I’ve been thinking about scope insensitivity, and wonder whether it can be mistaken for decreasing marginal value. Suppose a slice of pizza costs $1, and you’re willing to buy it for that much. That doesn’t mean that you’d be willing to buy a million slices of pizza for a million dollars—in fact, if you’re only hungry enough for one slice and don’t have a refrigerator handy, you may not want to pay more than $1 for any amount of pizza greater than or equal to once slice. The same can apply to donating to charity to save lives. You may value saving one life for $200, but maybe you’re not willing to pay $400 to save two lives.
You may value saving one life for $200, but maybe you’re not willing to pay $400 to save two lives
This is what scope insensitivity is. The original paper calls it “purchase of moral satisfaction”—the revealed preference in these experiments is for an internal state of moral satisfaction as opposed to the actual lives you’re saving. Like hunger, the internal state is quickly satiated and so exhibits diminishing returns, but actual lives do not exhibit diminishing returns (in the relevant range, for humans, on reflection).
That would be an effective demonstration of scope insensitivity in an ideal scenario where money has a flat conversion to utility in that range for the individual in question. If $200 is in the subject’s budget, but $400 is not, this may be entirely rational behavior. A donation which puts you into debt will have a much more dramatic effect on your own utility than one which leaves you solvent.
You’re right, the $200 vs $400 example isn’t ironclad—even a rational altruist will still have a limited budget. The reason I ascribed scope insensitivity to the example was that it’s worded in terms of valuing ‘saving lives’ as opposed to valuing lives, which, as I explain, is a hallmark of scope insensitivity.
Saving 10 times more birds would make me 10 times more happy. Therefore, I would make a bigger celebration afterwards. But the celebration also comes from my limited budget, so there would be less money left to contribute to saving the birds.
Eliezer’s answer (I can’t find it at the moment) is that there are many more lives around, and it seems unlikely that U(875,431 people die) - U(875,430 people die) is that different from U(875,430 people die) - U(875,429 people die).
(OTOH, once you’ve spent $200 to save a life, you have $200 less, and so the marginal utility of another $200 is larger.)
The decreasing marginal value is the amount of fuzzies your donation lets you feel.
If you get 10 utilons from doing something that saves one person’s life, perhaps you get only 12 from doing something that saves two people’s lives.
Decreasing marginal satisfaction from doing more to help a problem is the cause of scope insensitivity—people feel bad knowing that they aren’t helping, and feel good knowing that they are.
For the same reason you only value the pizza you eat, not just pizza eaten in general—because it’s utility derived from consumption of a good/service.
I’ve been thinking about scope insensitivity, and wonder whether it can be mistaken for decreasing marginal value. Suppose a slice of pizza costs $1, and you’re willing to buy it for that much. That doesn’t mean that you’d be willing to buy a million slices of pizza for a million dollars—in fact, if you’re only hungry enough for one slice and don’t have a refrigerator handy, you may not want to pay more than $1 for any amount of pizza greater than or equal to once slice. The same can apply to donating to charity to save lives. You may value saving one life for $200, but maybe you’re not willing to pay $400 to save two lives.
This is what scope insensitivity is. The original paper calls it “purchase of moral satisfaction”—the revealed preference in these experiments is for an internal state of moral satisfaction as opposed to the actual lives you’re saving. Like hunger, the internal state is quickly satiated and so exhibits diminishing returns, but actual lives do not exhibit diminishing returns (in the relevant range, for humans, on reflection).
That would be an effective demonstration of scope insensitivity in an ideal scenario where money has a flat conversion to utility in that range for the individual in question. If $200 is in the subject’s budget, but $400 is not, this may be entirely rational behavior. A donation which puts you into debt will have a much more dramatic effect on your own utility than one which leaves you solvent.
You’re right, the $200 vs $400 example isn’t ironclad—even a rational altruist will still have a limited budget. The reason I ascribed scope insensitivity to the example was that it’s worded in terms of valuing ‘saving lives’ as opposed to valuing lives, which, as I explain, is a hallmark of scope insensitivity.
Still doesn’t explain people paying less to save 20000 birds than to save 2000.
Saving 10 times more birds would make me 10 times more happy. Therefore, I would make a bigger celebration afterwards. But the celebration also comes from my limited budget, so there would be less money left to contribute to saving the birds.
(End of rationalization exercise.)
I never claimed that no effective demonstrations of scope insensitivity exist.
Eliezer’s answer (I can’t find it at the moment) is that there are many more lives around, and it seems unlikely that U(875,431 people die) - U(875,430 people die) is that different from U(875,430 people die) - U(875,429 people die).
(OTOH, once you’ve spent $200 to save a life, you have $200 less, and so the marginal utility of another $200 is larger.)
What would the decreasing marginal value be in the charity example? If we’re just talking fuzzies, isn’t that scope insensitivity?
The decreasing marginal value is the amount of fuzzies your donation lets you feel.
If you get 10 utilons from doing something that saves one person’s life, perhaps you get only 12 from doing something that saves two people’s lives.
Decreasing marginal satisfaction from doing more to help a problem is the cause of scope insensitivity—people feel bad knowing that they aren’t helping, and feel good knowing that they are.
That does raise the question of why you only value lives you saved. If all lives are important, marginal value won’t change much over one life saved.
For the same reason you only value the pizza you eat, not just pizza eaten in general—because it’s utility derived from consumption of a good/service.
It seems you’re confused about what scope insensitivity means.