Handbook of Cognitive Task Design

Erik Hollnagel

• Publisher: CRC Press
• Year: 2003
• ISBN:
  9781135642471
• Edition: 1
• Topic: Technology & Engineering
• Subtopic: Industrial Engineering
• Index: Technology & Engineering

This Handbook serves as a single source for theories, models, and methods related to cognitive task design. It provides the scientific and theoretical basis required by industrial and academic researchers, as well as the practical and methodological guidance needed by practitioners who face problems of building safe and effective human-technology s

PART I
Theories

1
Prolegomenon to Cognitive Task Design

Erik Hollnagel

University of Linköping, Sweden

Abstract

The design of artefacts, and in particular artefacts involving computing technology, is usually focused on how the artefact should be used. The aim of cognitive task design (CTD) is to go beyond that focus by emphasising the need to consider not only how an artefact is used, but also how the use of the artefact changes the way we think about it and work with it. This is similar to the envisioned world problem, that is, the paradox that the artefacts we design change the very assumptions on which they were designed. The ambition is not to make CTD a new discipline or methodology, but rather to offer a unified perspective on existing models, theories, and methods that can be instrumental in developing improved systems. In this context, cognition is not defined as a psychological process unique to humans, but as a characteristic of system performance, namely the ability to maintain control. The focus of CTD is therefore the joint cognitive system, rather than the individual user. CTD has the same roots as cognitive task analysis, but the focus is on macrocognition rather than microcognition; that is, the focus is on the requisite variety of the joint system, rather than the knowledge, thought processes, and goal structures of the humans in the system.

WHAT IS CTD?

In a handbook of CTD, it is reasonable to begin by defining what CTD is. It is useful first to make clear what it is not! CTD is not a new scientific discipline or academic field; nor is it a unique methodology.

CTD Is not a New Scientific or Academic Field

Since the mid-1970s, the terms cognitive and cognition have come to be used in so many different ways that they have nearly lost their meaning. In particular, there has been an abundance of more or less formal proposals for lines of activity or directions of study that as a common feature have included the words cognitive and cognition in one way or another. These range from cognitive ergonomics, cognitive systems engineering, and cognitive work analysis to cognitive tools, cognitive task analysis, cognitive function analysis, cognitive technologies, cognitive agents, and cognitive reliability—to mention just a few.

Some of these have thrived and, by their survival, justified their coming into the world as well as established an apparent consensus about what the word cognitive means. Others have been less successful and have in some cases languished uneasily between survival and demise for years. The situation does not in any way seem to improve, because the words cognitive and cognition are used with ever-increasing frequency and, inevitably, diminishing precision— although at times with considerably success in obtaining funding.

CTD is not put forward in the hope that it will become a new scientific field. Indeed, if that were the case it would signify failure rather than success. As this handbook should make clear, CTD is something that we already do—or should be doing—although we may lack a precise name for it.

CTD Is not a Unique Methodology

The widespread use of the terms cognitive and cognition is, of course, not the result of a whim but does indicate that there is an actual problem. Indeed, many people are genuinely concerned about issues such as, for instance, cognitive tasks and cognitive reliability, because these are salient aspects of human behaviour in situations of work, as well as of human behaviour in general.

It is because of such concerns that a number of specific methodologies have emerged over the years. A case in point is the development of cognitive task analysis as an “extension of traditional task analysis techniques to yield information about the knowledge, thought processes, and goal structures that underlie observable task performance” (Chipman, Schraagen, & Shalin, 2000, p. 3). Even though cognitive task analysis is used as a common name for a body of methods that may vary considerably in their aim and scope, partially because of the imprecision of the term cognitive, they do represent a common thrust and as such a distinct methodology.

Whereas cognitive task analysis is aimed at analysing cognitive tasks, that is, knowledge, thought processes, and goal structures, CTD is not aimed at designing cognitive tasks as such, at least not in the precise sense implied by Chipman et al. (2000). Instead of being a unique design methodology, CTD is proposed as a unified perspective or point of view that can be combined with a number of different methods. This is explained in the following section.

CTD Is a Unified Perspective on Design

By saying that CTD is a perspective, I mean that it designates a specific way of looking at or thinking about design. More particularly, it is a specific way of thinking about the design of systems in which humans and technology collaborate to achieve a common goal. These systems are called joint cognitive systems, and this concept is explained later in this prolegomenon. The way of thinking that is represented by CTD can be expressed by a number of principles:

• Every artefact that we design and build has consequences for how it is used. This goes for technological artefacts (gadgets, devices, machines, interfaces, and complex processes) as well as social artefacts (rules, rituals, procedures, social structures, and organisations).

• The consequences for use can be seen both in the direct and concrete (physical) interaction with the artefact (predominantly manual work) and in how the use or interaction with the artefact is planned and organised (predominantly cognitive work). Thus, introducing a new “tool” not only affects how work is done, but also how it is conceived of and organised. In most cases, this will have consequences for other parts of work, and it may lead to unforeseen changes with either manifest or latent effects.

