This paper presents a unifying approach to the issue of using computers and computerized mathematical models for decision analysis and support. The interpretative aspect of computer intelligence is an essential part of this approach; this paper presents a mathematical formalization of this aspect. Another essential part of this approach is a proposal to integrate the three major formalized frameworks for rational decision making: utility theory (mostly normative), aspiration theory (mostly behavioral), and the (mostly hierarchical) theory of goal- and program-oriented management. These two elements lead to the formulation of principles for a broad class of decision support systems, which includes many existing systems but which also opens new possibilities. Two main cases of decision support are discussed: joint decision making (i.e., involving only one decision maker or several decision makers who have to reach a consensus) and independent decision making (assuming a game-like situation with several decision makers). These cases are illustrated by reference to the DIDASS system, its applications and extensions.