Re: Is sentience an emergent brain behavior?


Wolf Kirchmeir wrote:
JGCASEY wrote:
Wolf Kirchmeir wrote:
JGCASEY wrote:
Wolf Kirchmeir wrote:

The denigrators of behaviorism call for understanding the
architecture of the brain, and claim that understanding this
will provide a better explanation of behaviour than study of
behaviour as such. Well, there it is: when we learn, the
architecture of the brain changes. But most of that change
(perhaps all) depends on environmental signals.

And there is a very good reason why development should depend
on environmental signals. However it is the DNA that sets the
rules not the environment.
This is simply wrong. Look, I've said several times that genes are
switched on and off by environmental factors.


But only if the genes *allow* the environmental factors to turn
them on or off. If the switch is stuck, or the key doesn't fit,
the enviroment can do nothing.



Absent or present a given
environmental factor, and the organism will develop one way or
another. Think about what that means about what "sets the rules.
" Think about what sense it makes to even talk of "setting the
rules."

The same thing it means when the rules are set in a computer.


The example I gave in another thread
as regards a light seeking mechanism. The light may appear to
control the mechanism. Move the light and the mechanism follows.
But the rule is in the mechanism, not in the light. Change the
rule to avoid light and the behavior changes. The light now
appears to cause the mechanism to avoid it. But that is not
the case it is the mechanism that "decides" what to do in the
way it is built.
Yeah, but the example is not a learning machine. It can't be
conditioned: it's internals structure cannot be changed by
experience, hence its behaviour cannot change. IOW, it cannot
develop. So how is it relevant?


I can give examples of learning machines that can change their
behavior, but that change depends on what the machine does with
the input, not just on the input itself.

Since "changing their behaviour" == "learning", your comment is more
than a little odd. Are you claiming that a system can learn and yet not
change its behaviour????

As for "The change depends on what the machine does with the input...."
Well, sure. So what's your point?

That it is the machine that "decides" (by the way it is built)
not the environment (input).


I mean, do you really believe that a behaviorist will claim that any old
input at all will have some effect on an animal? If so, you have such a
defective understanding of behaviorism that there's no point taking you
seriously. An animal can respond only to stimuli it can discriminate --
what else? A blind creature cannot respond to light, so it can't be
conditioned using light as a discriminant.


It could all be a comedy of misunderstanding for all I know.


If mechanism P seeks light and mechanism Q avoids light we
look for the cause in the mechanisms, not in the light.
That depends on whether P and Q are capable of learning.
AFAICT, you've assuming machines that cannot learn, so what's
your point?

Two machines, same environment. One has *internal* needs to
stay warm, the other has *internal* needs to stay cool. Both
have light direction input. Trial moves made. Outcomes then
determine that one machine seeks light the other avoids it.
Same environment. Two different machines. Different behaviors
correlate to different machines, not different environments.

Your example is still a bummer. Now the machines are not in fact
seeking/avoiding light, but heat. If you used "cold light" (ie, a light
source whose heat signal didn't rise above the general background), the
machines wouldn't move at all, 'cuz they couldn't detect the heat....

Your argument amounts to a tautology. Your example shows that "A machine
that cannot learn cannot learn". So?

As I say to the wife, don't listen to what I say, listen to what
I mean. You know, reading other minds. Do you really have trouble
understanding what I am saying, or what I am misunderstanding?

Heat, light, why do you think I used heat? Make it light if you
like it is not the issue.


Development may depend 100% on environmental signals and as I
wrote for a very good reason, let the environment provide as
much of the information for development as possible, all the
DNA than has to provide is the rules. Change the rules you change
the behavior, but the environment stays the same**. The behavior
is thus rule determined. Thus the animal determines the behavior
not the environment although the implementation of the behavior
may require the environment for information.
HUH?

e = 6
m1 = 2
m2 = 4
e + m1 = 8
e + m2 = 10

which determines the different outcomes, e or the values of m's?
I would say the difference is due to the m's as e stays the same.

m1 "stays the same." m2 "stays the same."

Your question about "which determines the outcomes" is pointless without
a function relating e and m. What's that function?

After sending I decided it was a silly example. Essentially I am
saying that machine + environment are coupled via the machines
input (for all practical purposes) and machines can all be different
but the environment is essentially the same. The differences in the
machines behaviors is due to the machines being different and the
environment being essentially the same. Looking for the reason one
machine behaves one way and another machine behaves another way is
in the machine not in the environment.



