Behaviour and "Behaviour" (was Re: -- a lot of things)

[Martin Taylor 2015.10.29.00.15]

Not how I have ever read it. I have always read it as "Behaviour is

the way perception is controlled" or, more shortly “Behaviour is for
the control of perception”. That phrase, at least in CSGnet
discussion, was often qualified by the insertion of “intentional”,
which disqualifies nervous tics and side-effects as behaviour; but
it makes clear that Behaviour, in PCT, maps onto the conventional
usage of “behaviour” exactly by analogy to the way Perception in PCT
maps onto the everyday use of “perception”. I can’t claim to know
what Bill was thinking when he created the book title, but I can
claim that my interpretation is natural, and makes logical sense in
the context of his use of “perception” to refer to the controlled
variable.
Exactly! Isn’t that what I tried to get across in my messages to
Boris? The behavioural output influences the Complex Environmental
Variable, and thereby influences the perception. The whole control
loop ensures (if all the parameters are suitable) that the influence
of the behaviour on the perception is in the direction of reducing
the error.
In the passage that Rick keeps trying to persuade people to read
(LCS I, 171-76) Bill restates the old point that there is never one
cause for anything that happens, but he restates it in the context
of perceptual control. It’s a general truth that there is no one
cause for anything, and the fact that, with different reference
values and different disturbance values, the control system needs
different output values to achieve the same result is just one more
example of that general truth. If I trip over a misplaced toy and fall, the cause is the toy, but I
wouldn’t have fallen if I had been in another room, so the cause of
my fall is my being in the same room as the toy. OK, that’s
nonsense, but maybe it makes the point. The cause of any observable
event is a confluence of many variables. When Bill talks about the
behaviour of opening the door, he lists a lot of things that could
vary, such as the force needed to cause the door to open under the
circumstances. That’s an error of kind. Opening a door is not a
force, and is irrelevant to the control loop for which the
controlled peception is the door angle. The applied force is a
behaviour relevant to a much lower-level controlled perception.
Opening the door is sui generis, the observable behaviour that
influences the (believed to exist) physical door-angle so as to
bring the door-angle perception to its reference value.
Bill lists “Behaviours” on p172 just as I would, despite Rick’s
assertion to the contrary and despite Bill’s own text on p 173. His
“means” column lists behaviours used in control loops for a few of
the necessary lower-level controlled perceptions. As I wrote (in a
message a couple of days ago that I haven’t yet seen appear on
CSGnet), when I control for producing the image of a certain letter
on my screen, my behaviour is hitting the right key, not moving my
finger to the right place, bending it appropriately, lowering it,
and then removing it; when I control my perception of the party in
Government, my behaviour is voting, not walking to the polling
station, not picking up a pencil, not putting an X on a piece of
paper; when I control my perception of others understanding my
meanings, my behaviour is constructing an argument, not formulating
words, phrases and sentences, and not executing typing movements on
a keyboard.
Behaviour is part of any control loop. It is not the whole of the
control loop, for if it were, why have the two words “behaviour” and
“control”. But I have said all this before, for which I apologise if
you already saw it.
Martin

···

On 2015/10/28 11:34 PM, Bruce Nevin
wrote:

Nice post, Martin.

In the title Behavior: the control of perception ,
the colon is equivalent to is in the sentence
"behavior is the control of perception.

      Consequently, in order to talk about "behavior" in this PCT

sense we have to talk about the entire loop, because it is the
entire loop that controls perception. The reference value
doesn’t control perception. The behavioral outputs don’t
control perception. Circular causation around the entire
closed loop controls the perceptual input in accord with the
reference value.

[Bruce Nevin (20151029.1310)]

BN: Nice post, Martin.

BN: In the title Behavior: the control of perception, the colon is equivalent to is in the sentence "behavior is the control of perception.Â

MMT (2015.10.29.00.15): Not how I have ever read it. I have always read it as “Behaviour is the way perception is controlled” or, more shortly “Behaviour is for the control of perception”.Â

This identifies ‘behavior’ with ‘behavioral outputs’ or ‘actions’. That is the usual sense of the word. From the Random House Unabridged Dictionary:

behavior

—behaavioral, adj. —behaviorally, adv.

  /bi hayv"yeuhr/, n.

    1. manner of behaving or acting.

    2. Psychol., Animal Behav.

       a. observable activity in a human or animal.

       b. the aggregate of responses to internal and external stimuli.

       c. a stereotyped, species-specific activity, as a courtship dance or startle reflex.

    3. Often, behaviors. a behavior pattern.

    4. the action or reaction of any material under given circumstances: the behavior of tin under heat.

  Also, esp. Brit., behaviour.

Behavior (in the usual sense) is “observable activity”. Observability extends probably inward as far as observable muscle tensions and outward to proximal environmental consequences of movements (e.g. bar presses), where it can get rather sloppy with unacknowledged recognition of intentions. (“He’s driving a nail. No, he’s adjusting the position of that rafter. Oh … no, he’s killing ants. He just gets really upset when he sees ants and the building’s not even closed in yet.”)Â

I hold to my ‘in your face’ interpretation. In support of my reading, there is abundant testimony that the conventional use of the colon is to express equivalence, e.g.: https://guinlist.wordpress.com/tag/equivalence/

I understand that you would prefer it to be otherwise in the title B:CP, but I hope that you will at least acknowledge this widely accepted convention and the legitimacy of reading it as “behavior is the control of perception”. If you do take this into consideration, e.g. by saying that you recognize that there are two possible readings, and that you prefer the other one, I think you may be able to control your perception of others’ understanding more successfully.Â

BN: Consequently, in order to talk about “behavior” in this PCT sense we have to talk about the entire loop, because it is the entire loop that controls perception. The reference value doesn’t control perception. The behavioral outputs don’t control perception. Circular causation around the entire closed loop controls the perceptual input in accord with the reference value.

