The Discursive Dilemma

The discursive dilemma arises because the aggregation of individual attitudes is subject to cer­tain inescapable limitations.¹⁰ Formal impossibility theorems demonstrate that there exists no aggregation procedure that could satisfy an intuitive set of criteria for deriving the collective view from those of the individual members.

Tenure Committee is a by now classic example of the discursive dilemma.¹¹ Three members of a university’s tenure committee have to decide whether to recommend awarding tenure to Dr. Borderline.¹² University by-laws for tenure state that excellence in areas of research, teaching, and service is required (and sufficient) for tenure. In the case of Dr. Borderline, in each of the three areas a majority of the committee vote that she has achieved the required standard of excellence. The committee therefore recommends awarding tenure to Dr. Borderline (Table 21.1).

The reason why the situation is dilemmatic is that all decision procedures available to the committee are problematic in terms ofhow they derive the collective decision from the judgments of individual members. In the previous paragraph the committee was described as using the so- called premise-based decision procedure. They took a majority vote on each criterion, letting the decision itself follow from these votes as the logical conclusion. The manifest problem with this decision procedure is that, as can be seen in Table 21.1, the committee recommends awarding tenure despite the fact that individually each member of the committee is against this.

The committee could use the conclusion-based procedure instead, where each member votes only on one question, namely whether or not to recommend awarding tenure to

Table 21.1 Tenure Committee

Dr. Borderline.

So, the committee faces a dilemma: if it embraces the premise-based procedure it will reach a decision not supported by any of its members; if it uses the conclusion-based procedure it violates consistency at the group level.¹⁴ Impossibility theorems generalize this finding showing that the difficulty is not restricted to the premise-based and conclusion-based procedures. More precisely, what these theorems show is that there exists no collective judgment aggrega­tion procedure which for any (rational) profile of individual judgments will both (i) guarantee responsiveness to the views of members on each of the issues involved, and (ii) yield collective judgments on these issues which are themselves consistent and complete (List & Pettit 2002).¹⁵

The negative result of impossibility theorems does not entail of course that decision procedures would meet various normative desiderata equally well. Among others, it has been argued that the premise-based procedure is under most circumstances a better “truth-tracker” than the conclusion-based procedure (Pettit & Rabinowicz 2001; Bovens & Rabinowicz 2006). Perhaps even more importantly, the use of the premise-based procedure may be required by moral, political, or legal norms (Pettit 2001; Chapman 2002; List 2006). Some also believe that only the use of the premise-based procedure enables the group to live up to fundamental demands of rationality (I will return to this last point in section 21.6 below).

At the same time, some other considerations actually favor the conclusion-based procedure. First, in some cases groups may want to put a premium on unanimity over simple majority. Second, the conclusion-based procedure may better safeguard against strategic voting (List 2006: 391). Third, the premise-based procedure is not always a better “truth-tracker” (Bovens & Rabinowicz 2006). Fourth, in some cases we may want to or have to take a stand on a logic­ally complex proposition as a whole without having a firm view on one or more of the simple propositions constituting it. Fifth, decomposing a conclusion into premises is to some extent an arbitrary matter because those very premises can themselves be decomposed. Sixth, in many cases the same conclusion might follow from different and logically independent sets of prem­ises. That is, different arguments may favour the same conclusion.¹⁶

21.5