TWO KINDS OF ‘SUBJECTIVE CONSEQUENTIALISM’

In his classic paper ‘Alienation, Consequentialism, and the Demands of Morality’, Peter Railton defines two kinds of consequentialism, subjective and objective:

We must immediately note that Railton here confuses two things, calling each of them ‘subjective consequentialism’. The first is the view that we ought always to use a subjective consequentialist decision procedure. That is, that we should always think consciously about our available information, and what it tells us about the likely consequences of our various actions, in order to determine which of these actions is most likely to produce the best consequences, and then perform this action. Railton argues that following such a theory produces alienated agents who are less likely to produce good consequences than typically agents who reject the theory. He also offers a regress argument suggesting that the theory can't even be followed at all, since an agent who tried to do so would be caught in an endless loop of trying to determine how much time she should take to make a decision, but first trying to determine how much time she should spend making that calculation, and so on.

These arguments seem decisive, but make no dent upon the second kind of subjective consequentialism, which is not a view about what decision-procedures we ought to use, but simply about what agents ought to do. As a complete ethical theory, this more properly deserves the name of subjectivism consequentialism. Howard-Snyder defines this theory as claiming that:

While she does not define ‘best consequences’, I will assume that this is equivalent to ‘maximum expected utility’, resulting in a consequentialism which is thoroughly subjective with respect to the agent's reasonable beliefs about the probabilities of various outcomes, their values, and the effect that his actions will have on these probabilities. So defined, subjective consequentialism in no way entails that we should use a decision procedure of the sort Railton mistakenly thinks subjective consequentialists must be committed to. In fact, his own arguments make it clear that they should not be, for we can easily see in advance that decision procedures which attempt to be too conscious and rigorous about the use of information available to us will certainly result in our inefficient use of that same information. Howard-Snyder's definition leaves open the possibility that we can use other decision procedures to respond to the information available to us to maximize the chance that our actions will produce consequences which, in light of that information, are likely to be maximally good.

Given the above definitions, there are several strong arguments against objective consequentialism. Howard-Snyder argues in the rest of her article that it is unreasonable to condemn agents for failing to perform their objective consequentialist duty when this would require the use of information which is subjectively unavailable. She also suggests that if an agent were, by chance, to perform some action which satisfied her objective consequentialist duty, when the information available to her implies that her subjective consequentialist duty lay elsewhere, the action isn't really hers. She was merely among its causal antecedents, but had no power of choice to intentionally pursue and attain her objective consequentialist duty. It was not her action as an agent.

Eric Wiland strengthens Howard-Snyder's argument by showing that the split between our objective and subjective consequentialist duties is not merely an occasional feature of our lives, but is massive and ever-present.Footnote ³ There almost certainly exist, and will always exist, facts about the future, both about impending disasters where thousands of lives could be saved with ample warning, and instructions for producing beneficial scientific wonders like the cure for cancer, which if recited today and delivered to the right persons under the right circumstances would lead to tremendously good consequences. Hence it is our perpetually unfulfilled objective consequentialist duty to produce and deliver texts with such information. Naturally, we are quite incapable of doing so, and even a chance recitation of such instructions could hardly be sensibly attributed to us as agents, any more than literary genius could properly be attributed to the one lucky monkey on a typewriter who happened to produce some pages of Hamlet after vast numbers of failed key pecks.

Still, the division between objective and subjective consequentialist duties has seemed attractive to many people. In the rest of this article I will show that Railton's further arguments in support of the former fail.

RAILTON'S RESISTANCE

Railton shows how we can avoid the alienation of subjective consequentialist decision-making with the exemplary case of a non-alienated but reasonably consequentialist person, Juan. Juan acts out of love and spontaneous feeling most of the time, but when pressed to justify his motivations, Juan can do so on the grounds that if he didn't he would become alienated, and hence wouldn't be of much use to anyone. Juan is an example of what Railton calls a sophisticated consequentialist, who

Railton seems to think that Juan's goodness gives support for objective consequentialism, because he is doing the sorts of things objective consequentialism would have him do. But this is a hasty conclusion. I would suggest that what is appealing about Juan is not that he leads an objectively consequentialist life, but that he is committed to doing so, and has reasons (based on evidence available to him and a reasonable process of deliberation) for thinking that he is acting in a way to produce the best consequences he knows how to produce. In other words, he is a subjective consequentialist in Howard-Snyder's sense.

