A number of philosophers (e.g., Brown 2004, Dreier 1993, and Louise forthcoming) have hypothesized that most, if not all, non-consequentialist theories can be “consequentialized.” More precisely, the conjecture is this (paraphrasing Brown 2004): For any moral theory M, there is some conceivable theory of the good that, when combined with the consequentialist principle “φ-ing is morally permissible iff φ-ing would produce the best available state of affairs,” yields moral verdicts that are, in every instance, identical to those of M. For example, suppose that M includes an agent-centered constraint against the commission of murder, such that agents are prohibited from committing murder even for the sake of minimizing the number of murders committed overall. To accommodate such a constraint, the consequentialist need only adopt a theory of the good that holds (1) the value of a state of affairs in which a murder has been committed is agent-relative such that its value varies according to whether or not the evaluator is the murderer, and (2) the disvalue in an agent committing murder herself is, from her position (that of the agent), greater than the disvalue in numerous others committing comparable murders. What’s more, if M prohibits an agent from committing murder for the sake of minimizing the number of murders she herself commits, then the consequentialist need only adopt a theory of the good that holds (3) the value of a state of affairs in which a murder has been committed is temporally-relative such that its value varies according to whether or not the murder in question would take place in the present or the future, and (4) the disvalue in an agent committing murder now is, from her present position, greater than the disvalue in her committing numerous other comparable murders in the future. Thus, by incorporating certain agent-relative and temporally-relative values in its theory of the good, the consequentialist can, it would seem, yield moral verdicts identical to those of any other moral theory.
Following Brown 2004, I’ll call the above-mentioned conjecture “Dreier’s Conjecture.” From Dreier’s Conjecture, some philosophers have reached some fairly bold conclusions. Jennie Louise concludes both that “all moral theories are consequentialist” (forthcoming, p. 2) and that “since we are now all under the consequentialist umbrella, the question now becomes, not whether we should be consequentialist or not, but whether we should be Value-Neutralists or Value-Relativists” (forthcoming, p. 38). Philip Pettit concludes that consequentialism will retain a distinctive profile of its own only if it is defined solely in terms of neutral value (1997, pp. 130-1). And Campbell Brown concludes that “consequentialism is empty” (2004, p. 35). (To be fair, I should note that Brown doesn’t think that this conclusion follows from Dreier’s Conjecture alone—see his 2004.)
In a series of three entries (including this one), I will contest Dreier’s Conjecture and the conclusions that Louise, Pettit, and Brown have drawn from it. In Part I (which appears below), I’ll argue that Dreier’s Conjecture is false. In Part II (which will be posted tomorrow), I’ll argue that analogues of Dreier’s Conjecture are true of most moral theories, including contractualism, virtue ethics, and divine command theory—the implication being that if Dreier’s Conjecture establishes that we’re all consequentialists, then these analogues establish that we’re all contractualists, virtue ethicists, and divine command theorists as well, and this is just absurd. In Part III (which will be posted on Thursday), I’ll argue that Louise, Pettit, and Brown presuppose a mistaken conception of what distinguishes one moral theory from another. I’ll argue that even if a consequentialist theory and a nonconsequentialist theory are extensionally equivalent, the two are still importantly distinct.
Part I:
In this section, I argue that Dreier’s Conjecture is false. I argue there is no theory of the good that, when combined with the consequentialist principle “φ-ing is morally permissible iff φ-ing would produce the best available state of affairs,” yields moral verdicts identical to those of a moral theory that includes agent-centered options—options to either safeguard one’s own interests or sacrifice those interests for the sake of doing more to promote the impersonal good.
