On the epistemic interpretation of voting

 

In this paper, I sketchily present the epistemic interpretation of voting and its (possible) consequences. My aim is to show that, even if one takes the epistemic interpretation in its best lights, one ends up in contradiction; plainly, my thesis is that epistemic interpretation is logically void.

The reason for advancing the epistemic interpretation of voting was to offer a solution to the question ‘what are the use and the meaning of vote?’. If voting is to justify the imposition of certain policies or regulations upon individuals that may oppose to this imposition, then voting should explain why the resolutions adopted by some electoral processes are morally justifiable and acceptable to all the citizens. For Coleman and Ferejohn, there could be three kinds of justification for imposing a decision coercively.[1] There is the contractarian (or rational) justification, the instrumentalist (or consequentialist) justification, and the proceduralist one. The epistemic interpretation of vote is a special case of proceduralist justification, but before analyzing this point, I shall present the differences between the three main kinds.

The contractarian justification of the collective decision making states that the outcome of voting may be coercively enforced even on those which are either negatively affected or against it if and only if the collective decision has been took by an ex ante unanimously agreed procedure. A first objection to this attempt is that such an unanimous agreement is not practically feasible. A second one is that it would be absurd to suppose that some rational individuals would ever preclude themselves to change their minds in the case that they would find out that they were wrong. Nevertheless, these objections against the contractarian justification overlook the fact that the ex ante agreement is a counter-factual supposition only. Precisely, what does it matter is that a given procedure could be justified by the fact that it is reasonable to imagine that all the rational individuals would unanimously agreed upon it in an initial or state-of-nature position. This is the reason why practical impediments do not count as objections to contractarian justification of a decision-making procedure. Nevertheless, one may say that contractarian justification does not ensure that a certain procedure of decision-making will be agreed upon, and therefore the outcomes based upon such a procedure may be arbitrary. For Coleman and Ferejohn, this is not a good argument against none of the possible justifications of decision making procedures (namely, they say that it is irrelevant to say that ‘if the rules would have been different, the outcome would have been different too’)[2].  For me it seems quite unclear the reason why they reject this kind of objection; in my view, at this point they simply abandon the task of giving arguments, and they simply assert that because there is no absolute or objective motive for having a certain rule instead of another rule (take the football’s rules), then one should not question the rule. First, I would say that suspending the reason due to this seemingly absence of a good motive for accepting a rule proves nothing but a lack of interest in explaining why a certain rule has been adopted (for instance, in football, the rules may be explained historically). Second, the comparison between a game and democracy is illegitimate, because the way the outcomes are produced is highly important for a democracy; given the fact that the authors agree with the fact that the way the outcomes are produced matters, the above-mention comparison should be illegitimate in their own terms.

The instrumentalist justification of voting allows the enforcement of collective decisions due to the fact that decided actions have desired consequences. The instrumentalist view is a non-cognitivist conception of voting, based upon preferences’ ranking. One could say that the instrumentalist interpretation of voting is faced with the difficulty of an outcome satisfying none of the voters. Now, this would not be a special problem if it would not imply that in the case that the outcome of decision-making obtained according to an instrumentalist procedure that is opposed to the individual preferences or interests of the voters should be justified according to non-instrumentalist criteria. This is so because the original claim was that the outcome is justified because it has good or desired consequences; but because the consequences are bad for or not desired by some (and, in the extreme cases, by all) voters, then either the outcome is unintelligible or it should be grounded into another theory. All in all, a pure instrumentalist justification is not feasible; it should either be based upon a different criteria, or to normatively restrict its availability to a certain set of outcomes.

The proceduralist justification of collective decision making is radically different from the first two because ‘proceduralism holds that what justifies a decision-making procedure is strictly a necessary property of the procedure - one entailed by the definition of the procedure alone’[3]. For comparison, the contractarian justification of the procedure rests upon its evaluation in an initial position, while the instrumentalist one depends on the evaluation of the expected consequences. These evaluations are done through criteria that are unrelated to the procedure; to put this into a strong formulation, in these two cases the procedure is chosen from a set of possible procedures in virtue of being fitted with the evaluative criteria. Before going further, I would like to point out that this distinction between the proceduralist justification, on one hand, and contractarian and instrumentalist justifications, on other hand, is committed to reducing contractarian justification to a special kind of instrumentalist justification. Plainly, Coleman and Ferejohn say that the contractarian justification depends on how the voting procedure rule’s consequences ‘would be evaluated ex ante’[4].  Therefore, contractarianism is consequentialism of a special sort - namely, that consequentialism that evaluates not some specific or already given outcomes, but an indeterminate set of possible future consequences.