Railton resists the move from praising the sophisticated consequentialist to advocating subjective consequentialism by claiming that unlike subjective consequentialism, objective consequentialism can distinguish the truth conditions of a theory of its acceptance conditions under various circumstances, a distinction which is a virtue in other fields such as science.Footnote ⁵ Hence it might be right for Juan to accept a certain set of moral principles or attitudes, or to perform actions which the principles lead him to do, even if it is false to say that such principles or attitudes, or the resulting actions, are morally good.

However, the grounds for distinguishing truth and acceptance conditions in other areas of knowledge do not apply to moral theories. In reference to a scientific theory, for instance, ‘truth conditions’ refer to facts in the world which the theory describes, while ‘acceptance conditions’ refer to conditions, primarily our available evidence, which justify our performance of a specific action: believing the theory to be true (or, more pragmatically, acting as if it was true). Obviously, these two conditions can come apart, for many theories are true which we have insufficient grounds for believing in, and we may also at times have sufficiently good evidence to justify our acceptance of a theory which is not actually true.

But moral theory is about when, in general, we are justified in performing certain actions, including actions which constitute believing in or accepting the theory or its consequences, or acting as if any of these were true. There is no separate subject matter for it to be ‘about’. Therefore its truth conditions and acceptance conditions needn't, and perhaps cannot, come apart the way they do for other kinds of theories.Footnote ⁶ Indeed, separating these two leads to some other rather absurd claims, as when Railton says that

Since Railton earlier said that our objective consequentialist duty defines right actions, it follows that Juan should not do what is right. Given Wiland's argument, Railton would even be compelled to concede that almost none of us should ever do what is right. Yet it is hard to see this as anything other than a very confusing distortion of our language in the name of a theory, which should only be allowed if the theory is especially compelling. If moral claims have no subject matter apart from justification, then the theory which apparently requires this redefinition of moral terms is quite unmotivated.

Trying to separate the concepts of rightness and what we should (or ought to, or can justifiably) do is not merely confusing. It leads to the irrelevancy of one side of the dichotomy: for if what I should do and what it is right or my duty to do are two different and incompatible things, then I cannot be required to do both, again using the principle that ‘ought implies can’ and the fact that I cannot do contradictory things. If I ought to have a character of a certain sort, and this conflicts with doing my duty or what is right, then the concepts of ‘duty’ and ‘rightness’ are simply superfluous, confusing, and misleading. We should take no further interest in them, since they don't tell us anything about what we ought to do. This is equivalent to adopting subjective consequentialism, but with an unfortunate and pointless jettisoning of the previously useful concepts of duty and rightness.

It might be objected that talk of achieving the objectively best consequences out of those logically possible to attain, and describing the possible ways of getting there with the terms ‘duty’ and ‘right action’, is useful in describing ideals which Juan may in the long run try to live up to more closely. But this is only useful if it is practically possible for Juan, or for observers evaluating Juan, to see how they should act in turn, to use such ideals to get closer to them. This in turn is only true if such ideals can enter into our conscious deliberation, or at the very least if we can be reliably responsive to the evidence about what the objective consequences of various actions are. But subjective consequentialism already allows all this, and indeed requires it, where possible. For if the evidence currently available to Juan indicates that his current conclusions about his moral duty are unreliable or doubtful, and that there is a significant likelihood that he can improve this by gathering still more evidence or thinking more carefully, then subjective consequentialism often would oblige him to do so, when this would be the best subjectively available way for him to increase the expected utility of his actions. If on the other hand his current evidence indicates that the chance of being wrong about his current choices is low with respect to the cost of more thinking and evidence-gathering, then he should not do so. Hence the ideals of objective consequentialism are of no practical value to Juan or anyone else attempting to do their consequentialist duties, beyond what subjective consequentialism already allows. We should conclude that to the extent that objective consequentialism posits an ideal of action describing ‘duties’ which Juan cannot practically use to guide his action, he is not only not to be blamed for not using them or for not achieving the theoretically better consequences, but he is in no sense wrong in failing to do better: for he cannot do better. If he did miraculously produce such goods they would be done randomly, by luck, just like winning a lottery ticket. The mere logical possibility that he could do better should actually be described as a case of his body executing certain motor actions, not as a case of Juan, as an intentional moral agent, producing better consequences than he already is.

