Philip Pettit is a bit of a hero of mine, though sadly for irrational, prejudicial and nationalistic reasons. He’s the only contemporary Irish philosopher of note (apologies to all the un-noted ones), and this gives me some hope. I am, after all, in this for the notoriety.
Pettit has written many interesting things over years. If you’re keen on contemporary republican political theory, you should start with his book Republicanism: a theory of freedom and government. And if you’re interested in the possibility of group or corporate agency, you should look at his co-authored book (with Christian List) called Group Agency. It is all fascinating, rigorous and well-crafted stuff. It is also influential, not just in philosophy but in the real world too. His political theory, for example, had an important influence on the policies of Zapatero’s Spanish government.
Anyway, that’s enough hagiography. Over the next two posts I want to take a critical look at an argument that Pettit made in a recent paper titled “The Inescapability of Consequentialism”. No prizes for guessing what that argument is (hint: it has something to do with the inescapability of consequentialism). The remainder of this post deals with the terminological and conceptual issues underlying that argument, setting out a fairly detailed account of the distinction between consequentialism and non-consequentialism. Part two will cover the argument proper. Although this may suggest that this post is quite dry and dull, I’d like to think that it isn’t, and even if it is the groundwork it lays for part two is pretty essential. So you can’t really afford to skip it.
1. Consequentialism in General
One of the nice things about Pettit’s article is that he gives a detailed, albeit highly abstract, account of consequentialism. He then marries this to a defence of a particular sub-type which I’ll call “standby” consequentialism (standby-CQ). I’ll briefly summarise both of these elements of his discussion.
Starting with consequentialism in general, Pettit offers the following definition:
Consequentialism: The right alternative in any choice is a promotional function of the agent-neutral (or non-indexical, impartial) good.
We could put this more straightforwardly by saying that consequentialism is the view that you should pick the option which promotes the good to the greatest degree, but this more straightforward definition would skip over some of the important complexities in Pettit’s definition. Indeed, each of the italicised terms in his definition is chosen for a particular reason and hence worthy of further elaboration.
First, we have the notion of “choice”, which covers any scenario in which an agent must pick between different options. An option is defined as a package of probable outcomes that are (however minimally) within the agent’s control. As Pettit notes, the contemporary preoccupation in consequentialist theory is with the choice between actions simpliciter. For example, the choice between drinking orange juice or drinking apple juice is a choice between actions simpliciter. But there are other kinds of options that can be brought within consequentialism. For example, an agent may choose between different plans, policies, dispositions, motives, and habits. In each case the actions underlying those options might be quite complex, requiring years of effort or coordination among other agents to bring about.
Second, we have the notion that the rightness of a choice is a “promotional function” of the good. This means that the option selected is right if and only if it is the one that does the most to promote the good. In other words, if it is the one that maximises the good, or leads to the optimal allocation of the good. This is a distinctive trait of consequentialism, and is to be contrasted with the non-consequentialist approach of “honouring” or “respecting” particular types of good in one’s actions. We’ll talk about this in more detail below.
Third, we have the notion of “agent-neutral” good. This is included in the definition to ensure that consequentialists never prioritise or privilege their own status, perspective or good when picking options. They must pick options that achieve the best outcome from a universal and impartial perspective. This is another distinctive characteristic of consequentialism, and, indeed, for Pettit’s purposes is perhaps the most distinctive characteristic. We’ll see this when we get to his actual argument for the inescapability of consequentialism in part two.
Fourth, and finally, we have the notion of the “good”. This is a placeholder for whatever kinds of things, or states of affairs, we deem to have value. It is a matter for axiological theory to determine what these might be. We can, of course, give some famous examples. The hedonic utilitarians, for instance, held that conscious pleasure or well-being was the only kind of good. The preference utilitarians held that the satisfaction of desires/preferences was the only kind of good. In both instances the theories picked out one thing as being the sole bearer of value. Not all theories of the good will be monistic in this way. For example, certain schools of natural law theory pick out seven or eight “basic goods”, which they deem equally valuable and incommensurate. This suggests that there could be many types of consequentialism, all varying in what they think is good.
2. Standby Consequentialism
Now that we have this general understanding of consequentialism, we can proceed to Pettit’s preferred variation, standby-CQ. This theory is constructed in response to a classic criticism of consequentialist theory.
Consider the following case:
Friendship Request: Your friend asks you to help them move apartments. You think about it from a consequentialist perspective, isolating the various goods associated with helping versus not-helping and eventually deciding that helping them would indeed promote the most good. Your main reason for thinking this is that maintaining and preserving bonds of friendship and loyalty is important. Assume this is correct. Can you be confident that you have done the right thing?
Superficially, we might say “yes”. But, so the critics argue, this may not be the case. It could, plausibly, be argued that bonds of friendship and loyalty are good only insofar as they are authentic, and that the authenticity of friendship depends (at least in part) on its sometimes spontaneous and unquestioning nature, particularly when it comes to responding to requests. Thus, the very fact that you had to think about the request from a consequentialist and agent-neutral perspective undermines the good that you are trying to promote. In this way, the consequentialist approach can seem self-defeating in certain cases.
You may or may not find this example compelling. It is merely an instance of a more general phenomenon, one that has to do with the complexity of evaluating the goodness of actions. You see, the goodness of an action is partly a question of its outcomes, but also partly a question of its causal origins, including the dispositions and motives from which it flows. If this is right, there may be other cases in which consequentialist deliberation undermines the good it is trying to promote.
