points

CSGnet Archive CSG1993
1

response to Tom Bourbon. I believe this will be undestood by others who do
not have copies of Tom's note to me

Tom,

I find the "control of perception" notion agreeable, and I believe that many
others (e.g., Chris Wickens, and several who follow the ecological
perspective) also find it agreeable. From there, I believe there is
considerable deviation, although again, most of these folks would, I think,
agree with some of your points.

On set points, I would not wish to claim that set points, affordances, and
attractors are the same. What I do believe is that where those who subscribe
to HPCT see a set point, others will often see an affordance or an attractor.
I also present this as a relatively casual observation and would not want to
defend it too rigorously.

The nature of an affordance is difficult for most, and even some ecological
people are not very clear on this. The best people are Len Mark (JEP: HP&P,
1987, 361-370) and Bill Warren (with Whang, JEP: HP&P, 1987, 371-383).
Dynamical systems is a little more difficult. Models abound, which raises a
concern that there are too many. Which is the right one? Is that even a
relevant question? For human behavior, the most serious attempts come out of
Kelso's laboratory.

Recent papers by Schoner et al. (Journal of Motor Behavior, 1992, 29-48) and by
Schmidt et al. (from Turvey's lab, JEP: HP&P, 1991, 635-651) are the best
leads I have. These latter papers will not give you much about the origins of
attractors. That aspect is still evolving, and you would need to turn to
examples from physical science to get the gist of this.

As far as I can tell, no one outside of HPCT worries overly much about
real-time implementation although in a book titled Swimmer by Uttall et al
there is an expression of concern via an admission that they have not achieved i
t.

Since my first note, some responses have clarified the purpose of the CROWD
program. It is now evident to me that there is a sharp divergence here
between HPCT and ecological/dynamical systems. In the former view (if I read
those comments correctly) the behavior of concern is that of the individual
creatures. In the latter view, these examples suggest how organized patterns
might be assembled (without instruction) from independent elements. As a
further note here, this is a sharp and apparently irreconciliable difference.
The particular view one takes is, I think, based more on one's foundational
assumptions than on evidence. I cannot see any point in arguing over the
direction we should take. Those who take the correct direction may solve the
problem of accounting for behavior, or at least may progress towards on
explanation. Those who take an incorrect direction will presumably flash
endlessly. The history of science suggests is that this sort of
thing is inevitable.

One further note on the Ecological view: as an example of the assembly
of independent elements into an organised pattern one might think of
coordinating body parts for a tennis serve. In this sort of action, errors
do not appear to informthe actor about the needed corrections. eg to
achieve a faster serve one may have to think of rotating the shoulders,
snapping the elbow or throwing the ball further forward versus hitting harder.
Gavan Lintern
Aviation Research Laboratory
University of Illinois at Urbana Champaign
# 1 Airport Rd., Savoy, IL 61874
TEL: (217) 244-8637/6905 FAX: 244-8647

2

This note is in response to a note from Rick Marken and other responses about
models.

Rick,

To respond to your comments on models and CROWD.

I detect that my use of models may be somewhat in line with your thinking, but
not identical. You seem to have a greater degree of faith in their power or
value. I suspect that some of our differences at this level will emerge from
the different experiences we have had with models. I gained almost no useful
experience of models in my own graduate program, and have had to bootstrap
myself in this area.
On the CROWD program. I was missing your point and insofar as I now get it, I
needed a considerable shift in frame of reference. Also, see my notes on
CROWD to Tom Bourbon.

P.S. In my mailing to you, I have also included a copy of a landing paper by
Gross and Wieringen.

Some General Notes on Models

There is a strong sense in many of the responses I have received that models
need to work in real time, that they need to be implemented (versus be
analytical), and that they need to be comprehensive.

I do not feel as strongly about the first two as most respondents appear to.
I am quite happy with an analytical model. I regard Gibson's 1955 paper on
optic flow as an acceptable model, which means I think I could build a machine
(probably a computer model) to do what Gibson claims. I do not need to do it
on every occasion to be satisfied. For me, the most substantive issue here is
whether people do respond to optic flow as claimed. Construction of a model
establishes feasibility, but does not demonstrate that it is done that way.

