Re: Emulation vs. Simulation

From: Damien Broderick (d.broderick@english.unimelb.edu.au)
Date: Mon Mar 26 2001 - 22:38:01 MST


At 10:38 AM 3/26/01 -0800, Robert replied to my bleat:

>> how quickly could such a monster look-up
>> table--listed, presumably, by a god, or someone else equal to the task--be
>> looked up ?

>you have to state whether you
>are running the zombies in the 'real' world or in the simulation.

In the real world, otherwise it isn't interesting (well, to me, anyway).

>In the 'real' world, the question is how much memory can you
>put in the brain

Part of my question implied doubt that *you* (anyone) can *put in* a lookup
table answering to the task. Unless one fell off the back of a god's truck.

> 10^28 bits
>Compare that with the "recall" memory of humans of
>less than 10^9 bits ("recognition" memory may be higher).
>So you can *theoretically* have walking around a lookup table
>of human memories of 10^19 bits.
>The recall times can be *very* fast for the reasons I mention
>in the response to Lee. Whole branches of computer science
>are based on concepts of data caches, locality of reference,
>organizing it hierarchically, multiple simultaneous accesses,
>etc. We know how to organize this stuff.

Great! Now go and find the sequence of canned bits you need *right now*,
using as many caching levels or whatever as you like. Quick, the impromptu
poetry contest is about to be won by that toad Blackadder--

>> BTW, this possibility seems to me to fall over dead in just the way that
>> Skinner's behaviorism did when famously assailed by Chomsky

>Could you elaborate on this a bit further? I'm not connecting
>the references and the discussion.

Sorry, I'm taking some shortcuts through the lookup table. Nearly 50 years
ago, Noam Chomsky (it is widely held) showed that finite-state grammars
were unequal to the task of explaining productive, creative human speech,
because we don't say one word then look for another and another after that.
Does caching solve this? Maybe it does; certainly Chomsky's
transformational grammar led people to expect the linguistic apparatus to
be modularised, which was later borne out by direct scans.

Chomsky also wrote a brutal review of one of B F Skinner's last books,
showing why empirical behaviorism failed on similar grounds. You can't see
the chunkiness of the world unless you're prewired by evolution to look for
the joints where nature is most readily carved (a view argued earlier by
Kant); that is, the mind can't be a tabula rasa. Few on this list would
deny that. But that doesn't mean the templates come pre-stocked, far from
it. Input and output require *generative* processes that Chomskyans would
say exceed any look-up table that we could realistically anticipate (in
nature). Stuart Kauffman, BTW, claims in INVESTIGATIONS that a biosphere is
emergently complex in *transfinite* measure, which might have some bearing
on this (or might not).

As I understand it, models of cognitive function these days assume at least
two classes of storage and access: some lookup tables or something like
them (where the brain checks the index and goes straight to the answer),
and some constructive imagery procedures where, eg, one wanders around on
an imaginary map looking for the answer (the time taken being a linear
measure of the amount of `mental space' walked through). I'm uncertain how
Freeman/Hopfield-style attractor models fit into this: I assume that model
is what neurally underlies the `lookup table' metaphor, even though it's
instantiated in a quite different structure from a computer, and works by
completely different principles.

Chomsky vs Skinner is briefly discussed in Howard Gardner's THE MIND'S NEW
SCIENCE.

Damien Broderick



This archive was generated by hypermail 2b30 : Mon May 28 2001 - 09:59:43 MDT