In any event, Pettit thinks that standby-CQ can help to avoid this problem. Standby-CQ holds that in certain cases — such as the friendship case — you should cultivate a general disposition to act spontaneously and automatically in response to the ordinary range of cases. Hence, when your friend asks you to move apartment, you respond automatically and without consequentialist deliberation. This allows you to promote the desired good without undermining it in the process.
This attitude only holds for the ordinary range of cases. When we move outside that range, the ordinary consequentialist mode of analysis prevails. Thus, for example, if your friend asks you to move a dead body and dump it in the sea, you should think about it in accordance with the general consequentialist principle. This is why the position is called “standby CQ”. Although it cedes active control over decision-making to sets of dispositions in many cases, the inner consequentialist is always on standby, ready to take over when things get a little strange. This is what distinguishes it from rule-CQ. A proponent of rule-CQ cedes control to rules, holding that we should follow rules even when they may lead to a sub-optimal outcome in a particular case. Standby-CQ never mandates this: the right alternative is always the one that promotes the most good.
I pass no comment on whether standby-CQ is successful. I mention here it because it is an interesting variant on consequentialism, and because Pettit appeals to it when making his main argument.
With that long introduction and overview of consequentialism out of the way, we can be a little briefer in our treatment of non-consequentialism. This is because non-consequentialism can be defined simply as the rejection of consequentialism. In other words:
Non-Consequentialism: The right alternative in any choice is not a promotional function of the agent-neutral good.
This isn’t particularly informative, but that’s by design. There are as many ways of being a non-consequentialist as there are ways of rejecting the consequentialist thesis. This is because the consequentialist is proposing the existence of a general moral constraint on choice, and there are at least three different ways in which to reject that general constraint.
Particularism: There are no generalisable moral constraints on choice. In other words, the rightness of an action is always determined in response to the particulars of the case at hand.
Prerogativism: The consequentialist constraint on choice holds generally, but there are some exceptional cases where one has the prerogative to discard it.
Alternativism: There are other, non-promotional, general moral constraints on choice.
The first two forms of non-CQ can be ignored here. They are really only included in the interests of completeness. It is the third form, which is really the name for a whole family of different theories, that warrants our attention. Theories within this family point to many different constraints on action. These constraints may sometimes coincide with the consequentialist constraint, but may diverge on other occasions.
Take two classic examples. The truth-telling constraint and the non-violence constraint. The truth-telling constraint holds that you should always tell the truth. In most cases this will lead to the optimal outcome, all things considered. But in some cases it won’t. For example, in cases where telling the truth leads to someone’s death, or irreparably damages a valuable relationship. The same is true for the non-violence constraint. In most cases it will be better, all things considered, not to harm another person. Still, there may be some cases where violence is needed to secure a better outcome.
There are two important features of non-consequentialism that need to be emphasised here. First, although there are many different kinds of non-consequentialist constraint, the values implicit in those constraints will often be similar to the values singled out by consequentialist theories. A consequentialist can appreciate the value of non-violence and truth-telling just as much as a non-consequentialist. The difference between the approaches comes not in relation to the goods they are concerned with, but ino the attitude they think one should have towards those goods. Consequentialists hold that you should promote the good with every act; non-consequentialists hold that you should honour and respect the good in every act.
Second, non-CQ drops the agent-neutrality requirement of consequentialism. Far from mandating an impartial and non-indexical attitude toward morality, non-CQ actually mandates an indexed and highly partial attitude (even Kantian universalism, as Pettit argues at some length). To be precise, it holds that whether an option or outcome is preferable varies depending on whether you are directly involved in the choice or not. I’ll quote from Pettit here:
Suppose that I am asked to assess the two following scenarios on the basis of the non-violence constraint. In one, an anonymous agent NN resorts to violence in a way that breaches the non-violence constraint but as a result there is less violence of that sort overall; in the other, the agent NN refrains from such violence but as a result there is more of that sort of violence overall. If I know that I am NN, then consistent with being committed to non-violence myself, I shall rank the second scenario above the first. If I know that I am one of the other agents [within NN’s society], then consistent with that commitment I shall rank the first scenario over the second; not to do so would be to look with equanimity on the prospect of my being violent.
("The Inescapability of Consequentialism", p. 55)
What Pettit is saying here is that one’s attitude toward different scenarios varies depending on the active role one plays in those scenarios. Consistent with the non-violence constraint, you would not wish yourself to do violence. But this means you may prefer a scenario in which someone else does violence if that reduces the likelihood of your doing violence. In this manner, non-consequentialism is committed to an identity-dependence criterion in moral assessment: it really matters who you are when engaging in moral assessment.
Now, I’m not sure that I find Pettit’s example entirely persuasive. It seems like there is something fishy going on when you ask a non-consequentialist to rank their preference for scenarios, and I’m not convinced he fully rebuts the case for the universalisation constraint. But I quite agree with him when he says that the identity-dependence criterion explains important features of non-consequentialism. In particular, it explains the importance that non-consequentialists attach to personal responsibility, and to the differences between doing and allowing. For instance, non-consequentialists are famous for clinging to a distinction between killing and letting die, and this can be explained by reference to the identity-dependence criterion.
In addition, the identity-dependence criterion is crucially important when it comes to understanding Pettit’s inescapability argument, which is something we’ll try to do the next day.