• The primary target of design is often the direct interaction with or use of the artefact, as in human–computer interaction and human–machine interaction. Interface design, instruction manuals, and procedures typically describe how an artefact should be used, but not how we should plan or organise the use of it. However, the realization of the artefact may affect the latter as much, or even more, than the former.

• As a definition, the aim of CTD is to focus on the consequences that artefacts have for how they are used, and how this use changes the way we think about them and work with them, both on the individual and the organisational level. The ambition is to ensure that CTD is an explicit part of the design activity, rather than something that is done fortuitously and in an unsystematic manner.

CTD is not new, because many of the principles have been recognised and used before. Some may be found among the roots of cognitive task analysis, as described later. Two more recent examples are the concept of “design for quality-in-use” (Ehn & Löwgren, 1997) and the notion of task tailoring (Cook & Woods, 1996). It is also a postulate that this perspective is unified. Perhaps it would be more correct to say that CTD should become a unified perspective on design. The argument of this prolegomenon, and of many of the chapters that follow, is that the design of artefacts and systems has very often neglected the aspect of cognitive tasks in this sense, that is, as the way we think about artefacts rather than as the way we use them. A considerable amount of work has been done on issues of usability and usefulness, and several schools of thought have established themselves. However, both usability and usefulness focus more on the direct use of artefacts than on how that use by itself changes how we come to see them.

THE NEED OF CTD

The 20th century has witnessed dramatic changes to the nature of human work, particularly since the 1950s. These changes led to the emergence of human factors engineering, ergonomics, and later developments such as cognitive ergonomics and cognitive systems engineering. The extent of the changes is amply demonstrated by tomes such as the Handbook of Human– Computer Interaction (Helander, Landauer, & Prabhu, 1997) and the Handbook of Human Factors and Ergonomics (Salvendy, 1997). In relation to CTD, the following two changes were particularly important.

First, work went from being predominantly manual work—or work with the body—to being predominantly cognitive work—or work with the mind. Many manual tasks have disappeared while new cognitive tasks have emerged. For those manual tasks that remain, technology has often changed them considerably.

This change to the nature of human work is the logical continuation of the development that gained speed with the industrial revolution, the essence of which is that machines or technology are used to amplify or replace human functions. Modern information technology has significantly increased the speed by which this development takes place and the types of work that are affected. Even for the kinds of work that are still largely manual, technology or machinery is usually involved in some way. (A checkout counter at a supermarket is a good example of that.) Today there are, indeed, very few types of work that depend on human physical power and abilities alone. Machines and technology have effectively become part of everything we do, whether at work or at home. This means that cognitive tasks are everywhere, and work design is therefore effectively CTD.

Second, cognitive tasks, even loosely defined, are no longer the prerogative of humans, but can also be carried out by a growing number of technological artefacts. Machines and information technology devices have, in fact, for many years been capable of taking over not just manual but also mental work. In other words, the ability to amplify human functions has migrated from the physical (motor) to the cognitive parts of work. It is not just that work has become more cognitive, but also that humans have lost their monopoly of doing cognitive work or cognitive tasks. When we conceive of and build machines, we must therefore also in many cases consider CTD.

The changes to the nature of human work require corresponding changes to the methods by which work is described, analysed and designed. The cognitive aspects of work have traditionally been addressed by cognitive task analysis, as an extension of classical task analysis. However, cognitive tasks are more than just an addition to manual tasks, and the changes affect work as a whole rather than what humans do as part of work. To be able to build efficient and safe sociotechnical systems, we must be concerned about CTD from the point of view of the individual who carries out the work as well as of the work system. CTD thus comprises the study of cognitive work—by humans, by machines, and by human–machine ensembles—and covers the whole life cycle of work from preanalysis, specification, design, and risk assessment to implementation, training, daily operation, fault finding, maintenance, and upgrading. CTD is about the nature of work as it is now and as we want it to be in the future.

THE MEANING OF COGNITION

As previously argued, the words cognitive and cognition have generally been used without much precision, and Wilson, Crawford, and Nichols make the same point in chapter 5 in this handbook. It is therefore necessary to provide a more precise definition of what cognition means. This is not an easy task, because etymology, semantics, and practice may be at odds with each other. Although it is tempting to accept Jim Reason’s definition of cognition as that which goes on “between the ears,” it behooves an editor to sound more scientific. This leads to the following line of reasoning.

Cognition is used to describe the psychological processes involved in the acquisition, organisation, and use of knowledge—with the emphasis on rational rather than emotional characteristics. Etymologically it is derived from the Latin word cognoscere, to learn, which in turn is based on gnoscere, to know. Following the example of Mr. Pott in the Pickwick Papers (Dickens, 1837), we could then define cognitive tasks as those that require or include cognition.

