This topic needs a title

[Bruce_Abbott]

[From Bruce Abbott (950608.1930 EST)]

Rick Marken (950608.1320)] --

Bruce Abbott (950605.1210 EST)

Reinforcement theory makes reinforcement the central explanatory principle
for behavior change; PCT makes it a side-effect of control. The argument is
not about the objective phenomenon of reinforcement but its theoretical
significance.

This statement suggests that the same phenomenon (behavior change) is
explained by both reinforcement theory and PCT; the two theories just explain
it in different ways: reinforcement theory says that behavior change results
from the "strengthening" effect of reinforcement; PCT says that behavior
change is a side effect of control.

Yep.

I think Bruce is claiming that PCT is better (more accurate, simpler, etc)
than reinforcement theory as a model of behavior change. In other words,
reinforcement theory (like Ptolmeic theory) is basically OK but PCT (like
Copernican theory) is much better.

Nope. I am claiming that current reinforcement provides a consistent and
generally compelling framework for understanding behavior (which is why it
has so many adherents) and is capable of handling the sort of data Bill P.
asserts it cannot (the ratio data). This is not the same as saying that it
is basically O.K., which seems to be your take on what I am saying. One can
grant that Ptolemaic theory correctly describes the apparent motions of the
planets through the heavens without implying that the theory is correct ("O.K.")

If the theory can handle such data then its adherents will see nothing
surprising or contradictory in such data, nor will they feel the need to
sweep such findings under the rug. This explanation eliminates the need to
posit any conspiracy of silence.

Bill P. noted that I had avoided the issue about performance on ratio
schedules when it came up earlier and I did not wish to appear to be ducking
it. So I attempted to respond while realizing that I had not really thought
the problem completely through. The result is that I did not really make a
good case and I knew it, but I still believe such a case can be made, once
I've given it more consideration.

Bruce is, therefore, understandably
puzzled by the response he is getting from those of us who presumably also
believe that PCT is better than current theories of behavior. This puzzlement
is obviously very frustrating for Bruce, as evidenced by:

"O.K., O.K., you caught me. All this time I've been part of a top secret,
high-level plot to undermine PCT and reestablish traditional reinforcement
theory as the "top dog" in the field of learning and behavior..."
                                            Bruce Abbott(950607.1245 EST)

Well, it is frustrating when your motives (er, attempts to correct
deviations of controlled perceptions from their reference values) are
misunderstood, as in this case, especially when one of the people doing the
misunderstanding claims to be able to read your mind (as in his mindreader
program). But the paragraph from which the above quote was taken was
intended to do something Rush Limbaugh likes to do: illustrate absurdity
with absurdity. [No, I'm not a big Limbaugh fan, thank you.] After all,
the above "admission" is also consistent with the "evidence," isn't it? And
if it is, how can you claim to know what I'm "really" trying to do? Hey, it
might even be true... (;->

I do agree that a part of the difficulty in challenging reinforcement theory
is that many of its proponents do evaluate it only in qualitative terms. It
is also the case that reinforcement theory is currently undergoing
challenges and modifications in response to those challenges, so that it
constitutes a moving target or, perhaps more accurately, a number of
alternative views. For this reason a given finding may be fatal to one
version but consistent with another.

Clearly, Bruce sees himself as a friend of PCT who is being treated as an
enemy.

Not as an enemy, but perhaps as someone whose viewpoint is sometimes
misunderstood and criticized for the wrong reasons. I would hope that there
are no "enemies" here, just people with sometimes differing opinions who are
willing to argue them and to listen carefully to the other point of view.
But yes, I do see myself as a friend of PCT.

It think what's going on is a conflict between two very different views of
PCT. One view (Bruce's) is that PCT is another theory of behavior -- like
reinforcement theory, various cognitive theories, motivational theories,
etc -- and is of interest to the extent that it can explain existing data
_at least_ as well as the other theories can. This is the "alternative
theory" view of PCT. The other view (mine) is that PCT is about the
phenomenon of control and that the goal of PCT is to understand this
phenomenon. According to this view, most existing data is irrelvant to
understanding control. This is the "alternative phenomenon" view of PCT.

I do think of PCT as another theory of behavior, but not as one alternative
among many equals. Rather, I think of it as the correct solution, at least
in broad outline (the details remain to be worked out, remember). Behavior
is what I'm interested in studying and understanding. I'm not interested in
studying the "phenomenon of control" unless it helps me to understand
behavior. It does, so I am. But the "phenomenon of control" is only a part
of the total picture: there is the phenomenon of perception, the phenomenon
of memory, the phenomenon of learning, the phenomenon of discrimination, and
many others, all worthy of study in their own right.

Once you know that the rat in an operant chamber is controlling food input
(you have identified the phenomenon of control), you know that only a control
model (like PCT) can explain the phenomenon; a reinforcement is simply not an
alternative.

Well, I'm convinced of that, but I wish to convince others who currently see
the reinforcement explanation as adequate, and to do so I need to understand
how the PCT model of the universe accounts for the apparent motions of the
planets, even though those motions are only a side-effect of not being at
the center of the universe.

Regards,

Bruce

[Bruce_Abbott]

[From Bruce Abbott (950625.1210 EST)]

Well guys, today I'm starting the second half of my first century. I don't
know how this has happened so soon; since when does 50 come right after 35?

Now, let's see if the remaining brain cells still work . . .

Bill Powers (950624.1005 MDT) --

    Bruce Abbott (950624.1030 EST)

    From the point of view of reinforcement theory, SOMETHING must be
    producing this behavior; the question is what. The answer, for the
    radical behaviorist, is to be found in the current environment and
    in the lasting effects of the organism's history of experience with
    the environment and its contingencies.

With these assumptions, there is nothing left to do but to try to
identify the current environmental causes and the history of
interactions. The behaviorists do not really consider that the causes of
behavior may be internal, and trace back to the environment only via the
long path through evolution. They have renounced that option.

. . .

If an inexplicable behavior occurs, the behaviorist does not say "Well,
the cause must lie inside the organism, since I can't find it in the
environment." That would contradict the basic belief that all causes are
in the environment. Instead, the behaviorist says "There must have been
some reinforcer acting that I didn't see." Thus both successes and
failures of prediction are taken into account, and the theory is
invulnerable against disproof.

