I take it that we should avoid moral daredevilism. Let’s understand a moral daredevil as someone who takes unnecessary moral risks. And let’s say that S takes an unnecessary moral risk in doing A iff. A does not maximize expected morality (I don’t say ‘expected value’). But how is the expected morality of an action determined? The expected morality of A is determined as follows, where M is any moral moral theory recommending or permittng A and VA is the moral value/disvalue of A.
P. EM(A) = [Pr(M) . V(A/M)] + [Pr(~M) . V(A/~M)]
I don’t intend (P) to be understood in consequentialist terms. The value of an action might be a function of the virtue it displays or a function of the strength of the deontic requirement it fulfills. The expected morality of action A then is the value of A given the truth of the moral theory that recommends A multiplied by the probability that the theory is true + the value of A given that the theory is false multiplied by the probability that the theory is false. I say, anyone who does not act in accordance with (P) is a moral daredevil.
Pr(~M) is just the disjunction Pr(M1) v Pr(M2) v … v Pr(Mn) for all alternatives to M on which A is not permissible. These are moral theories that are mutually incompatible and such that each has some chance of being true. Similarly, Pr(M) is just the disjunction of moral theories Pr(Mi) v Pr(Mii), v . . v. Pr(Mk) on which A is permissible.
How does (P) work? Here’s a rough idea. I seem to be moved by the following sort of argument. There are moral theories M on which fetuses are not persons and abortion is permissible. I don’t know of any moral theory on which abortion is something we want to maximize. But there are other moral theories on which the fetus is a person and abortion is murder, or something very close to that. It is one of the worst things you can do. It’s not rational—not that I can see—to take that moral risk. The fact (if it is one) that M is more probable than M’ does not in general make the action less than moral daredevilism. It is worth noting that this is not an argument for prohibiting abortion legally or otherwise. The argument is at best directed to those considering how to exercise their freedom. As I say, I seem to be moved by arguments along these lines. It is difficult to know finally what is morally permissible. But we do know when we’re in situations that involve huge moral risks. I can’t see the rationality in taking that sort of moral risk. Maybe someone can talk me down.
Hi Mike,
I don't think I'm the person to talk you down; but do you think this commits you to some sort of commensurability? I guess I worry a little about finding a single metric for all moral considerations--about equivocating on 'V' from one context to the next.
I find this a very plausible line of thought, and I think this is a very important thing to try to work out.
I worry about cases of very low probability moral theories. For instance, I am worried about cases where some action A is forbidden by a very low probability moral theory T1, and permitted but not required by all alternative moral theories.
Take the case of the Pythagoreans who allegedly thought that eating beans was immoral. Suppose there is some tiny non-zero probability that they're right. Suppose, further, that on no competing theory do I have any obligation to eat beans because there is an alternative that is equally nutritious and morally unproblematic. Then EM(eat beans) is slightly less than EM(eat alternative), and so I shouldn't eat beans.
This gets really bad if infinities come into play. Consider the gnostic theory that procreation imprisons a pure soul within a body. Maybe this is the ultimate offense against purity, an infinite offense. And maybe on no competing theory is it an infinite offense to fail to procreate. So, if the gnostic theory has probability greater than zero, EM(procreate) is minus infinity, and so one shouldn't procreate.
It would be nice if one could just dismiss crazy theories, maybe by having some lower probability threshold. But I don't see how to do that in a principled way.
It's tempting to say that it's wrong to do an action A if EM(A) is significantly less than EM(B), where B is some alternative. But, again, it's hard to get a clear story on the "significantly less".
Maybe one can overcome some of these problems by noting that among the theories, there are some with moderate probability on which prudential value implies some moral value (say, Aristotelian virtue ethics). If we allow that we can outweigh a moral wrong multiplied by a very small probability with a non-obligatory moral good multiplied by a moderate probability, we can get rid of some of the problems. Thus, if you like beans slightly more than the alternative, maybe virtue ethics implies that there is some moral value V in your eating beans rather than the alternative. And while on the Pythagorean view the moral disvalue -W of eating beans is very very bad, nonetheless the Pythagorean view is so improbable that V.P(virtue ethics) is much bigger than W.P(Pythagoreanism). That won't help with infinities, though.
Here is a different kind of problem. Suppose I apply this in every decision. Then I end up living a very crabbed life, a life where I am absorbed too much in avoiding what could possibly be a sin according to some moral view, and as a result I end up living a really distorted life where I end up neglecting important things that according to genuinely plausible theories are morally valuable. For instance, I get so worried about whether tofu falls under the Pythagorean bean prohibition, and whether fruitarianism has a positive or only infinitesimal probability, that I don't end up having any time for friendship, prayer and service to neighbor.
