I was discussing UDT yesterday and the question came up of how to treat uncertainty over your utility function. I suggested that this could be transformed into a question of uncertainty over outcomes. The intuition is that if you were to discover that apples were twice as valuable, you could simply pretend that you instead received twice as many apples. Is this approach correct? In particular, is it transformation compatible with UDT-style reasoning?

New Answer
New Comment

2 Answers sorted by

Scott Garrabrant

190

Utility functions are invariant up to affine transformation. I don't need to say how much I value a human life or how much I value a chicken life to make decisions in weird trolly problems involving humans and chickens. I only need to know relative values. However, utility uncertainty messes this up. Say I have two hypotheses: one in which human and chicken lives have the same value, and one in which humans are a million times more valuable. I assign the two hypotheses equal weight.

I could normalize and say that in both cases a human is worth 1 util. Then, when I average across utility functions, humans are about twice as valuable as chickens. But if I normalize and say that in both cases a chicken is worth 1 util, then when I average, the human is worth about 500,000 times as much. (You can still treat it like other uncertainty, but you have to make this normalization choice.)

This is precisely the issue discussed at length in Brian Tomasik's article "Two-Envelopes Problem for Uncertainty about Brain-Size Valuation and Other Moral Questions".

Indeed.

We tried to develop a whole theory to deal with these questions, didn't find any nice answer: https://www.lesswrong.com/posts/hBJCMWELaW6MxinYW/intertheoretic-utility-comparison

But if you can answer questions like "how much money would I pay to save a human life under the first hypothesis" and "under the second hypothesis", which seem like questions you should be able to answer, then the conversion stops being a problem.

8Scott Garrabrant
You are just normalizing on the dollar. You could ask "how many chickens would I kill to save a human life" instead, and you would normalize on a chicken.
3cousin_it
I'm normalizing on my effort - eventually, on my pleasure and pain as a common currency. That's not quite the same as normalizing on chickens, because the number of dead chickens in the world isn't directly qualia.
3Stuart_Armstrong
The min-max normalisation of https://www.lesswrong.com/posts/hBJCMWELaW6MxinYW/intertheoretic-utility-comparison can be seen as the formalisation of normalising on effort (it normalises on what you could achieve if you dedicated yourself entirely to one goal).

Thanks, very interesting. I guess when I said I was imagining a situation where oranges were twice as valuable as was imagining them as worth X utility in situation A and 2X in situation B and suggesting we could just double the number of oranges instead. So it seems like you're talking about a slightly different situation than the one I was envisaging.

waveman

20
if you were to discover that apples were twice as valuable, you could simply pretend that you instead received twice as many apples

No, because twice as many apples are not usually twice as valuable. This because utility functions are not linear.

You can kind of deal with uncertainty about utility by fudging expectations about outcomes but, trust me, it is the primrose path to hell.

If the utility function is the square root of the number of apples you could multiply the number of apples by four. The question is mainly about whether you can do that kind of adaption than about anything else.

2 comments, sorted by Click to highlight new comments since:

What if you decided that, actually, apples had negative value? Then would you pretend you received negative apples?

We could take apples away.