Now consequentialism can avoid being overly demanding if the value in an agent promoting her own interests is agent-relative (e.g., proportionately greater for her than it is for others). In fact, consequentialism won’t be very demanding so long as there are a number of instances in which the agent-relative value in an agent promoting her own interests outweighs the agent-neutral value in having the impersonal good maximized. But this only takes consequentialism so far. It lets consequentialism hold, as many other moral theories do, that it is permissible for agents to promote their own interests even where they could instead do more to promote the impersonal good. Many theories, however, hold that not only is it permissible to safeguard one’s own interests, but also that it is permissible to sacrifice those interest for the sake of doing more to promote the impersonal good, and this is something that consequentialism cannot accommodate simply by incorporating certain relative values. To illustrate, consider the following case—I call it the Weekend. I can either spend this weekend working on this entry or volunteering for Oxfam (and assume, for the sake of simplicity, that these are my only choices). Here, we have the intuition that I am permitted to work on this entry even if I could do more to promote the overall good by volunteering for Oxfam. Agent-relative consequentialism can accommodate this intuition so long as it is true to say that the state of affairs where I spend this weekend working on this paper is, from my perspective, better than the state of affairs where I spend this weekend volunteering for Oxfam. The problem, though, is that, in this case, agent-relative consequentialism now seems to imply that I am morally required to spend the weekend working on this paper, for doing so would bring about what is, from my perspective, the best available state of affairs. Yet, intuitively, we believe that I am permitted to sacrifice the interest I have in working on this entry in order to do something that would better promote the overall good.
So it seems that although consequentialism can avoid implying that I am morally required to volunteer for Oxfam, it can do so only at the cost of implying something just as implausible, namely, that I am prohibited from volunteering for Oxfam. Thus even if agent-relative consequentialism isn’t overly demanding, it seems that it will still be counter-intuitive, for it cannot accommodate our intuition that agents have a moral option in more than just those few rare instances in which two or more acts are equal in their propensity to bring about what’s best.
We’ve seen, then, that consequentialism cannot accommodate agent-centered options simply by incorporating certain relative values in its theory of the good. It follows, then, that Dreier’s Conjecture is false: non-consequentialist theories that incorporate agent-centered options cannot be consequentialized simply by adjusting the consequentialist’s theory of the good. For the record, let me say that I believe that consequentialism can incorporate agent-centered options, but in order to do so consequentialism must do more than just tinker with its theory of the good. Those interested in how consequentialism might do so should read my “Position-Relative Consequentialism, Agent-Centered Options, and Supererogation.”
References:
Brown, Campbell (2004). “Consequentialise This.” Working manuscript—draft of June
1, 2004. Research School of Social Sciences, Australian National University.
Dreier, James (1993). “Structures of Normative Theories,” The Monist 76: 22-40.
Louise, Jennie (forthcoming). “Relativity of Value and the Consequentialist Umbrella,” The Philosophical Quarterly.
Pettit, Philip (1997). “The Consequentialist Perspective.” In Marcia Baron, Philip Pettit, and Michael Slote, (eds.), Three Methods of Ethics, Blackwell, pp. 92-174.
Portmore, Douglas W. (2003). “Position-Relative Consequentialism, Agent-Centered Options, and Supererogation,” Ethics 113: 303-332.
I don’t follow your argument regarding the example of the Weekend. You write:
“Agent-relative consequentialism can accommodate [the intuition that my working on this paper is permissible] so long as it is true to say that the state of affairs where I spend this weekend working on this paper is, from my perspective, *better* than the state of affairs where I spend this weekend volunteering for Oxfam.” (emphasis mine)
But why must the former state of affairs be judged *better* from your persective? In order to accommodate the intuition, why would it not be sufficient to judge that the two states of affairs are equally good from your perspective? It seems that, if the agent-relative consequentialist makes the latter judgement, she can say that *either* course of action is morally permissible.
Doug,
I’d better come defend the conjecture that bears my name. Oh, Campbell has just spoiled my chance to be the first commenter. Oh well.
It follows, then, that Dreier’s Conjecture is false: non-consequentialist theories that incorporate agent-centered options cannot be consequentialized simply by adjusting the consequentialist’s theory of the good.
1. As Campbell says, you don’t explain what’s wrong with the theory that ranks the two options in Weekend (namely, volunteering for Oxfam and working on the paper) together. I don’t think that’s a good theory, myself, but explaining what is wrong with it suggests another theory that seems (to me) more plausible.
2. The difficulty, if it should prove to be a real one, raised by options does not seem to have anything to do with agent-centeredness. Imagine an aestheticist view according to which we are permitted to choose the option with less total welfare in its consequences as long as the chosen option contains much more aesthetic value than its alternatives. This theory is agent neutral, but if options are unconsequentialisable, than the aestheticist theory is.
Sorry, the sentence beginning with “It follows” was supposed to be in italics. I guess the comments section here doesn’t support html markers?
