London Ambulance Service) has led to a shift in emphasis from addressing technology to addressing human-device interaction problems. According to Long & Dowell, the link between Psychology and Engineering is more than a marriage of convenience, it has become essential in the wave of technological advancement that requires humans to interact with devices. As Shackel (this volume) suggests, the need to address problems has led to the emergence and shaping of the discipline. Long & Dowell argue for a problem-led approach and propose that the objective of this discipline should be to get human-computer systems to work effectively. Like Payne, Long & Dowell argue that the link between Cognitive Psychology and Information Technology is far from straightforward and they suggest that even Applied Cognitive Psychology fails to link these two disciplines together (coupling). Rather, Long & Dowell argue for a separate and distinct discipline of Cognitive Engineering which is analogous to the relationship that Software Engineering shares with its allied disciplines of Computer Science and Engineering.

Long & Dowell argue that this view proposes two different ways of conceiving the link between Cognitive Psychology and Information Technology (IT). The one-stream perspective suggests a direct link between Cognitive Psychology, Applied Cognitive Psychology and IT whereas the two-stream view suggests that Cognitive Psychology and Applied Cognitive Psychology exist in parallel to Cognitive Engineering and Information Technology (this is similar to the argument that Payne puts forward in favour of HCI). They are cautious about the relationship between these two streams. However, they show that the two-stream view is more realistic as developments in Cognitive Psychology do not directly translate into developments in IT even when mediated by Applied Cognitive Psychology. They suggest that this is because the initial developments in Cognitive Psychology did not directly address a problem in IT, whereas the focus of Cognitive Engineering is directly upon design problems in IT. Long & Dowell show that Cognitive Engineering and Software Engineering are very similar in principles, practices and approach but for one subtle and important difference: Cognitive Engineering emphasises that the design focus is upon the requirements of user populations whereas Software Engineering emphasises the design in terms of the functioning of the computer.

The Future of Engineering Psychology?

The vision offered by the perspectives are of a problem-driven focus of Engineering Psychology with concerns about the performance of human-device systems. Technological advances are likely raise issues in the areas of advanced transportation, co-operative work, teleworking, health, pollution and leisure. Recent research effort has called for more theory-based approach from the discipline, in the design practices and processes, in the evaluation and understanding of the way in which devices support human thought. There is an inextricable link between Engineering Psychology and the Science of Technology and is up to Engineering Psychologists to rise to these challenges.

Professor Neville Stanton,
BSc, MPhil, PhD, C.Psychol, FBPsS, FErgs

Transport Research Group
School of Civil Engineering & the Environment
University of Southampton
Highfield
SOUTHAMPTON SO17 1BJ
UK

'The Psychologist'

St Andrews House
48 Princess Road East
LEICESTER

LE1 7DR

UK

http://www.thepsychologist.org.uk