MMT: Exactly! Isn’t that what I tried to get across in my messages to Boris?Â

Please don’t assume that because I paraphrase what you said my aim was necessarily to disagree, or to put a finer point on it, or something like that. Paraphrase is a good demonstration of understanding. Paraphrase offers other readers a second way of expressing it in words, which may help their understanding. Paraphrase demonstrates that at least two people (author of the first utterance and author of the second) are collectively controlling perceptual variables in those two ways, which enhances others’ perception that there is something ‘real’ there into whose collective control they too might enter with their own ways of expressing their understandings.Â

MMT: That phrase, at least in CSGnet discussion, was often qualified by the insertion of “intentional”, which disqualifies nervous tics and side-effects as behaviour; but it makes clear that Behaviour, in PCT, maps onto the conventional usage of “behaviour” exactly by analogy to the way Perception in PCT maps onto the everyday use of “perception”. I can’t claim to know what Bill was thinking when he created the book title, but I can claim that my interpretation is natural, and makes logical sense in the context of his use of “perception” to refer to the controlled variable.

I read that as an acknowledgement that the PCT sense of behavior as control is at variance from the common usage (“observable activity in a human or animal”) while communicating to the stream of newcomers to PCT on CSG-net. I agree that ordinary usage is natural, i.e. the conventional norm. I agree that the PCT use of ‘perception’ (meaning a transform of the input quantity derived from the observed CV) aligns with the ordinary usage of that word; but the parallel term for the output quantity on the other side of the loop is not ‘behavior’, it is ‘behavioral outputs’ or ‘actions’.

Behavior (PCT sense) is, as you say, intentional, as distinct from observed activity that is caused by organism-internal disturbances (nervous tics, burps, & such). I agree that because it is what the entire loop does, and not just the observable activities, behavior (PCT sense) includes the intended environmental effects of those actions, and unintended consequences (side effects) are not included. (If side effects affect the CV as disturbances, perception of them may enter into another control loop which may then make adjustments to the first loop.) The ordinary definition of ‘behavior’ as “observable activity in a human or animal” does not exclude unintentional activity, although of course everyone informally does make that distinction because we have an intuitive understanding of intentions. It is not possible to talk about intentional behavior correctly without talking about control. Intentional behavior is control of behavior.

MMT: a message a couple of days ago that I haven’t yet seen appear on CSGnet …

I see it in  in [Martin Taylor 2015.10.26.16.23]. I agree with Rick’s reply the next day, [Rick Marken (2015.10.27.1000)]:

RM: Behavior – any behavior – is not just the visible Means we see a person using to get a Variable to the Reference state; “Opening the door” is not just the grasping/pulling forces exerted on the door. Nor is behavior just the varying state of the controlled Variable; “Opening the door” is not just the changing angle of the door as it is opened. Nor is behavior just the Reference state of the controlled variable;  “Opening the door” is not just the angle of the door when it has been open. “Behavior” is all three variables [as the controlled variable is] brought to [its] reference [state] by appropriate means –  which turn out to be the the three observable components of control. That’s why I say that “behavior is control”. “Behavior” is all three variables brought to reference states by appropriate means –  which turn out to be the the three observable components of control. That’s why I say that “behavior is control”.

Rick, I was puzzled that you said that all three variables–means, CV, and reference state of CV–are brought to reference states by appropriate means. I offer this friendly amendment:

“Behavior” is all three variables [as the controlled variable is] brought to [its] reference [state] by appropriate means – Â which turn out to be the the three observable components of control. That’s why I say that “behavior is control”.

Of course, the means of controlling CV = “door open” (grasping, pulling) themselves involve subordinate CVs controlled to reference values by subordinate means, and there are superordinate purposes for which opening the door is the means. Maybe that’ was confusing the picture as you wrote that sentence.

···

On Thu, Oct 29, 2015 at 1:12 AM, Martin Taylor mmt-csg@mmtaylor.net wrote:

[Martin Taylor 2015.10.29.00.15]

  On 2015/10/28 11:34 PM, Bruce Nevin

wrote:

Nice post, Martin.

In the title Behavior: the control of perception ,
the colon is equivalent to is  in the sentence
"behavior is the control of perception.

Not how I have ever read it. I have always read it as "Behaviour is

the way perception is controlled" or, more shortly “Behaviour is for
the control of perception”. That phrase, at least in CSGnet
discussion, was often qualified by the insertion of “intentional”,
which disqualifies nervous tics and side-effects as behaviour; but
it makes clear that Behaviour, in PCT, maps onto the conventional
usage of “behaviour” exactly by analogy to the way Perception in PCT
maps onto the everyday use of “perception”. I can’t claim to know
what Bill was thinking when he created the book title, but I can
claim that my interpretation is natural, and makes logical sense in
the context of his use of “perception” to refer to the controlled
variable.

      Consequently, in order to talk about "behavior" in this PCT

sense we have to talk about the entire loop, because it is the
entire loop that controls perception. The reference value
doesn’t control perception. The behavioral outputs don’t
control perception. Circular causation around the entire
closed loop controls the perceptual input in accord with the
reference value.

