Re: death of the mind.

From: Alex Green (dralexgreen_at_yahoo.co.uk)
Date: 09/01/04 

Date: 1 Sep 2004 02:44:26 -0700

Wolf Kirchmeir <wwolfkir@sympatico.ca> wrote in message news:<CA0Zc.9839$7i2.526314@news20.bellglobal.com>...
> Alex Green wrote:
> [...]
> > The reason I used this example was that there are three sets of
> > information present: the input information, the action, the effect of
> > the action on the environment. The effect of the action is obviously
> > not encoded in the input, it is also not encoded in the movement of
> > the paw because the SAME MOVEMENT occurs in all 3 cases. The
> > behaviourist knows the effect of the action by observing it but where
> > is the effect in the system input->cat->paw movement? It must be
> > modelled in the cat. As the behaviourist will tell you, the cat moves
> > its paw to kill the mouse, or move the ball, it does not just lob the
> > paw into space with no intention.
>
> The mouse is part of the input, Alex. It's one of the contingencies that
> controls the cat's behaviour. And the effect of the action feeds back to
> the cat - so it _is_ "encoded in the input". Etc. IOW, your analysis is
> simplistic and incomplete.

This seems to be equivalent to suggesting that the cat must touch the
mouse before it can decide to kill it.

This example of cat and mouse was brought to mind by a very large
Ginger Tom who moved into my workplace about 2 years ago and has been
in residence around lunch break ever since. He was shaking his rear
end slowly and peering round a drain pipe. He suddenly struck out with
his paw and a mouse went flying up in the air. It fell to earth
apparently lifeless and the cat pounced on it and ran off with a pink
tail protruding from its mouth.

There are two views possible for the cat's behaviour. In the
behaviourist view everything but a plan of action is subject to
feedback control. The muscles are allowed to be controlled by internal
feedback from spindles and tendon organs and the spinal ganglia by
central control systems. But in behaviourism the cortex is not allowed
to operate an internal control system. Even though the cortex and
thalamic nuclei have a similar relationship to the cerebellum and deep
nuclei. In contrast, in the physiological/cognitive view the cat forms
a model based on sensory input and this can be used for predicting the
world (planning).

The argument that the cat does not have an internal model is not
consistent with the way that the cats paw is directed at killing the
mouse. If the cat contained no model of "death of mouse" the sight of
the mouse would trigger a reflex response, not a hunting plan
continually modified by circumstance that involves hiding behind
drainpipes.

Weak behaviourism occurs, the sight or sound of the mouse triggers a
hunting 'plan' that is a set of skilled actions, not a hunting
'response'.

>
> The behaviorist (me) will tell you that "moving the paw" is not what's
> being explained. It's "moving the paw in the direction of the
> mouse/ball/etc" that's being explained. The behaviorist (me) will also
> point out that the cat's own movements/etc are all part of the "input"
> as you so quaintly put it. And so on.

It is the directedness of the paw that demonstrates that there is a
model in the cat's brain. The only information that the cat is
communicating to the behaviourist is the flight of the paw. The death
of the mouse is environmental information. The fact that the cat aimed
the paw at the mouse shows that it contained this information (death
of mouse) before the death, as a model. Furthermore and most
importantly, the cat used the environmental information plus the
intention to direct the paw. It modelled the flight of the paw to a
particular place for a particular purpose and all the biological
servosystems in the cat kept the paw on this course.

> Why is the behaviorist attitude to explanation of cat behaviour useful?
> Because so long as you focus on "intentions" and link them to "internal
> models", you are likely to overlook a slew of environmental factors
> (contingencies) that affect and control the cat's behaviour. The cat and
> its environment constitute a system.

I think weak behaviourism is a good approach to behaviour. An ill
treated dog will instantly cower when an arm is raised. There is no
need to explore the dog's cerebral modelling to understand this
response. It is only the dog lover who might attempt to do so but the
cowering could be extinguished most rapidly by always having fresh
meat in the raised hand.

The fact that behaviourism works as a crude understanding of animals
does not mean that there is no internal model. It just shows that the
animal's brain can be represented by a black box with nothing of
significance to the trainer inside it.

