The Behavioural Illusion and modelling

[MartinT]

[Martin Taylor 2016.08.30.13.12]

The idea of the behavioural illusion is deceptively simple on its

surface. If some perception is perfectly controlled, the influence
of the disturbance on the corresponding environmental variable (the
CEV or “Complex Environmental Variable”) is exactly compensated by
the output’s influence on the same variable, which we can call “x”.
The disturbance “d” can be seen as a “stimulus” and the output “o”
as a “response”. To a naive observer it looks as though the response
depends on the stimulus according to whatever processing is done
inside the organism. But since “x”, the effect of the output “o” on
the CEV, is exactly what is needed to balance the disturbance, the
relationship between “stimulus” and “Response” is actually
determined by the environmental feedback path between output and
CEV, which we can write as a function x = f(o). Whatever happens in
the rest of the loop must conform to this, so we can write the
“stimulus-response relationship” as o = f^-1 (x) (left
figure). Change f(.) by altering the feedback path, and you change f^-1 (.),
the relationship between “stimulus” and “response”, but you did not
change the internal processing or the “stimulus”. The naive
appearance of the output being determined by the processing applied
to the stimulus is the “behavioural illusion”.

![BehIllusion_1.jpg|482x250](upload://9RGgm2bjJB41Gs9E3f2cCpugpgF.jpeg)

The behavioural illusion is that f<sup>-1</sup>    (.) tells you

something about the internal processing done by the organism, when
it really tells you about the environmental feedback pathway between
the output and the CEV. All it can tell you about the internal
processing is that the processing is capable of producing both
versions of f^-1 (.) under the original and the changed
conditions. That allows for a very large number of possibilities,
all of which implement control. It is clear that f^-1 (.)
must implement control, because the whole argument is based on the
complementary relationship between d and x. The right-hand figure
shows a hypothesis that implements control in a very generic way.

The problem with all this is that control is never perfect, so it is

not exactly true that o = f^-1 (x). At the very least,
there is a time delay, however small, between “o” and “x”, so x =
f(o) should be written x(t) = f(o(t-tau)). In order to invert this,
f^-1(.) must have the opposite time delay: o(t) = f^-1 (x(t+tau)).
In other words, “o” depends on future values of “x”, which it cannot
do in any real world. Control can never be perfect, if for that
reason alone, but there are all sorts of reasons why it is cannot be
perfect in any specific situation.

There's an engineering maxim that to find out how something works,

look to see how it fails. That is exactly what PCT simulation
modelling does. A control structure like that suggested in the
right-hand figure is assumed, the output stage is assumed to be a
leaky integrator, and some transport lag is assumed to exist (as it
must in any real world). Various parameters of the model, such as
transport lag and integrator gain and leak rates are tried out, and
some set of values fails to control in very much the same way as
does the system being tested – such as a human tracking an object
on a screen. This set of values applied to the hypothesised
structure is said to be a “model” of the internal operations of the
subject system.

The behavioural illusion is not that the relation between f(.) and f<sup>-1</sup>    (.)

precludes discovery of anything about f^-1 (.), or said in
another way: “The so-called stimulus-response relationship tells you
only about the environmental feedback function”, both of which are
exaggerations. It is that when a perception is controlled you can
treat the stimulus-response relationship as a function only of the
internal processing.

No matter what the situation, one cannot determine whether an

observation is an example of the behavioural illusion without
knowing two things: the controlled variable, and whether the
relation between output and the controlled variable is being
asserted to represent exclusively the internal part of the control
loop. Even when some observation can be shown to be a behavioural
illusion, the imperfection of control allows simulation modelling or
other techniques to probe the internal processing used by the
organism, and if there is no control at all – meaning that the
output cannot influence the input, then any relationship between
“stimulus” input and “response” output does in fact depend entirely
on the internal processing. The behaviour, if PCT is correct,
controls some other, higher-level, perception, but that is a
different story.

Martin

[Alex_Gomez-Marin]

So, a scientist can commit the behavioral illusionwhen control is perfect (from Powers’ mathematical derivation of the Spadework paper, in which he had to make the perfect control approximation to show how/when function 1/g is confounded with f). Right?

But, when control is not perfect, the scientist sees the environment feedback function mixed with the organism function. In that case, its still seems feasible to learn something valuable about the organism by the stimulus-response approach (unless one decides to claim that the only valuable thing is the controlled variable(s) and not the function). Or is that too heretic to say?

When there is no control, re-organization starts and then the problem is not the behavioral illusion but the fact that any attempt to estimate the organism function must incorporate transients, and so what we try to say about the organism now is not a static copy of what we will try to say about it some minutes later.

If what I said is accurate, this still makes the cautionary notes of the PCT approach to behavior very relevant, yet leaves the existence-ubiquity of behavioral illusions as a very particular laboratory case.

Now, unleash the lyons…

···

On Wed, Aug 31, 2016 at 5:14 AM, Martin Taylor mmt-csg@mmtaylor.net wrote:

[Martin Taylor 2016.08.30.13.12]

The idea of the behavioural illusion is deceptively simple on its

surface. If some perception is perfectly controlled, the influence
of the disturbance on the corresponding environmental variable (the
CEV or “Complex Environmental Variable”) is exactly compensated by
the output’s influence on the same variable, which we can call “x”.
The disturbance “d” can be seen as a “stimulus” and the output “o”
as a “response”. To a naive observer it looks as though the response
depends on the stimulus according to whatever processing is done
inside the organism. But since “x”, the effect of the output “o” on
the CEV, is exactly what is needed to balance the disturbance, the
relationship between “stimulus” and “Response” is actually
determined by the environmental feedback path between output and
CEV, which we can write as a function x = f(o). Whatever happens in
the rest of the loop must conform to this, so we can write the
“stimulus-response relationship” as o = f^-1 (x) (left
figure). Change f(.) by altering the feedback path, and you change f^-1 (.),
the relationship between “stimulus” and “response”, but you did not
change the internal processing or the “stimulus”. The naive
appearance of the output being determined by the processing applied
to the stimulus is the “behavioural illusion”.

The behavioural illusion is that f<sup>-1</sup>    (.) tells you

something about the internal processing done by the organism, when
it really tells you about the environmental feedback pathway between
the output and the CEV. All it can tell you about the internal
processing is that the processing is capable of producing both
versions of f^-1 (.) under the original and the changed
conditions. That allows for a very large number of possibilities,
all of which implement control. It is clear that f^-1 (.)
must implement control, because the whole argument is based on the
complementary relationship between d and x. The right-hand figure
shows a hypothesis that implements control in a very generic way.

The problem with all this is that control is never perfect, so it is

not exactly true that o = f^-1 (x). At the very least,
there is a time delay, however small, between “o” and “x”, so x =
f(o) should be written x(t) = f(o(t-tau)). In order to invert this,
f^-1(.) must have the opposite time delay: o(t) = f^-1 (x(t+tau)).
In other words, “o” depends on future values of “x”, which it cannot
do in any real world. Control can never be perfect, if for that
reason alone, but there are all sorts of reasons why it is cannot be
perfect in any specific situation.

There's an engineering maxim that to find out how something works,

look to see how it fails. That is exactly what PCT simulation
modelling does. A control structure like that suggested in the
right-hand figure is assumed, the output stage is assumed to be a
leaky integrator, and some transport lag is assumed to exist (as it
must in any real world). Various parameters of the model, such as
transport lag and integrator gain and leak rates are tried out, and
some set of values fails to control in very much the same way as
does the system being tested – such as a human tracking an object
on a screen. This set of values applied to the hypothesised
structure is said to be a “model” of the internal operations of the
subject system.

The behavioural illusion is not that the relation between f(.) and f<sup>-1</sup>    (.)

precludes discovery of anything about f^-1 (.), or said in
another way: “The so-called stimulus-response relationship tells you
only about the environmental feedback function”, both of which are
exaggerations. It is that when a perception is controlled you can
treat the stimulus-response relationship as a function only of the
internal processing.

No matter what the situation, one cannot determine whether an

observation is an example of the behavioural illusion without
knowing two things: the controlled variable, and whether the
relation between output and the controlled variable is being
asserted to represent exclusively the internal part of the control
loop. Even when some observation can be shown to be a behavioural
illusion, the imperfection of control allows simulation modelling or
other techniques to probe the internal processing used by the
organism, and if there is no control at all – meaning that the
output cannot influence the input, then any relationship between
“stimulus” input and “response” output does in fact depend entirely
on the internal processing. The behaviour, if PCT is correct,
controls some other, higher-level, perception, but that is a
different story.

Martin

[Huddy_Vyv]

[Vyv Huddy (13:51.31.08.2016)]

···

VH: I have a few comments and queries as I’ve been wondering a lot recently about similar points.

AGM: So, a scientist
can commit the behavioral illusionwhen ** control
is perfect** (from Powers’ mathematical derivation of the Spadework paper, in which he had to make the perfect control approximation to show how/when function 1/g is confounded with f). Right?

VH: Is that where Bill Powers say LCS I p. 147 “as long as the polynomial U remains large enough the apparent behavioural law will be unaffected”?

AGM: But, when
control is not perfect, the scientist sees the environment feedback function
mixed with the organism function.

VH: I don’t understand - what the scientist sees an person or animal doing is their behaviour? Bar pressing, moving a mouse, drawing a curve (ahem)…? If we applying PCT these events
are always interpreted in closed loop terms in my understanding.

AGM: In that case, its still seems feasible to learn
something
valuable about the organism by the stimulus-response approach

VH: A “stimulus response approach” is a behavioural illusion as long as the response is interpreted as caused by the stimulus. So
if a living system is doing the behaving then S-R is never the case as Rick said the other day?

[From Rick Marken (2016.08.29.1610)]

RM: A person is experiencing the behavioral illusion when
a variable external to a control system appears to be the cause of that system’s output. When you see an ice cube near the thermostat cause the heater to go on, you are experiencing a behavioral illusion; when you see a puff of air to the eye cause a blink,
you are experiencing a behavioral illusion; when you see a curve in a person’s hand motion cause a particular velocity of movement, you are experiencing a behavioral illusion.

VH: Rick Marken’s Dancer and the Dance paper describes this; Figure 2 which shows how the
loop manifest in a standard S-R / IV-DV set up I think.

AGM:(unless one decides to claim that the only valuable thing is the controlled variable(s) and not the function).

VH: I don’t understand how we can separate the components of the loop in this way (controlled variable and the function). Negative
feedback occurs from the whole organisation doesn’t it? All is valuable?

AGM: Or is that too heretic to say?

VH: In LCS III p. 70 Bill Powers says models we only fit for “well learned models under normal conditions”. For me he is saying we should spend our efforts on variables that organisms
have under almost perfect control. This directly contradicts the majority of behavioural research which is conducted in circumstances where control is very poor. Unless you can think of counter examples of situations where control is very good indeed? I guess
the measures used in most studies are not sufficiently precise.

AGM: When there is no control , re-organization starts and then the problem is not the behavioral illusion but the fact that any attempt to estimate the organism function must incorporate transients, and so what we try to say about the organism now is
not a static copy of what we will try to say about it some minutes later.

AGM:If what I said is accurate, this still makes the cautionary notes of the PCT approach to behavior very relevant, yet leaves the existence-ubiquity of behavioral illusions as a very particular laboratory case.

VH: I don’t think behavioural illusions are laboratory case. I think they are the primary reason for the difficulties of that practitioner psychologists have (like me) in applying any of the theoretical ideas we have learned in the real world. Functional analysis
of behaviour is an example of this problem, at least in my experience of applying it.

AGM: Now, unleash the lyons…

On Wed, Aug 31, 2016 at 5:14 AM, Martin Taylor
mmt-csg@mmtaylor.net wrote:

[Martin Taylor 2016.08.30.13.12]

The idea of the behavioural illusion is deceptively simple on its surface. If some perception is perfectly controlled, the influence of the disturbance on the corresponding environmental variable (the CEV or “Complex Environmental Variable”) is exactly compensated
by the output’s influence on the same variable, which we can call “x”. The disturbance “d” can be seen as a “stimulus” and the output “o” as a “response”. To a naive observer it looks as though the response depends on the stimulus according to whatever processing
is done inside the organism. But since “x”, the effect of the output “o” on the CEV, is exactly what is needed to balance the disturbance, the relationship between “stimulus” and “Response” is actually determined by the environmental feedback path between
output and CEV, which we can write as a function x = f(o). Whatever happens in the rest of the loop must conform to this, so we can write the “stimulus-response relationship” as o = f^-1 (x) (left figure). Change f(.) by altering the feedback path,
and you change f^-1 (.), the relationship between “stimulus” and “response”, but you did not change the internal processing or the “stimulus”. The naive appearance of the output being determined by the processing applied to the stimulus is the “behavioural
illusion”.

