Back in March, Tyler Cowen encouraged bloggers to list the books that most influenced them. I didn’t think I could come up with a list that was a) plausible and b) would make me look cool, so I did not participate. But I enjoyed reading other people’s lists, and I noted, along with Steve Landsburg, that Derek Parfit’s Reasons and Persons seemed to be the most frequently cited book. Vowing that I will be prepared the next time the influential books meme comes around, I got a copy, which I finished this evening.
I found the book interesting throughout, but Part III on personal identity really stood out. I found it exhilarating and metaphysically challenging. I have spent the last couple weeks badgering my friends incessantly about corpus callosotomy (split-brain surgery), which results in two separate streams of consciousness.
The parts of the book that challenge classical utilitarianism are also interesting (though perhaps a little repetitive at times). The standard counterexample to utilitarianism is Nozick’s utility monster. This is a creature that, in economic jargon, has increasing marginal utility of wealth. If at least one utility monster exists, then the optimal utilitarian distribution of wealth is for one utility monster to have all the wealth. This is supposed to be abhorrant, but the standard reply, of course, is that no such creature exists, or is even imaginable. Utilitarians can bite this bullet with no real-world consequences.
Parfit tries to create a utility monster that is imaginable. His utility monster takes the form of the Repugnant Conclusion:
For any possible population of at least ten billion people, all with a very high quality of life, there must be some much larger imaginable population whose existence, if other things are equal, would be better, even though its members have lives that are barely worth living.
In classical utilitarian terms, a gazillion Muzak-and-potatoes lives are better than ten billion high quality lives. Parfit spends most of Part IV trying to avoid the Repugnant Conclusion. I think Cowen resolves the issue with his Ideal Participant construct, and in any case, a gazillion lives is scarcely more imaginable than the utility monster.
[Caution: untrained philosopher. Proceed at your own risk.]
Can we come up with a better utility monster? I propose, as an imaginable, realistic challenge to utilitarianism: abortion. I think the utilitarian position on abortion has implications that no one—whether pro-life, pro-choice, or ambivalent—would accept.
Let’s start with some assumptions. First, assume that the unwanted fetus will grow into a person who has a life well worth living. This life will be marginally worse if it is born to a 14-year old mother than if it is born to an older, economically-established mother. Whether or not the fetus is aborted, the mother will also have a life worth living. At least in some cases, this life will be somewhat worse if she gives birth to the child, though in all cases her reduction in utility from giving birth is less than the gain in utility to the born child. Assume also that no one is substantially harmed by the existence of any of these people.
Under these assumptions, isn’t the utility-maximizing outcome for the mother to always give birth to the child? No. It depends on how many children the mother is otherwise planning to have, or whether the child is a marginal or inframarginal child. If we suppose that a woman is planning to have two children in her life, and she gets pregnant with one of them at age 14, the utility-maximizing outcome would be for her to abort the child and then have two children when she is ready for them. She and the children she does have will have better lives than she and the children she does not have would have had. However, if a woman is planning to have two children in her life, has them, and then becomes pregnant for a third time, the utility-maximizing outcome would be for her to give birth to the child. This third person will not otherwise be replaced later on by someone else who will have as much or more utility.
To sum up, under utilitarianism, the abortion of inframarginal pregnancies is permissible, but the abortion of marginal pregnancies is not. Pro-lifers would not accept that the abortion of inframarginal pregnancies is permissible. Pro-choicers would not accept that abortion is impermissible in the case of marginal pregnancies. But the oddest result comes if we contrast a woman who is planning to have two children with a woman who is not planning to have any. Suppose they both get pregnant at age 14. It is permissible for the woman who wants children to have an abortion, but it is impermissible for the woman who does not want children to have an abortion. This seems like it would conflict with everyone’s intuitions.
Does my example work? Can you find an acceptable utilitarian solution to this dilemma?
9 Comments
Hi Eli. I get your point for sure, and it could be a novel addition to the stock list of situations in which act utilitarianism garners intuitively implausible results: punishing the innocent, not punishing the guilty, forced organ donation, giving to charity a la Peter Singer, etc.
But I think the example requires further description, or at least a *very* strong ceteris paribus clause. It could be the case that, if the 14 year old aborts and gives birth later, her life will still be well worth living, and her future baby’s life will be worth living, but that the future baby will have much worse of an impact on the total utility of the world than the impact of the aborted baby – e.g., first (aborted) baby would have been Gadhi, second baby turns out to be Hitler. That’s an extreme example, but it could be pretty ordinary – later baby turns out to be a defense attorney who gets dangerous criminals off the hook (creating disutility), first baby would have managed a restaurant, or whatever.
