Anticipatory network models of multicriteria decision-making processes

Abstract

In this article, we will investigate the properties of a compromise solution selection method based on modelling the consequences of a decision as factors influencing the decision making in subsequent problems. Specifically, we assume that the constraints and preference structures in the (k + 1)st multicriteria optimisation problem depend on the values of criteria in the k-th problem. To make a decision in the initial problem, the decision maker should take into account the anticipated outcomes of each linked future decision problem. This model can be extended to a network of linked decision problems, such that causal relations are defined between the time-ordered nodes. Multiple edges starting from a decision node correspond to different future scenarios of consequences at this node. In addition, we will define the relation of anticipatory feedback, assuming that some decision makers take into account the anticipated future consequences of their decisions described by a network of optimisers − a class of information processing units introduced in this article. Both relations (causal and anticipatory) form a feedback information model, which makes possible a selection of compromise solutions taking into account the anticipated consequences. We provide constructive algorithms to solve discrete multicriteria decision problems that admit the above preference information structure. An illustrative example is presented in Section 4. Various applications of the above model, including the construction of technology foresight scenarios, are discussed in the final section of this article.