Let's not confuse glib "explanations" with offerings of proof. If the
theory assumes that the causes of behavior are to be found in the
environment (both present and past), then that is where you look for them.
PCT's approach is not different: if the theory assumes that all action
results from error in a controlled perception, then that is what you look
for to explain the behavior.
If either theory founded its "proof" merely on the ability of its
practitioners to identify reasonable-sounding sources of behavioral
motivation, then both would be invulnerable against disproof. When asked to
speculate as to why someone is doing such-and-such, practioners in either
camp can point to variables which may be as work and would permit the theory
to "explain" the behavior.

However, this is not research. When scientific research is being conducted
in order to explain a given set of observations relating to behavior,
considerable effort is expended to test alternative hypotheses and, to the
extent possible, rule out those that fail the tests. This is done in PCT,
and it is done in behavior analysis.

    It's a common criticism of reinforcement theory in applied settings
    that reinforcement "doesn't work" because Johnny still misbehaves
    in class even though the teacher offers Johnny a bribe for his
    compliance. The problem, from the point of view of reinforcement
    theory, is not that reinforcement doesn't work, but that there are
    other, more powerful sources of reinforcement at work which are
    controlling Johnny's behavior, such as the attention he gets
    from misbehaving.

In a real science, when a prediction fails you don't excuse it by saying
that something unobserved must have happened in just the right way to
keep the theory true. You investigate why it failed and produce the data
that explains the failure. The "more powerful sources of reinforcement"
kind of argument can make ANY theory true no matter how often it fails.
The fact is that reinforcement theory, when applied in the classroom or
elsewhere, often DOES fail. That is what the _real_ data say.

In a real science, when a prediction fails you develop tests to determine
what is going on. For example, your observations may suggest that Johnny is
misbehaving because of the attention such behavior brings. If this
hypothesis is correct, then eliminating this source of reinforcement should
allow this behavior to extinguish. So, you implement a test by introducing
"time out" for misbehavior. Each time Johnny begins to disrupt the class,
he is immediately taken to a very boring room and has to sit there for a few
minutes watching Ms. Blatherstone type reports. If, after a few such
experiences, we observe that the frequency of Johnny's disruptive behavior
takes a nosedive, we have empirical support for the hypothesis. Well run
behavior mod programs don't offer excuses for the failure of the
manipulations to have their intended effects, they use systematic data
collection to identify the source of the problem and find a way to correct it.

The basic question is whether behavior really works as each side says it
works. Resolving that question requires thinking up tests that could
give preference to one theory over the other. It requires taking
failures seriously and using only observed data to explain them. If
Johnny doesn't respond to operant conditioning because of other more
powerful reinforcers, show what the other reinforcers were and prove
that they are more powerful; don't just say they must have existed. If
Johnny tries to control and fails, show what the detailed causes of the
failure were, and _prove_ that they will always cause such a failure.

Yes, I agree, and so would any competent behavior analyst. As I noted in my
post, my pointing to these putative sources of reinforcement was done only
to show that such an accounting is _possible_, not that these sources are in
fact the correct explanation for your behavior. Finding the correct
explanation would take research.

In fact one of the problems for both _applied_ behavior analysis and
_applied_ PCT is that you don't know all the factors at work in a given
individual, whether the relevant factors are thought to be the history of
reinforcement and current contingencies or the relevant set of controlled
perceptions up and down the hierarchy. This means that the job involves a
lot of educated guesswork, based on observation and perhaps a bit of
experimentation. Even then it's not always going to work, because
information is incomplete and your ability to exert influence is limited.

In a basic research setting, we often can simplify the situation enormously.
For example, by using naive rats, I can be fairly well assured that my
subjects will not be motivated by a desire to frustrate the experimenter, as
that bigger rat you mentioned appears to be, and I can do a pretty good job
of eliminating most alternative sources of reinforcement, other than the one
or two whose effects I wish to study.

    So, as a behavior analyst, my problem is that my paltry $1.00 per
    minute is no match, in terms of reinforcing power, for the
    reinforcement you receive from Rick's approval,

Pay closer attention. I said that I didn't choose to do it Rick's way
because of the reward, but for a different reason all my own. Of course
as a behaviorist you had to assume that the larger reward prevailed,
even if it didn't, and that my "reasons" were illusions.

I did pay close attention--did you? I didn't say you choose to do it Rick's
way because of the reward, I offered several possibilities, including that
you find it rewarding to frustrate reinforcement theorists. As a
reinforcement theorist, I would not assume that your reasons are illusions
(they may or may not be), but that, whatever they may actually be, those
reasons can be traced back to your experiences.

Regards,

Bruce

[Bill_Powers1]

[From Bill Powers (2011.07.04.1450 MDT)]

Rick Marken (2011.07.04.1230) --

RM: I just tried filtering the causal model output, using a linear and the exponential filter, and it barely improves things at all f(the R^2 for the causal model goes from .47 to .49, still less than the R^2 of .62 for the control model).

BP: I take it that these are both correlations with the real behavior. The causal model accounts for 79% as much variance as the control model, so it's pretty close. I still don't like "causal" but I know what you mean.

For the disturbance, try putting in a disturbance of the cursor just after the color changes, with an amplitude of 30 and 90 (at random) or -30 and -90, choosing the sign to move the cursor in the direction that the cursor is to go. That should produce mouse movements with a zero average, I think.

Best,

Bill P.

[rsmarken]

[From Rick Marken (2011.07.04.2200)]

Bill Powers (2011.07.04.1450 MDT)–

Rick Marken (2011.07.04.1230) –

RM: I just tried filtering the causal model output, using a linear and the exponential filter, and it barely improves things at all f(the R^2 for the causal model goes from .47 to .49, still less than the R^2 of .62 for the control model).

BP: I take it that these are both correlations with the real behavior. The causal model accounts for 79% as much variance as the control model, so it’s pretty close. I still don’t like “causal” but I know what you mean.

