Actions and Intentions

[From Rick Marken (2003.05.21.1550)]

Bruce gregory (2003.0521.1731)--

Rick Marken (2003.05.21.1400)

> Intentions concern the results of actions. Action depends on what is intended _and_ on
> disturbances to the intended result. So you can never tell what is intended by looking
> at actions but you can (if you have tested) tell intentions by looking at the results
> of actions (to see if they are the results that were found , by test, to be intended
> results)

I know this makes perfect sense to you. On the other hand, have you ever
watched a golf game on TV? Have you ever watched a golfer tee off? Did you
have any doubts about his or her intention? Did you need to see where the ball
went before you could infer the golfer's intention? Was the golfer's swing
determined by disturbances or by his or her intention? See the problem?

Yes. I guess what distinguishes what I'm talking about from this situation is that the
result of action has been produced; I said something about jaw movements and the question
was whether I was making the point Bill thought I was making or whether I was making another
point. That is, the question was _what was_ my intended result. Was it to make a point about
hierarchical relationships in PCT or was it to make a point Marc thought I was making. Bill
knew what result I was intending, not because he could tell from the words (actions) I
produced but because he saw the result (point I was making with my words) and knew that it
was the kind of result I would intend to produce if my understanding of PCT was similar to
his.

The analogous situation in golf is when the swing (like my words) produces a result (like my
point about PCT). In the golf situation a swing produces many different results (just as my
words can evoke many different meanings). For example, a ball that is going to the 1st pin
may also be moving away from the 18th pin. Both of these are results of the swing. But you
can't tell which result was intended by looking at the swing. You can't really tell which is
intended by looking at the result, either. But if you know the golfer the way Bill's knows
me, you could be pretty sure that the golfer's intention was to hit the ball towards pin 1
(if the golfer is Jack Nicklaus) or away from pin 18 (if the golfer is Groucho Marx). But
the swing per se tells you nothing about the golfer's intention unless you know the result
it produced.

Best regards

Rick

···

--
Richard S. Marken, Ph.D.
Senior Behavioral Scientist
The RAND Corporation
PO Box 2138
1700 Main Street
Santa Monica, CA 90407-2138
Tel: 310-393-0411 x7971
Fax: 310-451-7018
E-mail: rmarken@rand.org

[From Bruce Gregory (2003.0521.2147)]

Rick Marken (2003.05.21.1550)

But
the swing per se tells you nothing about the golfer's intention unless you know the result
it produced.

You seem to have overlooked one tiny detail: the context in which the
swing occurs. Life, fortunately or unfortunately, unfolds in a context.

···

--
Bruce Gregory lives with the poet and painter Gray Jacobik in the future

[From Rick Marken (2003.05.21.2100)]

Bruce Gregory (2003.0521.2147)--

Rick Marken (2003.05.21.1550)

> But
> the swing per se tells you nothing about the golfer's intention unless you know the result
> it produced.

You seem to have overlooked one tiny detail: the context in which the
swing occurs. Life, fortunately or unfortunately, unfolds in a context.

Yes. And part of that context is who the golfer is, Nicklaus or Groucho.

But I think I understand your point and I think it's one that's worth exploring. We are often
in situations where we seem to be able to tell, by looking at a person's actions, what the
intended results of those actions are. If we see a person swinging a golf club at a driving
range we can be pretty sure that the intended result of that swing is a ball that lands 300
yards straight ahead of the golfer. But this is still a _guess_ about the golfer's intention.
It's certainly a reasonable guess but it could be, and often is, wrong. The fact that this
guess is wrong sometimes is confirmed when you see, for example, the golfer swing and miss the
ball completely. It was a practice swing and the intention was (at least partly) to miss the
ball.

I agree that you can often guess, from context, what results a person intends to produce based
the actions you see. But you certainly can't build a science of purpose (intentionality) on
this notion. The chances of error are way too high, not only because your guesses about the
intended result will often be wrong in general but also because they they will almost _always_
be wrong in particular, since the actual intended result is a perception controlled by the
actor. The intended result of a golf swing includes perceptions that you can't even see, like
the satisfying proprioceptive thunk that is produced when you hit the ball right in the sweet
spot.

The idea that we can determine a person's purpose by looking at their behavior in context is, I
believe the position Daniel Dennett takes in the "Intentional Stance". PCT shows why the
Intentional Stance is nothing more than posturing. The only way to determine a person's
intentions -- not make vague, general guesses about them but to actually know them to a near
certainty -- is to use the test for the controlled variable. And in order to do the test you
have to know more than just what the person is doing (how they are acting) or what results
these actions are producing. You have to test hypotheses about the result the person might be
producing intentionally by applying disturbances to those results and watching for lack of
effect.

Best

Rick

···

--
Richard S. Marken
310 474-0313

from [ Marc Abrams (2003.05.21.0117)]

[From Rick Marken (2003.05.21.2100)]

But I think I understand your point and I think it's one that's worth

exploring.

Wow ! I got to hand it to you Rick, You might be a few days late with this,
but at least you realized it. Good for you. Welcome to the real world.

Marc

[From Bill Powers (2003.05.22.1027 MDT)]

Bruce Gregory (2003.0521.2147)--

Rick Marken (2003.05.21.1550)

But
the swing per se tells you nothing about the golfer's intention unless
you know the result
it produced.

