Skip to main content

Knowledge Acquisition and Use in Dynamic Task Environments

Objective

The aims of the KAUDYTE Action were to:
-compare learning and control of systems with various characteristics
-compare learning by people with various strategies and attitudes
-compare various methods of testing and assessing knowledge after training
-examine the benefits and limits of acquired knowledge when used in new situations
-construct an improved conceptual framework based on these findings
-state the impact of the findings upon applications.
The way that human beings learn and perform when in control of complex systems was studied. Complex systems are those involving a number of variables which show dynamic changes even in the absence of any decision by the operator. The programme of work focussed on the benefit or harm that may arise if experience in 1 situation is followed by performance in a slightly different one. The method is experimental and aims to produce a more integrated conceptual framework than hitherto.

The aims of the project were to:
compare learning and control of systems with various characteristics;
compare learning by people with various strategies and attitudes;
compare various methods of testing and assessing knowledge after training;
examine the benefits and limits of acquired knowledge when used in new situations;
construct an improved conceptual framework based on these findings;
state the impact of the findings upon applications.

The project used a predominantly experimental approach, implemented by constructing tasks and asking people to perform them. The 5 partners exchangedtasks and methods of measurement, and each applied them in a somewhat different field. The purpose was to avoid drawing conclusions from a narrow situation.

Experiments were undertaken on the control of abstract systems with no apparent connection to past experience, but also on reasonably realistic simulations of the mixing of asphalt, or of the control of a patient's physiological functions in an intensive care unit. One of the topics of study was the benefit found when a person gains experience in a realistic situation but then has to control an unfamiliar one with the same formal structure.

Comparisons were made of graphical and linguistic ways of presenting information about the structure of a task, and of the relative merits of attempting to succeed in the task, of monitoring the control of a system by a mechanical device, and of freely exploring the system with no obligation to do well.

The problems examined were chosen so as to shed light on the set of principles governing the human control of dynamic systems.
APPROACH AND METHODS
This Action used a predominantly experimental approach, implemented by constructing tasks and asking people to perform them. The five partners exchanged tasks and methods of measurement, and each applied them in a somewhat different field. The purpose wasto avoid drawing conclusions from a narrow situation.
Experiments were undertaken on the control of abstract systems with no apparent connection to past experience, but also on reasonably realistic simulations of the mixing of asphalt, or of the control of a patient's physiological functions in an intensive care unit. One of the topics of study was the benefit found when a person gains experience in a realistic situation but then has to control an unfamiliar one with the same formal structure.
Comparisons were made of graphical and linguistic ways of presenting information about the structure of a task, and of the relative merits of attempting to succeed in the task, of monitoring the control of a system by a mechanical device, and of freely exploring the system with no obligation to do well.
The problems examined were chosen so as to shed light on the set of principles governing the human control of dynamic systems.
PROGRESS AND RESULTS
The experimental work was completed in October 1991, and the final six months of the Action were spent in integrating the theoretical and applied conclusions of the whole Action. The final report (March 1992) contains a statement of the unified view that has reulted. Academic publications are now being prepared by the various partners.
Briefly, performance of system control must involve some representation, inside the person, of knowledge about the system. But, successful control can result from alternative sub-sets of knowledge. In particular, it can be lacking in structural knowledge about the system, and confined to knowledge of appropriate actions. That knowledge may benefit from the provision of a decision-advice system, from feedback about the success of action, and indeed from action rather than observation. For transfer to new problems, however, such a knowledge-base is less effective than a more structural representation. This is favoured by teaching principles, by displaying more task information than the minimum needed for performance. The importance of these points becomes greater as the system becomes more complex, and (regarding it as a finite-state automaton) capable of entering more states.
POTENTIAL
Chemical plants, power stations, and transport systems all provide examples of situations in which people attempt to intervene in complex dynamic systems. In each there is an increasing need for assistance and advice for the human operator; potentially, that need can be met by technology. If this advice is to be effective, however, it is imperative to better understand the nature of human decisions.

Coordinator

THE CHANCELLOR, MASTERS AND SCHOLARS OF THE UNIVERSITY OF OXFORD
Address
South Parks Road
OX1 3UD Oxford
United Kingdom

Participants (4)

Rheinische Friedrich-Wilhelms-Universität Bonn
Germany
Address
Römerstraße 164
53117 Bonn
UNIVERSITÄT DER BUNDESWEHR
Germany
Address
Holstenhofweg 85
22043 Hamburg
Universität Bayreuth
Germany
Address

Bayreuth
VRIJE UNIVERSITEIT BRUSSEL
Belgium
Address
Avenue F.d. Roosevelt, 50, 122
1050 Bruxelles