From: Richard Loosemore (rpwl@lightlink.com)
Date: Mon Apr 24 2006 - 12:44:58 MDT
Eliezer S. Yudkowsky wrote:
> Richard Loosemore wrote:
>>
>> Eliezer,
>>
>> I'm confused: was your point that Phil's comment was so foolish that
>> it deserved only sarcasm?
>
> I wouldn't go that far. I would say that it sounds plausible to
> computer scientists but conflicts with experimental observations in the
> brain sciences and in evolutionary biology.
>
Fair comment, then, if a bit harshly phrased. My response was in the
parallel reply to Robin Lee Powell (which crossed with your reply):
several factors may contribute to the level of intelligence in
cross-species and intra-human comparisons, so that what is happening in
the case of an intelligent human is that all those other factors are
finally cleared out of the way and the brain hits the seven-chunk
endstop that (hypothetically) could be determined by its volume.
I took Phil's argument to be that from here on out the limiting factor
might be the chunking limit, which could scale in a nasty way with
volume, or be the result of a tradeoff that admits of no solutions
whatsoever beyond a certain number of chunks.
Personally, I am not as pessimistic as Phil, so I don't think this is
going to be a problem, but I have only arguments like (a) chunking is
probably not the main story when it comes to intelligence, and (b) the
constraints on chunking anyway turn out to be polynomial and not
exponential (wiring connectivity considerations by themselves shouldn't
lead to exponentials). But I think it is an open question.
Richard Loosemore.
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:56 MDT