And that is the result of comparing the correlations (with real behavior) of the filtered causal (open loop, if you prefer) and control model sans disturbances to the cursor. If this pans out, doesn’t it suggest that the dynamics of the filtered output of the open-loop model is not equivalent to that of the control model? After all the control model does have three levels and there is a disturbance to the perception controlled by the top level (the logical relationship between output and cursor color).

For the disturbance, try putting in a disturbance of the cursor just after the color changes, with an amplitude of 30 and 90 (at random) or -30 and -90, choosing the sign to move the cursor in the direction that the cursor is to go. That should produce mouse movements with a zero average, I think.

I was actually planning to do that in my next series of experiments. I certainly have more work to do on this. But we can discuss this in greater depth at the meeting. I look forward to it.

Best

Rick

···

–
Richard S. Marken PhD
rsmarken@gmail.com
www.mindreadings.com

[PHILIP_JERAIR_YERANO]

Content-Type: application/pdf; name="Adlam_fundamental_essay_Jan.pdf"
Content-Disposition: attachment; filename="Adlam_fundamental_essay_Jan.pdf"
X-Attachment-Id: f_jjudxd880

Adlam_fundamental_essay_Jan.pdf (94.2 KB)

[Bruce_Abbott]

[From Bruce Abbott (950724.2150 EST)]

Rick Marken (950724.1800) --

I found a non-linear relationship between ratio requirement and
reinforcements/hr and a linear relationship between ratio requirement
and seconds/reinforcement.

Good! What does that linear relationship imply? Think now . . . what
could it POSSIBLY mean?

This finding is (as I noted above) irrelevant
to the problem with your approach to predicting reinforcements/hr
from seconds/reinforcment.

[Ah, so that's it: he thinks I'm trying to predict reinforcements/hr from
seconds/reinforcement. Where on earth . . . ?] Rick, please reread the
post--carefully. You're off jumping to confusions again. >:-<

Regards,

Bruce

[William_T_Powers4]

[From Bill Powers (950727.0600 MDT)]

Bruce Abbott (95062x) --

Having essentially capitulated yesterday, I decided to try to write the
complete equations for the relationships you proposed.

Let R = observed reinforcement rate
    B = observed behavior rate
    b = pressing rate within ratio
    c = collection time (constant)
    m = scheduled ratio

(1) R = 1/(m/b + c)

(2) B = m/(m/b + c)

  dividing (1) by (2) we get R/B = m, the ratio.

In your derivation, you proposed that 1/R = c + k*m. Since R is a
function of m, b, and c, we must eliminate R by using (1):

m/b + c = c + k*m,

from which it is evident that

k = 1/b.

What you have shown empirically, therefore, is that

1/R = c + m/b, or

R = 1/(m/b + c),

which is equation (1).

···

------------------------------------
Equation (1) gives us R as a function of b and the ratio. It describes
how the apparatus will react to behavior at a rate b alternating with
nonbehaving periods of duration c. As b increases, R will increase
nonlinearly to a limit 1/c. Therefore the maximum possible reinforcement
rate for any behavior rate and any ratio is 1/c.

This, however, gives us only a description of the environment. To solve
for the actual behavior rate and reinforcement rate, we must have a
second equation describing the operation of the organism.

The control-theoretic equation can be rendered simply as

b = g*(R' - R),

meaning that the behavior rate between collection periods is a gain
constant g times the difference between the desired reinforcement rate
R' and the received reinforcement rate R.

substituting from (1) for R, we have

b = g*R' - g/(m/b + c), or

b = g*R' - g*b/(m + b*c), or

b(1 + g/(m + b*c)) = g*R'

If c is zero, this equation becomes

b' = g*R'/(1 + g/m).

At a ratio of m = 1, the loop gain is just g, and if g is very large in
comparison with 1, b' = R. The loop gain becomes smaller as m increases.
If the collection time c is not zero, the gain becomes smaller still in
proportion to behavior rate.

OPCOND5 implements this control equation, also showing temporal details
due to individual behaviors and reinforcements.
-----------------------------------------------------------------------
In your analysis, you used only the environment equation showing how R
depends on b. No manipulation of this equation, with or without
empirical backing, can do any more than tell you how R depends on b.
What you proved using the data was that equation (1) is a very good
representation of the apparatus.

What you did not prove, however, is that there is no control. To do
that, you would have to propose an equation representing the way b
depends on R via a different path, the path through the organism rather
than the apparatus. Then you would have to show that this equation does
not result in solutions that fit the data.

It remains to be seen whether the control equation does fit the data.
Because the loop gain becomes so low when there is a collection period
and a high ratio, the model will predict very poor control over a large
part of the range, but it should fit the data over the entire range all
the way to FR-1.
-----------------------------------------------------------------------
Best,

Bill P.

[Susan_McCormack]

I was able to get a better perspective on some of the difficulties I have been facing and as a result my perceptions and belief in my ability to do something about them changed!

[Oliver]

[From Oliver Schauman (2011.08.26. 22:55GMT)]

Greetings CSGNET!

I hope you have had a fruitful annual meeting.

Question:

So I have set up two control loops- one with a reference value of 2 and the other a reference value of 6, controlling a loop on the level below.

I have attempted to add a reorganising system that alters the output gains of these two (conflicting systems). Might you experts be able to comment on whether this is looking like real PCT reorganization? Figure 1 is of the total mean sq error across the two systems and how it changes as the model is run. Figure 2 is the input and the error signal of the loop with a ref signal of 2.

Any helpful comments would be very much appreciated.

Best wishes,

Oliver Schauman

[rsmarken]

[From Rick Marken (2011.08.26.1615)]

Oliver Schauman (2011.08.26. 22:55GMT)–

Greetings CSGNET!

I hope you have had a fruitful annual meeting.

It was great. Best meeting in years!

Question:

So I have set up two control loops- one with a reference value of 2 and the other a reference value of 6, controlling a loop on the level below.

I don’t understand. Could you send a little diagram of what you modeled.

I have attempted to add a reorganising system that alters the output gains of these two (conflicting systems). Might you experts be able to comment on whether this is looking like real PCT reorganization?

I’d be happy to but I need to see a diagram of the model.

Figure 1 is of the total mean sq error across the two systems and how it changes as the model is run. Figure 2 is the input and the error signal of the loop with a ref signal of 2.

