The Bounded Rationality Constraint: Experimental and Analytical Results.
Abstract
The bounded rationally constraint sets an upper limit on the rate with which decisionmakers can process information satisfactorily. This rate is studied both experimentally and analytically. A simple computer game for a single decisionmaker was used in which subjects were asked to find the smallest of a set of ratios present on the screen for a limited amount of time. Both the amount of time (twelve values) and the number of ratios (two values) were varied. A Gompertz curve is used to model the experimental results and establish the existence of a time threshold beyond which performance decreased significantly. An information theoretic model of the cognitive workload is used to estimate the workload associated with the tasks. The time threshold T* and the cognitive workload lead to a value for the bounded rationality constraint for each subject and each number of ratios. The distribution of the bounded rationality constraint across subjects for each number of ratios is found to be normal. Also, the bounded rationality of each subject as the number of ratios is changed does not vary significantly. These results may be used in the design of multi-person experiments and eventually in the methodology for organization design. First, a single value of the bounded rationality constraint for each decisionmaker would be needed for similar types of tasks. Second, since the distribution of the bounded rationality constraint across subjects appears to be normal, establishing the threshold level for a sample of decisionmakers could be sufficient to estimate the level for a larger population. Keywords: Theses, Decision making. (jhd)
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 01, 1988
- Accession Number
- ADA195897
Entities
People
- Anne-claire A. Louvet
Organizations
- Massachusetts Institute of Technology