The behavioural illusion is that f^-1 (.) tells you something about the internal processing done by the organism, when it really tells you about the environmental feedback pathway between the output and the CEV. All it can tell you about the internal
processing is that the processing is capable of producing both versions of f^-1(.) under the original and the changed conditions. That allows for a very large number of possibilities, all of which implement control. It is clear that f^-1 (.)
must implement control, because the whole argument is based on the complementary relationship between d and x. The right-hand figure shows a hypothesis that implements control in a very generic way.

The problem with all this is that control is never perfect, so it is not exactly true that o = f^-1 (x). At the very least, there is a time delay, however small, between “o” and “x”, so x = f(o) should be written x(t) = f(o(t-tau)). In order to invert
this, f^-1(.) must have the opposite time delay: o(t) = f^-1 (x(t+tau)). In other words, “o” depends on future values of “x”, which it cannot do in any real world. Control can never be perfect, if for that reason alone, but there are all
sorts of reasons why it is cannot be perfect in any specific situation.

There’s an engineering maxim that to find out how something works, look to see how it fails. That is exactly what PCT simulation modelling does. A control structure like that suggested in the right-hand figure is assumed, the output stage is assumed to be a
leaky integrator, and some transport lag is assumed to exist (as it must in any real world). Various parameters of the model, such as transport lag and integrator gain and leak rates are tried out, and some set of values fails to control in very much the same
way as does the system being tested – such as a human tracking an object on a screen. This set of values applied to the hypothesised structure is said to be a “model” of the internal operations of the subject system.

The behavioural illusion is not that the relation between f(.) and f^-1(.) precludes discovery of anything about f^-1 (.), or said in another way: “The so-called stimulus-response relationship tells you only about the environmental feedback
function”, both of which are exaggerations. It is that when a perception is controlled you can treat the stimulus-response relationship as a function only of the internal processing.

No matter what the situation, one cannot determine whether an observation is an example of the behavioural illusion without knowing two things: the controlled variable, and whether the relation between output and the controlled variable is being asserted to
represent exclusively the internal part of the control loop. Even when some observation can be shown to be a behavioural illusion, the imperfection of control allows simulation modelling or other techniques to probe the internal processing used by the organism,
and if there is no control at all – meaning that the output cannot influence the input, then any relationship between “stimulus” input and “response” output does in fact depend entirely on the internal processing. The behaviour, if PCT is correct, controls
some other, higher-level, perception, but that is a different story.

Martin

VH: I
think reorganisation is difficult to study? There is a quote in B:CP. p.249 that the “we must first know a lot about the existing organisation and develop methods to
keep track of it continually”. Usually behaviour researchers
are interested in how people adapt or learn in rapidly changing experiments. Most behavioural experiments involving people take less than an hour sometimes a matter or minutes. I guess your field with animals
is different though?

[MartinT]

[Martin Taylor 2016.08.31.09.11]

Right.

I don’t know how heretical it is, not being inclined to see PCT in
religious terms, but it isn’t strictly accurate. The problem is that
in this situation you can’t disentangle the internal processing from
the environmental feedback path effects without modelling the
control process. A “stimulus-response approach” doesn’t get you very
far by itself. The argument is that whereas there are N ways in
which control can be effected, N is severely reduced by observation
of the ways control fails. If you look only at the relation between
“stimulus” and “response” you get a misleading impression of what is
going on inside, even though your observations may be perfectly
accurate. If you change the environmental feedback path, that
“stimulus-response” relationship also changes, which implies that
your earlier analysis left something out.
An engineer has a very hard time figuring out how a perfect machine
works, but observing how it fails tells a lot about the machine. As
a crude analogy, suppose Z = X + Y, but Y usually is very stable and
although one can observe all three variables, one sees only that Z
perfectly tracks variations in X. One would never find the correct
relationship among the three variables. Only when Y changes does Z
fail to track X perfectly, and then the observer has an opportunity
to discover the true relationship.
Not what I had in mind at all. I had in mind things such as
observing that the clouds are dark and getting darker, which
correlates with taking an umbrella when one goes out. There is no
possibility of controlling the darkness of the clouds, but one does
control a perception (in imagination) of one’s wetness-dryness, and
the means of controlling that perception is the umbrella. One does
not reorganize to be able to turn dark clouds white and fluffy. The
same applies to the kind of psychophysical experiment in which, say,
the subject tells the experimenter that a beep was or was not heard.
Nothing the subject does can affect that particular beep, but the
response to the experimenter does serve to control the subject’s
perception of the experimenter’s pleasure/displeasure with the
subject’s performance (and probably some other perceptions as well).
I don’t think it is so particular, but it shouldn’t be treated as a
catch-all. It’s pretty ubiquitous. For example, how often do you
read in the newspaper or see on a news broadcast that such-and-such
and event “forced” someone to do something? If you look at it from
the person’s viewpoint they did whatever they did because they were
controlling some perception to a value that would be closer to its
reference value if they did it than if they didn’t. The only real
exception to this is if the person’s motions were generated by
stronger forces than the person’s muscles could supply in
opposition. Even if someone is “forced” out of their house by a
fire, from their point of view they were just avoiding getting burnt
to death. Seeing an ice-cream truck on a hot day does not make one
buy an ice-cream, though a bystander might observe that one always
does so. The behavioural illusion in all these cases is that the
environmental “stimulus” created the environmental “response”.
I hope you don’t see this as the roar of one!
Martin

···

On 2016/08/31 7:25 AM, Alex Gomez-Marin
wrote:

      So,

a scientist can commit the behavioral illusionwhen ** control
is perfect** (from Powers’ mathematical derivation of the
Spadework paper, in which he had to make the perfect control
approximation to show how/when function 1/g is confounded with
f). Right?

      But,

when control is not perfect , the scientist sees the
environment feedback function mixed with the organism
function. In that case, its still seems feasible to learn something
valuable about the organism by the stimulus-response
approach (unless one decides to claim that the only valuable
thing is the controlled variable(s) and not the function). Or
is that too heretic to say?

      When

there is no control , re-organization starts and then
the problem is not the behavioral illusion but the fact that
any attempt to estimate the organism function must incorporate
transients, and so what we try to say about the organism now
is not a static copy of what we will try to say about it some
minutes later.

      If

what I said is accurate, this still makes the cautionary notes
of the PCT approach to behavior very relevant, yet leaves the
existence-ubiquity of behavioral illusions as a very
particular laboratory case.

      Now,

unleash the lyons…

      On Wed, Aug 31, 2016 at 5:14 AM, Martin

Taylor mmt-csg@mmtaylor.net
wrote:

          [Martin Taylor

2016.08.30.13.12]

          The idea of the behavioural illusion is deceptively simple

on its surface. If some perception is perfectly
controlled, the influence of the disturbance on the
corresponding environmental variable (the CEV or “Complex
Environmental Variable”) is exactly compensated by the
output’s influence on the same variable, which we can call
“x”. The disturbance “d” can be seen as a “stimulus” and
the output “o” as a “response”. To a naive observer it
looks as though the response depends on the stimulus
according to whatever processing is done inside the
organism. But since “x”, the effect of the output “o” on
the CEV, is exactly what is needed to balance the
disturbance, the relationship between “stimulus” and
“Response” is actually determined by the environmental
feedback path between output and CEV, which we can write
as a function x = f(o). Whatever happens in the rest of
the loop must conform to this, so we can write the
“stimulus-response relationship” as o = f^-1 (x)
(left figure). Change f(.) by altering the feedback path,
and you change f^-1 (.), the relationship between
“stimulus” and “response”, but you did not change the
internal processing or the “stimulus”. The naive
appearance of the output being determined by the
processing applied to the stimulus is the “behavioural
illusion”.

          The behavioural illusion is that f<sup>-1</sup>              (.) tells

you something about the internal processing done by the
organism, when it really tells you about the environmental
feedback pathway between the output and the CEV. All it
can tell you about the internal processing is that the
processing is capable of producing both versions of f^-1 (.)
under the original and the changed conditions. That allows
for a very large number of possibilities, all of which
implement control. It is clear that f^-1 (.) must
implement control, because the whole argument is based on
the complementary relationship between d and x. The
right-hand figure shows a hypothesis that implements
control in a very generic way.

          The problem with all this is that control is never

perfect, so it is not exactly true that o = f^-1 (x).
At the very least, there is a time delay, however small,
between “o” and “x”, so x = f(o) should be written x(t) =
f(o(t-tau)). In order to invert this, f^-1 (.)
must have the opposite time delay: o(t) = f^-1 (x(t+tau)).
In other words, “o” depends on future values of “x”, which
it cannot do in any real world. Control can never be
perfect, if for that reason alone, but there are all sorts
of reasons why it is cannot be perfect in any specific
situation.

          There's an engineering maxim that to find out how

something works, look to see how it fails. That is exactly
what PCT simulation modelling does. A control structure
like that suggested in the right-hand figure is assumed,
the output stage is assumed to be a leaky integrator, and
some transport lag is assumed to exist (as it must in any
real world). Various parameters of the model, such as
transport lag and integrator gain and leak rates are tried
out, and some set of values fails to control in very much
the same way as does the system being tested – such as a
human tracking an object on a screen. This set of values
applied to the hypothesised structure is said to be a
“model” of the internal operations of the subject system.

          The behavioural illusion is not that the relation between

f(.) and f^-1 (.) precludes discovery of anything
about f^-1 (.), or said in another way: “The
so-called stimulus-response relationship tells you only
about the environmental feedback function”, both of which
are exaggerations. It is that when a perception is
controlled you can treat the stimulus-response
relationship as a function only of the internal
processing.

          No matter what the situation, one cannot determine whether

an observation is an example of the behavioural illusion
without knowing two things: the controlled variable, and
whether the relation between output and the controlled
variable is being asserted to represent exclusively the
internal part of the control loop. Even when some
observation can be shown to be a behavioural illusion, the
imperfection of control allows simulation modelling or
other techniques to probe the internal processing used by
the organism, and if there is no control at all – meaning
that the output cannot influence the input, then any
relationship between “stimulus” input and “response”
output does in fact depend entirely on the internal
processing. The behaviour, if PCT is correct, controls
some other, higher-level, perception, but that is a
different story.

              Martin

[rsmarken]

[From Rick Marken (2016.08.31.1800)]

···

On Wed, Aug 31, 2016 at 4:25 AM, Alex Gomez-Marin agomezmarin@gmail.com wrote:

AGM: So, a scientist can commit the behavioral illusionwhen control is perfect

RM: The quality of control has nothing to do with the behavioral illusion. This is Martin’s hobby horse: a way he thinks he can study control systems as though they were input-output systems. I’ll post something on the behavioral illusion soon but in the meantime, while your here, Alex, could you please tell me what it means when it is said that movement “follows the power law”? How do you decide whether a movement fits the power law or not? Is it the R^2 fit of a power law to the data? For example, does a movement follow the power law if the R^2 for the regression of log(C) on log (A) is > some critical value, like .9? Or does a movement follow the power law if the beta coefficient of the regression is close to 2/3? Or does a movement follow the power law only if both R^2 is greater than some value and beta is close to 2/3?

Best

Rick

(from Powers’ mathematical derivation of the Spadework paper, in which he had to make the perfect control approximation to show how/when function 1/g is confounded with f). Right?

But, when control is not perfect, the scientist sees the environment feedback function mixed with the organism function. In that case, its still seems feasible to learn something valuable about the organism by the stimulus-response approach (unless one decides to claim that the only valuable thing is the controlled variable(s) and not the function). Or is that too heretic to say?

When there is no control, re-organization starts and then the problem is not the behavioral illusion but the fact that any attempt to estimate the organism function must incorporate transients, and so what we try to say about the organism now is not a static copy of what we will try to say about it some minutes later.

If what I said is accurate, this still makes the cautionary notes of the PCT approach to behavior very relevant, yet leaves the existence-ubiquity of behavioral illusions as a very particular laboratory case.

Now, unleash the lyons…

–
Richard S. Marken

“The childhood of the human race is far from over. We
have a long way to go before most people will understand that what they do for
others is just as important to their well-being as what they do for
themselves.” – William T. Powers

On Wed, Aug 31, 2016 at 5:14 AM, Martin Taylor mmt-csg@mmtaylor.net wrote:

[Martin Taylor 2016.08.30.13.12]

The idea of the behavioural illusion is deceptively simple on its

surface. If some perception is perfectly controlled, the influence
of the disturbance on the corresponding environmental variable (the
CEV or “Complex Environmental Variable”) is exactly compensated by
the output’s influence on the same variable, which we can call “x”.
The disturbance “d” can be seen as a “stimulus” and the output “o”
as a “response”. To a naive observer it looks as though the response
depends on the stimulus according to whatever processing is done
inside the organism. But since “x”, the effect of the output “o” on
the CEV, is exactly what is needed to balance the disturbance, the
relationship between “stimulus” and “Response” is actually
determined by the environmental feedback path between output and
CEV, which we can write as a function x = f(o). Whatever happens in
the rest of the loop must conform to this, so we can write the
“stimulus-response relationship” as o = f^-1 (x) (left
figure). Change f(.) by altering the feedback path, and you change f^-1 (.),
the relationship between “stimulus” and “response”, but you did not
change the internal processing or the “stimulus”. The naive
appearance of the output being determined by the processing applied
to the stimulus is the “behavioural illusion”.