Dear Campbell,
Thanks for your comment. You’re right: In the case that I call the Weekend, the agent-relative consequentialist could claim that the agent-relative value in my spending the weekend working on my blog entry exactly equals the agent-neutral value in my spending the weekend volunteering for Oxfam. So, if this claim is plausible (and I don’t think that it is), the agent-relative consequentialist could account for a moral option in this case. But the real issue is whether the consequentialist can account for a wide range of moral options, because many moral theories (e.g., many deontological theories) allow for a wide range of moral options, even where those options are not all on a par in terms of the value they would produce.
To illustrate, consider another case: Weekend II. In this case, I have four alternative actions available to me: (1) work on my blog entry, (2) donate blood, (3) volunteer for Oxfam, and (4) volunteer for NPR. Assume that in terms of agent-neutral value the ranking from best to worse is (3), (4), (2), and (1). Furthermore, assume that in terms of agent-relative value the ranking from best to worse is (1), (4), (3), and (2)—let’s pretend that I have a dreadful fear of needles. Many deontological theories would hold that all of (1)-(4) are morally permissible. But how is the consequentialist to accommodate these moral verdicts by adjusting her theory of the good? She can’t say that (1)-(4) are all equally good. Actions (2) and (4) are sub-optimal whether we’re judging in terms of agent-relative value or agent-neutral value or both. So how can we consequentialize moral theories that hold that (2) and (4) are morally permissible? I don’t see a way of doing it by means of tinkering with the consequentialist’s theory of the good. Thus I still maintain that Dreier’s conjecture is false in that we can’t consequentialize moral theories that allow for a wide range of moral options by adjusting the consequentialist’s theory of the good.
Dear Jamie,
I’m not sure I understand what you’re driving at. I agree that “the difficulty, if it should prove to be a real one, raised by options does not seem to have anything to do with agent-centeredness.” (I brought it up only because many moral theories include agent-centered options.) The problem for consequentialism in accommodating options lies with its commitment to maximization. Maximzing consequentialism will allow for options only in a very narrow range of cases, cases where two or more available acts are equally optimal in their value production. Now I should note that one might resort to satisficing consequentialism. According to satisficing consequentialism, agents are not required to bring about the best available state of affairs, but only one that is “good enough.” Thus, on satisficing consequentialism, there will be a moral option in every situation in which there is more than one way to bring about a sufficiently good state of affairs. But to resort to satisficing consequentialism is resort to consequentializing by some other means than by tinkering with the theory of good—hence, it’s not consequentializing as it’s typically understood. Regardless, it seems that satisficing consequentialism is unable to account for many typical options. For instance, we believe that I am permitted to spend this weekend working on this post even though it will contribute next to nothing to the impersonal good. However, as Slote admits, satisficing consequentialism must hold that the pursuit of such a personal project (i.e., one that does little to promote the impersonal good) is forbidden, that is, unless the standard for being “good enough” is held to be ridiculously low.
If by consequentialism one means the consideration of reasons and consequences that will produce the best act, rule, principle, or goal, then it’s not hard to see how Drier’s conjecture would follow. But the “thin” definition I provided is barely distinguishable from the giving of reasons for justification, or rationality in general, so it isn’t at all surprising that all other moral theories could fall under that umbrella. A stronger set of principles, axioms, or decision procedures won’t likely be as fungible.
Now I don’t see how satisficing consequentialism isn’t consequentialism because it relies on constraints outside of a general rule dictating the choice of the best outcome associated with the good_1_. If that follows, then Hare’s two level theory of utilitarianism isn’t utilitarianism._2_ The archangel and the prole is a story not only about the limits of knowledge, but human weakness as well. Limits on the will implies limits on action, implies limits in outcomes, implies limits on consequentialism (contra Hare, who seems to suggest it’s mostly lack of knowledge; weakness of the will is a problem for his account). But pessimism about the human capacity for altruism doesn’t eliminate striving for the ideal of the greatest good for the greatest number, or any other moral teaching_3_. It’s possible that versions of consequentialism taking into consideration the limits of human capacities will recommend different courses of action that are more practical and lead to better outcomes_4_, so it’s not simply a case of constraints imposed from outside “morality” or moral theory.