So, for instance, here:
“However, if a woman is planning to have two children in her life, has them, and then becomes pregnant for a third time, the utility-maximizing outcome would be for her to give birth to the child. This third person will not otherwise be replaced later on by someone else who will have as much or more utility.”
This point seems to require additional assumptions about the utility of people other than simply the mother and the child. The utility maximizing outcome will be for her to give birth to the child if and only if that particular child’s particular future life will contribute more to the total utility pie than she will subtract. This effect on the aggregate is not necessarily correlated to the utility that the *child herself* will experience.
So basically I just think some extra care in formulation is required. We are, in effect, assuming away the utility of everyone but the mother and children.
These possible birth scenarios are notoriously difficult because people have absolutely no idea what their possible future children will turn out to do or how their possible future children’s lives will affect other third or fourth or fifth future parties who will have effects on total utility, for the better or the worse -> knowledge problems to the max.
As for a utilitarian solution – I imagine the strategy for reply will be much like that taken towards utilitarianism’s other implausible results. One can either bite the bullet or go with some kind of rules or rights that are allegedly justified by appeal to utility. Both have problems.
Sorry so long! Hope that helps.
Thanks for the feedback. I did write, “Assume also that no one is substantially harmed by the existence of any of these people,” by which I meant all people in the example, though that certainly wasn’t clear since I hadn’t mentioned the possible later children yet. But of course, even if that were stated more clearly, that doesn’t solve the problem because the child conceived at age 14 could be the next Mother Teresa, even if the latter child is not Hitler.
One approach would be to add a ceteris paribus clause, as you say. But we need not have equal outcomes—an alternative would be to profess equal ignorance about both possible children. It’s true that the latter child might be Hitler. But the first child might be Hitler too. If we are equally ignorant about both situations, the ex ante expected utility is as I described. (This is how economists often deal with symmetric cases of imperfect information: add an error term with mean zero.)
In any case, you make a fair point about the need to state the argument with more precision. I will definitely be more careful when I include this example in my philosophical magnum opus. ;-)
hi very interesting post.
One thing though:
Wouldn’t it be a requirement on the mother plan to have the number of children that would maximise utility so that the aborted child would always be reduce utility, since she would have had the child at a time in her life that would not maximise utility – else she would have wanted the child.
again I enjoyed reading this post,
david
David, thanks for commenting. I think my example works even if the mother is not planning to have the socially optimal number of children. It works as long as having a child now reduces the number of children she would have later.
This is pretty picky, but… you write:
“One approach would be to add a ceteris paribus clause, as you say. But we need not have equal outcomes—an alternative would be to profess equal ignorance about both possible children. It’s true that the latter child might be Hitler. But the first child might be Hitler too. If we are equally ignorant about both situations, the ex ante expected utility is as I described. (This is how economists often deal with symmetric cases of imperfect information: add an error term with mean zero.)”
Professing equal ignorance would go some ways towards helping the formulation be more precise (and I’m sure you’re right as far as ex ante utility calculations go), but doing this may in effect be confusing act utilitarianism’s criterion of rightness and decision procedure. (Forgive me if I get a little lecture-ish here, I just TAed a course that focused heavily on this very topic, taught by a professor who is a consequentialist, with whom I was also taking a seminar on such matters).
A criterion of rightness establishes the fundamental right-making and wrong making feature(s) of actions. The clearest statement of act utilitarianism’s criterion of rightness is: “an act is right, if and only if and because, there is no other act available to the agent which would produce more utility”, where utility is impartially aggregated across persons (and across time).
The decision procedure is just the way that a moral theory would have agents decide what to do. Interestingly, an act utilitarian needn’t hold that agents should always- or even often – engage in explicit utility calculations as a way of making decisions regarding how to act. The act utilitarian may sanction or require the use of various rules of thumb (don’t lie, don’t steal, don’t murder) as a way of maximizing utility. Calculating utility, after all, takes alot of time and effort. Notice that the act utilitarian who endorses rules as a decision procedure is different than a rule utilitarian: the AU thinks that maximizing utility on any given occasion is the fundamental right making feature of an action, while the RU thinks that being in accordance with a rule is the fundamental right making feature of an action (when the rule is part of the set of rules which when followed reliably in a society would maximize utility).