The behavioural illusion is that f<sup>-1</sup>    (.) tells you

something about the internal processing done by the organism, when
it really tells you about the environmental feedback pathway between
the output and the CEV. All it can tell you about the internal
processing is that the processing is capable of producing both
versions of f^-1 (.) under the original and the changed
conditions. That allows for a very large number of possibilities,
all of which implement control. It is clear that f^-1 (.)
must implement control, because the whole argument is based on the
complementary relationship between d and x. The right-hand figure
shows a hypothesis that implements control in a very generic way.

The problem with all this is that control is never perfect, so it is

not exactly true that o = f^-1 (x). At the very least,
there is a time delay, however small, between “o” and “x”, so x =
f(o) should be written x(t) = f(o(t-tau)). In order to invert this,
f^-1(.) must have the opposite time delay: o(t) = f^-1 (x(t+tau)).
In other words, “o” depends on future values of “x”, which it cannot
do in any real world. Control can never be perfect, if for that
reason alone, but there are all sorts of reasons why it is cannot be
perfect in any specific situation.

There's an engineering maxim that to find out how something works,

look to see how it fails. That is exactly what PCT simulation
modelling does. A control structure like that suggested in the
right-hand figure is assumed, the output stage is assumed to be a
leaky integrator, and some transport lag is assumed to exist (as it
must in any real world). Various parameters of the model, such as
transport lag and integrator gain and leak rates are tried out, and
some set of values fails to control in very much the same way as
does the system being tested – such as a human tracking an object
on a screen. This set of values applied to the hypothesised
structure is said to be a “model” of the internal operations of the
subject system.

The behavioural illusion is not that the relation between f(.) and f<sup>-1</sup>    (.)

precludes discovery of anything about f^-1 (.), or said in
another way: “The so-called stimulus-response relationship tells you
only about the environmental feedback function”, both of which are
exaggerations. It is that when a perception is controlled you can
treat the stimulus-response relationship as a function only of the
internal processing.

No matter what the situation, one cannot determine whether an

observation is an example of the behavioural illusion without
knowing two things: the controlled variable, and whether the
relation between output and the controlled variable is being
asserted to represent exclusively the internal part of the control
loop. Even when some observation can be shown to be a behavioural
illusion, the imperfection of control allows simulation modelling or
other techniques to probe the internal processing used by the
organism, and if there is no control at all – meaning that the
output cannot influence the input, then any relationship between
“stimulus” input and “response” output does in fact depend entirely
on the internal processing. The behaviour, if PCT is correct,
controls some other, higher-level, perception, but that is a
different story.

Martin

[Alex_Gomez-Marin]

rick, to your diverting question, let me offer a diverting answer;, that is how you do it: you plot in one axis what you do and in the other what you say and if r2 is close to one in loglog space, then what you mean is a powerlaw, and so de facto a perceptual illusion and a behavioral confusion. there are other more sophisticated ways, like non linear regressions, etc. but you can create and demonstrate illusions and confusions with the simpler method.

···

On Wed, Aug 31, 2016 at 4:25 AM, Alex Gomez-Marin agomezmarin@gmail.com wrote:

AGM: So, a scientist can commit the behavioral illusionwhen control is perfect

RM: The quality of control has nothing to do with the behavioral illusion. This is Martin’s hobby horse: a way he thinks he can study control systems as though they were input-output systems. I’ll post something on the behavioral illusion soon but in the meantime, while your here, Alex, could you please tell me what it means when it is said that movement “follows the power law”? How do you decide whether a movement fits the power law or not? Is it the R^2 fit of a power law to the data? For example, does a movement follow the power law if the R^2 for the regression of log(C) on log (A) is > some critical value, like .9? Or does a movement follow the power law if the beta coefficient of the regression is close to 2/3? Or does a movement follow the power law only if both R^2 is greater than some value and beta is close to 2/3?

Best

Rick

(from Powers’ mathematical derivation of the Spadework paper, in which he had to make the perfect control approximation to show how/when function 1/g is confounded with f). Right?

But, when control is not perfect, the scientist sees the environment feedback function mixed with the organism function. In that case, its still seems feasible to learn something valuable about the organism by the stimulus-response approach (unless one decides to claim that the only valuable thing is the controlled variable(s) and not the function). Or is that too heretic to say?

When there is no control, re-organization starts and then the problem is not the behavioral illusion but the fact that any attempt to estimate the organism function must incorporate transients, and so what we try to say about the organism now is not a static copy of what we will try to say about it some minutes later.

If what I said is accurate, this still makes the cautionary notes of the PCT approach to behavior very relevant, yet leaves the existence-ubiquity of behavioral illusions as a very particular laboratory case.

Now, unleash the lyons…

–
Richard S. Marken

“The childhood of the human race is far from over. We
have a long way to go before most people will understand that what they do for
others is just as important to their well-being as what they do for
themselves.” – William T. Powers

On Wed, Aug 31, 2016 at 5:14 AM, Martin Taylor mmt-csg@mmtaylor.net wrote:

[Martin Taylor 2016.08.30.13.12]

The idea of the behavioural illusion is deceptively simple on its

surface. If some perception is perfectly controlled, the influence
of the disturbance on the corresponding environmental variable (the
CEV or “Complex Environmental Variable”) is exactly compensated by
the output’s influence on the same variable, which we can call “x”.
The disturbance “d” can be seen as a “stimulus” and the output “o”
as a “response”. To a naive observer it looks as though the response
depends on the stimulus according to whatever processing is done
inside the organism. But since “x”, the effect of the output “o” on
the CEV, is exactly what is needed to balance the disturbance, the
relationship between “stimulus” and “Response” is actually
determined by the environmental feedback path between output and
CEV, which we can write as a function x = f(o). Whatever happens in
the rest of the loop must conform to this, so we can write the
“stimulus-response relationship” as o = f^-1 (x) (left
figure). Change f(.) by altering the feedback path, and you change f^-1 (.),
the relationship between “stimulus” and “response”, but you did not
change the internal processing or the “stimulus”. The naive
appearance of the output being determined by the processing applied
to the stimulus is the “behavioural illusion”.

The behavioural illusion is that f<sup>-1</sup>    (.) tells you

something about the internal processing done by the organism, when
it really tells you about the environmental feedback pathway between
the output and the CEV. All it can tell you about the internal
processing is that the processing is capable of producing both
versions of f^-1 (.) under the original and the changed
conditions. That allows for a very large number of possibilities,
all of which implement control. It is clear that f^-1 (.)
must implement control, because the whole argument is based on the
complementary relationship between d and x. The right-hand figure
shows a hypothesis that implements control in a very generic way.

The problem with all this is that control is never perfect, so it is

not exactly true that o = f^-1 (x). At the very least,
there is a time delay, however small, between “o” and “x”, so x =
f(o) should be written x(t) = f(o(t-tau)). In order to invert this,
f^-1(.) must have the opposite time delay: o(t) = f^-1 (x(t+tau)).
In other words, “o” depends on future values of “x”, which it cannot
do in any real world. Control can never be perfect, if for that
reason alone, but there are all sorts of reasons why it is cannot be
perfect in any specific situation.

There's an engineering maxim that to find out how something works,

look to see how it fails. That is exactly what PCT simulation
modelling does. A control structure like that suggested in the
right-hand figure is assumed, the output stage is assumed to be a
leaky integrator, and some transport lag is assumed to exist (as it
must in any real world). Various parameters of the model, such as
transport lag and integrator gain and leak rates are tried out, and
some set of values fails to control in very much the same way as
does the system being tested – such as a human tracking an object
on a screen. This set of values applied to the hypothesised
structure is said to be a “model” of the internal operations of the
subject system.

The behavioural illusion is not that the relation between f(.) and f<sup>-1</sup>    (.)

precludes discovery of anything about f^-1 (.), or said in
another way: “The so-called stimulus-response relationship tells you
only about the environmental feedback function”, both of which are
exaggerations. It is that when a perception is controlled you can
treat the stimulus-response relationship as a function only of the
internal processing.

No matter what the situation, one cannot determine whether an

observation is an example of the behavioural illusion without
knowing two things: the controlled variable, and whether the
relation between output and the controlled variable is being
asserted to represent exclusively the internal part of the control
loop. Even when some observation can be shown to be a behavioural
illusion, the imperfection of control allows simulation modelling or
other techniques to probe the internal processing used by the
organism, and if there is no control at all – meaning that the
output cannot influence the input, then any relationship between
“stimulus” input and “response” output does in fact depend entirely
on the internal processing. The behaviour, if PCT is correct,
controls some other, higher-level, perception, but that is a
different story.

Martin

[MartinT]

[Martin Taylor 2016.09.01.10,15]

[From Rick Marken (2016.08.31.1800)]

I don't want to get into yet another unresolvable argument with

Rick, so I will say very little here other than that I fully agree
with the quoted statement, as my two messages on the behavioural
illusion will have shown to anyone who read them.

Like any illusion, the "behavioural illusion" refers only to how

people interpret observational data, not to the data or to the
analytical methods open to people who want to use the observations
to study some problem.

I leave it to others -- anyone who read my behavioural illusion

postings – to judge whether the following is an illusion or
corresponds to some alternate reality.

Martin

PS. Perhaps it is worth noting for readers who joined CSGnet over

the last 20 years that at the 1993 CSG meeting, Powers saluted me as
the first scientist to acknowledge that their own hard-won theory
was subsumed by PCT. I had previously been treating a particular
situation as due not to input-output systems but as a set of complex
feedback loops, which became conceptually simplified when seen as an
application of PCT outside its usual domain of study.

···

On Wed, Aug 31, 2016 at 4:25 AM, Alex
Gomez-Marin agomezmarin@gmail.com
wrote:

              AGM:

So, a scientist can commit the behavioral
illusionwhen control is perfect

          RM: The quality of control has nothing to do with the

behavioral illusion.

          This is Martin's hobby horse: a way he thinks he can

study control systems as though they were input-output
systems.

          I'll post something on the behavioral illusion soon

but in the meantime, while your here, Alex, could you
please tell me what it means when it is said that movement
“follows the power law”? How do you decide whether a
movement fits the power law or not? Is it the R^2 fit of a
power law to the data? For example, does a movement follow
the power law if the R^2 for the regression of log(C) on
log (A) is > some critical value, like .9? Or does a
movement follow the power law if the beta coefficient of
the regression is close to 2/3? Or does a movement follow
the power law only if both R^2 is greater than some value
and beta is close to 2/3?

Best

Rick

              (from

Powers’ mathematical derivation of the Spadework
paper, in which he had to make the perfect control
approximation to show how/when function 1/g is
confounded with f). Right?

              But,

when control is not perfect , the scientist
sees the environment feedback function mixed
with the organism function. In that case, its still
seems feasible to learn something valuable
about the organism by the stimulus-response approach
(unless one decides to claim that the only valuable
thing is the controlled variable(s) and not the
function). Or is that too heretic to say?

              When

there is no control , re-organization starts
and then the problem is not the behavioral illusion
but the fact that any attempt to estimate the organism
function must incorporate transients, and so what we
try to say about the organism now is not a static copy
of what we will try to say about it some minutes
later.

              If

what I said is accurate, this still makes the
cautionary notes of the PCT approach to behavior very
relevant, yet leaves the existence-ubiquity of
behavioral illusions as a very particular laboratory
case.

              Now,

unleash the lyons…

–
Richard S. Marken

                                    "The childhood of the human

race is far from over. We
have a long way to go before
most people will understand that
what they do for
others is just as important to
their well-being as what they do
for
themselves." – William T.
Powers

                  On Wed, Aug 31, 2016 at

5:14 AM, Martin Taylor mmt-csg@mmtaylor.net
wrote:

                      [Martin

Taylor 2016.08.30.13.12]

                      The idea of the behavioural illusion is

deceptively simple on its surface. If some
perception is perfectly controlled, the
influence of the disturbance on the
corresponding environmental variable (the CEV
or “Complex Environmental Variable”) is
exactly compensated by the output’s influence
on the same variable, which we can call “x”.
The disturbance “d” can be seen as a
“stimulus” and the output “o” as a “response”.
To a naive observer it looks as though the
response depends on the stimulus according to
whatever processing is done inside the
organism. But since “x”, the effect of the
output “o” on the CEV, is exactly what is
needed to balance the disturbance, the
relationship between “stimulus” and “Response”
is actually determined by the environmental
feedback path between output and CEV, which we
can write as a function x = f(o). Whatever
happens in the rest of the loop must conform
to this, so we can write the
“stimulus-response relationship” as o = f^-1 (x)
(left figure). Change f(.) by altering the
feedback path, and you change f^-1 (.),
the relationship between “stimulus” and
“response”, but you did not change the
internal processing or the “stimulus”. The
naive appearance of the output being
determined by the processing applied to the
stimulus is the “behavioural illusion”.

