Re: Finding useful functions- part 1

From: Stargazer (fuckoff_at_spammers.com)
Date: 11/08/04 

Date: Mon, 8 Nov 2004 16:36:29 -0200

Wolf Kirchmeir wrote:
> Stargazer wrote:
> > Wolf Kirchmeir wrote:
> >
> > > Stargazer wrote:
> > > [...]>
> > >
> > > > The control of reinforcement schedules in an experimental
> > > > setting (no matter if with fishes or mermaids) will inform you
> > > > about how the animal reacts to artificially controlled
> > > > reinforcement schedules. That's part of what I call "artificial
> > > > environments". It will not inform you much about how the animal
> > > > acts in natural environments, where contingencies are
> > > > completely different.
> > >
> > > It's not clear what you mean by contingencies, not why you think
> > > that natural ones would have different effects than artifcial
> > > ones. The reinforcemnt schedule? The reinforcers? The animal's
> > > responses to the reinforcers? The discriminators? Etc?
> > >
> > > Just what do you think you are talking about?
> >
> >
> > It is an unwarranted assumption to suppose that studying particular
> > schedules of reinforcement will generalize to natural settings,
> > where these schedules will never happen sequentially (besides
> > being intermixed with different sets of stimuli).
>
> Yoy're mistaken if you think that behaviorists can't/won't devise
> schedules that don't resemble natural ones. You're further mistaken if
> you believe that "intermixing" of stimuli doesn't happen in the
> experimental setting. In fact, one of the problems of any experimental
> analysis of behaviour is teasing out which stimuli actually produce
> the observed effect(s( on the animal's behaviour. It's essentiallu
> impossible to reduce the number of stimuli to one.

Behaviorism (and, as I said, other psychologies, including cognitive)
suffer from a problem known as "lack of ecological validity".
Laboratory-based schedules of reinforcement cannot easily duplicate
the kind of intense and unpredictable sequence of events in natural
environments. Any ethologist (who are used to observe animals in
natura) will gladly testify this.

> > Additionally,
> > what behaviorists study is not an animal's response to specific
> > stimuli, but the responses to what the animal _understands_
> > of those stimuli. This may lead to false conclusions, because the
> > experimenter observes ("perceives") the stimuli in a different
> > way than the animal being tested.
>
> See note above. And just how do you know what an animal
> "understands"????

That's exactly my point: no one knows what an animal "understands".
Therefore, any conclusion made using stimulus/response structures
is, in fact, using "experimenter-perceived stimulus/animal responses
to animal's perceived stimulus". This will, as I said, lead to
empirical and perhaps conceptual errors. Without grounding behavioral
experimentation in neuroscientific data and solid theoretical analysis,
experiments that behaviorists propose are just sophisticated ways
to mislead oneself. Besides, you have not commented on my point
regarding neural plasticity and the "control" of behavior.

>
> > If that is not enough, there's
> > the effect of neural plasticity. Schedules of reinforcement,
> > operant reinforcement, generalization and discrimination studies,
> > etc., all affect the organism's ability to respond to what the
> > experimenter set up. The organism becomes more "apt" to produce
> > particular behaviors that are a function of what the experimenter
> > contrived. This leads not to an understanding of the true behavior
> > of an animal, but to how behaviors of these animals change in
> > face of a predefined (and artificial) schedule, and not their
> > real capacity to handle stimuli from an unpredictable and hectic
> > natural environment.
>
> The notion that "predefined" schedules of reinforcement differ in any
> essential way from natural ones is simopky wrong.

The burden of proof is at your side. Tell us how is it possible that
a pigeon, in natural environments, will be subjected to anything
that barely resembles what is done in a laboratory-controlled
study. Or, conversely, how can behaviorists provide a laboratory
environment which mimicks what happens in a natural setting.

> As a matter of fact,
> it was EAB's attempts to refine understanding of how reinforcement
> worked that led to the insight that random schedules are the most
> powerful in shpaing behaviour. Random schedules are pretty well what
> happens in nature. So ---

Not exactly. In nature, you have "moderately random schedules". This
leads to very complex situations, because some classes of responses
are specific to the context (one pigeon at the ground is subject to
different circumstances than another in a tree or flying).

> I'm beginning to understand why DL gets annoyed with comments of the
> type you've written. They sound reasonable to anyone who has a
> limited/simplistic concept of EAB. Hell, I used to think this way,
> until I read some actual EAB.

Then read more EAB, but just don't stop there. Read also what has
been written *against* EAB. I guarantee that you will learn more about
behaviorism reading what the critics have to say. And you don't have
to look for recent things, because these points have been made at least
40 years ago. The scientific understanding of the whole issue has
changed dramatically since the sixties.

*SG*

Quantcast