The authors’ claim is not that there could be such a procedure that could justify its outcomes in virtue of its intrinsic properties only, but that the contractarian or instrumentalist justifications are not the obviously winners of the decision-making quest; moreover, the if one would accept, in the end, a contractarian or instrumentalist justification, then one should seriously revise the motives for that acceptance (nevertheless, I shall not scrutinize this second claim because my thesis is that the rationale that support it is logically void).

The social choice theory is one attempt to provide a proceduralist justification of voting, and the discussion started by the impossibility results given by the social choice theory seems to strongly affect the other two kids of justification as well. The social choice theory seems to be justifiable by its intrinsic properties because it is able to aggregate a collective decision from the individuals’ hierarchy of preferences. Nevertheless, for Riker this capacity is not that obvious; his observation is that ‘although each individual in the society [may have] a transitive ordering of preferences, the outcome of voting is not [necessarily] transitive’, that there are cases when we have ‘a coherent men and an incoherent society’. For example, we could suppose that an individual X prefers A more than B and B more than C (noted as A>B>C); that Y prefers C>A>B; that Z prefers B>C>A. The individual preferences of X, Y, and Z are transitive and coherent. However, their votes’ outcome is incoherent, as the figure shows: A>B>C>A>B>C.[5]

Coleman and Ferejohn’s reply is that the probability of mapping incoherent voting outcomes from coherent individuals’ preferences is quite small.[6] Notwithstanding, the principal lesson they withdraw out of Riker’s example is that contractarian and instrumentalist justifications are under the same threat. For Riker, a social choice procedure should be subject to both normative and semantic constraints. The normative constraints are ethical, requiring the procedures to be fair, while the semantic constraints are logical, requiring the procedures to be meaningfully related to the individuals’ wills or preferences. In Riker’s opinion, the threat for the purely proceduralist justification is the infringement of the semantic constrains, while for Coleman and Ferejohn the semantic constrains may be violated by contractarian or instrumentalist justifications, too.[7]

The semantic constraints are the stability of voting’s outcome and its uniqueness. On the one hand, the outcome of voting is said to be stable if from a set of transitive individuals’ preferences one could map out a transitive (non-paradoxical) ordering of collective preferences. On the other hand, the outcome of voting is said to be unique (or unambiguous) if from a given set of individuals’ preference ordering the same collective decision will be obtained regardless of the aggregation procedure.[8]

The first reply to Riker is that any aggregating procedure may generate unstable outcomes, regardless its justification. For instance, the fact that a certain aggregating procedure has been agreed upon ex ante does not protect it from generating unstable outcomes in some special circumstances; the same remark is true for the instrumentalist justification of an aggregating procedure.

For the above reason, what seems to distinguish the procedures based on contractarian or instrumentalist justifications is the fact that they do not produce ambiguous outcomes. As it was already said above, the procedures based upon contractarian or instrumentalist justifications are constrained in the sense that their outcomes should fit some certain external criteria. The range of the outcomes is, somehow, quite well defined, and due to this fact the contractarian-based and instrumentalist-based procedures would not produce ambiguous results. More exactly, in these cases the procedures are chosen exactly in order to produce a certain kind of outcomes; as the range of the possible outcomes is already and externally predefined, it is not the case to say that these procedures are generating ambiguous results.

Therefore, what seems to individualize and to weaken the proceduralist-based procedures is the fact that they may generate ambiguous results. Now, ambiguity may be understood in two ways. The first interpretation of outcome’s ambiguity is that if one uses different aggregation rules starting from the same set of individuals’ preference orderings, the results would not match. As I already mentioned, for Coleman and Ferejohn this interpretation of ambiguity is uninteresting and, in the end, it does not constitute a serious objection to a proceduralist justification of a decision making. The rationale for taking this interpretation as unimportant is the assumption that what matters is that when the individual voters know the aggregation rules, their collective decision should be unique. To ask what would have happened if they would have knew that the aggregation rules were different is an external question to the task of internally defending a certain set of aggregation rules. More exactly, if a certain set of aggregation rules may be internally (that is, procedurally) defended, then the fact that other set of aggregation rules are not defendable or that, though equally defendable, they are producing a different outcome starting from the same set of individuals’ preference ordering is not an objection. I should subscribe myself to this rationale;[9] however, the rejection of this objection disregards the normative constraints that a social choice procedure shall fulfill.

