Re: Qualia Question



Curt Welch wrote:

> You haven't been around enough to hear my thoughts I guess.
>
> Intelligent human behavior simply comes about because we are learning
> machines. We learn though simple operant conditioning. Pain and pleasure
> are fundamental to the learning processes. It's the purpose of the
> learning machine. It's the value measurement which the learning machine
> uses to select behavior. The "good" behaviors are the ones which are
> expected to produce the most long term pleasure and the least long term
> pain. Everything the body classifies as pain, acts as a punishment for
> behavior, and everything the body classifies as pleasure acts as a
> reinforcer to the learning system.

If you remove the teleological aspects implicit in "The "good"
behaviors are the ones which are expected to produce the most long term
pleasure and the least long term pain." then this would seem a
plausible model for behaviour in organisms of a certain kind - I'm
thinking of Sterelny's model for a stage in cognitive evolution that
maps many environmental inputs to many bavioural outputs as a 'step up'
from the one input (fire) to one output (recoil) that would describe
the Skinnerean machine I mentioned above. The physiological reactions
of pleasure and pain could be something like general measuring units in
the behavioral output decision process.

Including the teleological function suggests the kind of internal world
mapping (a virtual testing ground for potential behavioural responses)
that Dennet talks about. Is this the kind of model you are proposing?

I think both models might usefully describe ways in which the
physiological constituents of pain can be used to determine adaptaive
behaviour, but what neither of them do is account for (nor do they
require) the sensation of pain that I feel. These systems should be
just as happy functioning on the purely physiological aspects of pain.
They raise the curious question of: Why should pain be unpleasant?
Animal behaviour is affected by thousands negative input controls that
don't involve pain (or any conscious sensation at all) - why is pain
not like that?

> We are not directly conscious of the changes the learning hardware is
> making to our system. But if you simply give sensory inputs which monitor
> pain sensors, we become conscious of it.

Well if you give the sensory inputs to consciousness, yes we feel pain.
I just don't see consciousness described anywhere in the system you
propose.

> Hook the signal to an input, and
> we become conscious of it.

It would be nice if it were that easy, no?

> We react to pain in the ways we do because of the fact we are a learning
> machine motivated by pain. The way it "feels" to us is a result of how we
> learn to react to it.

I'd agree. But I don't think you've got anywhere near explaining why we
feel this as 'pain' rather than the learning system simply taking its
cues form the physiological data available to it. Why does pain hurt?
Why is pain not sufficiently described by the neronal activity that
goes with it?

.

Relevant Pages

  • Re: Hey Curt: Behavioral Systems Theory
    ... All "motivational drives" can be mapped to these two ... There is no other type of learning at the lowest level. ... An increase in the behavior of using pain the increase pleasure. ...
    (comp.ai.philosophy)
  • Re: Qualia Question
    ... > If we could arrive at an explanation of our senstaion of pain in terms ... Some form of Skinnerean reflex that caused the organism ... > differentiation' but arriving at consciousness is the goal of this ... Intelligent human behavior simply comes about because we are learning ...
    (comp.ai.philosophy)
  • OT; Sore Back, Pay Attention
    ... Over 100 million people suffer from back pain in one form or another. ... Many doctors agree that back pain can be minimized by learning how to ... They place their elbows on their knees and lean ... put excessive stress on your lower back and hips. ...
    (soc.culture.german)
  • Re: brain for AI
    ... Pain is a high level "opinion" the brain can have about itself ... And a love affair with the thalamus be overdone. ... our learning is simply a blocking or selection by the thalamus ...
    (comp.ai.philosophy)
  • Re: brain for AI
    ... Pain is a high level "opinion" the brain can have about itself ... And a love affair with the thalamus be overdone. ... our learning is simply a blocking or selection by the thalamus ...
    (comp.ai.philosophy)