A number of human tasks involve decisions under various information in the man-machine system and their influences are important in overall system performance.
One major purpose of this study is to examine human information processing capabilities and limitations in the estimation of Bayesian posterior probabilities.
Thirty subjects consist of two groups according to whether each has ever experienced any statistics or probability class or not. Each group performed a experiment, which was to be respond subjective posterior probabilities under given prior probabilities and Bayesian conditional probabilities by each subject. The group which has ever experienced statistics or probability class, shows patterns to overestimate posterior probabilities under given Bayesian conditional probabilities and low prior probabilities and to underestimate posterior probabilities under given Bayesian conditional probabilities and high prior probabilities. The other group shows no systematic pattern to estimate posterior probabilities under given various pair of Bayesian conditional probabilities and prior probablities.
Another result shows that risks and cost trade-offs influence human decision making behavior.