The second interpretation of ambiguity is more troubling for the proceduralist justification of collective decisions, because it addresses an internal uncertainty. The second understanding of ambiguity inquires if it is possible to state that what has been obtained from a given set of individuals’ order of preferences through a procedure is the collective decision of those individuals having those preferences’ ordering about a specific voted problem, or not. Before continuing, and in order to avoid discussing topics that are not in the scope of this paper, I shall point out that this second interpretation of ambiguity may affect both Riker’s liberal standpoint (that is, democracy understood as restricting the range of voting for amending officials) and populist standpoint (i.e., democracy understood as extending the scope of voting to discovering the general will that warrants the self-respect and self-realization of humans). This is so because it does not matter if the outcome of voting is understood as a general will or as a ban on official in order to formulate the question if the outcome is ambiguous (in its second meaning).[10]

Nevertheless, in the case of interpreting the outcome of voting as a general will, the 2^nd interpretation of ambiguity applies if and only if one assumes that there should be a unique general will in all electoral processes and that, moreover, this general will is defined by voting. To this problem, one may give three kinds of replies:

i.      One can deny that there should be a general will in all the instances when voting is required as a decision-making procedure.

ii.    One can simply refuse the idea that there is a unique general will. The argument would be as in the case of the contractarian justification of voting, namely that the general will only specify a set of acceptable outcomes.

iii.    One can say that the general will is not defined by voting, but that the votes are simple judgments about what it could be general will in a given electoral process. The judgments may be true or false, and due to the fact that votes have a truth-value, there are not to be confused with the voters’ preferences regarding the decided matter, because they are voters’ best opinions about what should be done. Simply said, the relation between voting and general will is an epistemic one.

The first reply to the ambiguity objection seems to be self defeating for the proceduralist justification of voting, simply because the criterion according to which it is decided when should be looked for a general will and when should not it is external for the proceduralist attempt. But this is exactly what was at stake, namely the proceduralist claim that the outcome of voting is holly justifiable in terms of the procedure used to produce that outcome from the given individuals’ preference orderings.

In what regards the second reply to the ambiguity objection, Riker objects that the set of possible outcomes is so large that the use of general will is vacuous.[11]

The third reply, namely the epistemic account, it takes the general will as transcending the act of voting. Therefore, there is no necessary connection between the results of voting and the general will, which implies that the impossibility results met by social choice theory do not apply here. In Cohen words, this third standpoint is to be distinguished from the other two procedural accounts by asserting the tightness of the relationship between voting and general will. More exactly, the first two accounts should be characterized as purely procedural, because they try to identify and/or define general will through the procedure employed in mapping out of individuals’ ranked preferences in the electoral process. On the other side, the epistemic interpretation of voting is a reasonable proceduralism, which states that “the judgments of majorities provide a reasonable, although imperfect procedure for determining the general will”[12]. Therefore, the reasonable proceduralism does not claim that there is a certain way to find out the objective general will, but only that is possible to be found out or approximated. All in all, “the general will is characterized in terms of an ideal procedure of deliberation or collective choice, while democratic decision making is construed as an imperfect procedure which, when suitably organized, has the property of providing evidence about how best to achieve the object of the general will”[13].