Maybe one could here make a move similar to that which some utilitarians make. These utilitarians observe that some people are going to lead lives productive of lesser utility if they remain act utilitarians. Thus, they ought to try to bring it about that they are not act utiltiarians. It could be that this theory could lead to a similar thing. Even though the morally optimal life would take into account all moral views of non-zero probability, it is morally better to acquire the vicious habit of neglecting all the moral views except the three (say) that have the greatest probability in one's moral calculations.
A related context that I've thought a bit about is what one should do if one is persuaded by Pascal's Wager kinds of arguments but one is worried about a multiplicity of non-zero probability religions. I think one ends up leading a life where in any case where an action is prohibited by one of these religions under pain of loss of eternal joy, and where no other religion requires that action on pain of loss of eternal joy, then one prudentially ought to refrain from the action. But, again, the worry is that this will result in a life that is too crabbed--the union of the prohibitions of the non-zero probability religions, even after one subtracts those actions that are required by some religions, will be so restrictive that the life it will leave us with will be one that none of the religions judges to be very good. Again, maybe the move to make is to deliberately neglect the low probability ones, on prudential grounds.
Mike,
I think I get at the type of argument you are working toward, I think?
I have a couple of questions/ comments.
1. When you talk of the probability that the theory is true What kind of probability do you have in mind? Epistemic? Objective?
2. Is "unnecessary risk" really the correct standard for moral daredevilism? If moral daredevilism is obviously wrong it seems it would be because it involves some kind of recklessness, Or some kind of action in defiance of a significant and unacceptable risk. If moral darevilism merely involves taking a risk that you did not have to take then why think that is wrong?
By analogy to darevilism in general. Suppose I have two comparable cars sitting in my driveway - The red one has side airbags, the green one does not. When I jump in the green car to drive up to the drug store to get a milkshake, I take an unnecessary risk that I will lose my life in a side impact crash. You would hardly say I was a 'daredevil' for doing so. Nor does it seem that I did something wrong for not choosing the green car. Even though there was really no reason for me to choose the green one.
And when I have a non-safety related reason for choosing the green one, then I certainly did nothing wrong.
In my real life, my wife and I own a very late model minivan and a 19 year old Miata. If my insurance rates are any indication, the minivan is much safer. However it is also more expensive, uses more gas, and is much less of a hoot to drive. And so I mostly drive the Miata.
I wonder if once you flesh out "moral daredevilism" in a robust enough way, whether it will still do the work that you want it to do.
3. Risking something huge is not the same as taking a huge risk. When you say But we do know when we’re in situations that involve huge moral risks. I can’t see the rationality in taking that sort of moral risk. I wonder what you have in mind. If you mean simply that you are risking the possibility that you action is very immoral, that does not seem irrational.
By analogy again... when I drive the Miata to the barber, I am risking death (and the resulting effects on my wife and young daughters). I am risking something huge, but not taking a huge risk.
4. When you say that There are moral theories M on which fetuses are not persons and abortion is permissible. I don’t know of any moral theory on which abortion is something we want to maximize. that seems correct if by 'maximize' you are referring to the number of abortions in society. However you do not seem to have public policy in mind, but rather individual moral agents contemplating having an abortion (correct me if I am wrong.) There seems to be no lack of moral theories M on which, for an individual s in some circumstance c, abortion is not only permissible, but recommended. (and even morally required for s in c on some moral theories) If this is right, it is not clear that the balance will tip the "right" way in your expected morality equation.
5. It might be that non-moral goods might still recommend an action even if that action does not maximize expected morality.
Just some late night thoughts about where I think the potential challenges to your line of thinking would lie. (Is that appropriate? I am new to this blogging world thing.)
Alex,
Lots of good points. I'm hoping that these are resolvable in some principled way. Recall that Pr(M) is not the probability of any single theory. It is the probability of all of those theories on which A is permissible. So, take the set of A-approving theories and how valuable A is relative to each. The partition of A-approving theories will yield some overall expected morality for A. Something like this,
P1 = Pr(M1).V(A/M1) + . . .+ Pr(Mn).V(A/Mn)
Compare that to the partition of ~A-approving theories and the value of ~A relative to each of those.
P2 = Pr(Mi).V(~A/Mi) + . . .+ Pr(Mk).V(~A/Mk)
If P1 greater than or equal to P2 then A is morally permissible.
We should find that peculiar recommendations (those recommended uniquely by some moral theory, or some very small number of them) such as the Pythagorean theory will get swamped by the sheer number of theories that permit bean eating. If the Pythagorean tries to make up for its peculiarity by placing an infinite disvalue on A, it will thereby make the recommending theory improbable in the extreme (one might reasonably put its probability at some infinitessimal).
The question about how to actually live this way is important. It is perhaps good to restrict (P) to cases in which there is some genuine disagreement over competing theories and one of them permits an action that is, on the other view, an extreme wrong.
. . .but do you think this commits you to some sort of commensurability? I guess I worry a little about finding a single metric for all moral considerations--about equivocating on 'V' from one context to the next.
