Cognitive Psychology Of Ergonomics Research Paper

Several subdisciplines of psychology address issues related to human work. To some degree ergonomics can be seen as a blend of engineering and of experimental psychology. It is not only the application of knowledge from experimental psychology to engineering, the fertilization is bidirectional. Among the influences which triggered and shaped the rise of cognitive psychology were developments in engineering such as information theory, control theory, and computer science. A core characteristic of these developments was the abstract nature of major concepts; they were functional and not linked to physical implementations of the functions. Hence they were also suited for the analysis of human performance. Among those who established the close relation between ergonomics and cognitive psychology which developed as human work became more and more ‘mind-work’ rather than ‘body-work’ were Paul Fitts in the USA and Donald Broadbent in the UK. Both of them contributed to the key issues of what is sometimes called ‘cognitive ergonomics’ or ‘psychological ergonomics.’ Three of these key issues will be briefly addressed.

1. Human–Machine Systems

Humans interact with machines to achieve certain goals at work. In some way human–machine systems are basic units of ergonomic analysis. They can differ vastly in terms of complexity, depending on the kind of machine involved. The term ‘machine’ in this context is to be understood broadly, ranging from simple tools such as hammers and screwdrivers to vehicles and computers and finally to largely automated technical systems such as can be found in power plants and in the chemical industry.

1.1 Manual Control

A basic characteristic of a machine is its transformation of human movements into machine output, e.g., of movements of a computer mouse into movements of the cursor on the screen, or of rotations of steering wheels into changes in the direction of vehicles. In spite of the tremendous differences in terms of complexity, there are commonalities among many human–machine systems. Formally skilled operation of a machine requires the solution of a control problem, that is, a solution of the problem to find an appropriate input to achieve a desired output.

There are different types of solutions to a control problem. Two basically different ways are a closed-loop solution and an open-loop solution, which can be combined. In both cases the transformation implemented by the machine is (approximately) inverted and applied to the desired output. With closed-loop control the approximate inversion is an emergent characteristic of the closed-loop structure, whereas open-loop control requires an internal model of the transformation by the machine.

Manual-control tasks are typically studied by means of tracking setups which have the basic structure of a closed-loop system. The development of an internal model mostly implies the learning of an intermodal relation, specifically of the relation between visually perceived output of the machine and proprioceptively sensed self-produced movements. When such a relation follows a power law (which includes proportionality), it is easy to learn, and internal models of the transformation can be fairly precise. This becomes apparent when the machine has to be operated in an open-loop fashion, with the feedback loop being interrupted. However, when the transformation involves other nonlinear relations, an inversion (e.g., move the mouse to the right for a cursor movement to the left), or dynamic components (such as an integration in the case of driving a vehicle), internal models tend to become less precise and the dependence on closed-loop control more pronounced.

Perhaps the most extensively studied variation of visuo-motor relations is called stimulus–response compatibility. For example, when a cursor moves to the right upon a mouse movement to the right, this would be a spatially compatible relation, and a directional reversal would produce an incompatible relation. More generally, the term ‘compatibility’ refers to the naturalness of a visuo-motor transformation, which is often fairly obvious to a user. Usually there are strong differences between compatible and incompatible relations when visual stimuli and responses share a common dimension (such as left– right). Whether or not a certain relation is compatible depends not only on the physical situation, but also on the mental representation of it. For example, when one has to respond to small and large numbers between 1 and 12 by pressing a right or left key, the assignment of small numbers to the left key and large numbers to the right key results in shorter reaction times (compatible relation) than the opposite assignment (incompatible relation); however, the normally incompatible relation becomes compatible when one imagines the numbers being arranged on the face of a clock.

1.2 Monitoring And Supervisory Control

Modern technology allows machines to become more and more complex, and manual-control tasks can mostly be automated. Modern airplanes are popular examples of this. There is a certain irony in such automation which should not be overlooked. On the one hand automation serves to avoid human errors and to reduce mental load, but on the other hand there are emergency situations in which the system changes from automatic to manual control. The irony is that a human operator who is deprived of the daily experience of controlling the machine in normal situations is required to master it in the difficult emergency cases. In the normal situation, the task of the operator in highly advanced transport and production systems becomes that of a supervisor. The task of the supervisor is no longer continuous control of the system output, but rather the monitoring of the system state, the discovery and diagnosis of abnormal states, and the taking of appropriate actions when abnormal states are discovered.

Monitoring can be a boring task: irregular system states are rare events, so there is little that happens. The classical problem which gave rise to the systematic study of maintained attention to displays which only rarely indicate critical events is that of the radar controller. In the course of a watch there is typically a fairly rapid decline of the probability of detecting a critical event, the so-called vigilance decrement. The same phenomenon can become a problem in industrial inspection tasks. In particular, the really rare events carry a risk of being overlooked. One of the factors which are likely to contribute to the vigilance decrement is temporary inactivation of the appropriate task set.

