Dynamic Fuzzy Sets

This paper presents a way of fuzzy inferencing that allows for the resulting judgment to change in time. The aim of the research is to explain the empirical evidence that people’s feelings about a person or a social situation sometimes oscillate between a highly positive value and a highly negative value even in the absence of new data. The proposed method of judgment generation employs the concept of fuzzy set, however, the idea of dynamic fuzzy sets, for which membership values are allowed to change in time, is introduced. A set of constant membership values has been used as the control parameter a in a dynamic model Xt+1 = F(a, Xt ) with the proposed function F implemented as a feed-forward neural network. Xt is the set of state variables. One of the state variables was interpreted as the resulting judgment. Simulation showed that the judgment produced by the proposed model sometimes oscillates like the social judgment produced by a human subject.