This easily leads us to an axiomatic position, which starts from the fact that humans are cognitive beings (or that humans have cognition); hence human performance has a cognitive component. However, following the same line of reasoning, we could also argue that human actions are driven by motives and emotions, and that human performance therefore has a motivational and an emotional component—which indeed it has. Although it is evidently true that humans have cognition, the axiomatic position makes it difficult to extend the notion of cognition to other entities, such as technological artefacts and organisations. It also begs the question of what cognition really is.

An alternative is to use a more pragmatic definition, which is based on the characteristics of certain types of performance. Human performance is typically both orderly (systematic and organised) and goal directed. This can be used as a provisional definition of cognition, and it can be extended to require cognitive tasks to have the following characteristics:

• Cognitive tasks are driven by goals (purposes and intentions) rather than by events. They include cause-based (feedforward) control as well as error-based (feedback) control. Cognitive tasks are therefore not merely responses based on algorithmic combinations of predefined elements, but require thinking ahead or planning over and above complex reactions.

• Cognitive tasks are not limited to humans; cognitive tasks can be found in the functioning of organisations, of certain artefacts (a growing number, but still not many), and of animals.

Polemically, the issue is whether the definition of cognitive tasks is based on an axiomatic definition of cognition, or a pragmatic characterisation of performance. In other words, cognitive tasks could be defined as tasks performed by a system that has cognition—which presumably only humans have (?). Alternatively, cognition could be defined as a quality of any system that has certain performance characteristics, and that therefore can be said to do cognitive tasks. This would base the definition on the characteristics of the tasks and of system performance, rather than on the possible constituents and explanations of internal mechanisms.

From this perspective, cognitive tasks are characteristic of humans, organisations, and some artefacts. (I leave out animals in this context.) CTD is consequently concerned with how functions and structures—of a cognitive system proper and of its environment—can be designed to further the system’s ability to perform in a purposeful manner and to let it keep control of what it is doing. Looked at in this way, CTD refers to (joint) cognitive systems as a whole, whether they are biological individuals, artificial intelligences, or organisations. CTD clearly also goes beyond cognitive tasks analysis, as the emphasis is on the potential (future) rather than the actual (past and present) performance.

The importance of CTD stems from the fact that any change to a system—such as the introduction of new technology, improved functionality, or organisational changes—inevitably changes the working conditions for the people in the system, and hence their cognitive tasks. All design is therefore implicitly or explicitly CTD. This is obviously the case for technological artefacts and information devices, because these directly affect user tasks. A little thought makes it clear that the same is true for any kind of design or intentional change to a system, because the use of the system, that is, the way in which functions are accomplished and tasks carried out, will be affected. CTD comprises the study of how intentional changes to system functions and structures affect the conditions for work, and hence the cognitive tasks, and the development of concepts and methods that can be used to improve design practices.

The outcome of this line of reasoning is that cognition is not defined as a psychological process, unique to humans, but as a characteristic of system performance, namely the ability to maintain control. Any system that can maintain control is therefore potentially cognitive or has cognition. In this way the focus of CTD is not just on the characteristics of putative human information processing or capabilities such as recognition, discrimination, and decision making that normally are seen as components of cognitive work. The focus is rather on descriptions of the performance of cognitive systems in the complex sociotechnical networks that provide the foundation of our societies, and how this performance must change to enable the systems to stay in control. CTD therefore struggles with the dilemma known as the envisioned word problem (Woods, 1998), which is how the results of a cognitive task analysis that characterises cognitive and cooperative activities in a field of practice can be applied to the design process, because the introduction of new technology will transform the nature of practice! Or, put more directly, the paradox of CTD is that the artefacts we design change the very assumptions on which they were designed.

COGNITIVE SYSTEMS AND JOINT COGNITIVE SYSTEMS

The concept of a joint cognitive system (JCS) has already been mentioned several times. Although classical ergonomics and human factors engineering have often emphasised the necessity of viewing humans and machines as parts of a larger system, the analysis nevertheless remains focused on the level of elements of the larger system, specifically humans and machines. A consequence of this seemingly innocuous and “natural” decomposition is that the interaction between human and machine becomes the most important thing to study. Cognitive systems engineering (CSE; Hollnagel & Woods, 1983) instead argues that the focus should be shifted from the internal functions of either humans or machines to the external functions of the JCS. This change is consistent with the idea that humans and machines are “equal” partners, and that they therefore should be described on equal terms. Humans should not be described as if they were machines; neither should machines be described as if they were humans. Whereas the former has been the case so often that it practically has become the norm, the latter has only rarely been suggested or practised. (Some examples would be anthropomorphism and animism. Thus most people tend to ascribe psychological qualities ...

Table of contents

• Cover Page
• Title Page
• Copyright Page
• Series Foreword
• Acknowledgments
• Contributors
• Acronyms
• Part I: Theories
• Part II: Methods
• Part III: Field Studies