                      The behavioural illusion is that f<sup>-1</sup>                          (.)

tells you something about the internal
processing done by the organism, when it
really tells you about the environmental
feedback pathway between the output and the
CEV. All it can tell you about the internal
processing is that the processing is capable
of producing both versions of f^-1 (.)
under the original and the changed conditions.
That allows for a very large number of
possibilities, all of which implement control.
It is clear that f^-1 (.) must
implement control, because the whole argument
is based on the complementary relationship
between d and x. The right-hand figure shows a
hypothesis that implements control in a very
generic way.

                      The problem with all this is that control is

never perfect, so it is not exactly true that
o = f^-1 (x). At the very least,
there is a time delay, however small, between
“o” and “x”, so x = f(o) should be written
x(t) = f(o(t-tau)). In order to invert this, f^-1 (.)
must have the opposite time delay: o(t) = f^-1 (x(t+tau)).
In other words, “o” depends on future values
of “x”, which it cannot do in any real world.
Control can never be perfect, if for that
reason alone, but there are all sorts of
reasons why it is cannot be perfect in any
specific situation.

                      There's an engineering maxim that to find out

how something works, look to see how it fails.
That is exactly what PCT simulation modelling
does. A control structure like that suggested
in the right-hand figure is assumed, the
output stage is assumed to be a leaky
integrator, and some transport lag is assumed
to exist (as it must in any real world).
Various parameters of the model, such as
transport lag and integrator gain and leak
rates are tried out, and some set of values
fails to control in very much the same way as
does the system being tested – such as a
human tracking an object on a screen. This set
of values applied to the hypothesised
structure is said to be a “model” of the
internal operations of the subject system.

                      The behavioural illusion is not that the

relation between f(.) and f^-1 (.)
precludes discovery of anything about f^-1 (.),
or said in another way: “The so-called
stimulus-response relationship tells you only
about the environmental feedback function”,
both of which are exaggerations. It is that
when a perception is controlled you can treat
the stimulus-response relationship as a
function only of the internal processing.

                      No matter what the situation, one cannot

determine whether an observation is an example
of the behavioural illusion without knowing
two things: the controlled variable, and
whether the relation between output and the
controlled variable is being asserted to
represent exclusively the internal part of the
control loop. Even when some observation can
be shown to be a behavioural illusion, the
imperfection of control allows simulation
modelling or other techniques to probe the
internal processing used by the organism, and
if there is no control at all – meaning that
the output cannot influence the input, then
any relationship between “stimulus” input and
“response” output does in fact depend entirely
on the internal processing. The behaviour, if
PCT is correct, controls some other,
higher-level, perception, but that is a
different story.

                          Martin

[rsmarken]

[From Rick Marken (2016.09.01.1110)]

···

Martin Taylor (2016.08.30.13.12)–

RM: This is a nice effort to explain the behavioral illusion but I’d like to give my own explanation. I’ll start by defining an illusion as an observation that is not what it appears to be. For example, here is an example of an optical illusion:

RM: What you observe are two parallel vertical lines that bulge out in the middle. This is an illusion because we know, for other reasons (such as from measuring the distance between the lines at the top, middle and bottom) that there is no bulge. Another illusion is the observation of a straight, rigid object, like a pencil, bending when placed in water, like so:

RM: Again, this is an illusion because we know, for other reasons (such as feeling along the pencil while it is immersed) that the object is not bent.

RM: Behavioral illusions are illusions in the same way these optical illusions are illusions: they are observations that are not what they appear to be. One example of a behavioral illusion is a “reflex”, such as the patellar of “knee jerk” reflex:

RM: What you see is a hammer tap below the knee causing the lower leg to kick up. The hammer tap looks like a “stimulus” that is the cause of the knee jerk “response”. It looks like stimulus causes response. The stimulus-response appearance of reflexes was not considered an illusion until it was shown to be, by other mean. The “other means” was PCT.

RM: The PCT analysis of the stimulus-response illusion starts by noting that behavior is a closed-loop control process organized around the control of the sensory effects of stimuli: controlled variables. When this loop is analyzed correctly it can be shown (as Powers did in his 1978 Psych Review paper) that the observed causal relationship between stimulus and response is actually the inverse of the causal relationship between the response and the sensory effect of the stimulus; this relationship is called the feedback function and the sensory effect of the stimulus is the controlled variable.

RM: So the “behavioral illusion” refers to the fact that when you observe the behavior of a closed loop control system – living or artificial – what you see is not necessarily what is actually happening. In the case of the stimulus-response illusion, the actual causal connection from stimulus to response is “masked” by the fact that it occurs as part of a closed loop control process.

RM: There are actually three versions of the behavioral illusion, which are described in my “Blind Men and the Elephant” paper in “More Mind Readings”. Reflexes, like the patellar, are examples of the stimulus-response illusion; operant behavior is an example of the reinforcement illusion; and the power law is an example of the emitted output illusion (called the cognitive illusion in the paper).

RM: How “illusory” any of these behavioral illusions are depends on many factors. For example, the stimulus-response illusion is much more pronounced when the stimulus (disturbance to the controlled variable) is abrupt and strong rather than gradual and weak (see “A Bucket of Beans” in LCS II); the emitted output illusion is more pronounced when there are no obvious disturbances to the controlled variable. These illusions are also more pronounced when the system under observation controls well. A system that controls poorly will make only a very weak “response” to a “stimulus”(disturbance) that affects the variable it is controlling. So the stimulus will not give the appearance of causing the response and there will be a very weak stimulus-response illusion.

RM: All behavioral illusions result from failure to understand the implications of the fact that one is observing at the behavior of a closed loop control system. It’s like looking at the bent pencil illusion above and not knowing the implications of the fact that the pencil is partly in water. This analogy should make it clear why the behavioral illusion is so important. A researcher who doesn’t know that the refraction of light is different in water and air will start looking for explanations for why the pencil is bent; a researcher who doesn’t know that a closed loop system acts to control its input will start looking for explanations of why stimuli cause responses, why reinforcement selects responses or how responses are emitted.

RM: The behavioral illusion is important because it shows that behavioral scientists have been studying the different flavors of the illusion – taking them for what is actually happening – for over 100 years.

Best

Rick

–
Richard S. Marken

“The childhood of the human race is far from over. We
have a long way to go before most people will understand that what they do for
others is just as important to their well-being as what they do for
themselves.” – William T. Powers

[FredNickols]

[From Fred Nickols (2016.09.01.1554 ET)]

FN: I really like your post below, Rick. It clears up a lot for me. I also spotted something that I think might present a problem when explaining PCT to others. Here’s the snippet I pulled from your longer post below:

[From Rick Marken (2016.09.01.1110)]

RM: So the “behavioral illusion” refers to the fact that when you observe the behavior of a closed loop control system – living or artificial – what you see is not necessarily what is actually happening.

FN: It’s the “what you see is not necessarily what is actually happening� that’s giving me pause. I think most people, especially some behaviorists I know, would snort or laugh and say something like, “Don’t tell me I’m not seeing what I’m seeing.� I think a better way of saying what you’re driving at is to say, “How you explain or account for what you see, especially if you do so in cause-effect or stimulus-response terms, is very different from the way those same observations are explained by control theory. More important, in light of control theory, those stimulus-response explanations are very misleading. In control theory, especially perceptual control theory (PCT), those misleading explanations are referred to as “the behavioral illusion.�

FN: At least that’s the way it seems to me, Rick.

FN: Anyway, thanks again for this post. I like it – a lot.

Regards,

Fred Nickols, Consultant

My Objective is to Help You Achieve Yours

DISTANCE CONSULTING LLC

“Assistance at a Distance�^SM

···

From: Richard Marken [mailto:rsmarken@gmail.com]
Sent: Thursday, September 01, 2016 2:09 PM
To: csgnet@lists.illinois.edu
Cc: Henry Yin
Subject: Re: The Behavioural Illusion and modelling

[From Rick Marken (2016.09.01.1110)]

cc: Henry Yin, because he understands the importance of the behavioral illusion and I’d like to hear what he thinks of this post.

Martin Taylor (2016.08.30.13.12)–

RM: This is a nice effort to explain the behavioral illusion but I’d like to give my own explanation. I’ll start by defining an illusion as an observation that is not what it appears to be. For example, here is an example of an optical illusion:

RM: What you observe are two parallel vertical lines that bulge out in the middle. This is an illusion because we know, for other reasons (such as from measuring the distance between the lines at the top, middle and bottom) that there is no bulge. Another illusion is the observation of a straight, rigid object, like a pencil, bending when placed in water, like so:

RM: Again, this is an illusion because we know, for other reasons (such as feeling along the pencil while it is immersed) that the object is not bent.

RM: Behavioral illusions are illusions in the same way these optical illusions are illusions: they are observations that are not what they appear to be. One example of a behavioral illusion is a “reflex”, such as the patellar of “knee jerk” reflex:

RM: What you see is a hammer tap below the knee causing the lower leg to kick up. The hammer tap looks like a “stimulus” that is the cause of the knee jerk “response”. It looks like stimulus causes response. The stimulus-response appearance of reflexes was not considered an illusion until it was shown to be, by other mean. The “other means” was PCT.

RM: The PCT analysis of the stimulus-response illusion starts by noting that behavior is a closed-loop control process organized around the control of the sensory effects of stimuli: controlled variables. When this loop is analyzed correctly it can be shown (as Powers did in his 1978 Psych Review paper) that the observed causal relationship between stimulus and response is actually the inverse of the causal relationship between the response and the sensory effect of the stimulus; this relationship is called the feedback function and the sensory effect of the stimulus is the controlled variable.

RM: So the “behavioral illusion” refers to the fact that when you observe the behavior of a closed loop control system – living or artificial – what you see is not necessarily what is actually happening. In the case of the stimulus-response illusion, the actual causal connection from stimulus to response is “masked” by the fact that it occurs as part of a closed loop control process.

RM: There are actually three versions of the behavioral illusion, which are described in my “Blind Men and the Elephant” paper in “More Mind Readings”. Reflexes, like the patellar, are examples of the stimulus-response illusion; operant behavior is an example of the reinforcement illusion; and the power law is an example of the emitted output illusion (called the cognitive illusion in the paper).

RM: How “illusory” any of these behavioral illusions are depends on many factors. For example, the stimulus-response illusion is much more pronounced when the stimulus (disturbance to the controlled variable) is abrupt and strong rather than gradual and weak (see “A Bucket of Beans” in LCS II); the emitted output illusion is more pronounced when there are no obvious disturbances to the controlled variable. These illusions are also more pronounced when the system under observation controls well. A system that controls poorly will make only a very weak “response” to a “stimulus”(disturbance) that affects the variable it is controlling. So the stimulus will not give the appearance of causing the response and there will be a very weak stimulus-response illusion.

RM: All behavioral illusions result from failure to understand the implications of the fact that one is observing at the behavior of a closed loop control system. It’s like looking at the bent pencil illusion above and not knowing the implications of the fact that the pencil is partly in water. This analogy should make it clear why the behavioral illusion is so important. A researcher who doesn’t know that the refraction of light is different in water and air will start looking for explanations for why the pencil is bent; a researcher who doesn’t know that a closed loop system acts to control its input will start looking for explanations of why stimuli cause responses, why reinforcement selects responses or how responses are emitted.

RM: The behavioral illusion is important because it shows that behavioral scientists have been studying the different flavors of the illusion – taking them for what is actually happening – for over 100 years.

Best

Rick

–

Richard S. Marken

“The childhood of the human race is far from over. We have a long way to go before most people will understand that what they do for others is just as important to their well-being as what they do for themselves.” – William T. Powers

[rsmarken]

[From Rick Marken (2016.09.01.1320)]

···

 Fred Nickols (2016.09.01.1554 ET)]

Â

FN: I really like your post below, Rick.Â

RM: Wow, that is so exciting Fred! Thanks. I think this is the first post of mine in weeks that anyone has liked!Â

Â

FN: I also spotted something that I think might present a problem when explaining PCT to others. Here’s the snippet I pulled from your longer post below:

Â

RM: So the “behavioral illusion” refers to the fact that when you observe the behavior of a closed loop control system – living or artificial – what you see is not necessarily what is actually happening.Â

Â

FN: It’s the “what you see is not necessarily what is actually happeningâ€? that’s giving me pause. I think most people, especially some behaviorists I know, would snort or laugh and say something like, “Don’t tell me I’m not seeing what I’m seeing.â€? I think a better way of saying what you’re driving at is to say, “How you explain or account for what you see, especially if you do so in cause-effect or stimulus-response terms, is very different from the way those same observations are explained by control theory.Â

RM: Yes! Much better  (and more correct) way of saying it! Thanks again Fred!

