> From: "Chatwin Judy" <JAC295@psy.soton.ac.uk>
> Date: Mon, 13 May 1996 14:02:59 GMT
>
> I have finally started to look at some of the exam questions in more
> depth and it seems to me that what I thought I knew to be easy and
> straightforward has proved more complex.
>
> Despite having spent many hours on the Skinner paper I am still
> having difficulty expanding fully on the question:
> 'What was wrong about behaviourism?'
>
> Am I right in understanding that:
>
> - there is no clear definition as to what is understood by
> behaviour,
Have a look at the rest of the "What Is Behaviour?" thread of discussion
in this skywriting archive. There is no clear boundary between what is
behaviour and what is body or brain activity (eye blinks, heart beat,
pupillary dilation, event-related potentials), but, overall, one can
still say that "behaviour" is what the organism DOES. You can observe
what it does. So behaviour is observable. Skinner did not want to
commit himself to a definition, but "what an organism does" covers most
of what he meant by behaviour too.
But it is not in DEFINING behaviour but in EXPLAINING it that cognitive
psychology parts company with behaviourism: Cognitive psychology tries
to explain behaviour, especially behavioural CAPACITY, by making
hypotheses (guesses, theories) about what is going on inside the head
that makes us able to do what we can do. These internal structures and
processes are sometimes called "internal representations" or even
"mental representations." Candidates for these internal structures and
processes include (1) analog processing, (2) computation (symbol
manipulation), and (3) neural nets.
Behaviourism was wrong to reject internal representations because they
were unobservable. All branches of science make hypotheses about things
that are not directly observable if it helps them predict and explain what
is observable (i.e., "if it helps them get lunch").
See also:
http://cogsci.soton.ac.uk/~harnad/Papers/Py104/millikan6.html
> - it makes no attempt to answer the 'how' question,
That's right: The "how" question is: How does the mind work? What is
going on in our heads that makes us able to do all the things we can do?
Behaviourism did not try to answer "how." At best it only tried to
answer "what?" and even that was focused on what we DO do, not what we
CAN do. Behavioural analysis describes behaviour, analyses its history,
to say what rewards and punishments "shaped" it. But it doesn't explain
how we work: it doesn't explain what kind of internal structures and
processes are needed to give us the behavioural capacities we have,
including the capacity to be shaped by reinforcement into doing the
many things we can potentially do.
> - it cannot be used to explain intelligence,
That's right. Behaviourism has great difficulty in explaining any
differences in capacity, whether between species or between individuals.
To explain differences in capacities, you need to explain the
differences between the internal structures and processes that give rise
to those capacities. Some people have much greater mathematical capacity
than others: Let us call that a difference in a special area of
intelligence. If these differences do not just arise from differences in
an individual's history (and evidence suggests that they do not just
arise from differences in an individual's history), then behaviourism
can tell us nothing about them. Nor is it just DIFFERENCES in
intelligence (apart from the ones that arise from history) that
behaviourism cannot explain: It cannot explain intelligence itself.
The beginning of an explanation of intelligence came with artificial
intelligence (AI): computers had the capacity to do "intelligent"
things, things that only people and animals had been able to do until
then, and, what's more, AI could tell you HOW they were able to do it.
So artificial intelligence was the first step in the explanation of
natural intelligence. Your natural intelligence is everything that you
are able to do; it is your behavioural capacity. Behaviourism could not
explain behavioural capacity because that required making hypotheses
about the unobservable internal structures and processes that generate
behaviour, and behaviourism rejected the unobservable and restricted
itself to the observable.
> - it cannot explain language,
Language is part of our behavioural capacity, perhaps the most
sophisticated and powerful part. Only our species is capable of
language. No amount of reward/punishment history has made any member of
any other species able to take part in this discussion with us. Skinner
called language "verbal behaviour," and he suggested that it was all
explained by the history of rewards and punishments that "shaped" it:
But why does the same history of rewards and punishments shape you and
me into being able to talk (well, write, but you know this could have
been a conversation too!) like this, whereas it won't do it for a
chimpanzee or a dog? It's not because it's "verbal," because we could
have been talking in morse code (and even a pigeon could peck morse
code, if it had the INTELLIGENCE, the capacity, to do so, but it
doesn't). It's because we have a capacity that they do not have, and
underlying this capacity are unobservable internal structures and
processes.
We will discuss language this week, but here's a preview: To have
language is to be able to name and describe all the things
we can name and describe, and to be able to give information to
one another with these names and descriptions. (Remember: information
reduces uncertainty between alternatives that matter to you, like what's
safe to eat for lunch, and where it is.) Behaviourism does not explain
how we can do these things.
See:
http://www.envmed.rochester.edu/wwwrap/behavior/jeab_htm/65/_65-185.htm
Language also has a grammar, and one of the most surprising things in
cognitive theory is Noam Chomsky's discovery that most of the grammar we
know we cannot have learned (through a history of trial and error,
reward and punishment, as Skinner would have thought). We were born with
a complicated structure called "Universal Grammar" UG already in our heads.
Behaviourism could never have told us that, nor could it have told us
the specifics of exactly what structure UG turns out to be; only
theoretical linguistics could do that, because UG is unobservable,
although you need to know what it is in order to explain our grammatical
capacity, including our capacity to learn what little of grammar we
actually do learn (since most of it is inborn).
See also:
http://cogsci.soton.ac.uk/~harnad/Papers/Py104/pinker.langacq.html
and
http://cogsci.soton.ac.uk/~harnad/Hypermail/Thinking.Psychologically/0081.html
> - it cannot explain complex perceptual mechanisms and processes such
> as pattern recognition?
That's right. Behaviourists can tell you that you call a cat "cat" because
you were rewarded in the past for doing so; but they cannot tell you
what structures and processes you must have inside your head to make you
ABLE to recognise and name cats, and to tell them apart from all the other
things you can recognise and name. This would require a theory of the
internal mechanisms that give you the capacity to recognise patterns.
Neural nets can do that; but behaviourism would not discover neural nets
because neural nets are not observable behaviour.
This archive was generated by hypermail 2b30 : Tue Feb 13 2001 - 16:23:41 GMT