Here’s where I’m going with this: You want the abortion case(s) to function as an objection to act utilitarianism’s criterion of rightness. Ceteris paribus clauses are used to assume away peripheral factors in order to more clearly see how a criterion of rightness functions. But directions about how to act under conditions of ignorance are substantive claims about act utilitarian’s *decision procedure*, not its criterion of rightness. And this could be bad because, while act utilitarianism’s criterion of rightness is pretty clear and well-established, its decision procedure is controversial.
Like I warned, that’s a picky (& kinda technical) issue. But the way you get better at philosophy is by having to deal with such things!
I would distinguish between two kinds of ignorance. Let’s call them Type I and Type II. Type I ignorance is ignorance that can be alleviated by paying some utility cost (doing research, asking around, etc.). Type II ignorance is fundamental in the sense that it cannot be reduced even by paying some utility cost.
It seems to me that Type I ignorance is relevant to the decision procedure. As you say, it is not optimal to eliminate all ignorance before acting, and therefore some decision must be made as to how much Type I ignorance is too much. There could be disagreement here as to the appropriate method of dealing with this question.
Type II ignorance is not relevant to the decision procedure. We can incorporate it into the criterion of rightness by adding the word expected: “an act is right, if and only if and because, there is no other act available to the agent which would produce more expected utility.” There is no ambiguity here about which acts are right.
Another way of making this point is to suppose that after each child is born, a die is rolled, and if it lands on 1, the child is Hitler. The same (fair) die is used in all instances. In this case, there is no ignorance that is relevant to the decision procedure. There can be no disagreement over the right utilitarian choice, because we know everything there is to know about the situation. The introduction of randomness should not undermine the criterion of rightness.
My claim is that the relative likelihood of a child being Hitler, Gandhi, etc. is what I am calling Type II ignorance. I therefore still see these cases as an attack on the criterion of rightness, and not on the decision procedure. You may object that I am unilaterally revising utilitarianism, but I think that any objection to this extremely modest revision is indefensible.
Ignorance distinction seems fine, and you certainly won’t find *me* objecting to act utilitarianism being modified as such. But it’s important to recognize, as you apparently do, that the expected utility version AU is indeed a modification of what appears to be the original version (Bentham stated it sloppily). In fact, it seems to me that many people who discuss utilitarianism outside of academic philosophy already either mistakenly or intentionally understand utilitarianism in the expected utility way. So it’s good that you made it explicit.
Well I’m a little late but…
The utility monster as a construct tries to create this paradox:
That, given that the right thing to do is X, should we do it?
By definition, as a utilitarian, the answer is yes. As such, the person posing the question can put anything inside of X, and the utilitarian usually tries to explain why action X does not, within our sphere of influence, ever come up as the right answer.
However, just saying that the issue never comes up is at best an incomplete answer. A better answer is, that if the issue were to come up, the world would have to be very different in order to accommodate the situation. Looking at the big picture, inside this newly fabricated completely different world, actions that seem abhorrent in our world would not look so terrible. In other words, this paradox is created by a fallacy similar to one trying to implicate a surgeon in terrible deeds by showing a series of pictures wherein the surgeon knifes people, without the context of his actions. Just because we’re used to knifing people being wrong doesn’t mean that it is fundamentally evil to cut people with knives.
Saying that, we can now look at the premises behind the abortion issue. If we reword the original statement to be less mathematical and more simplistic it reads something like this-
“Imagine two girls at the age of 14 both magically pregnant, both of which will live better lives if, during their life they have kids. Assume also that the kids will have good lives, and the population at large will benefit. Assume that the optimal time for these mothers to have their first child is at the age of 22. One parent is planning to start having children when she reaches 22 and intends to have two children total, but the other plans never to have any children.”
In the first case, if the girl has a child she will have difficulty taking care of her kid for the next eight years, and might end up dropping out of school or not going to college. Whatever benefit she receives from the child is lost due to opportunity cost because of the child she would have had later, and can be safely ignored. Abortion is the correct option. Assuming that this mother is wrong to settle on two children and should have even more, and that she was magically given two children on the spot, the answer would change, because the opportunity cost is no longer existent.
In the second case, the girl is preparing to make a very dangerous mistake that will jeopardize her overall utility. In order to save this girl from a sad lonely life with no kids, issues like school and college become unimportant. Abortion is now the wrong decision.