For Cohen, the interpretation of voting should accomplish two tasks.[14]

è The first task for the interpretation of voting is to explain the voting behavior of individuals. In Coleman and Ferejohn vision, voting is motivated because it ‘strengthens allegiance, increases competence, develops a sense of community, and the like’. Nevertheless, the epistemic interpretation of voting should explain why one would have the incentive to vote, if one’s vote is understood by one as a simple judgment about common good or general will. As put by Hurley, it seems that the epistemic interpretation of voting fails to explain why one would have any incentive to vote in this case[15]. The objection goes like this: in the case that one interprets one’s vote as one’s expressed preference, then one would have an incentive to vote because expressing preferences is directly related with one’s needs and interests; in contrast, if one sees one’s vote as a judgment, the incentives to vote are missing because there is no direct impact of voting upon one’s needs and interests, or even the opposite. This objection is related to the problems raised by the Condorcet’s jury theorems, which theorems support the epistemic interpretation of voting. The Condorcet’s jury theorems are based upon the judgmental competence of the members of the group; nevertheless, the judgmental competence requires that, on the one hand, the individuals who vote are good judges and, on the other hand, that the individuals who are good judges would vote according to their judgments, not according to their preferences. [16]To this difficulty, Hurley proposes the following solutions.

First, by institutional design, one may be both provided with incentives to vote and with necessary knowledge in order to make a good judgment; the institutional solution is that of dividing the electoral domains as to fit the personal competences of the voters with the specific of the problems supposed to be voted upon. Due to this divided domain, the voter will be somehow in her own sphere of interest, and therefore the impact of her vote will count for her - this is a partial incentive. This division of the domain has another secondary role, namely that of preventing or decreasing the probability of occurring the impossibility results. Second, the epistemic interpretation of voting should put more weight upon education, deliberative capacity, and public-spiritedness of the citizens in order to provide them with the required means to formulate as good as possible judgments about what should be done.

Second, the psychological appeal to public participation of collective deliberation is not only at least as successful as the appeal to self-interest, but it is very likely to extend the range of possibilities for collective actions in areas where this would not be possible through appeal to self interest. This is so because the incentives for epistemic cooperation are less egoistic and less affected by the impact of one’s vote upon the outcome.

è The second task for the interpretation of voting is to contribute to a normative political theory that ensures justice and democracy. The epistemic interpretation of voting claims that if there is a general will, then both public deliberations would be guided by the principles that characterize general will, and the act of voting would provide evidence about it. There are three elements of the epistemic interpretation of voting that ensure the task’s accomplishment.

The first element is the fact that there is an independent standard of correct decisions, independent in the sense that the standard does not depend upon a circumstantial outcome of voting. This independent standard is the general will, and a group of individuals may have a general will if ‘(a) the members of the group share a conception of the common good, and (b) the members regard the fact that an institution or policy advances the conception [of the common good] as a reason for supporting it, and (c) it is fully common knowledge that the conception [of the common good] is shared, and (d) the conception is consistent with the members of the group regarding themselves as free and equal”. [17]

The second element of the epistemic interpretation of voting is what indeed makes this account a cognitive one, namely that the votes expresses not personal preferences, but judgments according to and about the independent standard of correct decisions (i.e., judgments that try to approximate as much as possible the general will).

The third element is the understanding of decision making as a rational process of forming judgments about general will. The decision-making is based upon both discriminations among various levels of voters’ competence (i.e., democratic division of epistemic labor), and the education of citizens and their awareness of the democratic demands for public reasoning.

Now, for Coleman and Ferejohn, the problem with epistemic interpretation is that “the reliability of voting rules depends on our having independent access to the nature of the general will. Either we have such access or we do not. If we do, then voting is otiose; if we do not, then we cannot determine its reliability”[18]. To this objection, a first answer is that although one could not have an independent access to the general will, this does not necessarily imply that voting is useless. On the contrary, two basic elements make electoral process reliable. The first one is the fact that voting is based both upon a democratic and a rational decision-making[19]; the second one is that voting outcomes are rigorous due to the epistemic division of labor[20]. In my opinion, the first point should not count as a buttress for the epistemic interpretation of voting, simply because the original claim was that a proceduralist account, as the epistemic one is, should be distinguished from the contractarian and instrumental accounts exactly because in its case there is no need of appealing to external evaluations. However, in this instance, an appeal is made to some external normative evaluations of the procedure (i.e., rationality, fairness).

Nevertheless, the second point is acceptable on proceduralist grounds because the democratic division of epistemic labor is internally required by the attempt of reaching the best approximation of the general will. The problem with reliability is that even in the case that all the elements of the epistemic interpretation would be reliable, this it is not sufficient for being able to justify the fact that the collective decisions should be binding even for those individuals that oppose to them. What is complementarily needed is that the reliableness of the decision-making procedures is known by all the electors. The reason for this is that the procedures employed by the epistemic account should not limit themselves to producing outcomes that are more likely to be fitted to the individuals’ ordering options than other alternative procedures, but the epistemic procedures should provide legitimacy to the outcomes, too.