DISSECTING THE DEMON ARGUMENT

Railton suggests another reason for thinking that truth and acceptance conditions may diverge for a moral theory, especially a consequentialist theory. He asks us to

This seems to be a case where consequentialism is true (at least we shall suppose it is, among friends here), but where it should not be accepted. But the case is misleading, and actually entails no such separation between truth and acceptance in any of several possible sub-cases.

If there is no or very little gap between the time of the demon's threat and his execution of it, there is of course nothing we can do to avoid the horrors and we have no duty with respect to it. If there is a sufficient gap, then in this time, according to consequentialism, we should do what we can to avoid the future horrors. Now what would this amount to? We could perhaps simply shoot all the non-Kantians – including ourselves! But even so desperate a measure leaves open the possibility that a Kantian could later reflect upon morality and realize that consequentialism was the true theory after all (which, by hypothesis, it is, so the chance that this will happen is quite significant). Since this will again unleash the horrors, we must try to prevent this from happening, and Railton's tentative final suggestion is surely required here. We must brainwash everyone, not only to make them Kantians, but to erase from them any memory that we did this because of the demon. Furthermore, we must alter human consciousness (perhaps through sophisticated genetic manipulation) to remove the possibility of anyone's ever again becoming aware of or appropriately responsive to the evidence that any form of consequentialism is the right moral theory and Kantianism the wrong one.

Now are two possibilities: either we can do this successfully, or not. If we succeed, I submit that we have made the minds of the surviving humans impervious to reason in a very fundamental way, since they cannot be responsive to the appropriate evidence about when their actions are right or wrong, justified or not (note that I am not saying that current, non-brainwashed Kantians, are impervious to reason!). Whether this is really a world worth saving, I am not sure. Perhaps a world full of human-shaped animals mindlessly engaging in Kantian-approved behavior but incapable of doing otherwise is somewhat better than one of rational beings suffering excruciating punishment, knowing why yet unable to stop it. Incidentally, such beings would be acting, in Kant's words, in conformity with the moral law but not out of moral considerations or with moral worth. If the demon would not count such beings as ‘becoming Kantians’ then the brainwashing strategy is doomed to failure, and again it seems that there is nothing we can do to avoid the horrors (short of mass suicide), and hence we have no duty to try to do so. No divergence between truth and acceptance conditions for moral claims appears here.

Setting aside this issue, if we ask whether such beings should accept the true consequentialist theory, the question is meaningless at this point. We might as well ask whether rocks should be consequentialists, or if apes should accept the theory of chemical elements. These brainwashed ‘Kantians’ aren't rational beings and are not capable of accepting or rejecting consequentialism in any meaningful way. They are certainly not obliged to accept it, but this hardly justifies the claim that its truth and acceptance conditions differ, because presumably this would be the claim that the conditions under which a theory is true and under which a rational being should accept it are different, and this is not the case here. There are, in this scenario, no rational beings who should or should not accept the truth of consequentialism, even though it remains true, because there are no more rational beings at all, at least among the human race. Of course if there exist some other rational beings not under the demon's threat – including perhaps the demon itself – then there is no reason such beings shouldn't accept consequentialism as the right theory. Again, truth and acceptance conditions fail to diverge.

Now the other possibility is that we are not successful; despite our best efforts, some beings remain or are later born with enough rationality to potentially see that consequentialism is the correct moral theory. Should such beings accept consequentialism? Remember that all memory of the demon, and hence of the terrible things which will happen if they adopt consequentialism, have been erased. At that point I am prepared to say that persons faced with the evidence they have about morality should indeed become consequentialists (and it is right that they do so, their duty to do so, etc.), even though it would be better for them if they did not. The apparent paradox here is considerably reduced if we avoid the confusion of thinking that what an agent should do is also what wiser agents should advise the agent to do: I would hardly recommend to these later beings that they become consequentialists. Nevertheless I cannot fault them for doing this on their own, and indeed think they are doing the best they could be expected to do, and commend their choice of principles (notably, that of following the evidence to where it leads). The fact that some perverse agent intends to punish them for this choice is not a factor that can make their action wrong, and there is no sense (except that of bare logical possibility, which we earlier dismissed as insufficient) in which they can do any better. In this case truth and acceptance conditions remain the same. What they both diverge from are the conditions for an ideal outcome, but we hardly need assume a malicious agent for that to happen. This happens all the time due to the imperfect causal structure of the universe, as when I choose a mode of transportation statistically safer than the alternatives, intending to protect myself and my traveling companions, but as fate has it we become involved in a rare accident. In all such cases, our actions are not morally wrong, just very unlucky. To insist that we also use certain moral terms like ‘duty’ or ‘rightness’ to describe ideal outcomes which we can in no way steer ourselves towards invites confusion, because ‘duty’ and ‘right’ are traditionally associated with efforts to guide our action. But they cannot do this if strictly separated from the conditions under which we should accept (are justified in accepting) certain principles and/or perform certain actions.

