ON HUMAN ADAPTIVE STRATEGY IN AN UNPREDICTABLE ENVIRONMENT
Abstract
The interaction between human subjects and environment is studied in detail. The task is a hill-climbing problem in which the hill is simply a randomly pivoting ramp function. The human varies two quantities, x sub 1 and x sub 2, while he observes changes in a third quantity z. Constraints are imposed upon the manipulation of x sub 1 and x sub 2, and upon the slope and possible orientation of the ramp; also, transition probabilities are introduced which govern the rotation of the ramp from one orientation to another. Assumptions concerning the mechanism of human choice behavior in this situation lead to the development of a single-parameter stochastic behavioral model. The expected performance of the model is compared with the experimentally determined average performance of human subjects. The results indicate that human adaptive strategy is essentially stochastic but is better described by a three-parameter model than by a single-parameter model; and certain nonstochastic elements in human adaptive strategy enable human subjects to consistently exceed the performance of both the single- and three-parameter stochastic models by a small margin, especially in highly variable environments. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 01, 1961
- Accession Number
- AD0267388
Entities
People
- Peter Briggs
Organizations
- Massachusetts Institute of Technology