Two-facet human factors researches are being made dealing with human cognitive factors at man-machine interface (MMI). One is the experimental study to analyze human coginitve characteristics by psycho-physiological measurement, and the other is modeling on probabilistic human reliability evaluation.
A simplified experiment has been conducted to study generic characteristics of on-line human cognitive information processing at MMI. The cognitive tasks of learning and pattern classification were constructed using a simple state transition model, and they were given to subjects with conversation to CRT display of personal computer (PC). Four kinds of subjects¡Ç data were measured during the experiment: (i) eye movement data (saccade, eye attention, blinking, etc.) by eye mark recorder, (ii) bio-electric signals (ECG,CSR,EEG, etc.) by polygraph, (iii) verbal report and (iv) subjects¡Ç operational records. The validity of psycho-physiological measurement to the analysis of human cognitive characteristics was examined by those measured data. It was found that bio-informatic analysis would be useful for enhansing the objectivity of human cognitive analysis.
A numerical method was proposed for quantitative evaluation of human reliability by considering human cognitive factors at MMI. It takes into account of various human cognitive modes, introduces human characteristics parameters as various branching probability and transition time to obtain a probabilistic state transition model. The dunamic human reliability response is estimated by Monte Carlo simulation. The extention of the method to apply for practical NPP evaluation are also discussed.