Re: Evolved Preferences

Robin Hanson (
Tue, 29 Apr 1997 15:07:18 -0700 (PDT)

I wrote:
>Let me suggest that it is useful to distinguish the evolution of
>values from the evolution of "facts", or beliefs about facts. The
>space of possible basic values is much smaller than the space of
>possible fact beliefs;

Eric Watt Forste writes:

>This is an asymmetrical comparison because you use the word "basic"
>on only one side. I doubt that the space of possible basic values
>is much smaller than the space of possible general epistemic
>principles (reasons for choosing to treat a piece of information in your
>head as if it were factual). I also doubt that the space of possible
>specific valuations in concrete situations is much smaller than the
>space of possible beliefs.

Yes, the better distinction is between small spaces, which include
both basic values and general epistemic principles, and large spaces,
such as lots of little new facts. For evolution in small spaces, the
information sharing function of sex is less important. So I should
restate my claim as that evolutionary pressure on small spaces, such
as investment time-discount factors, should prefer asexual

I do think there is also an asymmetry between facts and values in that
it is plausible to think of context-specific values as generated
locally from facts and general values, while this is less plausible if
we switch fact and value. We really need lots of small fact info from
other organisms, but its not clear we need to share context-specific

>argues against the idea that current human values are in equilibrium.

Yes, we are likely way out of equilibrium. That is the motivation for
trying to analyze what a future equilibrium would be.

Robin D. Hanson