>
> There are no doubt internal states that change in response to the cat's
> environment. However, even if we grant that those state changes somehow
> result in observable behaviours, we cannot infer them, since different
> behaviours may be caused by the same or similar state changes.

>From a physiological viewpoint I would simply assume that except in
lower brain/spinal reflexes the cortex has nearly always modelled the
behaviour. I would infer the nature of the inner model from the
environmental state change it produces. The other approach, where
environment alone causes behaviour ignores the timing of neural
events, it can be 0.2 secs or more before the brain can respond to an
event. The prey would escape if the predator did not implement a plan
based on the execution of two or three skilled sequences and placed
itself in the correct location for these to be effective. It also
ignores the directness of actions and lumps the action and its
consequence together as 'behaviour'. Consequences are goals in a
cerebral control system, not reflexes.

> (In
> logical terms, you can't infer a consequence from an antecedent.) Nor
> can we infer how those state changes result in behaviours. Neurology
> will provide answers, or at least help, I think, but they won't be
> obvious ones. Labelling those internal states as "intentions" or
> "models" isn't helpful, since such labels bring with them all kinds of
> implicit assumptions, many of which are certain to mislead us. For
> example, one of my concepts of model is "something that is functionally
> similar to the prototype, such that functions of the model can be used
> to describe/predict functions of the prototype." That could be anything
> from a material object (a scale-model of a ship in a testing pool) to an
> abstract sytem of concepts (a set of mathematical equations describing
> the motions of the planets.) It might even be a neural net, though at
> this stage we simply don't know enough to be able to say so with any
> confidence.

This is a reasonable definition of 'model'.

> And if a neural net does in fact "model" (cat + mouse), it
> does so in ways that must be unlike the models we are accustomed to.
> Why? Because in this case the neural net is itself part of the system it
> models....

I cannot see the problem with the neural net modelling the effects of
its previous output on the environment. No one would suggest that the
neural net simply feeds back its output as an input. This
reverberation does not occur in any control system except one that has
a fault. Your objection that the neural net would model its own,
immediate output can be overcome by pointing out that such a thing
would have no purpose and is not observed, the neural net models the
animal in the world.

>
> Much of human psychology as "applied" in psychotherapy suffers from the
> same defect - there's all kinds of talk about attitudes, beliefs, etc,
> while what's actually happening in a given situation is being ignored.
> One of the consequences of this is that psychotherapy fails more often
> than it cures.

Weak behaviourism is a good approach. The brain models the organism in
the world, not itself, so various forms of desensitising/conditioning
are an excellent approach to therapy, particularly where reflex
responses are occurring such as panic attacks etc. However, idealistic
behaviourism, where the very possibility of an internal state is
rejected, seems more like a religious or political agenda (cf: some
forms of 'Marxist consciousness').

Best Wishes

Alex Green

Relevant Pages

  • Re: death of the mind.
    ... > the paw because the SAME MOVEMENT occurs in all 3 cases. ... As the behaviourist will tell you, the cat moves ... its environment constitute a system. ... behaviours may be caused by the same or similar state changes. ...
    (sci.cognitive)
  • Re: death of the mind.
    ... mouse" even though the movement of the paw may have been the same as ... Certainly if the mouse is killed and the cat eats it then the cat will ... probably repeat the action when the environment is the same again. ... >> behaviourist knows the effect of the action by observing it but where ...
    (sci.cognitive)
  • Re: death of the mind.
    ... > the action on the environment. ... What about the effect of the cat's action on cat? ... > the paw because the SAME MOVEMENT occurs in all 3 cases. ... internal state to you, and call that "your intention"; ...
    (sci.cognitive)
  • Re: death of the mind.
    ... paw is a direction, a momentum etc. ... present as an intention in the cat. ... The idealistic behaviourist who denies internal states in the cat ...
    (sci.cognitive)
  • Re: NBC: everybody is entitled to their opinion...
    ... pet dog is more than double that of a gas-guzzling sports utility vehicle. ... the Stockholm Environment Institute in York, Britain, to calculate ... so why shouldn't I be allowed to have a little cat to alleviate my ... "Our animals give us so much that I don't feel like a polluter at all. ...
    (rec.music.artists.springsteen)