BestÂ

Rick

Â

More important, in light of control theory, those stimulus-response explanations are very misleading. In control theory, especially perceptual control theory (PCT), those misleading explanations are referred to as “the behavioral illusion.�

Â

Â

FN: At least that’s the way it seems to me, Rick.

Â

FN: Anyway, thanks again for this post. I like it – a lot.

Â

Regards,

Â

Fred Nickols, Consultant

My Objective is to Help You Achieve Yours

DISTANCE CONSULTING LLC

“Assistance at a Distance�^SM

Â

Â

Â

From: Richard Marken [mailto:rsmarken@gmail.com]
Sent: Thursday, September 01, 2016 2:09 PM
To: csgnet@lists.illinois.edu
Cc: Henry Yin
Subject: Re: The Behavioural Illusion and modelling

Â

[From Rick Marken (2016.09.01.1110)]

Â

cc: Henry Yin, because he understands the importance of the behavioral illusion and I’d like to hear what he thinks of this post.

Â

Martin Taylor (2016.08.30.13.12)–

Â

RM: This is a nice effort to explain the behavioral illusion but I’d like to give my own explanation. I’ll start by defining an illusion as an observation that is not what it appears to be. For example, here is an example of an optical illusion:

Â

RM: What you observe are two parallel vertical lines that bulge out in the middle. This is an illusion because we know, for other reasons (such as from measuring the distance between the lines at the top, middle and bottom) that there is no bulge. Another illusion is the observation of a straight, rigid object, like a pencil, bending when placed in water, like so:Â

Â

Â

RM: Again, this is an illusion because we know, for other reasons (such as feeling along the pencil while it is immersed) that the object is not bent.Â

Â

RM: Behavioral illusions are illusions in the same way these optical illusions are illusions: they are observations that are not what they appear to be. One example of a behavioral illusion is a “reflex”, such as the patellar  of “knee jerk” reflex:

Â

Â

RM: What you see is a hammer tap below the knee causing the lower leg to kick up. The hammer tap looks like a “stimulus” that is the cause of the knee jerk “response”. It looks like stimulus causes response. The stimulus-response appearance of reflexes was not considered an illusion until it was shown to be, by other mean. The “other means” was PCT.Â

Â

RM: The PCT analysis of the stimulus-response illusion starts by noting that behavior is a closed-loop control process organized around the control of the sensory effects of stimuli: controlled variables. When this loop is analyzed correctly it can be shown (as Powers did in his 1978 Psych Review paper) that the observed causal relationship between stimulus and response is actually the inverse of the causal relationship between the response and the sensory effect of the stimulus; this relationship is called the feedback function and the sensory effect of the stimulus is the controlled variable.Â

Â

RM: So the “behavioral illusion” refers to the fact that when you observe the behavior of a closed loop control system – living or artificial – what you see is not necessarily what is actually happening. In the case of the stimulus-response illusion, the actual causal connection from  stimulus to response is “masked” by the fact that it occurs as part of a closed loop control process.Â

Â

RM: There are actually three versions of the behavioral illusion, which are described in my “Blind Men and the Elephant” paper in “More Mind Readings”. Reflexes, like the patellar, are examples of the stimulus-response illusion; operant behavior is an example of the reinforcement illusion; and the power law is an example of the emitted output illusion (called the cognitive illusion in the paper).Â

Â

RM: How “illusory” any of these behavioral illusions are depends on many factors. For example, the stimulus-response illusion is much more pronounced when the stimulus (disturbance to the controlled variable) is abrupt and strong rather than gradual and weak (see “A Bucket of Beans” in LCS II); the emitted output illusion is more pronounced when there are no obvious disturbances to the controlled variable. These illusions are also more pronounced when the system under observation controls well. A system that controls poorly will make only a very weak “response” to a “stimulus”(disturbance) that affects the variable it is controlling. So the stimulus will not give the appearance of causing the response and there will be a very weak stimulus-response illusion.Â

Â

RM: All behavioral illusions result from failure to understand the implications of the fact that one is observing at the behavior of a closed loop control system. It’s like looking at the bent pencil illusion above and not knowing the implications of the fact that the pencil is partly in water. This analogy should make it clear why the behavioral illusion is so important. A researcher who doesn’t know that the refraction of light is different in water and air will start looking for explanations for why the pencil is bent; a researcher who doesn’t know that a closed loop system acts to control its input will start looking for explanations of why stimuli cause responses, why reinforcement selects responses or how responses are emitted.Â

Â

RM: The behavioral illusion is important because it shows that behavioral scientists have been studying the different flavors of the illusion – taking them for what is actually happening – for over 100 years. Â

Â

Best

Â

Rick

Â

–

Richard S. MarkenÂ

“The childhood of the human race is far from over. We have a long way to go before most people will understand that what they do for others is just as important to their well-being as what they do for themselves.” – William T. Powers

–

Richard S. MarkenÂ

“The childhood of the human race is far from over. We
have a long way to go before most people will understand that what they do for
others is just as important to their well-being as what they do for
themselves.” – William T. Powers

[Alex_Gomez-Marin]

Beautiful analogies, Rick. Now, does this mean that one does not need to prove the behavioral illusion in each particular case because, by construction of argument, one already knows that it must be a behavioral illusion?

···

On Thu, Sep 1, 2016 at 10:18 PM, Richard Marken rsmarken@gmail.com wrote:

[From Rick Marken (2016.09.01.1320)]

 Fred Nickols (2016.09.01.1554 ET)]

Â

FN: I really like your post below, Rick.Â

RM: Wow, that is so exciting Fred! Thanks. I think this is the first post of mine in weeks that anyone has liked!Â

Â

FN: I also spotted something that I think might present a problem when explaining PCT to others. Here’s the snippet I pulled from your longer post below:

Â

RM: So the “behavioral illusion” refers to the fact that when you observe the behavior of a closed loop control system – living or artificial – what you see is not necessarily what is actually happening.Â

Â

FN: It’s the “what you see is not necessarily what is actually happeningâ€? that’s giving me pause. I think most people, especially some behaviorists I know, would snort or laugh and say something like, “Don’t tell me I’m not seeing what I’m seeing.â€? I think a better way of saying what you’re driving at is to say, “How you explain or account for what you see, especially if you do so in cause-effect or stimulus-response terms, is very different from the way those same observations are explained by control theory.Â

RM: Yes! Much better  (and more correct) way of saying it! Thanks again Fred!

BestÂ

Rick

Â

More important, in light of control theory, those stimulus-response explanations are very misleading. In control theory, especially perceptual control theory (PCT), those misleading explanations are referred to as “the behavioral illusion.�

Â

Â

FN: At least that’s the way it seems to me, Rick.

Â

FN: Anyway, thanks again for this post. I like it – a lot.<

Â

Regards,

Â

Fred Nickols, Consultant

My Objective is to Help You Achieve Yours

DISTANCE CONSULTING LLC

“Assistance at a Distance�^SM

Â

Â

Â

From: Richard Marken [mailto:rsmarken@gmail.com]
Sent: Thursday, September 01, 2016 2:09 PM
To: csgnet@lists.illinois.edu
Cc: Henry Yin
Subject: Re: The Behavioural Illusion and modelling

Â

[From Rick Marken (2016.09.01.1110)]

Â

cc: Henry Yin, because he understands the importance of the behavioral illusion and I’d like to hear what he thinks of this post.

Â

Martin Taylor (2016.08.30.13.12)–

Â

RM: This is a nice effort to explain the behavioral illusion but I’d like to give my own explanation. I’ll start by defining an illusion as an observation that is not what it appears to be. For example, here is an example of an optical illusion:

Â

RM: What you observe are two parallel vertical lines that bulge out in the middle. This is an illusion because we know, for other reasons (such as from measuring the distance between the lines at the top, middle and bottom) that there is no bulge. Another illusion is the observation of a straight, rigid object, like a pencil, bending when placed in water, like so:Â

Â

Â

RM: Again, this is an illusion because we know, for other reasons (such as feeling along the pencil while it is immersed) that the object is not bent.Â

Â

RM: Behavioral illusions are illusions in the same way these optical illusions are illusions: they are observations that are not what they appear to be. One example of a behavioral illusion is a “reflex”, such as the patellar  of “knee jerk” reflex:

Â

Â

RM: What you see is a hammer tap below the knee causing the lower leg to kick up. The hammer tap looks like a “stimulus” that is the cause of the knee jerk “response”. It looks like stimulus causes response. The stimulus-response appearance of reflexes was not considered an illusion until it was shown to be, by other mean. The “other means” was PCT.Â

Â

RM: The PCT analysis of the stimulus-response illusion starts by noting that behavior is a closed-loop control process organized around the control of the sensory effects of stimuli: controlled variables. When this loop is analyzed correctly it can be shown (as Powers did in his 1978 Psych Review paper) that the observed causal relationship between stimulus and response is actually the inverse of the causal relationship between the response and the sensory effect of the stimulus; this relationship is called the feedback function and the sensory effect of the stimulus is the controlled variable.Â

Â

RM: So the “behavioral illusion” refers to the fact that when you observe the behavior of a closed loop control system – living or artificial – what you see is not necessarily what is actually happening. In the case of the stimulus-response illusion, the actual causal connection from  stimulus to response is “masked” by the fact that it occurs as part of a closed loop control process.Â

Â

RM: There are actually three versions of the behavioral illusion, which are described in my “Blind Men and the Elephant” paper in “More Mind Readings”. Reflexes, like the patellar, are examples of the stimulus-response illusion; operant behavior is an example of the reinforcement illusion; and the power law is an example of the emitted output illusion (called the cognitive illusion in the paper).Â

Â

RM: How “illusory” any of these behavioral illusions are depends on many factors. For example, the stimulus-response illusion is much more pronounced when the stimulus (disturbance to the controlled variable) is abrupt and strong rather than gradual and weak (see “A Bucket of Beans” in LCS II); the emitted output illusion is more pronounced when there are no obvious disturbances to the controlled variable. These illusions are also more pronounced when the system under observation controls well. A system that controls poorly will make only a very weak “response” to a “stimulus”(disturbance) that affects the variable it is controlling. So the stimulus will not give the appearance of causing the response and there will be a very weak stimulus-response illusion.Â

Â

RM: All behavioral illusions result from failure to understand the implications of the fact that one is observing at the behavior of a closed loop control system. It’s like looking at the bent pencil illusion above and not knowing the implications of the fact that the pencil is partly in water. This analogy should make it clear why the behavioral illusion is so important. A researcher who doesn’t know that the refraction of light is different in water and air will start looking for explanations for why the pencil is bent; a researcher who doesn’t know that a closed loop system acts to control its input will start looking for explanations of why stimuli cause responses, why reinforcement selects responses or how responses are emitted.Â

Â

RM: The behavioral illusion is important because it shows that behavioral scientists have been studying the different flavors of the illusion – taking them for what is actually happening – for over 100 years. Â

Â

Best

Â

Rick

Â

–

Richard S. MarkenÂ

“The childhood of the human race is far from over. We have a long way to go before most people will understand that what they do for others is just as important to their well-being as what they do for themselves.” – William T. Powers

–
Richard S. MarkenÂ

“The childhood of the human race is far from over. We
have a long way to go before most people will understand that what they do for
others is just as important to their well-being as what they do for
themselves.” – William T. Powers

[rsmarken]

[From Rick Marken (2016.09.02.0930)]

···

On Thu, Sep 1, 2016 at 3:24 PM, Alex Gomez-Marin agomezmarin@gmail.com wrote:

AGM: Beautiful analogies, Rick. Now, does this mean that one does not need to prove the behavioral illusion in each particular case because, by construction of argument, one already knows that it must be a behavioral illusion?

RM: A very good question. I think the question is answered in Bill’s 1978 Psych Review paper (“Quantitative Analysis of Purposive Systems”, reprinted in LCS I) but I’ll try to give a quick summary specifically related to your question.Â

RM: I think another way of asking your question would be this: Do we need to prove (in the sense of “test” since we don’t prove propositions in science in the same way as we do in mathematics) in each particular case that the appearance of behavior as caused, selected or planned output is an illusion? If the behavior is that of a living system the answer would be a cautious “no”.