My thesis is that neither this rationale may save the epistemic account from the paradox stated above by Coleman and Ferejohn. The argument first states that if the electors falsely believe that the outcomes are produced by reliable procedures, then the epistemic account would not have cognitive value anymore, but only persuasive value. Having this point clarified, let me further assume that the electors are not misled in their belief that the epistemic procedures used for producing the binding collective decision are reliable. The problem that arises here is that if the epistemic procedures are indeed reliable, then they are based upon a division of epistemic labor, which division could not be explained in proceduralist terms. On the one hand, the division may not be explained in epistemic terms exactly because it will end up in a vicious circle. On the other hand, the division shall be explained either by appealing to external criteria of evaluation (e.g., the importance of that matter for society - but this criteria is not subject to epistemic inquiry), or by other voting procedures that are not epistemic.

Until here, I have shown that a purely proceduralist account for justifying binding collective decisions is either reducible to other sorts of justifications, or unsound. Moreover, the reasonable proceduralism (i.e., epistemic account of voting) is confronted with the same problems as the purely proceduralist account (in spite of Cohen’s claim). Therefore, the attempt to justify the enforcement of collective decisions through appeal to the epistemic and/or procedural interpretation of voting is a failure. In conclusion, the contractarian or instrumentalist justifications should not be revised on the grounds that the general will and the individuals’ judgments about it are relevant for the electoral processes.

However, this does not ensure that a contractarian or instrumentalist argument is clearly explaining why one shall accept some electoral upshots that one does not like. From a rational choice perspective, the first answer is that it is not rational to accept some electoral outcomes that you dislike at least in one case. Suppose that you are part of a minority group that will never be in the position to see its political views reflected within the mojoritarian decided policy. Nor will this group be in the position to have bargaining power in a coalition. This is the case, say, of an electorate that is so special that will ever be less than 0.1% from the entire population, like the group of philosophers in a population. To this electorate it cannot be argued that it is rational to accept the results of the vote, since neither from a one-shot game nor from an iterated-game perspective they draw no incentives for their submittal. I think that the reason why these people shall accept the results is left yet unexplained by both the remaining available theories, that is, contractarianism and instrumentalism.

Let me detail this before ending. From the contractarian perspective, one can argue that the small group of philosophers has agreed that regardless the outcome of elections, they will accept them. Then, their egoist rationale was that this compact is the best method in order to attain some basic goods, like security, a legal system, and so on. However, after a reasonable period of time, they discover that it’s irrational for them to accept the elections’ results, since it will never happen that they will be even close to be content with them. It would be difficult to explain them the reason why they should accept the perpetual imposition of electoral outcomes that are against their personal interest, that is, against their rational expectations.

On the other hand, from the instrumentalist perspective one can argue that the insignificant community of philosophers has reasons to accept the unwanted electoral outcomes because, at the end of the day, even these unwelcome outcomes will amount to some benefits for the philosophers (or that these benefits are somehow superior to the situation where no elections would take place). However, this may not be the case and the philosophers may never consider that they are better of with unwanted electoral outcomes than without. Given this, there seems to be no rational way to explain why this insignificant community shall accept the imposition of others’ wills or interests upon them.

The conclusion of my paper, however weak or minor, signals to the following fact. The connection between solving the collective action problems and the political solution (especially as it emerges from Lever’s Private desires, Political Action) has yet to be explained and secured, because it is not at all straightforward.

 

Cristian Vasilescu

2004

 

Bibliography:

 

Coleman, Jules, and John Ferejohn. Democracy and Social Choice. Ethics 97, 1986.

Cohen, Joshua. An Epistemic Theory of Democracy. Ethics 97, 1986.

Riker, W. H. Liberalism Against Populism. Prospect Heights, Ill: Waveland Press, 1982.

Waldron, Jeremy. "Rights and Majorities, Rousseau Revisited." In Majorities and Minorities, J. W. Chapman and A. Wertheimer, eds. Pp. 44-75. New York: New York UP, 1990.

Hurley, S. L. Natural Reasons. Personality and Polity. Oxford: Oxford University Press, 1989.