Exactly! Isn't that what I tried to get across in my messages to

Boris? The behavioural output influences the Complex Environmental
Variable, and thereby influences the perception. The whole control
loop ensures (if all the parameters are suitable) that the influence
of the behaviour on the perception is in the direction of reducing
the error.

In the passage that Rick keeps trying to persuade people to read

(LCS I, 171-76) Bill restates the old point that there is never one
cause for anything that happens, but he restates it in the context
of perceptual control. It’s a general truth that there is no one
cause for anything, and the fact that, with different reference
values and different disturbance values, the control system needs
different output values to achieve the same result is just one more
example of that general truth.

If I trip over a misplaced toy and fall, the cause is the toy, but I

wouldn’t have fallen if I had been in another room, so the cause of
my fall is my being in the same room as the toy. OK, that’s
nonsense, but maybe it makes the point. The cause of any observable
event is a confluence of many variables. When Bill talks about the
behaviour of opening the door, he lists a lot of things that could
vary, such as the force needed to cause the door to open under the
circumstances. That’s an error of kind. Opening a door is not a
force, and is irrelevant to the control loop for which the
controlled peception is the door angle. The applied force is a
behaviour relevant to a much lower-level controlled perception.
Opening the door is sui generis, the observable behaviour that
influences the (believed to exist) physical door-angle so as to
bring the door-angle perception to its reference value.

Bill lists "Behaviours" on p172 just as I would, despite Rick's

assertion to the contrary and despite Bill’s own text on p 173. His
“means” column lists behaviours used in control loops for a few of
the necessary lower-level controlled perceptions. As I wrote (in a
message a couple of days ago that I haven’t yet seen appear on
CSGnet), when I control for producing the image of a certain letter
on my screen, my behaviour is hitting the right key, not moving my
finger to the right place, bending it appropriately, lowering it,
and then removing it; when I control my perception of the party in
Government, my behaviour is voting, not walking to the polling
station, not picking up a pencil, not putting an X on a piece of
paper; when I control my perception of others understanding my
meanings, my behaviour is constructing an argument, not formulating
words, phrases and sentences, and not executing typing movements on
a keyboard.

Behaviour is part of any control loop. It is not the whole of the

control loop, for if it were, why have the two words “behaviour” and
“control”. But I have said all this before, for which I apologise if
you already saw it.

Martin

[Bruce Nevin (2015.10.29.23:15)]

RM: Behavior – any behavior – is not just the visible Means we see a person using to get a Variable to the Reference state; “Opening the door” is not just the grasping/pulling forces exerted on the door.

MMT: No, of course it isn’t. Those are lower-level behaviours used in controlling lower-level perceptions.

Behavior has a visible public aspect and a private aspect that can only be inferred from the model. The private aspect includes the perception by the actor and the reference for that perception. The public aspect includes the activity, the behavioral outputs, and the effects of the activity in the environment, including in particular the stabilization of the controlled environmental variable, all as perceived by the observer, Both aspects are part of behavior. If there is no reference, no purpose or intention, the activity, or what looks like activity to the observer, is not the visible aspect of behavior. If environmental effects of the activity are not perceived by the actor, and if the delta between the reference and that perception does not determine the course of the activity, then the activity is not the visible aspect of behavior. The word ‘behavior’ necessarily includes both the inferred, private aspect and the observable public aspect. In acquiescence to common usage, we often use the word ‘behavior’ as though it only referred to the public, observable behavioral outputs. This is careless, but readily understandable.

···

On Thu, Oct 29, 2015 at 6:27 PM, Martin Taylor mmt-csg@mmtaylor.net wrote:

[Martin Taylor 2015.10.29.17.11]

  (Extended quote to start, but

interspersed comments follow)

        [Bruce Nevin

(20151029.1310)]

              BN: Nice

post, Martin.

              BN: In the

title Behavior: the control of perceptio n, the
colon is equivalent to is in the sentence
"behavior is the control of perception.Â

MMT (2015.10.29.00.15 ): Not how I have ever read it. I
have always read it as “Behaviour is the way perception is
controlled” or, more shortly “Behaviour is for the control
of perception”.Â

        This identifies 'behavior'

with ‘behavioral outputs’ or ‘actions’. That is the usual
sense of the word. From
the Random House Unabridged Dictionary:

behavior

            —behavioral, adj.

—behaviorally, adv.

  /bi hayv"yeuhr/, n.

            Â  Â  Â  Â 1. manner of

behaving or acting.

            Â  Â  Â  Â 2. Psychol.,

Animal Behav.

            Â  Â  Â  Â  Â  Â  Â a.

observable activity in a human or animal.

            Â  Â  Â  Â  Â  Â  Â b. the

aggregate of responses to internal and external stimuli.

            Â  Â  Â  Â  Â  Â  Â c. a

stereotyped, species-specific activity, as a courtship
dance or startle reflex.

            Â  Â  Â  Â 3. Often,

behaviors. a behavior pattern.

            Â  Â  Â  Â 4. the action or

reaction of any material under given circumstances: the
behavior of tin under heat.

            Â  Â Also, esp. Brit.,

behaviour.