You seem to have overlooked one tiny detail: the context in which the
swing occurs. Life, fortunately or unfortunately, unfolds in a context.

You (and Marc) are arguing that there is some probability that a familiar
action done under familiar circumstances is intended by the actor to
produce a familiar result. Nobody can argue with that. The questions
remaining all have to do with _how_ probable your guess is. In cases where
the same thing has happened exactly the same way many times in the past,
and subsequent events in each case have proven that your assumption was
correct, the probability can be pretty high. Of course if you have simply
made the same assumption each time but without checking to see if you were
right, the probability is pretty low.

So the question is, how to you check to see if you were right? Generally,
you make a prediction that depends on the assumption's being right, and
that will fail if the assumption was wrong. Then you make the required
observation. If the observation fails, you change your interpretation
appropriately, and try again. When your predictions no longer fail in a
significant number of cases, you come to feel you have it right.

So you may assume from a series of tests that when the golfer strikes the
ball and it flies straight down the fairway toward the green, that was the
result he intended. You can check this very easily -- you ask him if that's
what he wanted the ball to do, and predict that he will say yes. After you
have seen this result many times and have obtained signs of approval from
the player each time, you probably don't have to go on asking every time.

However, there are also cases where the golfer executes what looks like the
same swing from the tee, and the ball curves gracefully left or right and
into the underbrush. Repeated checking will probably convince you that this
was NOT the effect the golfer intended. However, you have to notice the
_right_ result and ask specifically about it. Not "Did you intend the ball
to curve," but "Did you intend the ball to end up in the underbrush?" If
you ask the first question, you will become convinced that a curved path is
a mistake and the golfer intends only to hit the ball straight. You should
go on testing, in that case. When the golfer steps up to the tee of a par 4
with a dogleg to the left, and hits a long drive that curves to the left,
you should not say, "Oh, too bad." The ball ends up in the middle of the
fairway beyond the dogleg, which is precisely what the golfer will say he
or she intended.

So now the question is, how long should you go on checking your predictions
before you can decide to omit the testing? That's the critical point. If
you're claiming to do science, you have to recognize that nature is
variable and what seems to be the rule can change without warning. So you
would routinely set up your experiments to test as many assumptions as you
can _every time_. There never comes a time when you can say you don't have
to test any more because you have found the truth once and for all. Only
practical considerations of time and cost let you off the hook a little,
but as a scientist you will always know the irritating truth that if you
omit the check just once, that will be the time the answer would have come
out different.

In science, even when we feel we know the answer we pretend that we don't,
and do whatever we can to provide the answer anew, from the data. When Rick
or I say that you have to test to see what the golfer intended, we mean
that if it's not too hard to do so, or too socially inept, we would pretend
that we didn't think we knew, and would check. In reality there's always a
chance that we're wrong, so checking can't hurt. Just as you're about to
commiserate with Tiger Woods about the hook he just drove into the sand
trap, he turns to the kids behind him and says, "That's how you make it
hook. Now here's how you make it slice," and tees up another ball. So don't
jump to conclusions, no matter how often they've been right in the past. Or
go ahead and jump, but be prepared to be wrong.

Best,

Bill P.

[From Bruce Gregory (2003.0522.1434)]

Bill Powers (2003.05.22.1027 MDT)

So don't
jump to conclusions, no matter how often they've been right in the past. Or
go ahead and jump, but be prepared to be wrong.

Sound advice based on experience no doubt.

from [ Marc Abrams (2003.05.22.1713) ]

[From Bill Powers (2003.05.22.1027 MDT)]

You (and Marc) are arguing that there is some probability that a familiar
action done under familiar circumstances is intended by the actor to
produce a familiar result. Nobody can argue with that.

You've done a pretty good job so far.

The questions remaining all have to do with _how_ probable your guess is.

No. There are many other questions. Fred Nichols brought up a whole slew of
them.

In cases where the same thing has happened exactly the same way many times

in the past,

Nothing ever happens _exactly_ the same way, Nothing. PCT explains why this
is so. The similarities you happen to notice and perceive ( from
experience ) allow you to make assumptions about what you percieve. For the
most part, it really doesn't matter what that golfer is controlling for. We
know the rules of the game and he may try any number of different strategies
or plans to hit the ball the fewest number of times necessary to get the
ball in the designated hole. He may be more or less successful with his
plans or strategies in his attempte to do this.

Doing the PCT "Test" is done for 2 reasons only. 1) so the _tester_ can see
if he/she has guessed right on percieving a variable the testee might be
controlling for, which 2) Shows the Test is a viable way of testing for a
controlled variable.

and subsequent events in each case have proven that your assumption was
correct, the probability can be pretty high. Of course if you have simply
made the same assumption each time but without checking to see if you were
right, the probability is pretty low.

So far so good. We normally don't check because it doesn't matter.

So the question is, how to you check to see if you were right?

Who cares? I am simply concerned with how low a score the golfer can get. If
I'm _real_ interested I might want to know some of his "strategies" and
"plans", win or lose. What he might have been controlling for at any one
point in time is meaningless. Anyway, it has been shown empirically, that
peoples "explanations" for why they do things and what they actually do are
different. Your definitions In B:CP being an example of what you "thought"
you said vs. what you actually "intended" to say.

I believe Argyris has several effective ways of "validating" someone's
"intended consequences". And Arygris is not the only one.

Marc