|
|
|
Contents |
|
I have said that connectionist nets are good at grouping things together in a way in which we would intuitively group them together, and even at simply learning our categorization. For example:
Put on the input layer a female face plus code for female and teach it in the usual way to produce code for female head on the output layer.
The repeat for a number of female faces.
Put on the input layer a male face plus code for male and teach it in the usual way to produce code for male head on the output layer.
The input a new face.
You will find that it outputs with some reliability the correct categorization male or female.
This is just a property of some nets. You don't program anything in besides what I have just itemized. We don't know in detail how it does it.
In particular we don't have to teach it any rule on the model: "if a face has such and such a feature, and such and such a feature .... it's probably female." You just teach it the gender of a number of faces, and we find that hey presto the net has acquired the capacity to tell whether a face is likely to be male or female.
Just notice that in a von Neumann computer you would have to give it rules on the model I've sketched.
Fodor:
"It seems to me to be among the most important findings of philosophical
and psychological research over the last several hundred years (say, since
Locke first made the reductionist program explicit) that attempts at conceptual
analysis practically always fail.... There is 'bachelor', which is supposed
to mean 'unmarried man'; ...there are jargon terms, which are explicitly and
stipulatively defined; ...there is a handful of terms which belong to real,
honest-to-God axiomatic systems; ...and then there are the other half million
or so items that the OED lists. About these last apparently nothing much can
be done." (Fodor, "The present status of the innateness controversy",
in his Representations, Cambridge, Mass, Bradford Books/MIT Press,
1981, pp. 283-84)
| Different views of what a concept is |
And then notice how very difficult it always is to say what the rules are which govern a word or concept. It's frustratingly difficult to define such concepts as democracy or nationalism or beauty or topiary. But it's just as bad trying to give the rules which govern when it is right to use such words as knife or shirt or topiary.
It's almost as if our categorizations are not governed by rules at all. But what is the alternative? Could it be that the ordinary way we have of classifying things is to use connectionist nets to do the work that seems to come naturally to them? - ?
Whether a particular animal is classed as a fish, before we get analytic about it, is determined by how a tuned net responds to it. How has the net become tuned? By having lots of things put to it and having been taught by natural selection which ones to respond to and which to ignore.
We have spoken of folk psychology. Eliminativism proposes that folk psychology is a weak theory of human behaviour, destined to be replaced.
"Eliminative materialism is the thesis that our common-sense conception of psychological phenomena constitutes a radically false theory, a theory so fundamentally defective that both the principles and the ontology of that theory will eventually be displaced, rather than smoothly reduced, by completed neuroscience." Paul M. Churchland, 'Eliminative Materialism and the Propositional Attitudes', Lycan, 2nd edition, p.120.
Folk psychology is the name some theorists have given to our ordinary way of talking about human actions springing from what they want and what they believe will get them what they want. Why did I bring my keys? Because I want to get back in the house, and think the keys will be necessary for that.
Fodor on the enormity of folk psychology's beng wrong:
"if it isn't literally true that my wanting is causally responsible for my reaching, and my itching is causally responsible for my scratching, and my believing is causally responsible for my saying. ..if none of that is literally true, then practically everything I believe about anything is false and it's the end of the world. ( Fodor, A Theory of Content and Other Essays, Cambridge, Mass, Bradford Book/MIT Press, 1990, p. 156; quoted in Stich, Deconstructing the Mind, New York, 1996, OUP, p. 169)
Folk psychology is deeply embedded in human life. It is at the foundation of all our social and legal arrangements, and also in most of what we say and think about ourselves and others. So the claim that it is simply a mistaken theory is a dramatic and far reaching one. Entertaining it is an exercise in thinking the unthinkable. - And of course invaluable for a philosophical person for that reason.
Eliminativists believe it will be replaced by the theory of the workings of the brain and the central nervous system which is emerging from neuroscience. We will come to explain and predict what we do - better, our behaviour - not in terms of beliefs and desires, but in terms of neurones, synapses, ganglia, neurotransmitters, etc.
Peter was hypnotized. Under hypnosis he was given instructions about what he was to do once he had been released form the trance.
In his paper in the Reader, Paul Churchland spends the first part trying to show that folk psychology is indeed a theory. And that of course is the first critical step. It was a novel and surprising suggestion that our ordinary way of thinking about action was a theory. You may find it surprising that the suggestion has got established and gained ground. Churchland's article sprouts equations in the course of his discussion of this point. It is skippable if necessary. Pick it up again on page 123 (2nd edition; 1st edition p. 210) the beginning of Section 2 (Section II).
But he is only saying that folk psychology yields law like statements like this:
For all x and for all p
if x fears that p then x desires that not p
or
For all x and for all p
if x hopes that p and x discovers that p then x is pleased that p
He then argues that when you think about it is a poor theory. It fails in the following areas for example:
Prompt: Review this list. Do you agree that folk psychology has no light to throw on any of them?
Some people respond by saying that they wouldn't expect Folk Psychology to throw light on these things.
Do you agree?
Plenty of theories that seem common-sense at one time are abandoned later.
To suggest that folk psychology will be abandoned is not absurd.
Two fates for theories like folk psychology::
Reduction of one theory to another is said to take place when there are equivalent concepts in either theory - those in one correspond to those in another.
This is the kind of relationship envisaged by the identity hypothesis. There is an alleged equivalence between lightening and electrical discharge, clouds and moisture droplets, consciousness and brain activity.
| Brain implant may restore memory |
Eg Caloric theory
There was a theory of heat which postulated that it was a fluid - a special fluid, one a fluid which couldn't be felt, and didn't weigh anything, but occupied space. So the expansion of bodies when they became hot was accounted for, and the fact that heating a body didn't in itself increase their weight.
What replaced this theory was the hypothesis that heat was a form of energy.
Paul Churchland in the Reader article says that Folk Psychology could be heading for either fate.
Later he favours the latter. It is set to be eliminated by neuroscience.
This is eliminative materialism.
"Folk psychology suffers explanatory failures on an epic scale...it has been stagnant for at least 25 centuries, and ... its categories appear ... incommensurable with ... the categories of the background physical science whose long-term claim to explain human behaviour seems undeniable."
Churchland, in Lycan, p.124 (2nd edition, 1st edition p. 212).
END
Revised 24:02:03