        Behavior (in the usual

sense) is " observable
activity ".
Observability extends probably inward as far as observable
muscle tensions and outward to proximal environmental
consequences of movements (e.g. bar presses), where it can
get rather sloppy with unacknowledged recognition of
intentions. (“He’s driving a nail. No, he’s adjusting the
position of that rafter. Oh … no , he’s killing ants. He just gets
really upset when he sees ants and the building’s not even
closed in yet.”)Â

        I hold to my 'in your face'

interpretation. In
support of my reading, there is abundant testimony that the
conventional use of the colon is to express equivalence,
e.g.:Â https://guinlist.wordpress.com/tag/equivalence/

        I understand that you would

prefer it to be otherwise in the title B:CP, but I hope that
you will at least acknowledge this widely accepted
convention and the legitimacy of reading it as “behavior is
the control of perception”.

Bruce, I find this whole discussion rather weird. Up to now, I have

always assumed that Bill wrote B:CP for people not already familiar
with PCT. If that was not the case, and he was preaching to the
choir ( which I seriously doubt), then I could accept the
interpretation you and Rick put on “behaviour”. But if he was trying
to grab the attention of a new audience, as I believe, then that
interpretation makes no sense. “Behaviour” in the title has to mean
what his intended audience would have understood it to mean, as
illustrated in your “sloppy” examples.

Here's an expanded version of my "Behaviour is for the control of

perception": “Behaviour is the observable outward manifestation of
what an organism does when controlling a perception”. That’s just as
much “in your face” as your conflation of behaviour with the entire
control loop, which gives a novel in-group-only reading to
“behaviour”.

Years ago, we had a discussion about "You can't tell what someone is

doing by looking at what they are doing" and we used the example of
a man pushing a button set beside a door. I’m at the cottage and
have only my laptop, and on it I can’t find the multi-level diagram
that showed some of the possible behaviours the man might be doing,
or I would have included it at this point. Here are some
possibilities, anyway.

The behaviour might be:

Pressing a button

Testing whether a newly installed bell rings when the button is

pushed.

Visiting Aunt Maude.

Checking whether the house is empty.

Preparing to burgle the house.

Getting help after a traffic accident.

...

All of the above (though one has to imagine a rather strange nephew

and an odd coincidence for them all to be true simultaneously).

All of these are behaviours, possibly controlling perceptions of:

"the feel of the button pressure on the finger"

"the functioning of the bell"

"the pleasure Aunt Maude has when I visit"

"my uncertainty about whether the house is occupied"

"my uncertainty as to the vulnerability of the house to burglary"

"the presence of an ambulance at my car"

and so forth.



The behaviours, to me, are not the control loops. They are what

someone outside can see, and from which they might be able to infer
something about controlled perceptions, especially if they then
perform the Test for the Controlled Variable at the different
perceptual levels.

        MMT:

That phrase, at least in CSGnet discussion, was often
qualified by the insertion of “intentional”, which
disqualifies nervous tics and side-effects as behaviour; but
it makes clear that Behaviour, in PCT, maps onto the
conventional usage of “behaviour” exactly by analogy to the
way Perception in PCT maps onto the everyday use of
“perception”. I can’t claim to know what Bill was thinking
when he created the book title, but I can claim that my
interpretation is natural, and makes logical sense in the
context of his use of “perception” to refer to the
controlled variable.

      I read that as an acknowledgement that the PCT sense of

behavior as control is at variance from the common usage (" observable activity in a human or
animal ") while communicating to the stream of
newcomers to PCT on CSG-net. I agree that ordinary usage is
natural, i.e. the conventional norm. I agree that the PCT use
of ‘perception’ (meaning a transform of the input quantity
derived from the observed CV) aligns with the ordinary usage
of that word; but the parallel term for the output quantity on
the other side of the loop is not ‘behavior’, it is
‘behavioral outputs’ or ‘actions’.

Why? Why not use the word exactly by analogy with "perception", so

that it includes the everyday usage within the technical usage? Why
force it to mean something for which there is already a perfectly
good word, and thus require a new word to mean what the everyday
term means to everyone not on CSGnet, including people who might be
willing to learn PCT?

            MMT: a message a couple

of days ago that I haven’t yet seen appear on CSGnet …

          I see it in  in [Martin

Taylor 2015.10.26.16.23]. I agree with Rick’s reply the
next day, [Rick Marken (2015.10.27.1000)]:

I have Rick's reply, and my message is in my "Sent" box, but it

hasn’t appeared for me on CSGnet, as all my other postings have
done.

            RM: Behavior -- any

behavior – is not just the visible Means we see a
person using to get a Variable to the Reference state;
“Opening the door” is not just the grasping/pulling
forces exerted on the door.

No, of course it isn't. Those are lower-level behaviours used in

controlling lower-level perceptions.

Nor is behavior just
the varying state of the controlled Variable; “Opening
the door” is not just the changing angle of the door as
it is opened.

No, nor is it the force applied to change that angle.
            Nor is behavior just the

Reference state of the controlled variable; Â “Opening
the door” is not just the angle of the door when it has
been open.

No, it's part of a process, not the value of ANY variable or set of

variables. To equate “behaviour” with a variable is what is called
in engineering a “dimensional error”, of the same kind as equating
Force with Mass, on the grounds that when you apply force, the
amount of acceleration depends (in free space) only on the mass.

            "Behavior" is all three

variables [as the controlled variable is] brought to
[its] reference [state] by appropriate means – Â which
turn out to be the the three observable components of
control. That’s why I say that “behavior is
control”. “Behavior” is all three variables brought to
reference states by appropriate means – Â which turn out
to be the the three observable components of control.
That’s why I say that “behavior is control”.

