Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!ucbvax!ucsfcgl!cca.ucsf.edu!daedalus!brianc
From: brianc@daedalus (Brian Colfer)
Newsgroups: comp.ai
Subject: Re: Free will and responsibility.
Summary: QM+CT, behaviorsim, AI, etc.
Keywords: Behaviorism, materialism, dogma, science
Message-ID: <2134@ucsfcca.ucsf.edu>
Date: 22 Jun 89 22:17:42 GMT
References: <10333@ihlpb.ATT.COM> <3850@uhccux.uhcc.hawaii.edu> <52019@linus.UUCP> <533@orawest.UUCP> <2586@rice-chex.ai.mit.edu> <386@edai.ed.ac.uk> <149@unix.SRI.COM> <421@edai.ed.ac.uk> <2095@ucsfcca.ucsf.edu> <234@unix.SRI.COM>
Sender: news@cca.ucsf.edu
Reply-To: brianc@daedalus.ucsf.edu (Brian Colfer)
Organization: UCSF Dept. of Lab Med
Lines: 264

In article <234@unix.SRI.COM> ellis@chips2.sri.com.UUCP (Michael Ellis) writes:

>> Brian Colfer > Michael Ellis

Michael, 

    -  statement 

is my way of summerizing my understanding of your argument.

I. Link between QM & CT
    -  The brain is inherently chaotic due to fluctations caused by QM
	   effects in the molecules of the brain.

>    If you are decide to seriously answer this article, please address this
>    point: Chaos theory says that minor fluctuations at points of bifurcation
>    in suitably complex systems have massive global effects. If those minor
>    fluctuations are genuinely random (as in QM, not merely the classical lack
>    of knowledge) then the global behavior also is *genuinely* random. Not just
>    "too many variables": If we knew *all* the variables and had a computer
>    of infinite capacity -- if we were Laplace demons -- we still could not
>    predict. 

Where is the evidence for a link between QM and any CT effect seen in
the behavior of the brain?  

	1. CT has not been definitively shown to describe the behavior of
	   the brain.  Can you refer me to any articles which state that
	   CT has been shown (rather than just preliminary findings) to 
	   describe brain behavior.

	2. If CT and brain behavior is true why do you say that QM is the
	   source of the small fluctuations.  I think it is a pretty big
	   leap to state that QM effects are of sufficient magnitude (I am
	   not refering to their quantity) to generate chaos.

	3. Not all physical systems are chaotic (in the CT sense) yet all 
	   physcical systems have QM effects.

	4. While one cannot exclude QM as a source of flux since it is 
	   ubiquitous other sources may be dramatically more important 
	   for the system.  For example, in describing the system of the 
	   brain one can easily imagine that a single uncontrolled (in an 
	   experimental sense) stimulus of sufficient strength as a source 
	   of flux.  

II. Determinism
	-  QM+CT insures that there is an inherent non-determinism in the brain.

>    Ultimately, none of that stuff really is determined in the sense that
>    you want. And knock off the "mystical forces" crap. Would you prefer I
[...]
>>I'm saying that according to QM the brain is **just** as determined as 
>>everything else.  If we deal with everything else (but the brain) as if it 
>>was determined why not treat the Brain the same way...
>
>    And that's hardly very determined at all.
...
Note my additions in []

>    I *am* treating the brain like anything else. 
....
[I especially like this statement] 
....
>	                                                 Nothing physical in the
>    universe is completely predictable. This doesn't usually matter for 
>    phenomena whose attributes we care about [which] are approximated by linear
>    differential equations, like most ideal objects in engineering
>    books, or, similarly, those artifacts we have consciously designed after
>    their example. It is a great testimony to the cleverness of science that
>    we have designed things that behave so predictably under the press of
>    a button.
[...]
>    With QM, even if you knew *everything*, you could not predict. Quantum
>    randomness is "metaphysical" in most viable interpretations. But the ardent
>    determinist however, might still sleaze out by asserting that quantum
>    randomness operates only at levels so small that it has no macroscopic
>    consequences.
>
>    Now chaos theory predicts that microscopic fluctuations at points of
>    bifurcation have enormous effects on macroscopic global behavior. This
>    is a *hard* prediction of chaos theory. Those who doubt this are urged
>    to refer to any text on chaos theory. So Chaos theory + Quantum mechanics
>    are both required to assert the metaphysical and epistemic randomness
>    of global behavior for suitably complex systems.