Any helpful comments would be very much appreciated.

When I see the diagram I’d be happy to try to make some comments. A print out of the model code might be helpful as well; I can probably make it out no matter what language it was written in.

Best

Rick

···

Best wishes,

Oliver Schauman

–

Richard S. Marken PhD
rsmarken@gmail.com
www.mindreadings.com

[MartinT]

[From Oliver Schauman (2011.08.26.
22:55GMT)]

    Figure 1 is of the total mean sq error across the two systems

and how it changes as the model is run.

[Oliver]

[From Oliver Schauman (2011.08.28. 09:50GMT)]

[Martin Taylor 2011.08.28.00.28]

[From Oliver Schauman (2011.08.26. 22:55GMT)]

Figure 1 is of the total mean sq error across the two systems and how it changes as the model is run.

How did you manage to get a negative mean square error?

Oh sorry, Fig 1 is the change in total mean sq error.

Oliver

[Bill_Williams_UMKC]

Bill,

Its simple, all you have to do is inadvertantly insert a integration term in one
of the loops.

Once a model gets beyond a single loop I find that I encounter problems for
which there is no literature with readily accessible explainations. Such as the
problem generated by a budget loop switching on and off. Ordinarily we control
our budgets so as not to spend too much money. If spending is less than budget
constraint I don't expect the budget loop to increase expenditures. And,
modeling conflict between two goals is much more difficult than I would have
initially expected. If I was building analog circuits there is an enourmous
applicaions literature availible explaining how to implement various functions.
I've not found this to be true of control theory. Most of the circuit
literature was generated to promote the use/sale of a manufacture's componates.
Maybe Richardson's proposed book can serve the same purpose for control theory
code. And, while you may be reluctant to write a book on control theory there
is obviously a need for a text that would illustrate (with accompanying code)
basic functions in a fashion similiar to what you find in Burr-Brown or Analog
Devices catalogs.

After I make a few changes I'll send you my conflict demo. There's probablely a
better way of demonstrating conflict, but then I haven't seen anything in print
_anywhere_ about how to model conflict.

Best
  Bill

···

______________________________________________________________________
Do you want a free e-mail for life ? Get it at http://www.email.ro/

[Bill_Williams_UMKC]

[ From : Bill Williams Sunday 26 August 2001 18:30 CST ]

  CSGnet Members,

In the attached file there is "An International Open Letter to All Economics
Departments: An Invitation for Reconsideration" which requests that economics
departments reconsider their narrow orthodox approach to economic questions by
adopting a more open, pluralistic methodology including : 1) A broader
conception of human behavior, 2) Recognition of culture, 3) consideration of
history, 4) a new theory of knowledge, 5) empirical grounding, 6) Expanded
methods. In full, the first item in the list states: "The definition of
_Economic Man_ as an autonomous rational optimizer is too narrow and does not
allow for the roles of other determinants such as instinct, habit formation and
gender, class and other social factors in shaping the economic psycyhology of
social agents."

I am inviting comment on the use of the adjective "shaping" in the statement
above. "Shaping" is a term drawn from the psychology behaviorism. The
Institutionalist Summer School at UMKC in which the statement appears is a part
of a heterodox spectrum of traditions in economics which have, in part, adopted
psychological behaviorism as an alternative assumption to orthodoxy's assumption
that economic rationality consistes of maximization. The use of "shaping" would
appear to indicate a continuation among heterodox economists of an assumption
that behaviorism provides a psychological conception which is useful for
heterodox economic inquiry. If CSG members share my concerns in this respect I
would appreciate comments. If on the contrary it appears that my concern is
misplaced I would also wish to have adverse comments.

Best to all
  Bill Williams

···

______________________________________________________________________
Do you want a free e-mail for life ? Get it at http://www.email.ro/

        AN INTERNATIONAL OPEN LETTER TO ALL ECONOMICS DEPARTMENTS:

                     AN INVITATION FOR RECONSIDERATION.

Economics needs fundamental reform – and now is the time for change.

This document comes out a meeting of 75 students, researchers and
professors from twenty-two nations who gathered for week of discussion on
the state of economics and the economy at the University of Missouri -
Kansas City (UMKC) this June 2001. The discussion took place at the Second
Biennial Summer School of the Association for Evolutionary Economics
(AFEE), jointly sponsored by UMKC, AFEE and the Center for Full Employment
and Price Stability.

The undersigned participants, all committed to the reform of our
discipline, have developed the following open letter. This letter follows
statements from other groups who have similar concerns. Both in agreement
with and in support of the Post-Autistic Economics Movement and the
Cambridge Proposal, we believe that economic theory, inhibited by its
ahistorical approach and abstract formalist methodology, has provided only
a limited understanding of the challenging complexity of economic behavior.
The narrow methodological approach of economics hinders its ability to
generate truly pragmatic and realistic policy prescriptions or to engage in
productive dialogue with other social sciences.

All economics departments should reform economics education to include
reflection on the methodological assumptions that underpin our discipline.
A responsible and effective economics is one that sees economic behavior in
its wider contexts, and that encourages philosophical challenge and debate.
Most immediately, the field of economic analysis must be expanded to
encompass the following:

1. A broader conception of human behavior. The definition of economic man
as an autonomous rational optimizer is too narrow and does not allow for
the roles of other determinants such as instinct, habit formation and
gender, class and other social factors in shaping the economic psychology
of social agents.

2. Recognition of culture. Economic activities, like all social phenomena,
are necessarily embedded in culture, which includes all kinds of social,
political and moral value-systems and institutions. These profoundly shape
and guide human behavior by imposing obligations, enabling and disabling
particular choices, and creating social or communal identities, all of
which may impact on economic behavior.

3. Consideration of history. Economic reality is dynamic rather than static
– and as economists we must investigate how and why things change over time
and space. Realistic economic inquiry should focus on process rather than
simply on ends.

4. A new theory of knowledge. The positive-vs.-normative dichotomy which
has traditionally been used in the social sciences is problematic. The
fact-value distinction can be transcended by the recognition that the
investigator’s values are inescapably involved in scientific inquiry and in
making scientific statements, whether consciously or not. This
acknowledgement enables a more sophisticated assessment of knowledge
claims.