RM:  In his 1978 Psych Review paper Bill showed that living systems are very likely to be closed loop systems. We know that sensory inputs cause behavioral outputs via the nervous systems; and we can see that, at the same time, these behavioral outputs affect those same sensory inputs via the environment. And since the behavior of these systems is generally very stable, the feedback in the closed loop must be negative. So living systems are very likely to be negative feedback systems – Bill calls them N systems in the article – and N systems act to control their own inputs.Â

RM: So we can assume that we are looking at the behavior of a control system (rather than that of an open-loop causal system – Bill calls it a Z system in the article) when we are looking at the behavior of a living organism. That means when we see a living organism apparently reacting to stimuli, having its behavior “strengthened” by its consequences or producing a programmed set of outputs, we can be pretty sure that those appearances are “illusions”. We know this for the same reason that we know that the pencil doesn’t really bend when we place it in water. With living organisms we know that if they are N-systems these behavioral appearances are “side effects” of control of input; with the pencil we know that the appearance of bending is a “side effect” of differential refraction of light by water and air.Â

RM: But as in the case with the pencil in water, there is always the possibility that what is seen is not an illusion. The pencil might really have bent in water if, for example, one side of the pencil were made of a material that expands in water. And the behavior of an organism that we see might really be an example of caused, reinforced or computed output. But this very unlikely possibility is actually implicitly handled by the PCT approach to studying living control systems.Â

RM: PCT research aims to understand the behavior (controlling) of living organisms in terms of the input variables that are being controlled. So when a PCT researcher sees a living organism apparently reacting to stimuli, being manipulated by reinforcement or producing programmed output they try to guess at the aspects of the organisms sensory input around which this behavior might be organized; that is, they try to guess at what input (perceptual) variables might be under control.

RM: The general methodology used to do this kind of research  is called the Test for the Controlled Variable. An example of such research, using modeling to determine the perceptual variables under control is described in this paper: https://www.dropbox.com/s/1ajwvsybrfm6kop/Chasin%27Choppers.pdf?dl=0. In this research the object interception behavior observed could be seen as caused output; the view of the helicopter causes the actions of the pursuer. But this was assumed to be an illusion and, indeed, a model that controls aspects of the optical projection of the helicopter on the retina explains the behavior nearly perfectly.Â

RM: While the TCV is aimed at the discovery of the perceptual aspects of input around which an organism’s behavior is organized, it can also reveal the unlikely possibility that the observed behavior is not mediated by a controlled perception; that the apparently caused, reinforced or programmed output really is what it appears to be; that the behavior is really that of a Z system; that the pencil really is bent!Â

RM: So the bottom line answer to your question is, again, no; you don’t have to explicitly test to see whether every behavior of a living system that you observe is an example of a behavioral illusion. You go in assuming that apparently caused, selected or progreammed behavior is an illusion – that the pencil is not really bent. While you are doing the research aimed at understanding the behavior you see – under the assumption that it is organized around control of some perceptual variables – you might – though it is very unlikely – find that it is not the behavior of an N-system. So you might find that a behavior is actually a non-illusory caused output (the saccade is apparently an example). But the default assumption of PCT-based research is that behavior is the control of perceptual input; it is not caused, constrained, selected or programmed output.Â

BestÂ

Rick

–
Richard S. MarkenÂ

“The childhood of the human race is far from over. We
have a long way to go before most people will understand that what they do for
others is just as important to their well-being as what they do for
themselves.” – William T. Powers

On Thu, Sep 1, 2016 at 10:18 PM, Richard Marken rsmarken@gmail.com wrote:

[From Rick Marken (2016.09.01.1320)]

 Fred Nickols (2016.09.01.1554 ET)]

Â

FN: I really like your post below, Rick.Â

RM: Wow, that is so exciting Fred! Thanks. I think this is the first post of mine in weeks that anyone has liked!Â

Â

FN: I also spotted something that I think might present a problem when explaining PCT to others. Here’s the snippet I pulled from your longer post below:

Â

RM: So the “behavioral illusion” refers to the fact that when you observe the behavior of a closed loop control system – living or artificial – what you see is not necessarily what is actually happening.Â

Â

FN: It’s the “what you see is not necessarily what is actually happeningâ€? that’s giving me pause. I think most people, especially some behaviorists I know, would snort or laugh and say something like, “Don’t tell me I’m not seeing what I’m seeing.â€? I think a better way of saying what you’re driving at is to say, “How you explain or account for what you see, especially if you do so in cause-effect or stimulus-response terms, is very different from the way those same observations are explained by control theory.Â

RM: Yes! Much better  (and more correct) way of saying it! Thanks again Fred!

BestÂ

Rick

Â

More important, in light of control theory, those stimulus-response explanations are very misleading. In control theory, especially perceptual control theory (PCT), those misleading explanations are referred to as “the behavioral illusion.�

Â

Â

FN: At least that’s the way it seems to me, Rick.

Â

FN: Anyway, thanks again for this post. I like it – a lot./p>

Â

Regards,

Â

Fred Nickols, Consultant

My Objective is to Help You Achieve Yours

DISTANCE CONSULTING LLC

“Assistance at a Distance�^SM

Â

Â

Â

From: Richard Marken [mailto:rsmarken@gmail.com]
Sent: Thursday, September 01, 2016 2:09 PM
To: csgnet@lists.illinois.edu
Cc: Henry Yin
Subject: Re: The Behavioural Illusion and modelling

Â

[From Rick Marken (2016.09.01.1110)]

Â

cc: Henry Yin, because he understands the importance of the behavioral illusion and I’d like to hear what he thinks of this post.

Â

Martin Taylor (2016.08.30.13.12)–

Â

RM: This is a nice effort to explain the behavioral illusion but I’d like to give my own explanation. I’ll start by defining an illusion as an observation that is not what it appears to be. For example, here is an example of an optical illusion:

Â

RM: What you observe are two parallel vertical lines that bulge out in the middle. This is an illusion because we know, for other reasons (such as from measuring the distance between the lines at the top, middle and bottom) that there is no bulge. Another illusion is the observation of a straight, rigid object, like a pencil, bending when placed in water, like so:Â

Â

Â

RM: Again, this is an illusion because we know, for other reasons (such as feeling along the pencil while it is immersed) that the object is not bent.Â

Â

RM: Behavioral illusions are illusions in the same way these optical illusions are illusions: they are observations that are not what they appear to be. One example of a behavioral illusion is a “reflex”, such as the patellar  of “knee jerk” reflex:

Â

Â

RM: What you see is a hammer tap below the knee causing the lower leg to kick up. The hammer tap looks like a “stimulus” that is the cause of the knee jerk “response”. It looks like stimulus causes response. The stimulus-response appearance of reflexes was not considered an illusion until it was shown to be, by other mean. The “other means” was PCT.Â

Â

RM: The PCT analysis of the stimulus-response illusion starts by noting that behavior is a closed-loop control process organized around the control of the sensory effects of stimuli: controlled variables. When this loop is analyzed correctly it can be shown (as Powers did in his 1978 Psych Review paper) that the observed causal relationship between stimulus and response is actually the inverse of the causal relationship between the response and the sensory effect of the stimulus; this relationship is called the feedback function and the sensory effect of the stimulus is the controlled variable.Â

Â

RM: So the “behavioral illusion” refers to the fact that when you observe the behavior of a closed loop control system – living or artificial – what you see is not necessarily what is actually happening. In the case of the stimulus-response illusion, the actual causal connection from  stimulus to response is “masked” by the fact that it occurs as part of a closed loop control process.Â

Â

RM: There are actually three versions of the behavioral illusion, which are described in my “Blind Men and the Elephant” paper in “More Mind Readings”. Reflexes, like the patellar, are examples of the stimulus-response illusion; operant behavior is an example of the reinforcement illusion; and the power law is an example of the emitted output illusion (called the cognitive illusion in the paper).Â

Â

RM: How “illusory” any of these behavioral illusions are depends on many factors. For example, the stimulus-response illusion is much more pronounced when the stimulus (disturbance to the controlled variable) is abrupt and strong rather than gradual and weak (see “A Bucket of Beans” in LCS II); the emitted output illusion is more pronounced when there are no obvious disturbances to the controlled variable. These illusions are also more pronounced when the system under observation controls well. A system that controls poorly will make only a very weak “response” to a “stimulus”(disturbance) that affects the variable it is controlling. So the stimulus will not give the appearance of causing the response and there will be a very weak stimulus-response illusion.Â

Â

RM: All behavioral illusions result from failure to understand the implications of the fact that one is observing at the behavior of a closed loop control system. It’s like looking at the bent pencil illusion above and not knowing the implications of the fact that the pencil is partly in water. This analogy should make it clear why the behavioral illusion is so important. A researcher who doesn’t know that the refraction of light is different in water and air will start looking for explanations for why the pencil is bent; a researcher who doesn’t know that a closed loop system acts to control its input will start looking for explanations of why stimuli cause responses, why reinforcement selects responses or how responses are emitted.Â

Â

RM: The behavioral illusion is important because it shows that behavioral scientists have been studying the different flavors of the illusion – taking them for what is actually happening – for over 100 years. Â

Â

Best

Â

Rick

Â

–

Richard S. MarkenÂ

“The childhood of the human race is far from over. We have a long way to go before most people will understand that what they do for others is just as important to their well-being as what they do for themselves.” – William T. Powers

–
Richard S. MarkenÂ

“The childhood of the human race is far from over. We
have a long way to go before most people will understand that what they do for
others is just as important to their well-being as what they do for
themselves.” – William T. Powers

[MartinT]

[Martin Taylor 2016.09.01.16.32

[From Rick Marken (2016.09.01.1110)]

      cc: Henry Yin, because he understands the importance of the

behavioral illusion and I’d like to hear what he thinks of
this post.

Fred Nickols liked this post, and so do I (most of it). It

complements mine very nicely. I had thought of adding similar
references to optical illusions and mirages, but wanted to keep it
short.

I agree with Fred here. What you see is the closest you can get to

what is happening in a perceptual world. All you have is your
perception, and as Bill said often enough, it’s the only truth you
can truly rely on. What you perceive is what you perceive, without
question. What is hidden and must be imagined is the reason why you
perceive what you perceive. If you don’t know about control, you
think that the internal processing causes what you see in a apparent
stimulus-response situation. If control is very good, that’s an
illusion. If the “response” cannot influence the “stimulus” it isn’t
illusory. And if control is imperfect, it’s partly illusory. You
have to model the situation to figure out what’s going on. There’s
no one-size-fits-all black-and-whit dress to fit every situation.

According to PCT, the output behaviour is always action to influence

some controlled variable(s) toward its/their reference value(s), but
there are lots of conditions in which an uncontrolled or
uncontrollable perception affects the choice of action to control
that/those perception(s). When you are about to go out of the house
for a walk and you see the sky blue, you don’t take an umbrella, but
if the clouds are dark and getting darker, you probably will. You
can’t control your perception of the sky colour but you can use it
to control your (imagined) perception of being wet or dry later in
your walk.

Taking the umbrella isn't a stimulus-response action to a perception

of dark clouds. If you were simply going out to check the mailbox,
you wouldn’t be likely to take an umbrella unless it was actively
raining. It all depends on what is being controlled at a higher
level. Will you go out for a long walk or will you just pop out and
go back in? The action to implement those controlled perceptions
depends on the sky colour, a perception that is uncontrolled.

Martin

···

              Martin Taylor

(2016.08.30.13.12)–

            RM: This is a nice effort to explain the behavioral

illusion but I’d like to give my own explanation. …

          RM: So the "behavioral illusion" refers to the fact

that when you observe the behavior of a closed loop
control system – living or artificial – what you see is
not necessarily what is actually happening. In the case
of the stimulus-response illusion, the actual causal
connection from stimulus to response is “masked” by the
fact that it occurs as part of a closed loop control
process.

[Alex_Gomez-Marin]

So, your answer is like one of the Ten Commandments: “you shall regard all observed and interpreted output as illusion”.

Two more questions, then:

1. Is a living organism still “a closed loop system” even when the experimenter tries hard “to break the loop” so as to apply the stimulus-response strategy?

2. Why can’t then the experimenter be able to learn some aspects of the organism function, specially when the system is not an ideal system in terms of Bill’s classification?

···

On Fri, Sep 2, 2016 at 6:32 PM, Richard Marken rsmarken@gmail.com wrote:

[From Rick Marken (2016.09.02.0930)]

On Thu, Sep 1, 2016 at 3:24 PM, Alex Gomez-Marin agomezmarin@gmail.com wrote:

AGM: Beautiful analogies, Rick. Now, does this mean that one does not need to prove the behavioral illusion in each particular case because, by construction of argument, one already knows that it must be a behavioral illusion?

RM: A very good question. I think the question is answered in Bill’s 1978 Psych Review paper (“Quantitative Analysis of Purposive Systems”, reprinted in LCS I) but I’ll try to give a quick summary specifically related to your question.Â

RM: I think another way of asking your question would be this: Do we need to prove (in the sense of “test” since we don’t prove propositions in science in the same way as we do in mathematics) in each particular case that the appearance of behavior as caused, selected or planned output is an illusion? If the behavior is that of a living system the answer would be a cautious “no”.