	1. I am not saying that the universe is completely predictable.

	2. I am saying that some set of the brain's behavior is predictable.

	I think that you will agree with this ... the question that we are
	likely to disagree about is degree.  I am merely speculating but I
	think you would say that the set of brain behavior which is
	predictable is either insignificant because of quality and/or quantity.

	Note: brain behaviors are responses of the brain to its environment.
	      The characteristics of these responses are shaped by the
		  interacting effects of genetic and personal history of the
		  organism. For more on this see the section on beliefs and
		  desires.

	I see behaviorism as a systematic exploration of this set of behaviors.  
	Whether this set is insignificant is an empirical question.  While
	the scope of behaviorism is narrow its power is unrivaled in psychology,
	save physiological psych.  I think that the role of contemporary
	behaviorists is to explore the limits of the behavioral model both
	in theory and practice.  
	
	I also see that some of the current limits is the ideological repugnance 
	held by some regarding the application of science to humans.  This
	is a political limit not a technical one.

III. Levels of analysis 

	- behaviorism is too reductionistic to be relevant for the real
	  problems of people 

In this part of my discussion I flag speific points that I refer
to with [N] where N is the number of my response.

	>>Either the brain is *all* physical matter or it isn't ... which do think?
	>>If it is then the same "laws" applicable to the rest of the universe 
	>>apply to it also.  If it isn't then you must evoke spirtual metaphysical
	>>crap. (Flame throwers on..? Let me get my fire suit)
	>
    > I'm not into the metaphysics of substance. Let's just say I prefer to
    > bracket that question, especially considering how incoherent the notion
    > of "material substance" has become this century *in the scientific 
    > community*.
    >
    > Anyway, even if mind/brain identity theories are correct, we are probably
    > in deep trouble where meaningful predictions are concerned.
    >
    > To take an example from Fodor: Consider the function that maps physical
    > objects to money. Some money is in the form of valuable metal. Other
    > is in the form of slips of paper. And then there are electronic blips
    > in computers, and so on. Even if all this physical stuff *were* 
[1] > metaphysically deterministic can we ever reasonably expect physical
	> theories which predict the motions of all those physical objects to
    > tell us anything meaningful about economics? Except for the grossest of
    > predictions ("the gold itself will not suddenly transform into some
    > other metal"), physics can tell us very little about economics.
    >
    >    Token theories of mind/brain identity are sort of like that. Just
[2]-> what in the brain would correspond to the belief that your car needs
    > a valve job or the desire that your lover would stop cracking her 
	> knuckles? Even simple beliefs and desires like these almost certainly 
[3]-> correspond to a disjunction of an enormous number of radically dissimilar
   -> brain states (assuming "brain state" even refers to something real). Brain
   -> states can get us to gross predictions ("thoughts cease when we run
    > over this brain with a steamroller"), but it isn't clear that brain state
[4]-> talk will ever get us to meaningful predictions about beliefs that are
[5]-> any better than folk psychology.
	>
    >In other words, brain state language deals with natural kinds that "go
	>together"; mind state language deals with yet other natural kinds that
	>"go together". Both languages really talk about the same stuff if that
	>makes you feel better metaphysically. But the two different languages 
	>don't "go together" anymore than physics and economics (sort of like
	>a nightmarish version of applying Nelson Goodman's predicate "grue" to
	>emeralds). The realms of discourse are autonomous in that even if economics
	>is theoretically reducible to physics, the reduction is humanly impossible 
	>and unknowable. Throw in what QM+Chaos theory and the reduction may well
	>be theoretically impossible as well.