5. Empirical grounding. More effort must be made to substantiate
theoretical claims with empirical evidence. The tendency to privilege
theoretical tenets in the teaching of economics without reference to
empirical observation cultivates doubt about the realism of such
explanations.

6. Expanded methods. Procedures such as participant observation, case
studies and discourse analysis should be recognized as legitimate means of
acquiring and analyzing data alongside econometrics and formal modeling.
Observation of phenomena from different vantage points using various
data-gathering techniques may offer new insights into phenomena and enhance
our understanding of them.

7. Interdisciplinary dialogue. Economists should be aware of diverse
schools of thought within economics, and should be aware of developments in
other disciplines, particularly the social sciences.

Although strong in developing analytic thinking skills, the professional
training of economists has tended to discourage economists from even
debating – let alone accepting – the validity of these wider dimensions.
Unlike other social sciences and humanities, there is little space for
philosophical and methodological debate in the contemporary profession.
Critically-minded students of economics seem to face an unhappy choice
between abandoning their speculative interests in order to make
professional progress, or abandoning economics altogether for disciplines
more hospitable to reflection and innovation.

Ours is a world of global economic change, of inequality between and within
societies, of threats to environmental integrity, of new concepts of
property and entitlement, of evolving international legal frameworks and of
risks of instability in international finance. In such a world we need an
economics that is open-minded, analytically effective and morally
responsible. It is only by engaging in sustained critical reflection,
revising and expanding our sense of what we do and what we believe as
economists that such an economics can emerge.

Signed by:

Ricardo Aguado (Spain) Universídad del País Vasco

Dr. Stephen Dunn (UK) Staffordshire University

Dr. Eric R. Hake (USA) Eastern Illinois University

Fadhel Kaboub (Tunisia) University of Missouri- Kansas City

Nitasha Kaul (India) University of Hull, UK

Peter Kimani (Kenya) University of Nairobi

Meelis Kitsing (Estonia) London School of Economics

Agim Kukeli (Albania) Colorado State University

Joelle Leclaire (Canada) University of Missouri - Kansas City

Áine Ní Léime (Ireland) National University of Ireland - Galway

Hui Liu (China) University of Ottawa

Claudia Maya (Mexico) National Autonomous University of Mexico

Dr. Andrew Mearman (UK) Wagner College, USA

Jaime Augusto Torres Melo (Colombia) London School of Economics

Vassilis Monastiriotis (Greece) London School of Economics

Alfred Ng Yau Foo (Malaysia) University of Missouri -Kansas City

José Alfredo Pureco Ornelas (Mexico) National Autonomous University of
Mexico

Jairo J. Parada (Colombia) Penn State University

Franziska M. Pircher (USA) University of Missouri -Kansas City

David Pringle (Canada) University of Ottawa

Dr. James F. Smith (USA) University of Vermont

Pavlina R. Tcherneva (Bulgaria) Center for Full Employment and Price
Stability, UMKC

Ermanno Celeste Tortia (Italy) University of Ferrara

Eric Tymoigne (France) Université de Paris - Nord

Benton Wolverton (USA) University of Missouri - Kansas City

Questions, comments and critiques may be directed to David Pringle
(pringle_djg@hotmail.com) or Áine Ní Léime (ainenileime@hotmail.com).

[Bill_Powers1]

[From Bill Powers (2001.08.27.1307 MDT)]

Bill Williams Sunday 26 August 2001 18:30 CST --

I am inviting comment on the use of the adjective "shaping" in the statement
above. ... The use of "shaping" would appear to indicate a continuation

among >heterodox economists of an assumption that behaviorism provides a

psychological conception which is useful for heterodox economic inquiry.

OK, here are some comments you might make some use of:

You may be right, but I believe "shaping" is also used in a more general
sense to indicate that behavior is influenced or determined by the factors
named, through some causal process. Skinner's formal procedure called
"shaping" needs a purposive control system to carry it out; the inanimate
environment or abstract concepts can't carry out shaping in this technical
sense.

To conduct shaping in Skinner's sense it is necessary for the animal
trainer to observe (perceive) the animal's behavior and compare what is
observed with the behavior that is desired by the trainer (reference
signal). If the difference gets smaller, meaning that the animal has
changed its behavior in some way to make the animal trainer's perception
come a little closer to his or her reference condition, a "reinforcer", is
given to the animal for the purpose of making the observed behavior come
still closer to the desired behavior. By giving and withholding
reinforcers, the trainer controls the animal's actions, bringing them
gradually closer to the pattern the trainer wants to see.

Notice that this procedure requires judgments to be made as to whether a
given behavior of the other organism is closer to or farther from a
goal-behavior that has not occurred yet. It seems clear that "instinct,
habit formation and gender, class and other social factors" are not capable
of acting on an organism on the basis of such judgments, so the term
"shaping" used in this context clearly doesn't mean the procedure Skinner
worked out. The remaining meaning can only be the more general sense of
influencing or determining.

I would suggest that if heterodox economists wish to institute significant
reforms in their discipline, one of the most fruitful places to begin would
be in revising the theory of human behavior that has underlain not only
economics but all the social sciences since their beginnings. I would
suggest that the reforms not be limited to the question of _which_ social
or other influences cause human beings to behave as they do, but _whether_
such factors bear any causal relationship to human behavior at all. The
question of "whether" clearly takes precedence, because if the relationship
of any external factors to behavior is not one of causation, there is no
point in asking which factors cause behavior.

In the control-theory view of behavior, the immediate causes of behavior in
any organism lie inside the organism in the form of a structure of
goal-seeking control systems. The goals are not external to the organism,
although external situations and influences may have an identifiable
relationship to them. Rather, the goals for any given behavioral process
are in the form of reference conditions specified by higher-level systems,
or inborn systems. The goals do not define specific actions, either, but a
set of _perceived_ conditions that depend not only on the physical state of
the organism and its environment, but on the manner in which the organism
represents the environment to itself, inside itself. On its perceptual
systems, in other words.

