A Concept of adaptive interface is proposed which will work to adapt to both the mental and physiological state of the user at man-machine interface. The system concept is composed of the four blocks: (A)Bio-informatic sensor / transmitter / receiver to measure the user¡Çs visual information characteristics and his various physio-electric signals governed by autonomous nerval system, (B)Psycho-physiological signal analyzer to reduce various psycho-physiological measures by processing the measured bio-informatic signals, (C) Cognitive state estimator to estimate human cognitive state in his problem solving process, and (D)Interface controller which regulates the displayed information so that it may match well with the user¡Çs capability and his physical conditions. The general requirements for realizing such adaptive interface concept are firstly discussed and then a developmental study for cognitive state estimator is presented in this paper.
The cognitive state estimator in our experimental study is such a system that it processes time-series signals of various psycho-physiological measures and estimates temporal sequence of human cognitive state during problem solving process. The system consists of the two sub-systems: (a)automatic recognizer of data features appeared in short-time wave forms of psycho-physiological signals, and (b)cognitive state discriminator from thus identified data features. A promising result of exploratory study are presented in this paper, for the both sub-systems, neural network application for sub-system(a) while statistical quantification method for (b) , by using a subject¡Çs data obtained by a basic laboratory experiment on human cognitives utilizing eye mark recorder and polygraph as measuring devices for various psycho-psysiological signals from the subject.