Consequentialisms

Perhaps the paper would look like less of a joke if it weren’t entitled “Prank.pdf” — just a thought.

More seriously, I feel like we would prefer to live in a world where parents love their children based on our actual, not rational, beliefs: We value it because it was important in our own lives.

On the other hand, though, there’s probably some strong “human nature” that tends towards the usual parent-child bond. Since we’re dealing with happiness, it seems like human nature might be able to get us out of just about any jam if we’re able to say “Because of how we’ve evolved, people are a lot happier when they act in this manner. Thus it is moral.”

Of course, I’m not an ethicist, and I think I can fairly assume any thoughts I’ve had have already been had, better, by professionals.

I agree that’s a tougher version of the doctor problem, but I don’t really see how it affects my theory – unless some moderately spectacular assumptions are made.

If we let the lottery process just be the process of waiting to see who dies first, and all five agreeing to donate spare organs to the other four if they are first to go, that doesn’t seem icky at all. (At least not to me – but I have been accused of being insensitive on these matters.)

Now that lottery might not work if (a) we expect two or more of them will go at once, or close enough to it, or (b) that waiting for natural death will ruin the organs. But without one of those two assumptions there’s a very natural lottery to run.

Just why we should be happy with nature’s lotteries but not human’s is a harder question. Perhaps as anno says we’re just used to some things and know they work.

I’ve been trying to puzzle out what the theory might be and what it is a theory of and I’m not seeing it. So maybe you can help me out.

Is it a theory about the _rightness_ of actions?

Assuming that it is ….

Does it apply to action types or tokens?

Is it a theory that tracks those actions that are right, that explains why they are the right actions, that identifies a right-making property??

Perhaps a crisp canonical formulation would help.

Yep, it’s meant to be about which actions are morally right.

It’s meant to apply to action tokens not types, and for now I’d just be happy if it were extensionally correct – i.e. if it tracks those actions that are right. So the claim is just that an action is right iff it leads to a better world than its alternatives, where a is better world than b iff we’d prefer being in a to being in b from behind the veil of ignorance.

(I’ve assumed there the objective version rather than the subjective version. If need be replace would lead to with ‘would be expected to lead to’. I’ve also assumed something like causal determinism – that there’s one world an action would lead to. Removing that requires complicating the formulation, but only in familiar ways. It also requires us to have preferences over lotteries from behind the veil of ignorance. I think that’s OK, but others might disagree.)

I certainly don’t think it’s an account of the right-making features of actions. It’s right to visit your sick friend in hospital because she’s your friend, and she’s sick, and she could use the company. Those reasons, I think, set out the right-making features of the action. It’s possible that the consequentialist claim could be read as an explanation for why those things are right-making features. I’m mostly interested in the weaker claim that the consequentialist theory tracks the right actions.

There is. I think, a problem with the general statement of consequentialism and the articulation of the counter-example. The problem has to do with levels of description. The notion of a world populated by humans is one level of description, and on this level, one doesn’t want organs harvested from one human to benefit another human. But in your example, you can go down, without semantic loss or gain, from that level to the level in which the harvester is a doctor and the human is a poor person. I’d question that logical continuity, and the premise that these descriptions are unproblematically held within your general description of “humans.” I think that filling out of humans by their relationships one to another requires a thicker description of those relations before you can move from ethical judgements that are valid on level one to ethical judgments that are valid on the micro-level, level two. In fact, I would think that the mismatch between descriptions of worlds in terms of complete generality — Kant’s world, for instance — and worlds that are captured only by a thicker description of human relationships is the whole impetus behind some version of consequentialism. In a sense, this transposes the sorites paradox into ethical terms. Just as there is no formula for getting heaps out of sand, there is no formula for getting doctors and poor people out of humans.