      Rick, I was puzzled that

you said that all three variables–means, CV, and reference
state of CV–are brought to reference states by appropriate
means. I offer this friendly amendment:

            "Behavior" is all three

variables [as the controlled variable is] brought to
[its] reference [state] by appropriate means – Â which
turn out to be the the three observable components of
control. That’s why I say that “behavior is control”.

I would amend it slightly differently, for two reasons: the

dimensional error mentioned above, and that control is not always,
if ever, perfect. “Behaviour is what an organism can be observed to
do so as to bring a controlled variable to its reference state, as
manifest in the results on some observable property of the
observer’s environment.” If you observe someone turning a
doorhandle, pushing on the door, and when the door is open, passing
through the portal, you can reasonably infer that the person was
performing the behaviour of “going into another room” and to do so
performed the behaviour of “opening the door”, is support of which
he performed the behaviours of “turning the door handle” and
“pushing the door”. (Note all the "-ing"s).

        Of course, the means of controlling CV = "door open"

(grasping, pulling) themselves involve subordinate CVs
controlled to reference values by subordinate means, and
there are superordinate purposes for which opening the door
is the means. Maybe that’ was confusing the picture as you
wrote that sentence.

/Bruce

Not to mention confusing anyone trying to learn about PCT.

Martin
        On Thu, Oct 29, 2015 at 1:12 AM,

Martin Taylor mmt-csg@mmtaylor.net
wrote:

            [Martin Taylor

2015.10.29.00.15]

On 2015/10/28 11:34 PM, Bruce Nevin wrote:

Nice post, Martin.

In the title * Behavior: the control of
perception*, the colon is equivalent to is  in
the sentence "behavior is the control of
perception.

            Not how I have ever read it. I have always read it as

“Behaviour is the way perception is controlled” or, more
shortly “Behaviour is for the control of perception”.
That phrase, at least in CSGnet discussion, was often
qualified by the insertion of “intentional”, which
disqualifies nervous tics and side-effects as behaviour;
but it makes clear that Behaviour, in PCT, maps onto the
conventional usage of “behaviour” exactly by analogy to
the way Perception in PCT maps onto the everyday use of
“perception”. I can’t claim to know what Bill was
thinking when he created the book title, but I can claim
that my interpretation is natural, and makes logical
sense in the context of his use of “perception” to refer
to the controlled variable.

                  Consequently, in order to talk about "behavior"

in this PCT sense we have to talk about the entire
loop, because it is the entire loop that controls
perception. The reference value doesn’t control
perception. The behavioral outputs don’t control
perception. Circular causation around the entire
closed loop controls the perceptual input in
accord with the reference value.

            Exactly! Isn't that what I tried to get across in my

messages to Boris? The behavioural output influences the
Complex Environmental Variable, and thereby influences
the perception. The whole control loop ensures (if all
the parameters are suitable) that the influence of the
behaviour on the perception is in the direction of
reducing the error.

            In the passage that Rick keeps trying to persuade people

to read (LCS I, 171-76) Bill restates the old point that
there is never one cause for anything that happens, but
he restates it in the context of perceptual control.
It’s a general truth that there is no one cause for
anything, and the fact that, with different reference
values and different disturbance values, the control
system needs different output values to achieve the same
result is just one more example of that general truth.

            If I trip over a misplaced toy and fall, the cause is

the toy, but I wouldn’t have fallen if I had been in
another room, so the cause of my fall is my being in the
same room as the toy. OK, that’s nonsense, but maybe it
makes the point. The cause of any observable event is a
confluence of many variables. When Bill talks about the
behaviour of opening the door, he lists a lot of things
that could vary, such as the force needed to cause the
door to open under the circumstances. That’s an error of
kind. Opening a door is not a force, and is irrelevant
to the control loop for which the controlled peception
is the door angle. The applied force is a behaviour
relevant to a much lower-level controlled perception.
Opening the door is sui generis, the observable
behaviour that influences the (believed to exist)
physical door-angle so as to bring the door-angle
perception to its reference value.

            Bill lists "Behaviours" on p172 just as I would, despite

Rick’s assertion to the contrary and despite Bill’s own
text on p 173. His “means” column lists behaviours used
in control loops for a few of the necessary lower-level
controlled perceptions. As I wrote (in a message a
couple of days ago that I haven’t yet seen appear on
CSGnet), when I control for producing the image of a
certain letter on my screen, my behaviour is hitting the
right key, not moving my finger to the right place,
bending it appropriately, lowering it, and then removing
it; when I control my perception of the party in
Government, my behaviour is voting, not walking to the
polling station, not picking up a pencil, not putting an
X on a piece of paper; when I control my perception of
others understanding my meanings, my behaviour is
constructing an argument, not formulating words, phrases
and sentences, and not executing typing movements on a
keyboard.

            Behaviour is part of any control loop. It is not the

whole of the control loop, for if it were, why have the
two words “behaviour” and “control”. But I have said all
this before, for which I apologise if you already saw
it.

                Martin

[Bruce Nevin (2015.10.30.18:04 ET)]

Martin, my friend, we may secretly be in violent agreement. That’s often the case in terminological quarrels. Your main concern is to communicate clearly without undue barriers. I share that concern. Now let us reconsider just what it is that must be communicated. It doesn’t matter how clearly understood our words are if our readers understand from them the wrong thing.

