Newsgroups: comp.ai.philosophy
Path: cantaloupe.srv.cs.cmu.edu!das-news.harvard.edu!news2.near.net!MathWorks.Com!news.duke.edu!convex!cs.utexas.edu!uunet!psinntp!scylla!daryl
From: daryl@oracorp.com (Daryl McCullough)
Subject: Re: Is Common Sense Explicit or Implicit?
Message-ID: <1994Sep26.114409.4876@oracorp.com>
Organization: Odyssey Research Associates, Inc.
Date: Mon, 26 Sep 1994 11:44:09 GMT
Lines: 153

pindor@gpu.utcc.utoronto.ca (Andrzej Pindor) writes:

>In article <1994Sep21.131455.3228@oracorp.com>,
>Daryl McCullough <daryl@oracorp.com> wrote:
>>pindor@gpu.utcc.utoronto.ca (Andrzej Pindor) writes:
>>(in response to DrewDalupa <drewdalupa@aol.com>)
>>
>>>Why did a rock slide down a hillside? Perhaps it had a desire to lower
>>>its potential energy? Are you saying that "By denying yourself 
>>>the explanation in terms of desire, you leave yourself with that much 
>>>less understanding of the world around you"?
>>
>>Assuming that you don't mean your question about rocks to be simply
>>rhetorical, I think I can explain how the explanatory value of desires
>>is much better for people than it is for rocks. In the case of a rock,
>>saying that the rock desires to lower its potential energy is no more
>>informative than simply describing the behavior directly (say, using
>>physics). In contrast, consider a statement such as "The man bought a
>>gun because he wanted to kill the pesky raccoon." Surely, you must
>>admit that this explanation in terms of desire or purpose is much
>>better, and more full explanation than saying: "The man bought the gun
>>because such and such neuron fired in this or that part of his brain..."
>>
>You have created a strawman here: since we are unable (at least at present)
>to provide such a 'more full explanation' - so the other one is 'better'.

My point is that any "first-principles" account of human behavior,
starting with chemistry, or even quantum electrodynamics, is going to
be such a strawman. We already know what the first-principles account
goes: there is no reason (as far as I know) to think that there is
anything beyond ordinary electrical forces at work inside a human
brain. The only thing lacking is a way to cut through the
mind-boggling complexity of the interactions of billions and billions
of atoms to get anything like a predictive theory. Well, in my
opinion, we already have such a theory, which is folk psychology.

>>Using folk psychology to describe human behavior in terms of desires,
>>beliefs, emotions, and plans provides *enormous* explanatory and
>>predictive power. Of course, there are limits to folk psychology; we
>>can't deduce from first principles what desires a person has, or what
>>beliefs they hold, and perhaps a more reductionist approach would give
>>us a better understanding of such foundations. I don't believe that a
>>fuller understanding of how human brains work, even (or especially)
>>down to the level of atomic structure would ever supplant folk
>>psychological explanations.
>>
>Depending what you mean by 'supplant'. Knowing what are physico-chemical
>reasons of certain behaviors is already being utilized, for instance by
>certain medication, prozac being one example, or pharmacological treatments
>of sexual offenders - another. In these cases a particular behavior is
>explained not by beliefs or desires, but by a chemical state of the brain.

That's exactly what I do *not* mean by supplant. What you are talking
about is using physico-chemical explanations as *complementary* to
folk-psychology. There is no conflict between an description in terms
of beliefs and desires and a description in terms of chemistry. There
are low-level and high-level descriptions of phenomena, and they are
not in conflict. You are missing a key part of the picture if you only
see the low-level.

For example, if you are observing a computer chess program in action,
then (assuming the program is working well) one would describe the
resulting game using something like "The computer took the pawn so as
to open up a path for his rook". A description in terms of programming
language constructs "It took the pawn because it encountered an
if-then-else statement, and the condition turned out to be true" is
pretty worthless in normal cases. In unusual cases, for example, when
the computer does something bizarre, one might turn to a lower-level
explanation, because the high-level explanation breaks down. The same
thing is true of human beings. In normal circumstances, the high-level
explanation of behavior in terms of folk-psychology is the right level.

>Physical theories get any respect only if they cross
>bounds of a particular field in which they were postulated as an
>explanation.

I'm not claiming that folk psychology is respectable. I only claim
that it is useful, and that everyone uses it, even if they are too
embarrassed to admit it.

>If a theory is unable to cross such bounds or runs
>into contradictions when extended outside these bounds, it is
>abandoned (e.g. theory of ether). In case of behavior, beliefs,
>desires etc. present such flexible combination that nearly any
>behavior can be explained, perhaps by postulating yet another
>belief. And if in spite of this flexibility you cannot come up with
>an explanation in terms of beliefs etc., you can always say "he/she
>is out of his/her mind :-) (or drunk, etc). So the theory is
>unfalsifiable.

It isn't supposed to be a physical theory. We already *have* the
physical theory which describes human behavior---quantum
electrodynamics.  Folk psychology is a metatheory, a framework for
describing behavior, and it is an enormously useful
framework. *Particular* psychological explanations are falsifiable. I
gave the example of the explanation of DrewDalpa's behavior of going
to a store because he is infatuated with the check-out lady.

>>>Desires explain behavior, but we know just from behavior what the
>>>desires are! Isn't this sadly circular? 
>>>
>>No, it is *happily* circular. Let me try on the corresponding claim
>>about, say, physics:

>> 
>> Some folks (Einstein, Newton, among others) claim 
>> that there are these "physical laws" governing the
>> universe, and that if we only knew these laws, we 
>> could predict and understand our observations about the 
>> universe. But how do people know what these laws 
>> are? I am sure that the only way is from observations. 
>> Then, however, what is the explanatory power of the 
>> notion of "physical law"? Laws of physics explain 
>> observations, but we only know from observations what 
>> the laws are. Isn't this sadly circular?
  
> See above.

I think my analogy stands pretty well. Folk psychology is not
to be compared with a particular physical law (such as the law
of universal gravitation) but with the metaphysical assumption
that the universe is described by physical laws. This metaphysical
assumption is not falsifiable, although particular theories within
the frame work are.

Right, and it was useful to that extent. I don't have any disagreement
with you about objectivity. A belief is a theoretical construct, not
a physical object.

>>I don't have any problem with the idea that a robot has
>>beliefs. However, the whole point of ascribing beliefs and
>>desires is that it provides a very compact way to both explain past
>>behavior and predict future behavior (in novel contexts). If the
>>folk psychology explanation of the behavior simply restates the
>>behavior, then it is not worth much.

>OK, if you do not have problems with robot's "beliefs" then we have
>less disagreement that I've thought. However, I think that most
>proponents of beliefs, desires etc, would protest. Yes, there are
>regularities in behavior and correlations between various behaviors,
>and beliefs etc. do have a role in a 'folk' context, but perhaps not
>in a scientific context.

Oh, I agree completely that folk psychology is not science. It is
metaphysics. However, to me it is folk psychology that makes the
scientific study of the mind a distinct field. To me the applicability
of folk psychology is what distinguishes things with minds from things
without minds.

Daryl McCullough
ORA Corp.
Ithaca, NY