In other words, the correct decision is made based on marginal utility. Opportunity costs are important in any form of calculation.
The question raised by this paradox is this-
Assume that action X>Y>Z.
Given that you can decide for a given actor whether action X or Y is decided upon, which do you choose?
The obvious answer is X.
Now assume a new actor with a choice between Y or Z.
The answer is now Y. X is no longer an option, so it doesn’t matter.
We should now look at the second part of the paradox- that these answers force the actors into doing the express opposite of what they want to do. This is easy for a utilitarian to answer- what people intend to do is not the definition of morality or even particularly related to it. In other words, I am allowed as a utilitarian to force people into activities against their will. It’s curious that people object to this principle of utilitarians so often when almost every branch of philosophy has some form of coercion that it is willing to condone. Unless you hold stridently to the non-aggression principle, I shouldn’t need to justify bending people’s wills in return for a higher overall utility.
Okay, so let’s move on to the issue of population growth. My answer here is three fold-
A (the issue does not come up within our universe): Remember that time is also a dimension. Since time is infinite, and thus so is space. Life, too, is infinite. As such population cannot be increased or decreased. The only thing that can be modified is life’s density, which is a morally neutral object. As such, population levels should be held at a level wherein the denizen’s each reach an individual maximum of utility. Assuming this population is held for eternity, the maximum utility available has been achieved, and anything else detracts from overall utility.
B: Thresholds exist. Imagine that, for a person to start producing any utility at all, he must at least have some time spent doing activities besides minimum sustenance. Imagine also, that for each X% of time spent doing higher order activities his utility increases by Y% with his gains diminishing at a rate of Z%. Given that each additional person decreases the amount of time that each group member has for X, then there should be an optimal population based on X, Y, and Z which will result in maximum utility.
C: Negative utility also exists. Adding in negative numbers to a utility function has some very interesting results, and can help explain any number of phenomena. If we assume that each new person not only has good experiences such as eating potatoes and resting, but also has to suffer through hot days in the sun and tragic losses of family to disease and disaster, then it becomes important that positive utility X be higher than negative utility Y. As each individual’s standard of living goes down, X approaches Y. Assuming that Y is less elastic than X, we should limit population to an optimal number to achieve an ideal utility result.
Okay, now for Nozick. The problem with his utility monster is simple- first we must look at just what such a utility monster would have to be. In order for an entity to exist that receives an eternally increasing utility from goods, it would have to a number of features, amongst which would be the ability to aware of everything he was enjoying. In order to partake in an infinite wine tasting session he would need infinite awareness to take in the full substance of what he was doing. Being a being of infinite awareness, this entity is in effect not one person but an entire alien civilization. The difference between one person and two people is the ability to be aware of two separate lines of reality at once. An entity that is capable of doing this is equal in worth, and in fact physically the same as two people. So, the true nature of the utility monster can be rephrased like this:
“Assume that two societies exist. One society is in all ways better than the other. Should the inferior society give all of its resources to the superior society and die?”
The answer is yes. This becomes more readily apparent if I rephrases the question to this-
“Where the settlers of America correct when they decided to displace the Aztects and thus allow Americans to live in their place?”
If we use the universal principle it can be said that if the settlers were right to kill the Aztects, then it would also have been good if the Aztects had killed the Aztects. Of course, if the Aztects had realized how inferior they were, they could have corrected their ways, and would therefore have ceased being inferior anymore. So, while in theory it would be best if all inferior groups were to commit suicide and hand over everything to their betters, this is theoretically impossible to implement within this universe.
The paradox falls apart if you do not add false assumptions to the premise. When people imagine a utility monster, they see just one person, who can not compete with an entire civilization. However, the utility monster is, by definition, an existence far above any one person, and can thus only be compared with other transcendentals. Or, in modern terms, entire civilizations. It is much easier to accept that a civilization should die so as to allow another civilization to live than to accept that a civilization should die so a single person can live.
There is one utility monster which is notorious for its behaviour, exists in large numbers and is well defined in law and custom. It is the business corporation, and every dollar it makes is as important as the first dollar it makes – because if it does not make money, it will “die”.
We must live in a utilitarian world because indeed, politicians move heaven and earth to defend these behemoths from regulations, laws, taxes or anything else that might curtail their activities. And they are indeed consuming the entire earth.