The Role of Fuzziness in Decision Making

In this paper we discuss the true objective of fuzzy decision making models. In particular, we stress that fuzzy models should focus their attention on decision processes, instead of referring to the final output of a decision, which is a crisp act. In fact, while Probability Theory can properly model crisp acts, this is not the case for fuzzy decisions. But in fact human beings manage poorly defined arguments and alternatives, and what we usually call a decision is always ill defined, although we still expect they should produce consistent acts as their consequence. These acts, still being consistent with the previous fuzzy decision, are somewhat unpredictable, being extremely dependant on the specific circumstances at the moment an act is required by decision makers. Therefore, in many cases pursuing consistency of these subsequent crisp acts may be misleading. The main issue should be in principle to check consistency between those acts and the true decision behind them, usually poorly formalized. But this objective may sometimes be unrealistic, since there may be few chances to repeat the experience. We should also focus our attention on the arguments that led us to such a poorly defined decision. This should be a relevant role of fuzziness in decision making, viewed as a decision support problem (acts are supported by a fuzzy decision which is supported in fuzzy arguments).