Re: brain thoughts

On Nov 24, 10:58 am, c...@xxxxxxxx (Curt Welch) wrote:
casey <jgkjca...@xxxxxxxxxxxx> wrote:

You imagine it is nothing but conditioning. So how is
conditioning achieved in a neural circuit?


Well, that's a good question for the neuroscientists.
But frankly, I don't give a shit because I already
have it working with transistors and I see no reason
to switch to trying to build AI out of neurons.

I wasn't suggesting you build AI out of neurons.


But how do we find the mechanisms of natural selection
or reinforcement learning?


John, it doesn't need to be found. We already have it
it working in 100's of examples already. You are the
only one here that seems to "need to find it".


We know how it's implemented.


We don't know how RL is implemented in a cat let alone
a human otherwise we would have copied it. The mechanisms
used in TD-Gammon or balancing a pole do not scale up.

You can see something like RL as being a solution and yet
be mistaken about what is required to actually implement
it at a higher level. Darwin for example thought that
inheritance was the result of analog blending which in
fact wouldn't have worked, it had to be digital.


The only thing wrong with "stimulus response" is that
when you use that term, most people think the "stimulus"
is a short term event (like a light flash). In truth,
the stimulus that drives human behavior is EVERYTHING
that has happened to the brain since birth. As long as
you understand "stimulus" in theory extends back to the
machine's creation (all past inputs), then stimulus
response FULLY explains brain behavior and is not in
the least bit simple. It's actually "overly complex".
In other words, if you actually understand stimulus
response, you would understand it's not possible to go
beyond it.


Although there is a lot we don't know about the brain it
seems churlish to ignore what we do know. None of it
reflects the views you have expressed about how nodes
should behave or how they should be wired.

The chicken wire network of nodes of the hydra brain did
not evolve into anything more complex. More behaviorally
complex brains for some reason formed clusters of neurons
called ganglia which are asymmetrical in their distribution.

And if it is simply "conditioning all the way up" why
didn't the old conditioning circuits just get bigger?
Why did new networks with different circuitry evolve?


JC
.

Relevant Pages

  • Re: A request for information please.
    ... Its all fine to say that you have designed a robot, ... The brain picks up on all the sensory data it's been receiving. ... The decision to speak a word wasn't a result of just one stimulus (like ... and possible when we have so little first hand experience of the ...
    (comp.ai.philosophy)
  • Re: What is conditioning?
    ... Kirchmeir's argument about innate walking ... There are all kinds of innate learning ... that you would consider is an example of conditioning ... A "simple" stimulus like a bell or odor also requires ...
    (comp.ai.philosophy)
  • Re: A Touch of Thermo
    ... doesn't understand conditioning I am not in a position ... and it has been a while, "Reinforcement Learning" ... way you train a dog, then you need to know in some detail ... time behavior comes under "stimulus control", which it must, ...
    (comp.ai.philosophy)
  • Re: Consciousness: whats the problem?
    ... A "viewing" behavior. ... and not all dancing is a particular dance. ... static image in the brain as it would be in a camera. ... a result of someone else "conditioning" me, ...
    (comp.ai.philosophy)
  • Re: Turing Machines and Minds (Was: The Demise of Computationalism?)
    ... could be elicited from a stimulus. ... this stimulus and the repertoire of responses could be ... Er, no, that's not operant conditioning as I understand it. ... the thermostat only the complex behavior of an animal. ...
    (comp.ai.philosophy)