A simple example may be the complex path of an ant which may be
more a reflection of the complex obstacle course it follows rather
then the rule determining how it is to react to the environment.
If the path is a "reflection of the complex obstacle course", then it is
rule determined. IOW, your example refutes your thesis. "Return to Go.
Do not collect $200."

I said it was rule determined, and the rules are in the machine
not the environment.

A) The phrase "the path is a reflection of the complex obstacle course"
is the equivalent of "the path is rule determined." IE, the path cannot
"reflect the obstacles" if there are no rules.


Of course it can, like a stone rolling down a mountain side. The
difference is that an animal being affected by the obstacles has
a goal and will arrive at the same point at the bottom of the hill
whereas the stone will not. The stone is controlled by the environment
whereas the animals actions are also controlled by internal rules.



B) If you change the arrangement of the obstacles, you change the path.


And if you change the ant's internal rules you change the path.


C) "Rule determined" does not mean that one entity's behaviour governs
another. It just means that we can abstract rules about their
interaction: ie, if we know the one, we can infer the other - either way
round.

Well if you can infer why animal_x reacts one way and animal_y reacts
another way simply by observing their behavior then explain it. If
you can deduce the internal connections and predict all possible
behaviors without opening up the internals of a chess program I would
like to see it.


-- You once said you had trouble with maths: your comment above shows
one of your troubles: you are using a mathematical term ("rule
determined") in a non-mathematical way. But you don't seem to be aware
of your error in doing so.

I use it in the normal sense that we all use rules. They can be
simple or complex. In a computer program we use the IF THEN ELSE
construct as rules as to what to do in any given circumstance.

if i<80 then
outp(i) = int(rnd(1)*2)
else
outp(i) = 0
end if


As an added note the developmental rules also depend on the
laws of physics. DNA is necessary but not sufficient to explain
how animals develop.
Depends on the level of analysis.

** Actually the environment may not stay the same. The organism
can change the environment which in turn changes how the rules
will be implemented in the next unit of time.
Precisely.

Nice try, JGC - you've refuted yourself. :-)

Don't see how? It is the organism doing the doing not the environment.

The organism determines what changes to make to the environment not the
environment deciding how the organism should change the environment.

--
JC


Your error begins in your use of the word "allow." That's a metaphor,
and it misleads you throughly. You said once that you had trouble with
maths: I suspect it's because you have trouble discarding the
connotations of terms, which leads into you into hidden assumptions.
You cannot see that an interaction is not a master-servant relationship or
process. (Actually, I think the common understanding of "master-servant"
is an illusion, but I'll leave the proof as an exercise for the
interested student.)

The genes "allow" the environment to affect their "switches," then the
environment "allows" the genes to act on it. See?

I am capable of dropping the short hand of mentalese and give a
long winded explanation instead. We can eliminate the idea of
organism vs. environment by simply dealing with two coupled machines
and then asking ourselves what we mean by control, behavior etc.

I am talking about the relationship between coupled systems.


--
JC

.

Relevant Pages

  • Re: Is sentience an emergent brain behavior?
    ... I can give examples of learning machines that can change their ... you have such a defective understanding of behaviorism that there's no point taking you seriously. ... Same environment. ... then the rule determining how it is to react to the environment. ...
    (comp.ai.philosophy)
  • Re: Local Admin
    ... I am assuming that you are currently in an Active Directory environment and that your users are local administrators of their own machines once logged into their domain account. ... Download FREE whitepaper on how a managed service can ... Download FREE whitepaper on how a managed service can help you: http://www.cenzic.com/news_events/wpappsec.php And, now for a limited time we can do a FREE audit for you to confirm your results from other product. ...
    (Pen-Test)
  • Re: A critique of test-first...
    ... > network, by parallel I mean multiple CPUs in the same nest). ... > a controlled environment which can be exactly characterized for MTBF, ... > may be hardware on the links which are not under your control. ... > be interesting to hear how they spec'd the machines and network to get ...
    (comp.programming)
  • Re: A critique of test-first...
    ... network, by parallel I mean multiple CPUs in the same nest). ... In a distributed environment the environment is much less fixed. ... may be hardware on the links which are not under your control. ... be interesting to hear how they spec'd the machines and network to get ...
    (comp.programming)
  • Re: Alternative tof Microsoft Windows vis a vis Lisp
    ... >> The Lisp machines of years ago provided an integrated operating ... >> environment and the kind of user interface many of whose features are ... >> to design and maintain - but we are now 20 years in the future. ... >> So any attitudes which abjures the importance of these Lisp machines, ...
    (comp.lang.lisp)