Am I completely misunderstanding the discussion?
_1_ the impartiality axiom in Slote’s version is probably to blame
_2_ http://www-phil.tamu.edu/~gary/bioethics/ethicaltheory/hare.html
_3_ an ideal theory expressed through dilemmas such as the oxfam case is compelling because of our capacity for compassion even if there is a gap between reasons and actions
_4_ readers will groan if I mention the recent work of Steven Pinker, but I can’t think of anything comparable written by philosophers at the moment
Doug,
Maybe there is something,some premise or condition, left implicit in your argument. You are supposed to be arguing against the claim you call “Dreier’s Conjecture”. So, you are supposed to be showing that there is *NO* consequentialist theory that delivers the verdicts you want in the Weekend examples. But it’s just obvious that there *is* a consequentialist theory that delivers those verdicts, namely, one that ranks together all of the alternatives you declare to be permissible.
You assert:
“She can’t say that (1)-(4) are all equally good. Actions (2) and (4) are sub-optimal whether we’re judging in terms of agent-relative value or agent-neutral value or both.”
But whether they are suboptimal is obviously relative to a theory of the good. No doubt you can think of a theory of the good according to which (2) and (4) are worse than (1) and (3). But we can all think of a different theory of the good according to which all four options are ranked together.
Presumably I have missed some crucial point. What is it?
(For the record: I agree that satisficing is very puzzling for a ‘True Consequentialist’. I am inclined to think that satisficing is *per se* incoherent, so the fact that it doesn’t cohere with True Con. doesn’t seem to me to place Dreier’s Conjecture in serious jeopardy. But for the moment, I am just worried about the logic of your argument.)
Dear Jamie,
I was relying on an implicit assumption, namely, that the consequentialist couldn’t appeal to an utterly implausible theory of the good. But, as I now admit, I’m not entitled to such an assumption, for “Dreier’s Conjecture,” as I’ve formulated it, says that “For any M, there is some theory of the good…,” not “For any M, these is some remotely plausible theory of the good….” But before I cry “uncle,” let me make take another stab at producing a counterexample to Dreier’s Conjecture.
So suppose, as you suggest, that the consequentialist adopts a theory of the good where acts (1)-(4) are ranked such that they all tie for first place in terms of their value production. But now suppose that M is a deontological theory that holds not only that (1)-(4) are all morally permissible, but also that (2), for instance, is supererogatory. So, according to M, (1) and (2) have different deontic statuses. Act (1) is merely morally permissible, whereas act (2) is supererogatory. How can the consequentialist account for these differing moral verdicts given that it will have to rank the states of affairs produced by acts (1)-(4) together in order to get the result that they are all morally permissible?
So it seems that I was wrong to focus on options alone. The problem is, if there is a problem, that non-consequentialist theories that include both a wide range of options and supererogatory acts can’t be consequentialized by merely adjusting the consequentialist’s theory of the good.
Dear Shai,
I wasn’t claiming that satisficing consequentialism isn’t consequentialism. But if the consequentialist has to appeal to a satisficing function, then Dreier Conjecture is, as I’ve formulated it, false.
I think that Jamie, Campbell, and myself all agree that there is, for any M, a consequentialist theory that is extensionally equivalent to M. The point of contention is whether the consequentialist has to do more than just adjust her theory of the good.
I see. But that looks like a cheap victory, to me. The reason you were able to find a counterexample is that you defined consequentialism so as to allow it only two possible verdicts, and then you produced a theory with three possible verdicts, and the conjecture as you stated it is that some consequentialist theory will yield the same verdicts as any given theory.
You didn’t merely paraphrase (Brown 2004). I will now paraphrase (Brown 2004).
Conjecture:
For any moral theory M, there is some conceivable theory of the good, G, such that an action is right according to M iff it is best according to G.
This Conjecture stands unrefuted (by your examples, anyway).
I do think that supererogation looks strange from the perspective of consequentialism; I also think supererogation looks strange, simpliciter. So, the fact that it looks strange from the perspective of consequentialism does not mean that the consequentialist perspective is parochial. It may still be utterly universal.
Wow, thanks Jamie. You’re going to really keep me on my toes, I see. I hope that I’m up to the task after getting creamed at golf today by my fellow bloggers, Josh and Dave. I do have a number of things to say in response to your comment.