This is a problem. I doubt there is any such metric as well, but I think intuitive comparisons are possible and useful. We might consider letting V be a moral preferability measure, if we can make sense of moral preferability.
Erik, you write,
When you talk of the probability that the theory is true What kind of probability do you have in mind? Epistemic? Objective?
Right, I think it would have to be epistemic. But I think something like the principal principle is right, so your epistemic probability for M should equal the objective probability for M, where that objective probability is available.
Is "unnecessary risk" really the correct standard for moral daredevilism? If moral daredevilism is obviously wrong it seems it would be because it involves some kind of recklessness. . .
I have in mind a risk that is not morally necessary. Obviously, there are risks that are morally necessary, and I have no worries about those. But if you are not so sure that the moral theories that recommend A are true, and you know that there are moral theories on which A is terribly wrong, then it might well not be worth the risk to do A.
If you mean simply that you are risking the possibility that you action is very immoral, that does not seem irrational.
I tried to explain what I mean by appeal to (P).
There seems to be no lack of moral theories M on which, for an individual s in some circumstance c, abortion is not only permissible, but recommended.
There are no non-controversial moral theories on either side of this issue. This is why it matters a lot to consider the possiblity that you've endorsed the wrong theory.
Mike:
If A is a contrastive action, eating beans rather than alternative X, then on the bean-eating-approving theories the value of A is zero, since eating beans is of no greater moral worth than eating X. So P1 is exactly zero. On the other hand, there is one theory on which eating beans is bad, and so P2 is greater than zero. So we shouldn't eat beans.
And I do think one needs to look at contrastive actions, or at actions contrastively, for these kinds of evaluations.
Related to Alex's worries about living a crabbed moral life, I wonder if the view might also saddle one with a crabbed life simpliciter. (I'm thinking of Susan Wolf on the sacrifices of moral sainthood.) Maybe you're thinking that any consideration that could make a life go better (aesthetic, whatever) will be captured in one of the moral theories that enters into the calculus. If not, though, we may be able to come up with cases where maximizing expected morality requires you to sacrifice some great non-moral good; and so where you could plausibly be thought to be better off (on the whole, in all realms of life, morally and non-morally), or in general justified, taking the moral risk.
If A is a contrastive action, eating beans rather than alternative X, then on the bean-eating-approving theories the value of A is zero, since eating beans is of no greater moral worth than eating X.
I'm not sure we why we should focus on the 'better than' relation among actions rather than the 'valuable' property. I don't see the motive to make such a technical move.
Luke,
I don't see this as obviously more burdensome than adhering to other moral views. Similar things have been directed to trying to be a utilitarian agent or trying to be a Kantian. I'm not persuaded. Of course there will be limitations on how well an agent can live up to theoretical demands, but this comes in at the level of agent evaluation. I'm talking only about action evaluation.
Mike,
This is really an interesting post. I suppose I'm sympathetic but have concerns similar to those raised by Alex. In addition to worries about beans, I had worries about contraceptives. These worries are more worrisome given arguments that I've not yet had time to read (this summer by the pool, I hope) that lend support to the view that we ought to be identified with the egg prior to the egg's fusion with a sperm cell (here).
Anyway, my reason for writing. You said, "I don’t know of any moral theory on which abortion is something we want to maximize." Well, my summer reading the summer before last was likely more interesting than yours! I'll loan you my copy of Better Never to Have Been: The Harm of Coming into Existence. From Amazon's product description:
Better Never to Have Been argues for a number of related, highly provocative, views: (1) Coming into existence is always a serious harm. (2) It is always wrong to have children. (3) It is wrong not to abort fetuses at the earlier stages of gestation. (4) It would be better if, as a result of there being no new people, humanity became extinct. These views may sound unbelievable--but anyone who reads Benatar will be obliged to take them seriously.
Better Never to Have Been argues for a number of related, highly provocative, views: (1) Coming into existence is always a serious harm. (2) It is always wrong to have children. (3) It is wrong not to abort fetuses at the earlier stages of gestation. (4) It would be better if, as a result of there being no new people, humanity became extinct.
I've heard of the book. I can save him some time. I have a reductio for any hypothesis H that entails (1)-(4), viz., H entails (1)-(4).
I have a reductio for any hypothesis H that entails (1)-(4), viz., H entails (1)-(4).
It's one of the better reductio proofs. The book is great for the jokes. Who's lucky enough never to have been born? Hardly one in a million.
It's one of the better reductio proofs. . . .
I'm almost not kidding. There are bottom line or (near) non-negotiable moral propositions, and (1)-(4)come pretty close to those any sane person would deny. But I've no doubt that it is a good read (and it does sound funny). I'll definitely read it.
I'd agree that the _conjunction_ of (1), (2), (3), and (4) is pretty bad. But too much Woody Allan and Philip Larkin has me on the fence about (2). Isn't there a Seanna Shiffrin paper defending (2)?