Relationships between human behavior and many environmental factors have
been found, with rather low reliability, to exist. But if human beings are
actually organized as control systems, the most obvious interpretation of
such relationships is not necessarily the correct one. What seems on the
surface to be a causal relationship may actually conceal a quite different
relationship in which the organism's perceptions and goals play a
non-obvious part.

Consider, for instance, a simple cause-effect relationship in which a
person throws a punch at an experimental subject, and we observe that as
soon as the subject sees the fist approaching he ducks. The
simplest-seeming interpretation is that the sight of the approaching fist
was a stimulus that caused the response of ducking. We can even imagine a
mechanism: the visual stimulus caused neural impulses to travel from the
eyes to the brain, after which they were relayed to the muscles that
produce ducking movements.

There are several problems with this interpretation, however. The biggest
problem is to explain how it is that the muscles reached by the neural
signals happened to be just the ones required to evade exactly that
direction of punching, and that speed, using the body in the exact
orientation and configuration it happened to be in at the moment the punch
started. The moment we begin asking about details, the proposed
interpretation stops looking so simple. Now we have to bring in such things
as past experience, practice, generalization from one case to slightly
different cases, modification of causal links by other stimuli that alter
the link in just the way required, and so forth. As we begin to see all the
ramifications and complications needed to make this "simple" explanation
work, we might be more willing to consider alternatives.

One alternative, the one suggested by control theory, is that the subject
had set a very low reference level for the experience of being punched. He
didn't want to be hit. So the person took action designed to prevent being
hit, the direction and amount of action being influenced, of course, by the
direction and speed of the incipient blow, but being even more crucially
influenced by _the subject's desire not to be hit_. In fact,if the subject
had wished for the punch to reach his chin, the same "stimulus" would have
been followed by a very different "response." The stimulus itself, the
sight of the approaching fist, in fact, has no inherent or necessary
relation to any behavior that may follow it. Even if the behavior is quite
predictable, this does not mean that the stimulus "caused" it. Under
control theory, the purpose of any ensuing behavior is to modify the effect
that the person is experiencing or expects shortly to experience. And
behind that is the person's goal for having or not having a particular
experience. If most people duck a punch, this is not because punches cause
ducking, but because most people other than Hollywood stunt men and
masochists wish to avoid being hit.

All instances in which it seems that some stimulus, condition, influence,
situation, or factor "causes" a certain kind of behavior can be
reinterpreted as control behavior. You say you are angry because a thief
stole your money; would you be angry if you had wanted him to steal your
money? You say you worked harder to get a good recommendation from your
boss; would you have worked harder if you weren't trying to get a good
recommendation? Is a reward rewarding if you don't want it (meaning: set a
high level of the reference condition for having it)?

The route to any new interpretation begins with trying out an alternative.
If the alternative seems plausible on first view, the next step is to look
for experimental ways of testing both the original and the alternative.
There are methods for testing the idea that a person is controlling some
particular kind of variable, and for finding out approximately what the
reference condition is -- just how much of that variable is wanted (or how
strongly any of it is rejected). If something is found that is actually
under control by a person, then external influences and organismic actions
that affect the controlled variable will be seen to have a cause-like
relationship to each other, but the actual relationship will have been
shown not to be causal in the simpler sense.

The language of control theory (and the quantitative methods that underlie
it) offer a new interpretation of economic life, and indeed of all life
whether related to economics or not. Control theory offers not just a
choice between one set of causal factors and another thought to be more
important. It offers a radically new view of behavior itself, and of its
relationship to all supposedly causal factors and events.

Best,

Bill P.

  If CSG members share my concerns in this respect I

···

would appreciate comments. If on the contrary it appears that my concern is
misplaced I would also wish to have adverse comments.

Best to all
Bill Williams

______________________________________________________________________
Do you want a free e-mail for life ? Get it at http://www.email.ro/

       AN INTERNATIONAL OPEN LETTER TO ALL ECONOMICS DEPARTMENTS:

                    AN INVITATION FOR RECONSIDERATION.

Economics needs fundamental reform � and now is the time for change.

This document comes out a meeting of 75 students, researchers and
professors from twenty-two nations who gathered for week of discussion on
the state of economics and the economy at the University of Missouri -
Kansas City (UMKC) this June 2001. The discussion took place at the Second
Biennial Summer School of the Association for Evolutionary Economics
(AFEE), jointly sponsored by UMKC, AFEE and the Center for Full Employment
and Price Stability.

The undersigned participants, all committed to the reform of our
discipline, have developed the following open letter. This letter follows
statements from other groups who have similar concerns. Both in agreement
with and in support of the Post-Autistic Economics Movement and the
Cambridge Proposal, we believe that economic theory, inhibited by its
ahistorical approach and abstract formalist methodology, has provided only
a limited understanding of the challenging complexity of economic behavior.
The narrow methodological approach of economics hinders its ability to
generate truly pragmatic and realistic policy prescriptions or to engage in
productive dialogue with other social sciences.

All economics departments should reform economics education to include
reflection on the methodological assumptions that underpin our discipline.
A responsible and effective economics is one that sees economic behavior in
its wider contexts, and that encourages philosophical challenge and debate.
Most immediately, the field of economic analysis must be expanded to
encompass the following:

1. A broader conception of human behavior. The definition of economic man
as an autonomous rational optimizer is too narrow and does not allow for
the roles of other determinants such as instinct, habit formation and
gender, class and other social factors in shaping the economic psychology
of social agents.

2. Recognition of culture. Economic activities, like all social phenomena,
are necessarily embedded in culture, which includes all kinds of social,
political and moral value-systems and institutions. These profoundly shape
and guide human behavior by imposing obligations, enabling and disabling
particular choices, and creating social or communal identities, all of
which may impact on economic behavior.

3. Consideration of history. Economic reality is dynamic rather than static
� and as economists we must investigate how and why things change over time
and space. Realistic economic inquiry should focus on process rather than
simply on ends.

4. A new theory of knowledge. The positive-vs.-normative dichotomy which
has traditionally been used in the social sciences is problematic. The
fact-value distinction can be transcended by the recognition that the
investigator�s values are inescapably involved in scientific inquiry and in
making scientific statements, whether consciously or not. This
acknowledgement enables a more sophisticated assessment of knowledge
claims.

