I feel certain someone has already discussed the following problem, but I'm frustrated that I don't remember who and where. Anyway:
Suppose there is some action whose consequences, it appears, are on balance slightly bad. (Perhaps a boy steals a candy bar from a store.) It seems possible for there to be an omniscient person who enjoys immoral things
for their own sake. It also seems possible that this person is a consequentialist. Suppose she reads about this theft in the neighborhood newspaper. She enjoys this immorality. It's also plausible that her enjoyment of immoral things is
good. But because one consequence of this theft is that she enjoys reading about it, the total consequences of his theft are now, on balance, good. But then she no longer believes that his action is immoral. So she doesn't enjoy it. So, his theft does not have on balance good consequences. So his action is slightly bad………..
This pattern can oscillate infinitely.
What's going on here? It appears as though there is some kind of conflict between 1) enjoying immorality for its own sake, 2) believing in consequentialism, 3) believing enjoyment is good, 4) being omniscient. (I'm not even sure that #4 is necessary to generate the oscillation, but it certainly helps.)
I recall a discussion of this as well but cannot remember where. I don’t think the oscillation in this case is vicious. It seems instead rather accidental to me. You have to get the values of the consequences and the amount of enjoyment the person gets from immorality exactly right in a way that will be quite rare. If the action is immoral enough and/or she enjoys immorality just a bit, no oscillation follows. So it doesn’t seem to be that there is anything in principle conflicting between 1-4 but rather there can be contingent situation in which silly things happen as a result of them.
Eric,
For the sake of simplicity, let’s assume that determinism is true. Either the boy’s stealing the candy bar will maximize the good or it won’t. An act maximizes the good iff there is no available alternative that would produce more good than it would — that is, iff it is a maximizing option. You say that the omniscient person will take enjoyment in the fact that the boy has stolen the candy bar if the boy was wrong in doing so. In that case, she will take enjoyment in the fact that the boy has stolen the candy bar if her enjoyment is insufficient to make it a maximizing option. Where’s the conflict?
Also, if she is omniscient and believes that consequentialism is correct, then doesn’t it follow that consequentialism is correct? Or do you think that she can know that consequentialism is false and yet be a consequentialist (i.e., one who believes that consequentialism is true)?
Oh, dear! I didn’t mean to actually post that yet….I was trying merely to save it for editing later. Oh, well!
JS: I agree that the situation would be rare. Still, I wonder what to think about it.
DP: I included omniscience so that the person who enjoys immorality would know about all the other consequences of the theft. Obviously, I don’t want to assume that the person who enjoys immorality knows which moral theory it true! Thanks for pointing that out.
I’m not sure what to think about your other point. Maybe we shouldn’t assume that she can predict accurately how much she will enjoy the immorality of the theft. That would get the oscillation up and running.
Maybe we shouldn’t assume that she can predict accurately how much she will enjoy the immorality of the theft. That would get the oscillation up and running.
Either her total enjoyment, given all her oscillations, will be sufficient to make the boy’s stealing the candy bar a maximizing option or it won’t. If it is, then she will, during certain periods within her oscillations, falsely believe that the act is immoral. If it isn’t, then she will, during certain periods within her oscillations, falsely believe that the act is moral. Where’s the conflict, though?
Strange – I just had a message vanishing just like that.
Anyway, Eric, I think you can still edit the post.
Come to think of it, I also think something that Doug says implicates a consequentialist solution.
The oscillation you describe seems to be a temporal process or a continuous action of thinking in a certain way. The consequentialist could say that whether the agent should do this action, i.e., go on oscillating in the given way should be assessed by consequentialist criteria. Now assume that the agent goes on thinking in this way. Seems like her life will quickly get dull and bad. So if she has better options which would make things go better, the agent should pretty quickly stop the oscillation process. All of this seems to follow from what the consequentialist already thinks so she does seem to have a way of putting an end of whatever is going on.
If her enjoyment is a continuous function of the degree of immorality, and the immorality of the action is a continuous function of her enjoyment, and the immorality-space is the closed interval [-∞, ∞], then there’s an equilibrium point where her enjoyment is exactly appropriate to the all-things-considered immorality of the action.
(Here’s a simple case. Say the immorality I is 1 – a*E, where a is a positive constant and E is the amount of enjoyment. Say enjoyment is b*I, b another positive constant. Then the equilibrium comes at I = 1/(ab+1), E = b/(ab+1).)
If her enjoyment is at equilibrium, there’s no conflict. She is a bit weird, though. (She enjoys not enjoying things too much?)
The paradox is cleaner (avoiding Doug’s worries) if we use desire satisfactionism rather than hedonism.
(I wrote some blog posts about this a few years ago, and I think Ben Bradley wrote a paper on it more recently.)
Ah! Thanks, Richard (and everyone else)!
Let me explain why this was bugging me. Some people think that if it’s immoral to tell a particular joke, then (the telling of) that joke isn’t funny. Others think that the two issues–morality and humor–really have nothing to do with each other.
But I’m tempted to think that some jokes are funnier just because they are immoral. I enjoy them more. Of course, one shouldn’t tell them! But if even just part of the moral value of the telling of such a joke depends upon how much enjoyment it produces, I find myself potentially locked in the kind of puzzle I described above. I see that there are ways out of it–some of which we have already noted–but I still have some unresolved suspicion that something has gone amiss.