Let’s back up. I believe the origin was in discussion of Rick’s spreadsheet. This spreadsheet is organized according to the table on p. 172 of LCS I and the discussion on pp. 171-176. I note that on p. 171 Bill called it “a partial listing of activities”. The first column is labeled “Behavior” and contains “phrases of a kind used both in ordinary discourse and in scientific psychology to denote what an organism is doing.” For lay people and for most scientists “No analysis takes place.”

A partial analysis is laid out in the other three columns. Bill says (p. 173, 4th paragraph) “We can see immediately that the real actions of the driver … are all in the Means column. What is so casually called behavior results from the conjunction of many forces, only one of which is contributed by the driver. The Behavior column really lists consequences of the driver’s actions, consequences that are not determined by the driver’s actions, but are only influenced by them.” As an example, “Grasp, pull” gives more detail to “open [the] door”. Of course, yet more detail could be given, e.g. for “grasp”: extend fingers, reduce distance between fingers and handle, flex fingers around handle, etc. This is a partial analysis.

If the real actions are in the Means column, the intended consequences of the actions are in the last two columns, the controlled variable (“angle of door”) and the reference state for that variable (“80 degrees”). Because these consequences of actions are not just any consequences.

I apologize for belaboring this. This is the stuff of a PCT 101 class, or even a Pre-PCT cass. (Wouldn’t that be an excellent thing for our children to attend!) I know you have no quarrel with anything that Bill wrote here. You just feel that we should use the word “behavior” to mean “observable activity”, because it is the usage of the word that is most familiar to everyone. Why erect barriers to understanding, you ask. I laud your desire to be easily understood. I recently quoted a bit of your writing as an example of good communication over difficulties a lot of people have with the word “control”.

But just as you helped your readers understand that control is not what they might think, we also must help people understand that behavior is not what they might think. Behavior is not just observable activity. Behavior is always purposeful. (Setting aside the metaphoric usage in e.g. “the behavior of sodium chlorite in the presence of an acid” or “the behavior of trees in a high wind”.) Behavior has consequences and (unless it is thwarted) those consequences are intended.

You say we have a perfectly good word for this, namely, control. Well and good, but we also have perfectly good words for that to which you want to limit “behavior”, namely, observable actions, behavioral outputs, etc. There are good reasons why Bill and others have always employed such words to be clear, rather than referring to them as “behavior”, which the naive reader would naturally understand in the old way.

That reason is that the nature of behavior has been misunderstood until the advent of PCT. This is why Bill said in 2011, and I quote him again, “behavior is control of perception”. Or as he wrote in the 1973 Preface to B:CP, “Behavior is the process by which organisms control their input sensory data” (that’s in the last paragraph, on p. xiv of the 2005 edition). Not just the observable activities; the entire process.

It is central to the success of PCT to communicate that behavior is not what we have been led to believe. “Behavior is the process by which organisms control their input sensory data.” “Behavior is control of perception.”

···

On Fri, Oct 30, 2015 at 11:12 AM, Martin Taylor mmt-csg@mmtaylor.net wrote:

[Martin Taylor 2014.10.30.10.57]

[Bruce Nevin (2015.10.29.23:15)]

        RM: Behavior -- any behavior

– is not just the visible Means we see a person using to
get a Variable to the Reference state; “Opening the door” is
not just the grasping/pulling forces exerted on the door.

        MMT: No, of course it isn't. Those are lower-level

behaviours used in controlling lower-level perceptions.

      Behavior has a visible public aspect and a private aspect

that can only be inferred from the model. The private aspect
includes the perception by the actor and the reference for
that perception…

I don't agree that this is a good use for the word "behaviour".

“Control” is a perfectly good word for what you describe. “Control”,
like other words with technical meanings in PCT, also means
something to a PCT-naive person. As with “perception”, the everyday
meaning of “control” is closely related to the technical meaning.
The way I prefer to use the word “behaviour”, it also is closely
related to the everyday meaning.

Nothing in this discussion has persuaded me that it is a good idea

(from the viewpoint of trying to propagate PCT) to (1) destroy the
relation between the technical and the everyday meaning, and (2) use
the word to mean something different for which a perfectly good
technical word already is in ordinary usage. Even eliminating use of
the word “behaviour” from PCT discourse (which I do not favour)
would be preferable.

I see little chance of reconciliation, but little does not mean

none. I think it’s a bad idea to have PCT-lingo split into mutually
unintelligible dialects, and a worse idea to widen unnecessarily the
inevitable split between PCT-lingo and its parent language –
English as understood by those we hope will learn and develop PCT.

Martin

[Bruce Nevin (2015.10.30.22:45 ET)]

Yes. The word has both meanings. Newcomers only know one. And even that one (behavior=actions) is incorrectly understood without the other (behavior=control).

···

On Fri, Oct 30, 2015 at 7:31 PM, Bruce Abbott bbabbott@frontier.com wrote:

[From Bruce Abbott (2015.10.30.1930 EDT)]

Â

Martin Taylor 2015.10.30.11.21

MT: PS. In one of the messages Bruce [Nevin] referred to “You can’t tell what someone is doing by observing what they are doing” as evidence that “behaviour” should include the whole of control. I would read the mantra differently, without using the word “behaviour” or any PCT-technical words: “You can’t tell what someone wants to achieve by observing their actions.” In that, to make it technical, I would make these substitutions: “what someone want to achieve” -> “what perception someone is controlling with what reference value”, and “actions” -> “behaviour”. In everyday speech, those two elements are quite distinct, and indeed the ordinary observer might say something like “If he wants to get what he wants, that kind of behaviour won’t work”. Translated the way I would like to use the term: “That behaviour is ineffective for control of this perception”, as in “Riding a bike in a snowstorm at night isn’t a good way to stay safe.”
BA: As the risk of getting stuck to this “tar baby,� my own reading of the title to Bill’s book matches yours, Martin: “Behavior: the Control of Perception� I understand to mean that the function of behavior is to control perception. Read in this way, the title constitutes a broadside against the prevailing open-loop view of behavior as mere output: the product of environmental contingencies acting upon the organism.

