This series of posts is looking at the arguments in Philip Pettit’s paper “The Inescapability of Consequentialism”. Part one set out the core features of consequentialism and non-consequentialism (as Pettit sees them). I offer only the briefest of summaries here.
Consequentialism holds that the right alternative in any choice is a promotional function of the agent-neutral good. Non-consequentialism rejects this thesis, most commonly by imposing a set of constraints on action that do not require the promotion of the good. The long-standing debate in normative theory concerns which of these two approaches is correct. Cases can be constructed in which non-consequentialism seems right; but cases can also be constructed in which consequentialism seems right. Where does this leave us?
Pettit suggests it leaves us in a stalemate. His inescapability argument attempts to break that stalemate by reorienting our perspective on the theories. The inescapability argument suggests that, at a certain level of analysis, one cannot avoid consequentialist reasoning. I’m going to discuss the argument in four sections. First, I’ll offer my own formal reconstruction of the argument. Then, I’ll look at three objections to the argument addressed by Pettit.
1. The Inescapability Argument
As mentioned in part one, non-consequentialism adopts a partial and agent-centred account of moral assessment. For the non-consequentialist it really matters whether you are the agent performing an action or not. Thus, although the non-consequentialist can recognise the fact that, sometimes, the death of one person leads to greater overall good, they cannot sanction taking active steps to kill that person. Instead, they draw a distinction between killing and letting die, the former being impermissible and the latter being permissible. In this manner, non-consequentialism incorporates an identity-dependence criterion for moral assessment.
The key move in Pettit’s inescapability argument is to exploit this reliance on the identity-dependence criterion. As he sees it, the overarching focus on agential duties and agential assessments is a vulnerability in the non-consequentialist framework. Why? Because there are circumstances in which we must morally assess scenarios, and that do not turn on or implicate individual choices. In these scenarios non-consequentialism seems to run into a brick wall. It is at this point that consequentialism is inescapable.
The example Pettit uses to illustrate his argument is the assessment of political institutions. These are institutions that determine the options available to individual agents across a broad set of domains. They include institutions of marriage and civil partnership, civil respect, group incorporation, market exchange, and, of course, institutions of law-making and adjudication. Pettit argues that if our assessment of such institutions is to be something other that mere expressions of aesthetic preference, then we have to resort to consequentialism.
Let’s formalise this argument before commenting on it any further. The formalisation is, as always, my own take:
- (1) Non-consequentialism holds that the moral assessment of options must be sensitive to the identity of the assessor/agent in those options (the identity-dependence criterion); this is the distinguishing characteristic of non-consequentialism.
- (2) Therefore, non-consequentialism does not apply when assessing options that do not implicate or involve the choices of individual agents.
- (3) There are important (morally salient) cases in which we have to assess options that do not implicate or involve the choices of individual agents, for example the assessment of political institutions.
- (4) If our assessments of such cases is to be something other than the reflection of subjective preference, consequentialism is the only game in town.
- (5) Therefore, in some morally salient cases — such as the assessment of political institutions — consequentialism is inescapable.
I’ve collapsed the first stage of the argument into a simple inference from a single premise to a conclusion. This is mainly because I don’t think that’s where the objections lie. The main objections, as I see them, come with premises (3) and (4). Is it really true that the assessment of political institutions does no implicate individual choice? And is it really true that there is no other way of assessing such scenarios?
Pettit’s initial gloss on this is as follows. The assessment of political institutions could be re-described so as to implicate individual choices, but this would miss the point of such assessments. For example, we could say that political institutions depend on the choices of some original group of founders, or we could say that political institutions depend on the ongoing choices and coordination of choices between groups of agents. In this sense, we could open up some room for non-consequentialism. But, argues Pettit, this ignores the fact that the moral merits of political institutions do not depend on such choices. It does not matter, for instance, whether the original founder (if such a person exists) was motivated by greed or malice; what matters is whether the institutions he or she founded are just or equitable and so on. In other words, the quality of such original choices, and the motivations behind them, is not what counts in the assessment of political institutions.
I’m not sure whether this is an argument so much as it is a stipulation. Pettit seems to be just making it a condition for the assessment of political institutions that they do not implicate individual choice. That trespasses on the territory of begging the question. Still, I sort of agree with him: I don’t think that the moral quality political institutions is a function of the choices of those who create them. But I doubt a non-consequentialist would agree.
Leaving that aside, there are three other objections to the argument worthy of consideration. Pettit addresses these at length in his article. I summarise and provide some comments below.
2. The Rule-Consequentialist Objection
A rule-consequentialist rejects the notion that the right alternative is, always and everywhere, a promotional function of the agent-neutral good. Instead, the rule consequentialist holds that we should opt for rules that, in the average, promote the agent-neutral good, even if they may mandate sub-optimal actions in particular cases.
Could the rule-consequentialist provide an alternative criterion for the assessment of political institutions? Intuitively, the notion has some appeal. One of the main things that political institutions do is to set rules and obligations for individual action. So why not choose those rules in accordance with the strictures of rule-consequentialism? Take the “institution” of civil marriage. It prescribes a set of rules that must be followed in order to enter into a legally recognised marital union. It is very tempting to say that we should pick the particular configuration of the institution that creates rules that, on average, lead to the optimal outcomes.
