The reason I believe Pascal’s wager is flawed is that it is a false dichotomy. It looks at only one high utility impact, low probability scenario, while excluding others that cancel out its effect on expected utility.
Is there anyone who disagrees with this reason, but still believes it is flawed for a different reason?
This is an argument for why the argument doesn’t work for theism, it doesn’t mean the argument itself is flawed. If you would be willing to multiply the utility of each belief times the probability of each belief and proceed in choosing your belief in this way, then that is an acceptance of the general form of the argument.
If you assume that changing your belief is an available action (which is also questionable), then the idealized form is just expected utility maximization. The criticism is that Pascal incorrectly calculated the expected utility.
Right, one flaw in the idealized form is that it’s not clear that you can simply choose the belief that maximizes utility. But in some cases a person can, and does.
I think that an incorrect calculation, because one person considered 2 cases instead of N cases, is very different from being flawed as an argument.
PeerInfinity was writing about applying Pascal’s wager to atheism—so he must have been referring to the general form of the argument, not a particular application. Matthew B wrote that “Pascal’s Wager is a flawed argument for believing in ANY belief system”. Well, what about a belief system in which there are exactly two beliefs to choose from and the relative probabilities are (.4, .6) and the relative utilities of having the beliefs if they are true are (1000, 100) ? I would say the conclusion of the idealized form of Pascal’s wager is that you should pick the belief that maximizes utility, even though it is lower probability.
I would distinguish between the general form and the idealized general form. One way to generalize Pascal’s wager for belief B, is to compare the expected utilities of believing B and believing one contradictory Belief D in the conditions that B is true and that D is true. This is wrong no matter what belief B you apply it to.
The utility of having a belief is what is being considered in Pascal’s wager, and is quite different from the utility of the belief itself.
The utility of a belief itself wouldn’t sway you to choose one belief over another. Suppose againyou have the two beliefs X and Y, and they each have a certain utility if they are true. If X is true, then you “get” that utility, independently of whether you believed it or not, by virtue of it being true. For example, if there is utility to God existing, then there is that benefit of him existing whether you believe in him or not.
In contrast, there is also utility for having a belief.
To complicate things, there is a component of the utility that is independent of whether the belief is true or not, and there is a component of the utility that depends on the belief being true. In the case of theism, there is a utility to being a theist (positive or negative, depending on who you ask) regardless of whether God exists, and there would also be an extra utility for believing in him if he does exist (possibly zero, if he doesn’t care whether you believe in him or not).
You mean the case of the argument applied to theism? I would be willing to forfeit the applicability of the argument for this case, since I’m just interested in discussing the validity of the general argument.
I don’t like discussing general cases when I don’t have some concrete examples. The only ones I can think of are boring cases of coercion involving unethical mindreaders.
Yes, I agree: the utility of having a belief only makes sense when for some reason you are rewarded for actually having the belief instead of acting as though you have the belief.
OK, since theism is unique in this aspect, in order to generalize away from the theistic, let’s use the utility for acting-as-though-you-believe instead of the utility for actually believing, because in most cases, these should be the same.
… but then, as soon as you do this, the argument become just about choosing actions based on average expected utility and there’s nothing controversial about it. So I guess PW might just suffer from lack of application: there are few cases where you are actually differentially rewarded for having a belief (instead of just acting as though you do), and these cases (generalizing from theism) involve hypotheses that are too complex to parametrize (Silas’ argument).
Back to the immediate object level: PeerInfinity wrote about applying Pascal’s Wager to atheism. However, atheism doesn’t make a utility distinction between having a belief and acting as though you do. Or does it? Having beliefs motivate actions and make them easier to compute.
When PeerInfinity said he chose to believe atheism because it seemed to maximize utility, he might have been summarizing together that acting as though atheism was true was deemed utility maximal, and believing in atheism then followed as utility maximal.
The reason I believe Pascal’s wager is flawed is that it is a false dichotomy. It looks at only one high utility impact, low probability scenario, while excluding others that cancel out its effect on expected utility.