RM:  In his 1978 Psych Review paper Bill showed that living systems are very likely to be closed loop systems. We know that sensory inputs cause behavioral outputs via the nervous systems; and we can see that, at the same time, these behavioral outputs affect those same sensory inputs via the environment. And since the behavior of these systems is generally very stable, the feedback in the closed loop must be negative. So living systems are very likely to be negative feedback systems – Bill calls them N systems in the article – and N systems act to control their own inputs.Â

RM: So we can assume that we are looking at the behavior of a control system (rather than that of an open-loop causal system – Bill calls it a Z system in the article) when we are looking at the behavior of a living organism. That means when we see a living organism apparently reacting to stimuli, having its behavior “strengthened” by its consequences or producing a programmed set of outputs, we can be pretty sure that those appearances are “illusions”. We know this for the same reason that we know that the pencil doesn’t really bend when we place it in water. With living organisms we know that if they are N-systems these behavioral appearances are “side effects” of control of input; with the pencil we know that the appearance of bending is a “side effect” of differential refraction of light by water and air.Â

RM: But as in the case with the pencil in water, there is always the possibility that what is seen is not an illusion. The pencil might really have bent in water if, for example, one side of the pencil were made of a material that expands in water. And the behavior of an organism that we see might really be an example of caused, reinforced or computed output. But this very unlikely possibility is actually implicitly handled by the PCT approach to studying living control systems.Â

RM: PCT research aims to understand the behavior (controlling) of living organisms in terms of the input variables that are being controlled. So when a PCT researcher sees a living organism apparently reacting to stimuli, being manipulated by reinforcement or producing programmed output they try to guess at the aspects of the organisms sensory input around which this behavior might be organized; that is, they try to guess at what input (perceptual) variables might be under control.

RM: The general methodology used to do this kind of research  is called the Test for the Controlled Variable. An example of such research, using modeling to determine the perceptual variables under control is described in this paper: https://www.dropbox.com/s/1ajwvsybrfm6kop/Chasin%27Choppers.pdf?dl=0. In this research the object interception behavior observed could be seen as caused output; the view of the helicopter causes the actions of the pursuer. But this was assumed to be an illusion and, indeed, a model that controls aspects of the optical projection of the helicopter on the retina explains the behavior nearly perfectly.Â

RM: While the TCV is aimed at the discovery of the perceptual aspects of input around which an organism’s behavior is organized, it can also reveal the unlikely possibility that the observed behavior is not mediated by a controlled perception; that the apparently caused, reinforced or programmed output really is what it appears to be; that the behavior is really that of a Z system; that the pencil really is bent!Â

RM: So the bottom line answer to your question is, again, no; you don’t have to explicitly test to see whether every behavior of a living system that you observe is an example of a behavioral illusion. You go in assuming that apparently caused, selected or progreammed behavior is an illusion – that the pencil is not really bent. While you are doing the research aimed at understanding the behavior you see – under the assumption that it is organized around control of some perceptual variables – you might – though it is very unlikely – find that it is not the behavior of an N-system. So you might find that a behavior is actually a non-illusory caused output (the saccade is apparently an example). But the default assumption of PCT-based research is that behavior is the control of perceptual input; it is not caused, constrained, selected or programmed output.Â

BestÂ

Rick

–
Richard S. MarkenÂ

“The childhood of the human race is far from over. We
have a long way to go before most people will understand that what they do for
others is just as important to their well-being as what they do for
themselves.” – William T. Powers

On Thu, Sep 1, 2016 at 10:18 PM, Richard Marken rsmarken@gmail.com wrote:

[From Rick Marken (2016.09.01.1320)]

 Fred Nickols (2016.09.01.1554 ET)]

Â

FN: I really like your post below, Rick.Â

RM: Wow, that is so exciting Fred! Thanks. I think this is the first post of mine in weeks that anyone has liked!Â

Â

FN: I also spotted something that I think might present a problem when explaining PCT to others. Here’s the snippet I pulled from your longer post below:

Â

RM: So the “behavioral illusion” refers to the fact that when you observe the behavior of a closed loop control system – living or artificial – what you see is not necessarily what is actually happening.Â

Â

FN: It’s the “what you see is not necessarily what is actually happeningâ€? that’s giving me pause. I think most people, especially some behaviorists I know, would snort or laugh and say something like, “Don’t tell me I’m not seeing what I’m seeing.â€? I think a better way of saying what you’re driving at is to say, “How you explain or account for what you see, especially if you do so in cause-effect or stimulus-response terms, is very different from the way those same observations are explained by control theory.Â

RM: Yes! Much better  (and more correct) way of saying it! Thanks again Fred!

BestÂ

Rick

Â

More important, in light of control theory, those stimulus-response explanations are very misleading. In control theory, especially perceptual control theory (PCT), those misleading explanations are referred to as “the behavioral illusion.�

Â

Â

FN: At least that’s the way it seems to me, Rick.

Â

FN: Anyway, thanks again for this post. I like it – a lot./p>

Â

Regards,

Â

Fred Nickols, Consultant

My Objective is to Help You Achieve Yours

DISTANCE CONSULTING LLC

“Assistance at a Distance�^SM

Â

Â

Â

From: Richard Marken [mailto:rsmarken@gmail.com]
Sent: Thursday, September 01, 2016 2:09 PM
To: csgnet@lists.illinois.edu
Cc: Henry Yin
Subject: Re: The Behavioural Illusion and modelling

Â

[From Rick Marken (2016.09.01.1110)]

Â

cc: Henry Yin, because he understands the importance of the behavioral illusion and I’d like to hear what he thinks of this post.

Â

Martin Taylor (2016.08.30.13.12)–

Â

RM: This is a nice effort to explain the behavioral illusion but I’d like to give my own explanation. I’ll start by defining an illusion as an observation that is not what it appears to be. For example, here is an example of an optical illusion:

Â

RM: What you observe are two parallel vertical lines that bulge out in the middle. This is an illusion because we know, for other reasons (such as from measuring the distance between the lines at the top, middle and bottom) that there is no bulge. Another illusion is the observation of a straight, rigid object, like a pencil, bending when placed in water, like so:Â

Â

Â

RM: Again, this is an illusion because we know, for other reasons (such as feeling along the pencil while it is immersed) that the object is not bent.Â

Â

RM: Behavioral illusions are illusions in the same way these optical illusions are illusions: they are observations that are not what they appear to be. One example of a behavioral illusion is a “reflex”, such as the patellar  of “knee jerk” reflex:

Â

Â

RM: What you see is a hammer tap below the knee causing the lower leg to kick up. The hammer tap looks like a “stimulus” that is the cause of the knee jerk “response”. It looks like stimulus causes response. The stimulus-response appearance of reflexes was not considered an illusion until it was shown to be, by other mean. The “other means” was PCT.Â

Â

RM: The PCT analysis of the stimulus-response illusion starts by noting that behavior is a closed-loop control process organized around the control of the sensory effects of stimuli: controlled variables. When this loop is analyzed correctly it can be shown (as Powers did in his 1978 Psych Review paper) that the observed causal relationship between stimulus and response is actually the inverse of the causal relationship between the response and the sensory effect of the stimulus; this relationship is called the feedback function and the sensory effect of the stimulus is the controlled variable.Â

Â

RM: So the “behavioral illusion” refers to the fact that when you observe the behavior of a closed loop control system – living or artificial – what you see is not necessarily what is actually happening. In the case of the stimulus-response illusion, the actual causal connection from  stimulus to response is “masked” by the fact that it occurs as part of a closed loop control process.Â

Â

RM: There are actually three versions of the behavioral illusion, which are described in my “Blind Men and the Elephant” paper in “More Mind Readings”. Reflexes, like the patellar, are examples of the stimulus-response illusion; operant behavior is an example of the reinforcement illusion; and the power law is an example of the emitted output illusion (called the cognitive illusion in the paper).Â

Â

RM: How “illusory” any of these behavioral illusions are depends on many factors. For example, the stimulus-response illusion is much more pronounced when the stimulus (disturbance to the controlled variable) is abrupt and strong rather than gradual and weak (see “A Bucket of Beans” in LCS II); the emitted output illusion is more pronounced when there are no obvious disturbances to the controlled variable. These illusions are also more pronounced when the system under observation controls well. A system that controls poorly will make only a very weak “response” to a “stimulus”(disturbance) that affects the variable it is controlling. So the stimulus will not give the appearance of causing the response and there will be a very weak stimulus-response illusion.Â

Â

RM: All behavioral illusions result from failure to understand the implications of the fact that one is observing at the behavior of a closed loop control system. It’s like looking at the bent pencil illusion above and not knowing the implications of the fact that the pencil is partly in water. This analogy should make it clear why the behavioral illusion is so important. A researcher who doesn’t know that the refraction of light is different in water and air will start looking for explanations for why the pencil is bent; a researcher who doesn’t know that a closed loop system acts to control its input will start looking for explanations of why stimuli cause responses, why reinforcement selects responses or how responses are emitted.Â

Â

RM: The behavioral illusion is important because it shows that behavioral scientists have been studying the different flavors of the illusion – taking them for what is actually happening – for over 100 years. Â

Â

Best

Â

Rick

Â

–

Richard S. MarkenÂ

“The childhood of the human race is far from over. We have a long way to go before most people will understand that what they do for others is just as important to their well-being as what they do for themselves.” – William T. Powers

–
Richard S. MarkenÂ

“The childhood of the human race is far from over. We
have a long way to go before most people will understand that what they do for
others is just as important to their well-being as what they do for
themselves.” – William T. Powers

[Alex_Gomez-Marin]

Martin, I also like all these analogies and images, which are great for didactic presentations, etc.

The point is that if “According to PCT, the output behaviour is always action to influence some controlled variable(s) toward its/their reference value(s)” then there is nothing else to say… Namely, if BY CONSTRUCTION OF INTERPRETATION (and I agree there is truth in that interpretation, but not absolute truth!), if output is always illusion, then, of course, the power-law is an illusion (namely, it is indeed output…). If you all prefer to use the word ILLUSION to refer to OUTPUT, alright — you could have chosen something even more condescendeent to the rest of the non-PCT word, something like behavioral OUTPUT = behavioral RUBBISH.

I think I am finally getting it…

···

On Fri, Sep 2, 2016 at 8:31 PM, Martin Taylor mmt-csg@mmtaylor.net wrote:

[Martin Taylor 2016.09.01.16.32

[From Rick Marken (2016.09.01.1110)]

      cc: Henry Yin, because he understands the importance of the

behavioral illusion and I’d like to hear what he thinks of
this post.

Fred Nickols liked this post, and so do I (most of it). It

complements mine very nicely. I had thought of adding similar
references to optical illusions and mirages, but wanted to keep it
short.

I agree with Fred here. What you see is the closest you can get to

what is happening in a perceptual world. All you have is your
perception, and as Bill said often enough, it’s the only truth you
can truly rely on. What you perceive is what you perceive, without
question. What is hidden and must be imagined is the reason why you
perceive what you perceive. If you don’t know about control, you
think that the internal processing causes what you see in a apparent
stimulus-response situation. If control is very good, that’s an
illusion. If the “response” cannot influence the “stimulus” it isn’t
illusory. And if control is imperfect, it’s partly illusory. You
have to model the situation to figure out what’s going on. There’s
no one-size-fits-all black-and-whit dress to fit every situation.

According to PCT, the output behaviour is always action to influence

some controlled variable(s) toward its/their reference value(s), but
there are lots of conditions in which an uncontrolled or
uncontrollable perception affects the choice of action to control
that/those perception(s). When you are about to go out of the house
for a walk and you see the sky blue, you don’t take an umbrella, but
if the clouds are dark and getting darker, you probably will. You
can’t control your perception of the sky colour but you can use it
to control your (imagined) perception of being wet or dry later in
your walk.

Taking the umbrella isn't a stimulus-response action to a perception

of dark clouds. If you were simply going out to check the mailbox,
you wouldn’t be likely to take an umbrella unless it was actively
raining. It all depends on what is being controlled at a higher
level. Will you go out for a long walk or will you just pop out and
go back in? The action to implement those controlled perceptions
depends on the sky colour, a perception that is uncontrolled.

Martin

              Martin Taylor

(2016.08.30.13.12)–

            RM: This is a nice effort to explain the behavioral

illusion but I’d like to give my own explanation. …

          RM: So the "behavioral illusion" refers to the fact

that when you observe the behavior of a closed loop
control system – living or artificial – what you see is
not necessarily what is actually happening. In the case
of the stimulus-response illusion, the actual causal
connection from  stimulus to response is “masked” by the
fact that it occurs as part of a closed loop control
process.