I see that there is a difference in the relevance of different levels
of analysis.  Subatomic physics tells us very little about the working
of the brain.  This is the same sort of argument that Skinner made
when he described why he didn't think that phsyio-psych made behavioral
analysis irrelavent.   He said that who knows, some day someone may be
able to link up all the chemical/synaptic connections and explain the
micro mediation of behavior. What behaviorism essentially describes is a 
set of relationships between behaviors and their consequences.  He is
looking at the organism as a whole;  the analysis [reduction and 
catagorization] is of the behaviors not the organism.   

Specific points: (from above)

[1] Don't need to... physics is like physiology and behaviorism is like
    economics.  The important thing is that at any level of inquiry
	we are using systematic, rigorus methodologies for analysis.
	Although,  behaviorism has been much more empirical than
	economics.  In this sense it might be at midpoint between these two
	levels of inquiry.

[2] Probably some sort of complex synaptic response but we don't need
    to know this to predict and control this behavior.

[3] Discussions about brain states (CA, MA and maybe NM) is probably silly.
    This type of discussion should be on synaptic nexuses ... see the
	current issue of Scientific America...  but this is not my point 
	anyways... one does not need to know the micro-medaition of behavior 
	to describe contingencies of reinforcement.

[4] Probably a trivial point but, what do you consider a meaningful prediction?

[5] What meaningful predictions does folk psychology make?

IV. What are behaviors
    - ? 
	 I don't understand your position. You do not state that beliefs, etc.
	 are behaviors nor do state that they are not.  You say that's its 
	 obvious.

>>Are thoughts, beliefs, desires etc. behaviors?
>
>    Not if behavior only counts what is publically observable. There is
>    a crucial part of beliefs and desires that you have to be the subject in
>    order to know.

Skinner says that there are covert as well as overt behaviors and I
agree with this.  Private experiences are only avialable to the person
experiencing them.  These experiences are inherently excludable from
any verifiable analysis since they can only be indirectly examined by
self report.

>    And not if behavior leaves out intentionality: 

Why cannot intentionality be a behavior? albiet an unexplorable behavior.

>												    What you are trying to do
>    is an intrinsic part of what you are doing even if B.F. Skinner does
>    not count it as scientifically important.

Intentions are things we can not directly control or measure.  While
they exsist and are important and are intrinsic to the organism they
cannot be a part of science.  Science explores public experience it is
a process of communicating inductions and deductions in a verifiable
way.  Scientific claims without possiblilty of verification for validity 
are nonsense.  Speculations (e.g. thought experiments) are very different 
than claims (e.g. specification of the relationship between matter and energy 
or behavior and reinforcement).   We can speculate about others'
intentions but we can never really know about them.


>    Surely you know what beliefs and desires are. What nobody really knows
>    is what they "are made out of" (if indeed, they are made out of anything)
>    or "what they are caused by" or "how they are realized in our brains".

If they are behaviors then why not treat them as being caused by the
same type of things that cause other behaviors ... that is, environmental and 
genetic events.

V. application of Behaviorism to AI
----------------------------------------------------------
Two assertions:

	1.  I think that a neural-net simulating the change in the future
		probability of behavior due to the consequences of the behavior
		is what AI research need to approach AI.

	2.	Skinner's views on verbal behavior can significantly improve
		the strategies for AI.

----------------------------------------------------------

=============================================================================
Brian  | UC San Francisco        | E-mail: USENET, Internet, BITNET
Colfer | Dept. of Lab. Medicine  |...!{ucbvax,uunet}!daedalus.ucsf.edu!brianc
       | S.F. CA, 94143-0134 USA | brianc@daedalus.ucsf.edu 
       | PH. (415) 476-2325      | BRIANC@UCSFCCA.BITNET
-----------------------------------------------------------------------------
       "We are here because you are there." --- The Tubes, 1981
=============================================================================