Self-organizing neural network (SONN) is known to be able to extract features in input samples [Kohonen, 1995]. By updating not only the weight vector of the winning neuron in the self-organizing layer but also those of its neighboring neurons, neighboring neurons would eventually become to respond similarly to a specific input vector. Then the distribution of winning neurons for a class may be distinguished from those for other classes. Luttrell proposed a SONN which can inherently use the correlation between input vectors of separate clusters and he called it self-supervised adaptive neural network [Luttrell, 1992]. In this report, we propose the use of the selfsupervised adaptive algorithm in analyzing the correlation between cognitive style and the accuracy of intuitive time-series forecasting, and suggest a way to compare the relative degree of correlation between each of cognitive style, subjective emotion and physiological phenomenon and the accuracy of intuitive time-series forecasting.