First, I defined consequentialism as I did, because that’s the way Brown defines it. Brown defines consequentialism as follows: “an action is right if and only if its outcome is at least as good as that of every alternative action, and an action is wrong if and only if it’s not right” (2004, p. 4).
Second, I took myself to be applying the principle of charity in paraphrasing (I should have said “revising”) Brown’s formulation of Dreier’s Conjecture as I did. As Brown formulates it, Dreier’s Conjecture shows only that, for any moral theory M, there is some theory of the good that, when combined with consequentialism, yields the same moral verdicts as M but *only* in regards to moral permissibility. As Brown, puts it: for any plausible moral theory M, “there must be some theory of the good G such that, when combined with consequentialism, G yields the that distribution of right and wrong actions which is implied by M” (2004, p. 9). (N.B., by ‘right’, Brown means ‘morally permissible’, not ‘morally required’—see p.31 where two alternative acts are both right.) But it seems to me that in order to show that the C/NC distinction is empty, one would have to show that, for any (plausible?) moral theory M, there is some theory of the good that, when combined with consequentialism, yields the same moral verdicts as M *across the board*.
Here’s why. Imagine two moral theories, M1 and M2, and suppose that M1 and M2 agree both that A1 and A2 are morally permissible and that A3 is impermissible. But suppose that M1 holds that A1 is not merely morally permissible but is in fact morally required, whereas M2 holds that A1 is merely morally permissible and not morally required. Isn’t there, in this case, an important distinction between M1 and M2? The fact that M1 and M2 disagree about which acts are morally required seems to constitute an important distinction between the two. Hence the distinction between them can’t be an empty one. So I revised Brown’s version of Dreier’s Conjecture so that it says that, for every M, there is a version of C that’s extensionally equivalent to M in all its moral verdicts as opposed to in only those moral verdicts that concern permissibility. I take it that Brown needs the stronger version of Dreier’s Conjecture if he’s going to establish that the C/NC distinction is an empty one.
Now if M1 and M2 also disagree about whether, say, A2 is supererogatory or not, that too seems to be an important difference. And note that it’s not just that supererogatory acts look strange from the perspective of maximizing consequentialism, it’s that there can be no supererogatory acts on maximizing consequentialism. So moral theories that include supererogatory acts (e.g., Kantianism) cannot be consequentialize by merely tinkering with the theory of the good.
Doug,
Maybe Campbell’s formulation (of the Conjecture) leaves something to be desired, but your example (with A1, A2, A3, M1, M2) makes no sense to me.
Are A1 and A2 supposed to be *alternatives*, or are they supposed to be compatible (so that the agent might do both)? I assume they are alternatives (so that each excludes the other).
You say that the moral theory M1 *requires* A1 and also *permits* A2. How can this be? How can the theory permit me to do something that is incompatible with something it requires me to do? Is this supposed to be a moral dilemma? Or what’s going on?
Yeah, I see your cause for confusion. This is what I had in mind, though: A1 is some token act of caring for one’s child, A2 is some token act of taking care of one’s heath, and A3 is some token act of murder. Two moral theories that agree that A1 is morally permissible in the given circumstances might disagree on whether or not A1 is morally obligatory. Thus they give the same moral verdict as to whether A1 is morally permissible or not, but different moral verdicts as to whether A1 is morally required or not. I hope this makes sense. In any case, that’s what I had in mind.
Doug,
It makes sense, but I still don’t see how the example is getting any real traction.
Look, two theories that agree on what is permitted will, of necessity, agree on what is required. Permission and requirement are duals, and agreement and disagreement comprise negation. Ok, that’s too cryptic. Here’s what I mean.
Let ‘P’ be the permission operator, ‘R’ the requirement operator, and let them operate on whatever you think they operate on as long as the things are also negatable. (I like propositions, but you might like states of affairs or actions or properties. I take it all of these bear some kind of negation.) I’ll use ‘s’ as a schematic letter for these things.
Then ‘~Ps’ is equivalent to ‘R~s’. We can eliminate every ‘R’ in a theory in favor of ‘~P~’. So if two theories agree on what’s permitted, they must agree on what is required.
It’s another question altogether what is supererogatory.
You’re absolutely right. Thanks.