[Huddy_Vyv]

[Vyv Huddy (2130.02.09.2016)]

AGM: If you all prefer to use the word ILLUSION to refer to OUTPUT, alright — you could have chosen something even more condescendent to the rest of the
non-PCT word, something like behavioral OUTPUT = behavioral RUBBISH.

VH: For me illusion is the correct term because it descibes someone providing the wrong answer to the question of why a living organism acts following an external event. The answer stimulus event causes response event not only ignores the controlled variable
but it is also too restricted. One has to know what the organism is controlling for to understand the action. So on one level the controlled variable might be keeping relationship between one event and another at the reference. But for a person that cannot
be the fundamental purpose as there are many higher levels.

VH: Take an example of a participant in an psych experiment pressing a button when a light comes on. In this situation the participant might be controlling their perception of being a good participant
(a priniciple); this is achieved by fulfilling experimental task instructions (a program perception), this is achieved by controlling the relationship between one event (e.g. a light) and another
(e.g. a bar pressed) being held true (see Marken and Mansell, 2013). The latter relationship, only if interpreted as cause and effect, is the illusion. This illusion neglects both the controlled variable and also the higher purpose
that is the final cause of the light on - bar press relationship being held true. I think??

···

---

From: Alex Gomez-Marin agomezmarin@gmail.com
Sent: 02 September 2016 20:51
To: csgnet@lists.illinois.edu
Subject: Re: The Behavioural Illusion and modelling

Martin, I also like all these analogies and images, which are great for didactic presentations, etc.

The point is that if "According to PCT, the output behaviour is always action to influence some controlled variable(s) toward its/their reference value(s) "
then there is nothing else to say… Namely, if BY CONSTRUCTION OF INTERPRETATION (and I agree there is truth in that interpretation, but not absolute truth!),
if output is always illusion, then, of course, the power-law is an illusion (namely, it is indeed output…). If you all prefer to use the word ILLUSION to refer to OUTPUT, alright — you could have chosen something even more condescendent to the rest
of the non-PCT word, something like behavioral OUTPUT = behavioral RUBBISH.

I think I am finally getting it…

On Fri, Sep 2, 2016 at 8:31 PM, Martin Taylor
mmt-csg@mmtaylor.net wrote:

[Martin Taylor 2016.09.01.16.32

[From Rick Marken (2016.09.01.1110)]

cc: Henry Yin, because he understands the importance of the behavioral illusion and I’d like to hear what he thinks of this post.

Martin Taylor (2016.08.30.13.12)–

RM: This is a nice effort to explain the behavioral illusion but I’d like to give my own explanation. …

Fred Nickols liked this post, and so do I (most of it). It complements mine very nicely. I had thought of adding similar references to optical illusions and mirages, but wanted to keep it short.

RM: So the “behavioral illusion” refers to the fact that when you observe the behavior of a closed loop control system – living or artificial – what you see is not necessarily what is actually happening. In the case of the stimulus-response illusion,
the actual causal connection from stimulus to response is “masked” by the fact that it occurs as part of a closed loop control process.

I agree with Fred here. What you see is the closest you can get to what is happening in a perceptual world. All you have is your perception, and as Bill said often enough, it’s the only truth you can truly rely on. What you perceive is what you perceive,
without question. What is hidden and must be imagined is the reason why you perceive what you perceive. If you don’t know about control, you think that the internal processing causes what you see in a apparent stimulus-response situation. If control is very
good, that’s an illusion. If the “response” cannot influence the “stimulus” it isn’t illusory. And if control is imperfect, it’s partly illusory. You have to model the situation to figure out what’s going on. There’s no one-size-fits-all black-and-whit dress
to fit every situation.

According to PCT, the output behaviour is always action to influence some controlled variable(s) toward its/their reference value(s), but there are lots of conditions in which an uncontrolled or uncontrollable perception affects the choice of action to control
that/those perception(s). When you are about to go out of the house for a walk and you see the sky blue, you don’t take an umbrella, but if the clouds are dark and getting darker, you probably will. You can’t control your perception of the sky colour but you
can use it to control your (imagined) perception of being wet or dry later in your walk.

Taking the umbrella isn’t a stimulus-response action to a perception of dark clouds. If you were simply going out to check the mailbox, you wouldn’t be likely to take an umbrella unless it was actively raining. It all depends on what is being controlled at
a higher level. Will you go out for a long walk or will you just pop out and go back in? The action to implement those controlled perceptions depends on the sky colour, a perception that is uncontrolled.

Martin

[Alex_Gomez-Marin]

put the you are saying you know the right answer even before looking.

and that whatever is incomplete is wrong.

again, a stim-resp approach that does not respect the organism autonomy and fails to accomodate to its point of view is still able to extract info of the organism itself!

and, second, environment is not mere contingency, but “the house” of the organism… why insisting in not wanting to learn anything about that in the process? i suspect that pct fails to appreciate the importance, value and interest of the brain-body-world confluence by over-stressing the organism’s references. it is like the usual from on pole to the other dynamics in science. time to see our own biasses!

···

On Fri, Sep 2, 2016 at 8:31 PM, Martin Taylor
mmt-csg@mmtaylor.net wrote:

[Martin Taylor 2016.09.01.16.32

[From Rick Marken (2016.09.01.1110)]

cc: Henry Yin, because he understands the importance of the behavioral illusion and I’d like to hear what he thinks of this post.

Fred Nickols liked this post, and so do I (most of it). It complements mine very nicely. I had thought of adding similar references to optical illusions and mirages, but wanted to keep it short.

I agree with Fred here. What you see is the closest you can get to what is happening in a perceptual world. All you have is your perception, and as Bill said often enough, it’s the only truth you can truly rely on. What you perceive is what you perceive,
without question. What is hidden and must be imagined is the reason why you perceive what you perceive. If you don’t know about control, you think that the internal processing causes what you see in a apparent stimulus-response situation. If control is very
good, that’s an illusion. If the “response” cannot influence the “stimulus” it isn’t illusory. And if control is imperfect, it’s partly illusory. You have to model the situation to figure out what’s going on. There’s no one-size-fits-all black-and-whit dress
to fit every situation.

According to PCT, the output behaviour is always action to influence some controlled variable(s) toward its/their reference value(s), but there are lots of conditions in which an uncontrolled or uncontrollable perception affects the choice of action to control
that/those perception(s). When you are about to go out of the house for a walk and you see the sky blue, you don’t take an umbrella, but if the clouds are dark and getting darker, you probably will. You can’t control your perception of the sky colour but you
can use it to control your (imagined) perception of being wet or dry later in your walk.

Taking the umbrella isn’t a stimulus-response action to a perception of dark clouds. If you were simply going out to check the mailbox, you wouldn’t be likely to take an umbrella unless it was actively raining. It all depends on what is being controlled at
a higher level. Will you go out for a long walk or will you just pop out and go back in? The action to implement those controlled perceptions depends on the sky colour, a perception that is uncontrolled.

Martin

Martin Taylor (2016.08.30.13.12)–

RM: This is a nice effort to explain the behavioral illusion but I’d like to give my own explanation. …

RM: So the “behavioral illusion” refers to the fact that when you observe the behavior of a closed loop control system – living or artificial – what you see is not necessarily what is actually happening. In the case of the stimulus-response illusion,
the actual causal connection from  stimulus to response is “masked” by the fact that it occurs as part of a closed loop control process.

[amatic]

and, second, environment is not mere contingency, but "the house" of the organism.. why insisting in not wanting to learn anything about that in the process? i suspect that pct fails to appreciate the importance, value and interest of the brain-body-world confluence by over-stressing the organism's references. it is like the usual from on pole to the other dynamics in science. time to see our own biasses!

Modeling the environment with more details might produce an interesting simulation. Technically, the environment of the nervous system (in a model of a tracking task) *is* present even in a simple control system with a leaky integral - it just happens to be a very simple model. The nervous system loop that is doing visual tracking is a relatively high-level loop, there are muscle-control loops on levels below and muscle dynamics in the output of that loop, but they seem to produce a very linear response when combined, so they don't always need to be modeled in detail.
On the other hand, the are also models of arm movement that include the lower levels and muscle dynamics and gravitational force and stimuli in the environment (the Little Man, that is an older demo). Maybe also friction or density of the medium can be simulated. There might be some unexpected, non-obvious constraints on curved motion when accurate physics is combined with particular control schemes. There are some difficulties with that approach, the simulation could get quite complex, depending on the desired level of precision, for example what specific muscle model should be used, maybe each tendon needs to be modeled separately, how anatomically accurate does the model-hand need to be, and so on. Similar with other animals like fly larvae - should there be an anatomically accurate muscle-system model, does the perception of skin-pressure need to modeled?

[Alex_Gomez-Marin]

Good points, Adam! Some of Bill’s analytical derivations, because they are mathematically carried out for ideal control systems (to have an explicit analytical solution, they only work in ideal systems) seem to imply that, about the super complexity of nerve-muscle-tendon-bone-world interactions, one only needs to care and estimate the value of a linear constant (like in the rubber band case, where the whole universe ends up in the coupling constant “K”…). If one includes reorganization, then one needs to estimate the integration time and the leaking constant, but not much more. So, there is something deep there about dimensionality because from the “outside” there are so many parameters to potentially take into account, whereas from the animal perspective, perhaps in some (but not all) cases, all the the animal needs to know from the world is just how to tune the constant that represents the linear approximation in the U and V functions.

···

On Sat, Sep 3, 2016 at 3:52 PM, Adam Matic adam.matic@gmail.com wrote:

and, second, environment is not mere contingency, but “the house” of the organism… why insisting in not wanting to learn anything about that in the process? i suspect that pct fails to appreciate the importance, value and interest of the brain-body-world confluence by over-stressing the organism’s references. it is like the usual from on pole to the other dynamics in science. time to see our own biasses!

Modeling the environment with more details might produce an interesting simulation. Technically, the environment of the nervous system (in a model of a tracking task) is present even in a simple control system with a leaky integral - it just happens to be a very simple model. The nervous system loop that is doing visual tracking is a relatively high-level loop, there are muscle-control loops on levels below and muscle dynamics in the output of that loop, but they seem to produce a very linear response when combined, so they don’t always need to be modeled in detail.

On the other hand, the are also models of arm movement that include the lower levels and muscle dynamics and gravitational force and stimuli in the environment (the Little Man, that is an older demo). Maybe also friction or density of the medium can be simulated. There might be some unexpected, non-obvious constraints on curved motion when accurate physics is combined with particular control schemes. There are some difficulties with that approach, the simulation could get quite complex, depending on the desired level of precision, for example what specific muscle model should be used, maybe each tendon needs to be modeled separately, how anatomically accurate does the model-hand need to be, and so on. Similar with other animals like fly larvae - should there be an anatomically accurate muscle-system model, does the perception of skin-pressure need to modeled?

[rsmarken]

[From Rick Marken (2016.09.03.0930)]

···

Martin Taylor (2016.09.01.16.32)–

MT;  If you don't know about control, you

think that the internal processing causes what you see in a apparent
stimulus-response situation. If control is very good, that’s an
illusion. If the “response” cannot influence the “stimulus” it isn’t
illusory.

RM: The “response” (output) of a control system never has an effect on the “stimulus” (disturbance). The “response” affects the same sensory variable that is affected by the “stimulus”.

MT: And if control is imperfect, it’s partly illusory.

RM: The S-R illusion exists regardless of how good control is. The illusion is just more compelling when control is good. When control is poor you see only a statistical relationship between S and R. Sound familiar?

MT: You

have to model the situation to figure out what’s going on. There’s
no one-size-fits-all black-and-whit dress to fit every situation.

RM: This is what PCT research is about. Since you know that living systems are closed-loop, when you see apparent S-R, reinforcement or output generation behavior you start looking for the variable that is under control. In the unlikely event that you find that there is no controlled variable around whch the behavior is organized then you are not seeing the behavior of a control system; you are dealing with behavior that is described by physics (such as the saccade) rather than by control theory.

MT: Taking the umbrella isn't a stimulus-response action to a perception

of dark clouds.

RM: Right. But to the extent that you see it that way, it’s an example of a behavioral illusion. Can you think of the variable that is disturbed by the sight of dark clouds and corrected by taking an umbrella?

MT: If you were simply going out to check the mailbox,

you wouldn’t be likely to take an umbrella unless it was actively
raining. It all depends on what is being controlled at a higher
level.

RM: No, the observed relationship between dark clouds and taking an umbrella depends on the perception that is the disturbed by the sight of dark closed and brought to the reference by taking the umbrella.

Best

Rick

Will you go out for a long walk or will you just pop out and

go back in? The action to implement those controlled perceptions
depends on the sky colour, a perception that is uncontrolled.

Martin

–
Richard S. Marken

“The childhood of the human race is far from over. We
have a long way to go before most people will understand that what they do for
others is just as important to their well-being as what they do for
themselves.” – William T. Powers