I do see the problem, I think.
First, can I say that Jeff Sanford Russell’s fixed point point (so to speak) is cool, and strangely comforting?
But second, suppose the observer’s enjoyment isn’t a continuous function of immorality. She might get a huge kick out of immoral acts and situations, no matter whether they were just slightly or grossly immoral; and neutral or morally good situations, meh. I also think Eric should allow his observer to know that the particular consequentialist theory is true. I see no harm in this as one of the assumptions.
Now the point is not really that the value and enjoyment oscillate over time. It’s that there is no coherent valuation of the situation or act. This is somewhat paradoxical, because the assumptions seem to be perfectly coherent. Could the lesson be that there is one of the infamous knowability blind spots here? The consequentialism in question can be true, but an observer of the right kind who comes to believe it will render it false? That would be pretty surprising.
Also, Eric, do you know about Egan & Weatherson’s paper, Prankster’s Ethics? (The link is to a pdf. The paper is in Philosophical Perspectives 18.)
I think I see the problem, though I’m not sure what oscillation, determinism, omniscience, the target agent’s beliefs, or consequentialism have to do with it. So, on second thought, I probably don’t see the problem that Eric sees. In any event, the following are inconsistent premises:
1) Agent A enjoys action e, if e is wrong.
2) If A does not enjoy e, e is wrong.
Consequentialism helps to explain how (2) can be true, but presumably other ethical theories can do so as well.
From (1) we get its contrapositive:
3) If A does not enjoy e, e is not wrong.
This contradicts (2). That’s a bit weird, for they kinda sound mutually consistent. This reminds me of the barber paradox and related set-theoretic paradoxes. (1) can be true, and consistent with (2), so long as we restrict the wrong/right-making features of e to features other than A’s possible enjoyment of e.
Is there story coherent after all so as to generate the problem?
Here’s the crucial bit:
“It seems possible for there to be an omniscient person who enjoys immoral things for their own sake. It also seems possible that this person is a consequentialist. Suppose she reads about this theft in the neighborhood newspaper. She enjoys this immorality”
On what grounds can she come to make this judgment that the action is immoral? That judgment would require that there is a coherent valuation of the situation. But, surely, the omniscient consequentialist would know that there isn’t one (if Jamie is right).
Matt,
Hm, I don’t think your two premises are inconsistent.
Jussi,
I’m not seeing your point.
I was saying: the assumptions of the scenario must be inconsistent, but they do not appear to be so. (That’s a typical paradox, right? Think of the Surprise Examination, for instance.)
Is this omniscient woman a hairdresser who styles the hair of all and only those women who do not style their own?
Matt, your premise two is:
If A does not enjoy e, e is wrong.
But that doesn’t fit Eric’s case, since it is right actions that A does not enjoy.
Glad you think it’s cool, Jamie. In defense of the continuity assumption: it follows from physicalism, if the laws of physics turn out to be much at all like we think. So if physicalism is true, the stepwise values you suggested would be physically impossible (though of course arbitrarily close approximations are fine).
Tim Maudlin and some other people have used similar continuity-and-fixed-point considerations to deal with time travel paradoxes.
This sort of problem is discussed at length in Ben Bradley’s recent paper “A Paradox for Some Theories of Welfare”:
http://web.syr.edu/~wbradley/Paradoxes%20of%20welfare.pdf
Perhaps this paper was what Eric had in mind when he wrote the first sentence of his original post.
Jamie,
I’m not sure if I had a point. I just was (and am) very curious about the status of the initial judgment which the agent is supposed to make in the situation. When the consequentialist omniscient agent makes the first judgment that the action is immoral what valuation is this judgment based on? It is starting to seem to me that there is some assumption here about this judgment that could be incoherent.
Hurka has a relevant discussion in Virtues, Vices, and Values. As I remember it, he maintains that the value of the reactions must be less than the value of what is reacted to in order to avoid some of the paradoxes you come up with. So there are, on his view, some things that are basic goods, then there are futher things that are good such as having the right attitude towards the basic goods. But having the right attitude (hating) towards a basic bad must not have so much value that it is better to have the combo than to have neither. Not endorsing, just sayin …
Jamie-
You must be reading
1) Agent A enjoys action e, if e is wrong. and
2) If A does not enjoy e, e is wrong.
as material conditionals. Given Eric’s setup, I was reading them as stronger than this – A enjoys e because it is wrong and e is wrong because A does not enjoy it (his enjoyment would make it right). But then I should be more careful about using contraposition (or something like it)! So let me just read them as material conditionals, and add another premise:
*) A does not enjoy e
1, 2 and * are mutually inconsistent. Right? That is, unless 1 is implicitly relativized to include the wrong/right-making features of e *other than A’s enjoyment of e or lack thereof*, and 2 includes all wrong/right-making features of e, including A’s enjoyment of it or lack thereof. So that’s my solution – 1 really says that A enjoys e if it is wrong, taking into account all the normatively relevant features of e into account *except A’s reactions, if those are wrong/right-making features*.
Dale-
I thought we were working with a case where A’s enjoyment of e is sufficient to make e right (because of its good consequences), and A’s non-enjoyment of e is sufficient to make e not right (here, wrong, because the consequences aren’t good enough).
All-
Sorry if I’m missing something!
Matt,
D’oh! (In other words, I follow you now.)