Some comments on the paperMindsets, rationality and emotion in Multi-criteria Decision Analysis

Abstract

I was very happy to read this paper. It is rare that somebody takes the care to investigate the philosophical foundations of MCDA and despite the fact I have several divergent points of view with the author I was pleased in exercising some abstract considerations about our research and practice. Let me start explaining why I consider the issues addressed by the paper under a different perspective. What is the scope(s) of our research in MCDA? Are we trying to convince decision makers how to do best or are we trying to help people assisting decision makers to do best? Who is our audience? ‘Decision makers’ or ‘assistants’ to the decision makers? Perhaps both, but my personal impression is that who really uses decision analysis tools are what I call ‘analysts’ and these deserve particular attention. Decision theory and F. Wenst p’s analysis remains within a frame where the audience is instead decision makers. This is legitimate, but I feel that decision makers rarely are users of decision aiding tools. Instead analysts are great users of what I call the ‘decision aiding methodology’ (Tsoukias, 2003). I will therefore try to read the paper’s contribution under such a perspective. The result is that part of the author’s analysis loses interest. The fact that decision makers use values, norms, ethics, rules or whatsoever makes little difference. For the ‘analyst’ these are all ‘reasons’ supporting or opposing a certain recommendation. The use of multiple criteria becomes a natural way to represent the fact that the reasons to be considered (and modelled) are several, conflicting, and not necessarily readily available and clear. Further on, if we focus on the interactions between the decision maker and the analyst (what I call a ‘decision aiding process’ (Tsoukias, 2003b)) then I would rather talk about ‘constructing’ reasons instead of ‘eliciting’ them. As already shown in other fields where decision aiding is practiced (such as in psychotherapy, Capurso and Tsoukias, 2003) the reality within which we look in order to ‘solve’ a problem is constructed by the interaction between the client and the analyst (see also Watzlawick, 1984; Watzlawick et al., 1967, 1974). Under the above perspective, the issue is not to know whether we have to use norms, values or duties, but how to introduce such reasons within a formal model. It is not a matter of choice nor is it a question of advice. A decision maker has his/her own reasons and these naturally have different origins. We can discuss with the client whether in a precise problem situation it is appropriate to use a certain reason, but not whether we should use a precise type of reasons. The analyst’s problem is how to put together such different reasons taking into account that these may correspond to different concerns of the client. A decision aiding methodology cannot limit itself in considering a certain type of reasoning (deontic, value based, heuristic or normative). It has to be able to consider any reason. This does not seem to be the concern of the paper which is limited in claiming a wider range of reasons. There are also some further specific points to be discussed in this thought-provoking paper.