Re: Neural darwinism (Re: Human intelligence is obviously absurd)

From: Michael Wilson (
Date: Mon Jan 31 2005 - 11:05:05 MST

> The brain's construction of chained memories and actions is probably
> another tree, though a more functional metaphor might be the
> candelabra-shaped railroad marshaling yard, with words for cars

Firstly Calvin's and Edelman's versions of 'neural darwinism' aren't
the same. You're talking about Calvin's runtime DE, wheras Edelman
focuses on selection of neuron group with useful patterns, which more
closely resembles the way the human immune system operates than the
multi-generational selection and mutation needed for evolution.

Edelman's stuff is quite plausible, since it beings to explain the rate
and dynamics of neural dieoff and how the genetic encoding of human
intelligence achieves such remarkable compression. OTOH while Calvin's
ideas struck me as reasonable when I first read it, I can now see that
it just wouldn't work as a cognitive design. As a method of primary
cognition evolution is both radically inefficient in wetware and (when
adjusted for plausibility) time requirements, plus if it's anything
like the DE I've experimented with unstable and prone to severe
reasoning pathologies (which don't match actual, observed flaws in human
cognition). The 'evidence' for Calvin's view can in fact be explained by
a much less radical view of cognition; simply the inevitable consequences
of the massive, stochastic parallelism required to operate at such low
clock rates and reliabilities in the first place. When the brain implements
multiple action generation pathways triggered by the same context and
funneling into the same motor control area, it must by necessity place
them in parallel with each other and the logic that will decide which
subnetwork (equivalent to a Minsky agent here) to permit to take control
of the output. That control logic is generally pretty good; we don't often
attempt to issue two contradictory responses at once. Sometimes however if
the decision is finely balanced both outputs may be active for a short
while before the unstable feedback loop of cross-suppression causes one to
be selected. This is a simpler and plausible explanation of the behavior
Calvin attributes to competition between patterns, and the mechanism also
extends to cover several other common errors.

> imagine that many trains are randomly constructed on the parallel tracks.
> but only the best is selected to be let loose on the 'main track' of
> consciousness and speech.

That's a good metaphor, but I can't imagine why it makes you think of
armies of copied thoughts competing for neural real estate instead of
(a neural version of) several speculative logic chains that operate in
parallel with the control logic that selects between them; after all
the latter is a common feature of modern microprocessors where clock
cycles are again at a premium.

 * Michael Wilson

ALL-NEW Yahoo! Messenger - all new features - even more fun!

This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:50 MDT