Thursday, August 27, 2015

Stop talking about the "computer analogy" -- It was never a computer analogy!

From the very beginning cognitive scientists were clear that the analogy is the computer program not the computer itself.  Even Newell et al. made this point in their Logic Theorist paper.
Excerpt from The Myth of Mirror Neurons:
Newell and company were careful to point out that their theory does not imply that humans are digital computers, only that humans appear to be running a program similar to LT. 
We wish to emphasize that we are not using the computer as a crude analogy to human behavior—we are not comparing computer structures with brains, nor electrical relays with synapses. Our position is that the appropriate way to describe a piece of problem-solving behavior is in terms of a program: a specification of what the organism will do under varying environmental circumstances in terms of certain elementary information processes it is capable of performing. This assertion has nothing to do—directly—with computers. Such programs could be written (now that we have discovered how to do it) if computers had never existed.  A program is no more, and no less, an analogy to the behavior of an organism than is a differential equation to the behavior of the electrical circuit it describes. Digital computers come into the picture only because they can, by appropriate programming, be induced to execute the same sequences of information processes that humans execute when they are solving problems. Hence, as we shall see, these programs describe both human and machine problem solving at the level of information processes.


ikbol said...

The problem with the program analogy/paradigm is that programs don't run without a human mind to drive/operate them. Humans are clearly bringing a kind of intelligence, essential for real world functioning, that algorithms don't have. But cognitive science in practice treats programs as autonomous, independently operating, which they are not, as a matter of undeniable fact - and treats humans as run by their programs.

We need - can't do without - algorithms as an analogy, because we clearly do have all kinds of automatic routines.

But human-computer systems are a much better analogy of how the human conscious mind incorporates routines into its activities, than computer programs alone.

A human wordsmith using a word-processing program is a much better analogy of how humans write texts (with/without computers) than a wordprocessing program alone.

Also, the two parts of human-computer systems function completely differently. The word processing program is automatic. The human wordsmith composing a text consciously is deliberate, stuttering, struggling and anything but completely fluent like the program.

Proceeding on this basis should produce a great number of ideas for further neuroscientific basis which the pure program analogy excludes.

William Matchin said...

There is an interview with Chomsky somewhere where he explicitly rejects the idea that any of his ideas were born out of an analogy with computers. I'll have to track it down, but he was quite emphatic and annoyed that everyone kept recasting the history of the cognitive revolution as scientists being impressed with the power of physical computers.

ikbol said...

William, Would Chomsky reject his ideas re syntax being dependent on logic? If that is true, it makes little difference - algorithms are an extension of logic and maths. And I dunno the historical facts, but the 1960s timing of his first ideas means it is very unlikely he wasn't heavily influenced by burgeoning AI, whether he was aware of it or not.

P.S. The whole of linguistics is massively dependent on logic - its whole terminology implicitly assumes that words can be treated as words/symbols irrespective of their real world referents, (which is wildly wrong) - in the spirit of logic.

William Matchin said...

He would not reject this idea, and he has explicitly stated that linguistics grew out of research in computation (i.e. Turing and Church), rather than the progress in building physical computers.

I don't think anyone is disagreeing here. The point of Greg's post is that the embodied folks are arguing against a straw man - cognitive scientists weren't making an analogy to physical computers, they adopted computational theory.

ikbol said...

I don't really see the importance of the distinction between program and machine. I think the imnportant issue here is that all programs to date are symbolically/logicomathematically based, and presuppose that thought can be disembodied, contra cog. embodied sci. There are no truly CONCEPTUAL programs, only ones that handle VARIABLES like those of logic & maths. Living creatures are conceptual systems/computers. Handling concepts in the real world - being able to think in terms of HAND ME THAT KNIFE - LET'S GO HOME etc is a necessarily embodied business. Concepts are creative and can embrace the endless new objects or actions of a kind met in the real world - endless new HAND-shapes, KNIFE forms, new forms of GOING-travel. Current computer programs with their concept graphs can't embrace a single new form. This is one of the central unsolved problems of AGI. To be able to imagine endless new forms of any given body or action, you have to have a body that itself can adopt endless new forms. There is no true thought that is not embodied.