Â

BA: Back in the 1950’s when Bill was developing his ideas, Behaviorism was in its heyday, and for behaviorists, a major task of psychology was to identify functional relationships between the environmental conditions to which an organism is exposed and its behavior. The appearance of the term “behavior� in Bill’s title was sure catch the attention of behaviorists, especially with its bold assertion that behavior, far from being the mere product of a set of environmental contingencies acting on the organism through its physiological mechanisms, exists for the purpose of controlling certain of the organism’s perceptual inputs.

Â

BA: Unfortunately the term “behavior� in ordinary usage can mean different things, leading to such statements as “you can’t tell what a person is doing by observing what they are doing,� in which the first “doing� refers to what changes the person is trying to bring about, and the second “doing� refers to what activities the observer sees the person engaged in. Bill liked to distinguish these two meanings by substituting the term “actions� for “behavior� when referring to the behavioral output of a control system. Such actions (or more accurately the perceptual consequences of them) can themselves be under control at a lower level, although they need not be. Thus the action of flipping a light (what the person is observed to “do�) may be the means by which the person controls the intensity of the lighting (what the person is “doing�  by manipulating the switch).

Â

BA: In ordinary speech we often talk about behavior in terms of its perceptual effects, e.g., “I’m turning up the heatâ€? means that I’m doing something that I believe will increase the magnitude of my perception of the temperature. This usage tends to mask the role of the behavior as actions – outputs of a control system that act to bring about an intended state of a perception. We’re stuck with these two ways of talking about behavior, so the best we can do is to be clear as to our intended meaning in any given case: behavior as control, or behavior as the means of control.

Â

Bruce

[Martin Taylor 2014.10.30.10.57]

[Bruce Nevin (2015.10.29.23:15)]

Â

RM: Behavior – any behavior – is not just the visible Means we see a person using to get a Variable to the Reference state; “Opening the door” is not just the grasping/pulling forces exerted on the door.

Â

MMT: No, of course it isn’t. Those are lower-level behaviours used in controlling lower-level perceptions.

Â

Behavior has a visible public aspect and a private aspect that can only be inferred from the model. The private aspect includes the perception by the actor and the reference for that perception…

I don’t agree that this is a good use for the word “behaviour”. “Control” is a perfectly good word for what you describe. “Control”, like other words with technical meanings in PCT, also means something to a PCT-naive person. As with “perception”, the everyday meaning of “control” is closely related to the technical meaning. The way I prefer to use the word “behaviour”, it also is closely related to the everyday meaning.

Nothing in this discussion has persuaded me that it is a good idea (from the viewpoint of trying to propagate PCT) to (1) destroy the relation between the technical and the everyday meaning, and (2) use the word to mean something different for which a perfectly good technical word already is in ordinary usage. Even eliminating use of the word “behaviour” from PCT discourse (which I do not favour) would be preferable.

I see little chance of reconciliation, but little does not mean none. I think it’s a bad idea to have PCT-lingo split into mutually unintelligible dialects, and a worse idea to widen unnecessarily the inevitable split between PCT-lingo and its parent language – English as understood by those we hope will learn and develop PCT.

Â

[From Rick Marken (2015.10.29.1515)]

···

Bruce Nevin (20151029.1310)–

BN: I agree with Rick’s reply the next day, [Rick Marken (2015.10.27.1000)]:

RM: Behavior – any behavior – is not just the visible Means we see a person using to get a Variable to the Reference state; “Opening the door” is not just the grasping/pulling forces exerted on the door. Nor is behavior just the varying state of the controlled Variable; “Opening the door” is not just the changing angle of the door as it is opened. Nor is behavior just the Reference state of the controlled variable; “Opening the door” is not just the angle of the door when it has been open. “Behavior” is all three variables [as the controlled variable is] brought to [its] reference [state] by appropriate means – which turn out to be the the three observable components of control. That’s why I say that “behavior is control”. “Behavior” is all three variables brought to reference states by appropriate means – which turn out to be the the three observable components of control. That’s why I say that “behavior is control”.

BN: Rick, I was puzzled that you said that all three variables–means, CV, and reference state of CV–are brought to reference states by appropriate means.

RM: And well you should be (puzzled, that is). That’s probably the most important paragraph I wrote in this crazy discussion and I completely blew it. Your amendment helps but I think it would be better to re-write the whole thing. Here’s what I meant to say:

RM: Behavior – any behavior – is not just the visible Means we see a person using to get a Variable to the Reference state; “Opening the car door” is not just the grasping/pulling on the door. Nor is behavior just the varying state of the controlled Variable; “Opening the car door” is not just the changing angle of the door as it is opened. Nor is behavior just the Reference state of the controlled variable; “Opening the car door” is not just the angle of the door once it gets to the state “open” (80 degrees relative to the side of the car). “Behavior” is all three variables which are the three observable components of control; Opening the car door" is grasping/pulling on the door (Means) to change the angle of the door (controlled Variable) and bring it to the state “open” (the Reference State). That’s why I say that “behavior is control”. And named “behavior” can be seen as an example of control because it can be seen to involve these three variables – Means, controlled Variable and Reference State".