The drowning child example is a good case for the transformations that thicker descriptions can bring about. From one end, the rescue of 99 children is equivalent to the rescue of 99 children. But introduce a variable into the mix: what if — looking into the minds of the participants — there was an intention, on the part of his parents, to murder him? The intention, however, is not to murder him by drowning, but murder him in some other way. Accidentally, his drowning, in spite of the maximum effort put in by all parents to save the children, fullfilled that parental will to murder. One would want to say that the result, then, could be judged to be wrong vis a vis the intent of the murderous parents. On the other hand, the same result could have happened without any bad intentions in the minds of the rescuing parents. Now, a consequentialist, trying to get out of the tar baby posed by motive, has a problem here. This problem is invisible from the macro level, but very visible from the micro level.

I guess I’m sceptical. If the consequentialist test that you advance here merely tracks the right answers then we don’t have anything like a consequentialist theory in the traditional sense. And it looks like you’ve just found a way of incorporating some deontological constraints within a system that looks consequentialist.

I’m also puzzled about the connection (if any) to ordinary ethical judgements and decisions. So take the decision to visit your friend in hospital. You rightly identify the reason-giving (and right-making)features of the situation. But it looks as if you should want to say that there are circumstances where that decision should be overridden. No doubt there are. But I doubt that the best way of capturing those circumstances is to say that you should do something else instead just in case doing that other thing leads to a better world in the sense specified…

But then perhaps you want to say that from behind the VoI we’d judge that we don’t want commitments to be too easily overridden by the application of the test itself. But when the test gets self-referential in that way I find myself staring into a hole with n-different meta-levels, regresses, no proper control on what is and isn’t allowed to count &c).

Backword Dave hits on something that bothers me about the drowning-children example: How do you calculate consequences of your action when many people are acting? Do you simply hold everyone else’s actual action fixed, or calculate what they would have done if you had done it–in which case the policy with the best consequences is to save the child that no one else is actually saving?

The problem with this view comes when many people do things that collectively overdetermine something very bad, but individually have small positive consequences past the threshold for the bad thing. If you calculate the consequences of an individual agent’s actions, holding the rest fixed, the consequences are positive–the bad thing would have happened anyway. But the agents collectively did something bad.

(A canned example is: Two people are pointing guns at a third. Each simultaneously resolves a minor itch by squeezing their trigger finger. If A had not squeezed the finger, the target would have died anyway, so A’s action has the net consequence of relieving the itch, and so is right.–That can’t be right.)

You could say–we wouldn’t choose to live in a world where so many people sully themselves by collaborating in a collective bad action. But that seems like smuggling in deontology, as I think Chris B is saying.

This is much treated in the literature I’m sure, so I’d be happy if more clued-in people spoke up.

Robin, I am not averse to consequentialism, but I do think that it too hastily decouples motive from act in order to reach an admirable goal: embedding ethics in practice. To do this, I think you have to move from talk about worlds and humans to talk about situations and human relationships. As I said, I think that the assumption that you can move logically from one level of description reproduces all the problems of a more Kantian moral theory — and this is a problem, insofar as consequentialism is supposed to be an alternative to Kantian moral theory.

As an example of what I mean: when you say “As the film Minority Report I think shows, we should not be condemned for “evil thoughts” that we might have – no matter how horrible – before we have had the chance to enact them…,” the scarequotes don’t rescue “evil” from its moral value. How, actually, can these thoughts be evil if evil is defined solely on actions? Of course, you can speak of thought acts that are like speech acts, but at that point you have dissolved the distinction you are trying to preserve. That you, as an Other, embodied in some social institution, shouldn’t punish evil thoughts tells us only, or at least tells me only, that punishment and reward are not the necessary coefficients of moral judgement. If I were to pinpoint the one problem I have with both Kanian and consequentialist moral theory, it is the move that identifies judgments of good and bad with punishment and reward.

As to mixing in the motive of the parents, this comes from positing that the parents are rescuing children regardless of their personal relationship to the rescued child. For 99 parents, this comes from a consequentialist p.o.v.; for one, it comes from a small homicidal wager: that their child won’t get rescued.