Is there anyone who disagrees with this reason, but still believes it is flawed for a different reason?
This is an argument for why the argument doesn’t work for theism, it doesn’t mean the argument itself is flawed. If you would be willing to multiply the utility of each belief times the probability of each belief and proceed in choosing your belief in this way, then that is an acceptance of the general form of the argument.
If you assume that changing your belief is an available action (which is also questionable), then the idealized form is just expected utility maximization. The criticism is that Pascal incorrectly calculated the expected utility.
Right, one flaw in the idealized form is that it’s not clear that you can simply choose the belief that maximizes utility. But in some cases a person can, and does.
I think that an incorrect calculation, because one person considered 2 cases instead of N cases, is very different from being flawed as an argument.
PeerInfinity was writing about applying Pascal’s wager to atheism—so he must have been referring to the general form of the argument, not a particular application. Matthew B wrote that “Pascal’s Wager is a flawed argument for believing in ANY belief system”. Well, what about a belief system in which there are exactly two beliefs to choose from and the relative probabilities are (.4, .6) and the relative utilities of having the beliefs if they are true are (1000, 100) ? I would say the conclusion of the idealized form of Pascal’s wager is that you should pick the belief that maximizes utility, even though it is lower probability.
I would distinguish between the general form and the idealized general form. One way to generalize Pascal’s wager for belief B, is to compare the expected utilities of believing B and believing one contradictory Belief D in the conditions that B is true and that D is true. This is wrong no matter what belief B you apply it to.
Why would having the beliefs have utility? Isn’t utility a function of actions, as a rule?
There’s no contradiction in thinking “A is unlikely” and yet acting as if A is true—otherwise no-one would wear seat belts.
The utility of having a belief is what is being considered in Pascal’s wager, and is quite different from the utility of the belief itself.
The utility of a belief itself wouldn’t sway you to choose one belief over another. Suppose againyou have the two beliefs X and Y, and they each have a certain utility if they are true. If X is true, then you “get” that utility, independently of whether you believed it or not, by virtue of it being true. For example, if there is utility to God existing, then there is that benefit of him existing whether you believe in him or not.
In contrast, there is also utility for having a belief.
To complicate things, there is a component of the utility that is independent of whether the belief is true or not, and there is a component of the utility that depends on the belief being true. In the case of theism, there is a utility to being a theist (positive or negative, depending on who you ask) regardless of whether God exists, and there would also be an extra utility for believing in him if he does exist (possibly zero, if he doesn’t care whether you believe in him or not).
SilasBarta has pointed out a relevant argument regarding that case.
You mean the case of the argument applied to theism? I would be willing to forfeit the applicability of the argument for this case, since I’m just interested in discussing the validity of the general argument.
I don’t like discussing general cases when I don’t have some concrete examples. The only ones I can think of are boring cases of coercion involving unethical mindreaders.
Yes, I agree: the utility of having a belief only makes sense when for some reason you are rewarded for actually having the belief instead of acting as though you have the belief.
OK, since theism is unique in this aspect, in order to generalize away from the theistic, let’s use the utility for acting-as-though-you-believe instead of the utility for actually believing, because in most cases, these should be the same.
… but then, as soon as you do this, the argument become just about choosing actions based on average expected utility and there’s nothing controversial about it. So I guess PW might just suffer from lack of application: there are few cases where you are actually differentially rewarded for having a belief (instead of just acting as though you do), and these cases (generalizing from theism) involve hypotheses that are too complex to parametrize (Silas’ argument).
Back to the immediate object level: PeerInfinity wrote about applying Pascal’s Wager to atheism. However, atheism doesn’t make a utility distinction between having a belief and acting as though you do. Or does it? Having beliefs motivate actions and make them easier to compute.
When PeerInfinity said he chose to believe atheism because it seemed to maximize utility, he might have been summarizing together that acting as though atheism was true was deemed utility maximal, and believing in atheism then followed as utility maximal.