5. Empirical grounding. More effort must be made to substantiate
theoretical claims with empirical evidence. The tendency to privilege
theoretical tenets in the teaching of economics without reference to
empirical observation cultivates doubt about the realism of such
explanations.

6. Expanded methods. Procedures such as participant observation, case
studies and discourse analysis should be recognized as legitimate means of
acquiring and analyzing data alongside econometrics and formal modeling.
Observation of phenomena from different vantage points using various
data-gathering techniques may offer new insights into phenomena and enhance
our understanding of them.

7. Interdisciplinary dialogue. Economists should be aware of diverse
schools of thought within economics, and should be aware of developments in
other disciplines, particularly the social sciences.

Although strong in developing analytic thinking skills, the professional
training of economists has tended to discourage economists from even
debating � let alone accepting � the validity of these wider dimensions.
Unlike other social sciences and humanities, there is little space for
philosophical and methodological debate in the contemporary profession.
Critically-minded students of economics seem to face an unhappy choice
between abandoning their speculative interests in order to make
professional progress, or abandoning economics altogether for disciplines
more hospitable to reflection and innovation.

Ours is a world of global economic change, of inequality between and within
societies, of threats to environmental integrity, of new concepts of
property and entitlement, of evolving international legal frameworks and of
risks of instability in international finance. In such a world we need an
economics that is open-minded, analytically effective and morally
responsible. It is only by engaging in sustained critical reflection,
revising and expanding our sense of what we do and what we believe as
economists that such an economics can emerge.

Signed by:

Ricardo Aguado (Spain) Univers�dad del Pa�s Vasco

Dr. Stephen Dunn (UK) Staffordshire University

Dr. Eric R. Hake (USA) Eastern Illinois University

Fadhel Kaboub (Tunisia) University of Missouri- Kansas City

Nitasha Kaul (India) University of Hull, UK

Peter Kimani (Kenya) University of Nairobi

Meelis Kitsing (Estonia) London School of Economics

Agim Kukeli (Albania) Colorado State University

Joelle Leclaire (Canada) University of Missouri - Kansas City

�ine N� L�ime (Ireland) National University of Ireland - Galway

Hui Liu (China) University of Ottawa

Claudia Maya (Mexico) National Autonomous University of Mexico

Dr. Andrew Mearman (UK) Wagner College, USA

Jaime Augusto Torres Melo (Colombia) London School of Economics

Vassilis Monastiriotis (Greece) London School of Economics

Alfred Ng Yau Foo (Malaysia) University of Missouri -Kansas City

Jos� Alfredo Pureco Ornelas (Mexico) National Autonomous University of
Mexico

Jairo J. Parada (Colombia) Penn State University

Franziska M. Pircher (USA) University of Missouri -Kansas City

David Pringle (Canada) University of Ottawa

Dr. James F. Smith (USA) University of Vermont

Pavlina R. Tcherneva (Bulgaria) Center for Full Employment and Price
Stability, UMKC

Ermanno Celeste Tortia (Italy) University of Ferrara

Eric Tymoigne (France) Universit� de Paris - Nord

Benton Wolverton (USA) University of Missouri - Kansas City

Questions, comments and critiques may be directed to David Pringle
(pringle_djg@hotmail.com) or �ine N� L�ime (ainenileime@hotmail.com).

[Bruce_Gregory]

[From Bruce Gregory (2001.08.27.1743)]

Bill Powers (2001.08.27.1307 MDT)

Very elegant. Very clear.

[Wolfgang_Zocher3]

[From Wolfgang Zocher (970823.1945 MEZ)

[Bill Powers (970820.1353 MDT)

Hi, Bill --

as an addendum to my first short post on LOGO:

LOGO is a very high-level list processing language that is,
in actuality, a special version of LISP. (Both developed at
MIT). Like its ancestor LISP is has a small discrete vocabulary
of primitive function words that allow student programmers to
build their own programming vocabulary. Children can define
other primitives with just one command and these new primitives
will operate exactly like the other primitives already in the
language.
The most fascinating aspect of LOGO is turtle graphics. With this
feature, the language contains what educators recognize as an
immediate feedback mechanism. The result of all programming efforts
is directly recognizable on the screen.
During the first lesson of my LOGO-courses, I told the children,
that they have to teach the turtle what they want it to do. So
the children are turned into teachers who learn for themselves.
I started with a limited repertoire of turtle commands and as
a first step, the children had to give commands to move the
turtle around, rotoate the turtle ... to get a feeling of what
the turtle can do AND where the limits are.
All my further efforts went in the direction of teaching
"structured programming" and "top down design" and some basic
features common to all programming languages (repeating parts
of programs and selection of program paths (if then else)).
Adding new words to the LOGO-language is a basic need for programming
in LOGO.
Most children immediatly recognize that

FD 100 RT 90 FD 100 RT 90 FD 100 RT 90 FD 100 RT 90

is FD 100 RT 90 repated four times and ask how to define an
"abbreviation word" for this sequence which the turtle "understands".

the first step is

REPEAT 4 [FD 100 RT 90]

and the second step

TO SQUARE
REPEAT 4 [FD 100 RT 90]
END

defines this new word. Result: the children have taught the
new word SQUARE to the turtle. And with this experience the base
for creating lots of buildings blocks (like LEGO's) for
complex graphics (means: complex programs) is defined.

From my experiences with LOGO as a first programming language

I can state that if only the basics of program structure and
program design are understood you can switch to each other
programming language (I prefere PASCAL) or stay LISPish:-)
[ During the past 15 years I also did programming courses
  (PASCAL, C, BASIC, LISP) for adults and recognized, that the main
  difficulties arose from the unability to have an idea of
  "structure" in programs and structure in the data. They
  all were able to learn the syntax of a new language, but
  laerning a language is not the same as programming]

From my point of view the main advantages of LOGO are:

  - the visual feedback of programming
  - the child as a teacher for the turtle.

