Current Scientific Approaches to Decision Making in Complex Systems.
Abstract
Twelve papers on the subject of human decision making were presented and discussed. There were several recurrent themes to the conference proceedings: (1) The otherwise confusing picture of human decision making is somewhat simplified by the existence of individually and culturally consistent cognitive styles and methods of recall. Hence, given a kit of decision tools, including computer aids as well as mathematical models, it is possible either to adapt the tools to suit the human decision maker or, vice versa, to train the decision maker to use decision tools that are chosen on grounds of efficiency or elegance; (2) It is possible that people may be trained to respond in a particular way that yields performance data: for example, to use confidence estimation and rating facilities. However, this does not mean that they can be trained to decide; (3) Attempts to fit simple minded models to human performance are generally unsuccessful, especially for sequential and temporally dependent choice. More predictive models are available at the price of increased sophistication and properly accounting for underlying cognitive processes; (4) Decision analysis is not, as sometimes supposed, a prescriptive exercise, introducing the decision maker to what he should do. Nor, as sometimes supposed, is it descriptive. The chief merit of the method, which has attestable value when used in this spirit, is to establish a compromise between the decision maker and the expert, so that, as a result of learning and dialogue, it is possible to construct a compromise model, which is on the one hand acceptably rational and communicable, on the other hand, tailored to the decision makers' ingrained habits.
Document Details
- Document Type
- Technical Report
- Publication Date
- Apr 01, 1976
- Accession Number
- ADA077920
Entities
People
- G. Pask