SUBJECTIVE CONSEQUENTIALISM: A NEW HOPE

Much confusion has been sown, by both consequentialists and their critics, by the supposed view that, for consequentialists, only consequences matter. As a moral theory, this view is perfectly absurd, since it requires agents to do things which they cannot possibly do. It violates our well-entrenched intuition that agents who intend to do good, and act in ways that could be reasonably expected to do good, are morally in the right, regardless of what happens. But we have been bewitched by the word ‘consequentialism’ into believing that only the noun's root – consequences themselves – could make this theory distinct, as if the attempt to achieve good consequences wasn't a perfectly good form of consequentialism, and quite distinct in its own right from deontological or aretaic theories.

Some consequentialists like Railton have attempted to reintroduce intention as an ethical factor by drawing distinctions between when an agent is objectively ‘wrong’ and when he is subjectively ‘blameworthy’ (or doing as he ought not, etc.). But once we abandon the idea that consequentialist ‘right’ can only be about consequences sans intention, this distinction is more clearly seen to be unmotivated, as well as confusing. There is a perfectly sensible way in which consequentialists can acknowledge the value of intent, without turning into virtue ethicists or Kantians. That is by becoming subjective consequentialists, and evaluating agents by the degree to which they respond to their environment, including their finite reasoning powers and empirical evidence, in ways which, given that evidence, aim to maximize expected utility (or consequences more generally). If we develop subjective consequentialism further, we might be able to break down some of the barriers between consequentialism and other ethical views. Many philosophers who have rejected consequentialism because of the excessive attention historically given to its objective version may then be motivated to give it a renewed consideration.

Objective consequentialism might be seen as a form of ideal observer theory, which in general suggests that ethical rightness is based upon judgments which might be made by some idealized agent, rather than those made by actual imperfect agents. Such theories are well-meaning; no doubt the actual choices we make are often wrong, attesting to our imperfection. But there is a difference between saying that our actual choices are wrong, and saying that we utterly lack the resources to do the right thing, and hence must appeal to resources (including inaccessible facts, and calculative capacities which we utterly lack) in order to do the right thing. There remains the underexplored possibility that when we act wrongly, this is not because some ideal observer would judge so, but because we have ignored some facts or failed to take advantage of some decision-making power which our current, non-ideal selves possess. What an ideal observer (who is smarter, and knows more, etc. than I do) judges that I ought to do can neither motivate nor justify my actions unless I, in my imperfect state, have some good reasons for adopting the views of this or that ideal observer, or trying to become more like it, etc. But such reasons must be accessible to me as I am, if it is in any way reasonable to expect me to respond to them, or condemn me if I don't. In other words, we need a non-ideal theory of ethical obligation to explain why the judgments of the ideal observer apply to my non-ideal self, which means that ideal observer theory cannot be the ultimate foundation of morality, and may even be superfluous, for reasons parallel to my arguments at the end of my section on ‘Railton's Resistance’, above. Subjective consequentialism may be developed into such a theory.

It might be feared that making right and wrong functions of our subjective knowledge and decision-making capacities will subjectivize morality to an unacceptable degree. While there may be reason for concern, there may be a way to show that this fear is not warranted. Certain types of evidence may be universally available to all agents. No agent can plausibly deny the pervasive evidence that she is an agent, that she is finite, fallible, similar in many ways to other agents, etc. If certain general moral claims could be shown to follow from such facts, then these may be universal objective moral truths derivable from any rational agent's subjective evidence, although doubtless many particular conclusions about what should be done in specific instances will justifiably vary amongst agents according to their evidence sets. But further development of this and other aspects of subjectivism consequentialism go beyond this article's scope.