Since I'm familiar with different programming languages I
avoid discussions about "THE best" language. When I'm planning
a program I decide which language is best for the problem I
have to solve and so I do my programming tasks in very
different languages. I like FORTRAN as much as C and LISP (LOGO)

Best,
Wolfgang

[Bill_Powers1]

Hi, Wolfgang --

as an addendum to my first short post on LOGO:

LOGO is a very high-level list processing language that is,
in actuality, a special version of LISP. (Both developed at
MIT). Like its ancestor LISP is has a small discrete vocabulary
of primitive function words that allow student programmers to
build their own programming vocabulary. Children can define
other primitives with just one command and these new primitives
will operate exactly like the other primitives already in the
language.

This is what Forth is supposed to be, too. The existence of such languages
must be telling us something about how the human "program level" (as I call
it) really works. If we knew what the actual "primitive" operations of the
program level were, we could see all other kinds of program-like thinking
as simply applications of the basic operations.

The most fascinating aspect of LOGO is turtle graphics. With this
feature, the language contains what educators recognize as an
immediate feedback mechanism. The result of all programming efforts
is directly recognizable on the screen.

Of course this has nothing to do with the structure of LOGO itself, does
it? The same visual display could be pre-programmed as a library procedure
in any programming language.

During the first lesson of my LOGO-courses, I told the children,
that they have to teach the turtle what they want it to do. So
the children are turned into teachers who learn for themselves.

That's a nice approach: give them control, for a change.

TO SQUARE
REPEAT 4 [FD 100 RT 90]
END

defines this new word. Result: the children have taught the
new word SQUARE to the turtle. And with this experience the base
for creating lots of buildings blocks (like LEGO's) for
complex graphics (means: complex programs) is defined.

Again, while this is a very good teaching method, it doesn't seem unique to
Logo.

From my point of view the main advantages of LOGO are:

- the visual feedback of programming
- the child as a teacher for the turtle.

But my point is that if you provided similar library routines for any other
language, you could say the same thing about them. The main feature of Logo
that I see is that it's interpretive and can be run one "word" at a time,
which is very helpful for seeing what the expressions accomplish. I once
had a copy of "Magic Pascal" or some such name, which was written as
multiprocessing compiler that would execute each line of code as it was
written, compiling at the same time (which can be done in Pascal because
it's a one-pass compiler). That would allow Pascal to be taught the same way.

Best,

Bill P.

[Lloyd_Klinedinst2]

[From Lloyd Klinedinst (980818.20.40)]

Hi Bill,

As there seemed to be some talk about making the private lists public on
CSG I'll post to this site.

Thanks very much, Tim - and thanks, Bill, for the reply.

[_Avery_Andrews]

[from Avery Andrews (920815)]
(penni sibun 920814, on marken 920808)

on my view, you talk like a cognitivist. partly cause you oppose pct
to behaviorism. more tellingly, cause you explicitly locate control
in the head, just as cognitivists do. ...

What I'd say to this is that PCT is not cognitivist, because the *phenomenon*
of control is only found when an organism (or other `Agent') is put in an
appropriate environment (e.g. if the road is covered with an oil-slick,
one ceases to observe control of the direction in which the car is going).
The phenomenon of control is just what Rick says it is: when you introduce
disturbances that you would expect to change some aspect of the environment,
but that aspect doesn't change, because the Agent consistently does things
that `mysteriously' counteract your attempted disturbances.

There is a terminological difficulty here in that in normal usage,
`control' does *not* imply any capacity to cope with unpredictable
disturbances as they arise (I have retained a copy of one of Randy
Beer's postings that documents this point extensively, which I could send
to anyone who'se interested in looking at it). So for example,
most people would call the spikey cylinder in a music box its
`control unit', but this object does not effect control in the PCT
sense, since it doesn't counteract disturbances in any systematic way.

PCT goes on to say (at least) two further things:

  a) an Agent that is efficacious in the real world must effect control

  b) the only way control can be effected is by means of certain kinds
     of internal arrangments - perceptors, comparators & effectors
     appropriately connected so as to constitute negative feedback loops
     in the context of the actual environment.

The argument for (a) is that the sorts of things that serious
Agents might need or be designed to achieve are going to be subject to
constant and unpredictable disturbances, which will then need to be
counteracted as they arise, and the argument for (b) is that no alternative
arrangments that effect control have been proposed (for example, the reason
that Sonja could cope with conveyerbelts, whirlwinds, and, let us say,
crunchbirds that would occasionally eat her monster-killing missiles in
mid-flight, is that her central architecture is full of little control
systems, even if Chapman didn't think of them that way).

So I'd see PCT as a subtype of interactive AI, which makes some additional
claims about how to make sense out of how the internal structure of
an Agent is related to its `behavior' (e.g. what happens in environments
that contain it). I'd take what I've come up with from cogitating about Sonja
so far as supportive of these claims (and am encouraged that Penni finds it
interesting) but there's still a lot I don't know, & there might be all
sorts of genuinely efficacious but not-PCT organization in Sonja that I have
missed. But my experience so far is that PCT ideas really are
helpful in figuring these things out (recall that I got on to Chapman
& Agre when I showed Penni my proposed control system organization for
getting beer, & she said that it was like their stuff).

What I would like to do is design and/or analyse some relatively large-scale
agent along PCT lines - the crowd people and astro are a start, but they
are obviously too rudimentary to convince people of the utility of
the approach, in part because they don't have any `tactics'. Crowd
people, for example, avoid obstacles, but they don't really navigate
in the way that Sonja does. if you set things up so that there is a
wall of stationary people between a mobile crowd person and her goal:

                               P
                               P
             G P Goal
                               P
                               P
                               P
                               P

(you can position stationary people by making them `active', but setting
their gains to zero), G will head straight for the Goal until she is about
to crash into the row of P's, and will then veer off one way or the
other, not consistently taking the shortest way. But Sonja (or my dog)
would head immediately to whichever end of the wall would give the shortest
path around. I conjecture that the way to get the crowd people to navigate
would be to give them more sophisticated perceptual functions (maybe even
tuning the ones they have would do the trick - tho I haven't managed
to do this), but to say this is not of course the same as to actually
do it (Sonja in effect controls for not having a wall between where she
is and where she wants to be going, according to my current understanding).

Avery.Andrews@anu.edu.audi