On termite nest construction. Peter Kugler did develop a model of this on an
IRIS work station. He has not published or documented it in any form. His
published treatments of this are analytical and descriptive.

To those who request references to models that have been implemented, I reply
that behavioral researchers who think in ecological or dynamical systems terms
have not, as far as I know, implemented their models. There are many
implementations in the physical and biological areas of dynamical systems
research, but none I know of that are specific to human behavior.

The people who follow the Simon and Newell Symbol System Hypothesis do a lot
of implementation and Simon (e.g., see Vera and Simon, Cognitive Science,
1993, 7-48) make claims of comprehensiveness, and implementability that sound
much like those coming from HPCT. I have a book right now called Swimmer that
makes some of these claims. I tend to ignore this work because I find it dull
and I am having a difficult time maintaining interest . In
addition, we regularly have people in AI coming by the University of Illinois
who claim to have implemented models of learning. I cannot refer you to much
of this because I fail to generate enough interest to keep track of it. I
would, however, supply a full reference for Swimmer on request although I suspec
t
you too will be disapointed.
Gavan Lintern
Aviation Research Laboratory
University of Illinois at Urbana Champaign
# 1 Airport Rd., Savoy, IL 61874
TEL: (217) 244-8637/6905 FAX: 244-8647

3

The following is my response to notes from Bill Powers & Rick Marken,
Although I refer later to some of the comments in Bill's note.

Rick, Bill:

The organization of my response to comments on Chris Wickens' book
came from a perspective as follows:

1. Nobody yet understands human behavior in all its complexity.
There are many contending theories. After dismissal of the cleary trashy
entries, a few remain that require some serious thought. I have selected
an approach that makes sense to me but I do not believe it constitutes the
final answer. The most I hope for is that it will come to represent at
least a part of a stable foundation for our science. There are several
views that I cannot subscribe to but I would not wish to dismiss because
they may contribute key pieces to the puzzle. I do not think anyone can
be certain about where a breakthroughs will or will not come from.

2. All theories have unsupported assumptions. These assumptions represent
the interface at which the truly innovative work must be done. The
development of a theory depends on this type of work.

3. All theorists have unsupported commitments. My own reading of the history
and philosophy of science suggests to me that at least some of these
foundational assumptions are, in fact, unsupportable. The resolution
apparently lies in the eventual success of the theory rather than any
ability to rationalise the approach one is taking.

4. From all this I take the position that arguments are of little use but
clarifications are of value.

Now for some specifics regarding Bill's note:

"What's wrong with control of action." Many recognize your concern
(Bernstein for one). Others do not. It is a valid and important issue.

I am happy to think about physically distinct levels but this remains as an
unsupported assumption. It is a crucial assumption, and it is one that many
will question.

Instructed
I use instructed here as Edelman introduced in his book on Neuronal Group
Selection. To be instructed is to have no degree of choice.

A few points to ensure there is no misunderstanding.

I would not wish to be seen as one who is trying to merge HPCT and
dynamical systems theory or ecological theory. By suggesting that an
affordance and an attractor might be viewed as a set point, I was attempting
to show a point of contact, from where the divergence would be apparent.
Ecological people would not like affordances to be characterized as set
points, and dynamical systems people would not like attractors to be so
characterized. .

Your discussion of equilibrium points is fair. Your dismissal of equilibrium
points as a contender in explanations of behavioral theory is too casual for
me.

On nonlinearity, some believe that human behavior is nonlinear at many
levels. They would (I think) argue that they are talking within an
unrestricted universe of discourse and that oscillations, limit cycles,
and much more permeate behavior.

I believe that the view of nonlinearity in dynamicals systems theory is
conceptually different to the view of nonlinearity in control system theory.

I should withdraw my references to mental models as constituting a form of
set point. Given that no one knows what a mental model is, that was a loose
characterization.

Control parameter (in dynamical systems theory) is certainly different to the
meaning carried in PCT.

To be sure there is no misunderstanding in relation to your second last
paragraph, it was my choice to think of set point and attractor in the same
terms. I know of no dynamical systems person who speaks
of set points, or who talks much about control theory.
Gavan Lintern
Aviation Research Laboratory
University of Illinois at Urbana Champaign
# 1 Airport Rd., Savoy, IL 61874
TEL: (217) 244-8637/6905 FAX: 244-8647