Oh, and here’s one further problem for consequentialism — it does make it rather difficult to pick out moral agents. After all, it is only because human beings have something like consciousness and they act that we distinguish their killings and charity from the acts of, say, a guppy. On the level of pure action, however, a guppy killing its thousand children and a homo sapiens killing its one aren’t morally distinguishable.

Robin, I’m sympathetic to the motives for consequentialism — which I take to be the effort to embed ethical judgments in the world of practice. But I don’t think consequentialism quite gets there.

1. Levels of description. The problem with starting out with on the level of the most general description and then descending to the next level, in which human relationships are fleshed out, is that it isn’t logically clear how you make that move. In other words, the hierarchy Brian presents is linear — you merely add things to worlds and to humans and you get doctors and parents. But I think that it is more likely that the moral world is non-linear — that there isn’t a “human being” to which you add the “doctor” formula and you get a doctor. This actually vitiates the whole point, which is to understand practice — how doctors are made, etc. etc.

2. When you write: “As the film Minority Report I think shows, we should not be condemned for “evil thoughts” that we might have – no matter how horrible – before we have had the chance to enact them,” the scarequotes around evil don’t rescue it from its moral status. But of course, if evil — or any moral judgment — only attaches to acts, then how can there be such a thing as evil thoughts. It would be like thoughts that are colored blue. I think that there are bad thoughts — and that the thinker reacts to them as bad thoughts. This is the whole phenomenon of conscience. A moral theory that has no place for, or explanation of, conscience is not without problems.

3. Perhaps by condemnation you meant merely that we describe something as evil — but there is a larger sense in which this is the flaw in consequentialist reasoning. I think we have to disjoin punishment or reward from good or bad. The consequentialist seems to think that the essence of bad is punishable. This, I think, reproduces the worst habit of utilitarianism. Ultimately, we know that punishment and reward stem from a different class of motives — motives circulated within various social institutions — then the motives that prompt us to describe an act or a thought as good or bad. It has long been a philosophical platitude that a good act is not good because it makes us feel good — but philosophers have a much harder time decoupling bad acts from punishment.

4. I presented a confusing interpretation of the parent example, you are right. Here’s what I meant. If the one hundred parents decide to rescue children regardless of relationship, the homicidal parents could then rescue a child — something good — only to cover up their wager — admittedly a long shot — that their own child wouldn’t be rescued. Now, would we say this was bad?

5. Speaking of which — if we do exclude all mention of motive or thought from our moral talk, it is puzzling how we pick out moral actors. Why distinguish humans from guppies, morally? It seems that you need some way of packing in certain distinguishing things about humans to begin with. And surely that can’t be derived from their acts — it has to be derived, somehow, from what they think of acts, and that they think of acts.

Matt – You bring up a very important point about collective action which I don’t have a good answer to. I remember reading Parfit’s “Reasons and Persons”, which is supposed to be a very good book on consequentialism and personhood, and I remember putting the book down because he completely fudged and/or ignored the vital question that you have raised. It was a good book up to that point, but I just couldn’t continue after I’d thought through the implications of that hole in the theory. It seemed to me then, and still seems, insoluble.

Roger’s parent example – in its clarified form – seems to me to be a fascinating, more convuluted example of the kind of collective action problem that consequentialism does badly at.

Roger – I don’t see what follows from taking into account a person’s personal history, or whatever, which isn’t necessarily relevant. Where does this lead?

As for bad thoughts, let me modify my absolutist and simplistic “anti-thought-crime” stance, which as you rightly pointed out is untenable. Yes, a thought _can_ be bad, but only because of what it predisposes us to *do* , if anything – and it can’t ever be as bad as the actual act. If it is only a “joke thought” that does not have any serious intention to do anything in reality attached to it, then it’s less harmful still.

For me, moral actors are distinguished by a very simple pragmatic criterion – their ability to comprehend moral judgements. It is pointless to upbraid a lion for “inhumanely” killing its prey.

As an aside, the “ability to comprehend” has to be defined quite loosely if you want to include full-blown psychopaths as moral agents – because they appear to have no functioning conscience, but we still want to be able to talk about them doing bad things.