As the manually controlled machines become complex, the controlling ability of the human operator becomes an important factor for stable and safe operation, in the design of man-machine systems. With an emergency or a difficult situation in mined, the present paper discusses how far one can count on the controlling ability of the human controller for a stable operation. Some new result of experiments are presented together with some problems yet to be solved.
The present study is carried out from the standpoint of man-machine interaction as information processing through interface devices and interior computer logic in the simulated CAI situation. Unimodal criterion functions in a two-dimensional space were given to a learner to find out the optimum point as the CAI problem. For this purpose, the pattern search method was modified and developed to analyze the learner's searching process. First, the learner chose one type of test hill among the criterion functions having each different optimum points and then set a trial point. The computer told the height of the hill corresponding to the point and also offered the informations of the next trial point based on the calculation by the modified pattern search. From these experimental results, there are some remarkable characteristics in the process of searching the optimum point. These basic analysis will give us some useful suggestions on the evaluation of CAI network which can be adaptive to each learner's response.