Monitoring can also be a task which places too high a burden on a human operator. There can be hundreds of state variables which have to be monitored, so the decision as to which state variable to inspect next can be difficult. In their inspection patterns operators adjust to the characteristics of the state variables. For example, state variables which change rapidly tend to be inspected more frequently than state variables which change only slowly. In addition, interdependences between state variables guide the inspection pattern. However, in the case of a failure the normal characteristics of the state variables may no longer be present, so that inspection patterns can be misguided.

The detection of abnormal system states can take minutes or hours, depending on a number of conditions. Among them is the rate of change of system states, which can be slow. Again, as does the pattern of inspection behavior, fault detection relies heavily on the operator’s internal model of the system. Unfortunately the concept of an ‘internal model’ is somewhat elusive when it comes to a precise characterization. One of the reasons is that the internal model on which monitoring and supervisory control is based is not necessarily identical with the reportable knowledge of an operator. Both incidental observations and formal studies suggest that at least part of the knowledge on which the operators’ monitoring and control actions are based is implicit in nature.

2. Workload And Multiple-Task Performance

One of the classical and continuing problems is the assessment of workload. Its importance results from the notion that overload or underload can result in errors, accidents, and/or health risks. There are three types of measurements that are employed for workload assessments, physiological measurements (such as heart rate and heart-rate variability), subjective ratings, and behavioral data.

The assessment of workload by means of behavioral data is closely linked to the study of multiple-task performance. Everyday observations suggest that the mind can be more or less dedicated to the performance of a certain task. To bring this observation into the realm of scientific treatment, the concept of capacity or of resources has been introduced: the more capacity is dedicated to the performance of a certain task, the higher is the performance in this task, and the less capacity is available for the performance of other concurrent tasks. From this general idea, it follows that there should be a tradeoff between performance levels in two concurrent tasks when performers vary their respective emphasis and, theoretically, the proportions of total capacity allocated to each of them. Performance tradeoffs indeed exist. For practical purposes, these simple concepts suggest that performance in a subsidiary task could be used to assess mental workload. However, so far all attempts to find a subsidiary task that could be used as a kind of yardstick in the assessment of mental load have failed.

The limits of multiple-task performance are important not only in the context of workload assessment, but also in the context of work design. For example, in supervisory control situations can arise which require different actions concurrently. Sometimes concurrent actions can be physically impossible, so that the problem becomes one of scheduling tasks according to priorities. Even when concurrent activities are physically possible, there is not only interference because of limited mental capacity, but also interference of a more specific kind, related, for example, to cross-talk within the perceptual or motor systems. A basic principle is that interference becomes more pronounced as the perceptual information for different tasks, the effectors involved, and the nature of the information (e.g., verbal vs. pictorial) become more similar.

3. Skill Acquisition

Human performance can be improved by practice, and the time needed for a certain action (such as the classical example of rolling a cigar) declines. Practice curves can be described by a power function for a great variety of skills; this fact is referred to as the ‘power law of practice.’ However, skilled performance has a number of additional characteristics. It tends to be less variable, and it tends to be associated with a reduced workload and less interference with the performance of concurrent tasks. The latter aspect of skill acquisition is called automatization. However, the reduction of dual-task interference is not a universal characteristic of skill acquisition, but there can also be certain secondary tasks for which dual-task interference increases.

In many cases the skills required to operate complex human–machine systems are acquired by means of special training devices, so that there are degrees of freedom in designing the training procedure. The design of training procedures can be a fairly complex process. Often it involves the identification of task elements which are practiced separately, special instructions, and special feedback. The training procedure shapes the internal model of the machine. Thus, a critical factor for the success of training procedures is the match of the practice situation and the final ‘test’ with the real system, in particular the match with respect to functional characteristics and with respect to the information available. Somewhat ironically, too much can be done to support the learner during training in terms of guidance and feedback; performance can become dependent on such additional guidance and feedback so that it will deteriorate as soon as it is no longer available.

Bibliography:

1. Damos D L (ed.) 1991 Multiple-task Performance. Taylor and Francis, London
2. Matthews G, Davies D R, Westerman S, Stammers R B 2000 Human Performance. Cognition, Stress, and Individual Differences. Psychology Press, Hove, UK
3. Proctor R W, Dutta A 1995 Skill Acquisition and Human Performance. Sage, Thousand Oaks, CA
4. Proctor R W, van Zandt T 1994 Human Factors in Simple and Complex Systems. Allyn and Bacon, Boston
5. Sanders A F 1998 Elements of Human Performance: Reaction Processes and Attention in Human Skill. Erlbaum, Mahwah, NJ
6. Wickens C D 1992 Engineering Psychology and Human Performance, 2nd edn. Harper Collins, New York