Re: Skinner

From: Harnad, Stevan (
Date: Mon Nov 20 1995 - 21:20:04 GMT

> Date: 20 Nov 95 22:26:33 EST
> From: Judith Economos <>

sh> Ask yourself this question: If someone were trying to build a robot that
sh> could do what we can do -- get about in the world, learn, categorise,
sh> name, describe, respond to descriptions -- how much would it help him in
sh> his project if Skinner told him that the robot's behaviour is shaped by
sh> its consequences, by its history of rewards and punishments?

je: It would suggest to him that he didn't need to know how it was done
je: if he'd build an alife machine and put it through evolution and
je: select ruthlessly for certain kinds of behavior, and got
je: categorization: that's how it would help.

By fumble-and-find combinatorics? Lots of luck! You may as well tell a
would-be airplane designer (before the Wright Bros.) the same thing:
Here's how you do it: Encode the environment (somehow), encode
the genome (somehow) and set Alife going: It'll eventually evolve the
Wright Bros., who will come up with the plane design...

I think a roboticist would want something more, ahem, helpful than

je: It would say "I don't
je: know the chemistry of building an apple pie out of atoms, but I
je: can give you the recipe for making an apple pie out of apples and
je: sugar and salt and so on (there's a lot of so on)".

A levels-of-analysis problem here: People make apple pies, and if the
question is how to make an apple pie, the answer's around. Nature (so
far) is the only maker of systems with our behavioural capacities, so
no one has the recipe. The recipe need not speak of atoms, just what it
takes to make systems that can do what we can do. Ditto for airplanes
(to be able to fly one, or to be able to give its flight history, is
not to know how it works).

je: I agree with you that
je: Skinner was trying to rescue psychology from a mess of subjective,
je: unverifiable, speculative and bootless "theorizing". As is usually the
je: case with revolutionaries (reactionaries) he went too far, and threw out
je: possible babies with the dirty bathwater. His job was to see how far you
je: could take descriptive, observable science of animal behavior and I
je: expect he would say it could still go farther. But while there is
je: no reason some of the plenitude of students should not pursue his
je: program if they want to, it is probably out of date and out of favor,
je: and most of them will probably want to use more modern techniques.
je: But this does not make him less than salutary and bracing to study,
je: and he did some memorable experiments with amazing results. Besides
je: which, he turned into a world-class eccentric as good as Russell
je: became.

I am less impressed by what Skinner did and found. Neither his findings
nor his concepts were very broad or deep; their reach far exceeded
their grasp, while giving the opposite impression. Rather than stay in
that rut, I think it's better if students keep the little from the
behaviourist legacy that was valid (that the psychologist's data will
always be behavioural data) and reject the rest (that the data would
speak for themselves, and that there is something wrong with
theorising about internal causes: indeed, theorising about internal
causes is what it is all about, as long as the causes are physical and
functional, rather than mental).

This archive was generated by hypermail 2b30 : Tue Feb 13 2001 - 16:23:56 GMT