We can grant that this is an appealing notion. The problem is that it is straightforwardly consequentialist. After all, even though they may not be consequentialists about individual actions, the rule-consequentialist is still a consequentialist about rules. When selecting institutions which set rules, they would be doing so on the basis that the rules set down did the most to promote the agent-neutral good. Those rules may be non-consequentialist in character, but that doesn’t mean the assessment of the institution is non-consequentialist.
3. The Rule-Contractualist Objection
A somewhat similar, though arguably more subtle, objection comes from the contractualist school of thought. Roughly, contractualism holds that the right alternative in a given choice is the one that reasonable people can agree upon. The precise form of the “agreement condition” varies across different contractualist theories. One of the most famous comes from Thomas Scanlon. He holds that the principles of right action are the ones that cannot be reasonably rejected. Ostensibly, this is non-consequentialist in nature since people could reasonably reject the consequentialist principle.
Contractualism looms large in debates about political morality. Indeed, if I were to pick the one word that is most readily associated with political morality, I would pick “contractualism”. It seems plausible to suggest, then, that contractualism could provide a non-consequentialist basis for picking political institutions. Instead of looking to the institutions that do the most to promote the good, we should look for institutions that nobody can reasonably reject. That would seem to undermine premise (4) of the inescapability argument.
Pettit’s response to this is a little suspicious. He holds that such an approach to picking institutions would, contrary to what we have just said, be consequentialist in nature. This is because, unlike non-consequentialism proper, it assesses institutions from an impartial and agent-neutral perspective. I’ll just quote what he says here (“Nagel” refers to Thomas Nagel who has advocated a kind of political contractualism):
“But the political theory to which Nagel directs us is not like that [i.e. not non-consequentialist]. It holds up an ideal for any society or polity: that its norms and laws should conform to the principles for the enforced, general regulation of behavior that no one could reasonably reject. And that ideal offers a neutral, identity-independent criterion for ranking systems of norms and laws — social and political institutions — on the basis of how far they realize it.
To put it more succinctly: “not reasonably rejectable” is just a good-making property of institutions, and contractualists are simply urging us to maximise that good-making property.
To make this plausible, it seems to me like Pettit has to collapse consequentialism into agent-neutrality. In other words, he has to say that any moral theory that holds to an agent-neutral system of evaluation is consequentialist. I guess that might be a legitimate move. But it seems like a stretch to me. I admit, however, that this may simply be because I find contractualism a plausible basis for political morality, and yet still think there are good-making properties other than “not reasonably rejectable” that are equally important, just not important in the political context. Maybe this just reveals confusion on my part.
4. The Group Agency Objection
The final objection is one that is close to Pettit’s heart. With the help of Christian List, he has written an entire book defending the possibility of group agents, i.e. individual corporate agents that are created through the aggregated behaviours of individual human beings. But, of course, if such group agents are possible, one has the basis for an objection to premise (3) of the inescapability argument.
How might this work? Well, very simply. Assume that the body politic is a group agent. Assume further that political institutions are chosen by this group agent. It would then follow that the assessment of political institutions does implicate individual choices. The group agent could be under non-consequentialist constraints, and those constraints could determine which institutions are morally acceptable and which are not.
Pettit has a rebuttal. As he sees it, there is a desideratum that any compelling normative theory of social arrangements must satisfy. The desideratum is that when assessing institutional options, the theory should enable us to consider all possible institutional forms. For example, when considering institutions of marriage, we should be able to consider institutions that recognise all different kinds of relationship: heterosexual, homosexual, polyamorous, polygynous, polygamous, and so on. We may reject some; but we should be able to consider all.
The group agency objections falls foul of this desideratum because it cannot allow us to consider institutional arrangements that require the death or destruction of the group agent. For example, if we assume the group agent is the democratic electorate, we cannot consider non-democratic institutions. This, Pettit submits, makes it an inadequate basis for the assessment of social institutions, and hence an unsuccessful reply to his argument.
To sum up, Pettit’s argument is that consequentialism is inescapable at a certain level of analysis. This is for two reasons. First, because non-consequentialism rules out the agent-neutral assessment of options. Second, because agent-neutral assessment is inescapable in at least some cases, particularly when it comes to assessing political institutions.
This argument works largely off the assumption that agent-neutral assessment is the hallmark of consequentialism. This might be questionable. It also works off the assumption that the assessment of political institutions is agent-neutral, which is debatable. Nevertheless, if the argument works, there is the question of its broader implications.
At first glance, the claim that consequentialism is inescapable at a certain level of analysis has no implications for other levels of analysis. In other words, we could be consequentialists about the assessment of political institutions, but non-consequentialists about the assessment of individual action. Pettit, however, suggests that his argument provides some support for a more global consequentialism. Specifically, he thinks that if we desire theoretical unity, we should prefer to apply the consequentialist model at all levels.