RM: This is what Table 1 on p. 172 of LCS I and the “Behavior is Control” Spreadsheet are all about. They show that various examples of “behavior”, like “Opening a car door” (in row 24 of the “Behavior is control” spreadsheet) can be seen to be examples of control; that they all involve Means used to bring a controlled Variable to a Reference state. I think this will become clearer if people think of everyday “behaviors” – eg. lifting a book, going to school, crossing the street, etc – and analyze them into these three visible components of control (Means, controlled Variable(s), and Reference State(s)) and enter them into a row of the “Behavior is Control” spreadsheet at:

https://docs.google.com/spreadsheets/d/1JmS6tOjt_nvrpmD5sGySwup0ZZCU_hYtZqlHxW80dME/edit#gid=0

RM: I hope that doing this will give you a better sense of the fact that what we call “behavior” in living systems is control.

BN: Of course, the means of controlling CV = “door open” (grasping, pulling) themselves involve subordinate CVs controlled to reference values by subordinate means, and there are superordinate purposes for which opening the door is the means. Maybe that’ was confusing the picture as you wrote that sentence.

RM: Perhaps. But you make an excellent point here. Control is clearly hierarchical, again in fact, not just in theory. That means that every entry in the “Means” column of the “Behavior is Control” spreadsheet could also be a new entry in the “Behavior” column, with a Means, CV and Reference State associated with it. So, as you say, “grasping/pulling”, the Means components of the behavior “Open the door”, are themselves “behaviors”; The controlled Variable in grasping, for example, is the relationship between the configuration of the hand and the door handle, the Means is muscle forces in the fingers and the Reference state is “handle grasped”.

RM: It might be a good exercise for someone to add “behaviors” to the “Behavior is Control” spreadsheet that are the Means of controlling the variables in involved in other behaviors in the spreadsheet.This could help in achieving Rupert’s goal of analyzing hierarchical relationships between the “behaviors” (and, thus, the CVs) in the spreadsheet

Best

Rick


Richard S. Marken

www.mindreadings.com
Author of Doing Research on Purpose.
Now available from Amazon or Barnes & Noble

On Thu, Oct 29, 2015 at 1:12 AM, Martin Taylor mmt-csg@mmtaylor.net wrote:

[Martin Taylor 2015.10.29.00.15]

  On 2015/10/28 11:34 PM, Bruce Nevin

wrote:

Nice post, Martin.

In the title Behavior: the control of perception ,
the colon is equivalent to is in the sentence
"behavior is the control of perception.

Not how I have ever read it. I have always read it as "Behaviour is

the way perception is controlled" or, more shortly “Behaviour is for
the control of perception”. That phrase, at least in CSGnet
discussion, was often qualified by the insertion of “intentional”,
which disqualifies nervous tics and side-effects as behaviour; but
it makes clear that Behaviour, in PCT, maps onto the conventional
usage of “behaviour” exactly by analogy to the way Perception in PCT
maps onto the everyday use of “perception”. I can’t claim to know
what Bill was thinking when he created the book title, but I can
claim that my interpretation is natural, and makes logical sense in
the context of his use of “perception” to refer to the controlled
variable.

      Consequently, in order to talk about "behavior" in this PCT

sense we have to talk about the entire loop, because it is the
entire loop that controls perception. The reference value
doesn’t control perception. The behavioral outputs don’t
control perception. Circular causation around the entire
closed loop controls the perceptual input in accord with the
reference value.

Exactly! Isn't that what I tried to get across in my messages to

Boris? The behavioural output influences the Complex Environmental
Variable, and thereby influences the perception. The whole control
loop ensures (if all the parameters are suitable) that the influence
of the behaviour on the perception is in the direction of reducing
the error.

In the passage that Rick keeps trying to persuade people to read

(LCS I, 171-76) Bill restates the old point that there is never one
cause for anything that happens, but he restates it in the context
of perceptual control. It’s a general truth that there is no one
cause for anything, and the fact that, with different reference
values and different disturbance values, the control system needs
different output values to achieve the same result is just one more
example of that general truth.

If I trip over a misplaced toy and fall, the cause is the toy, but I

wouldn’t have fallen if I had been in another room, so the cause of
my fall is my being in the same room as the toy. OK, that’s
nonsense, but maybe it makes the point. The cause of any observable
event is a confluence of many variables. When Bill talks about the
behaviour of opening the door, he lists a lot of things that could
vary, such as the force needed to cause the door to open under the
circumstances. That’s an error of kind. Opening a door is not a
force, and is irrelevant to the control loop for which the
controlled peception is the door angle. The applied force is a
behaviour relevant to a much lower-level controlled perception.
Opening the door is sui generis, the observable behaviour that
influences the (believed to exist) physical door-angle so as to
bring the door-angle perception to its reference value.

Bill lists "Behaviours" on p172 just as I would, despite Rick's

assertion to the contrary and despite Bill’s own text on p 173. His
“means” column lists behaviours used in control loops for a few of
the necessary lower-level controlled perceptions. As I wrote (in a
message a couple of days ago that I haven’t yet seen appear on
CSGnet), when I control for producing the image of a certain letter
on my screen, my behaviour is hitting the right key, not moving my
finger to the right place, bending it appropriately, lowering it,
and then removing it; when I control my perception of the party in
Government, my behaviour is voting, not walking to the polling
station, not picking up a pencil, not putting an X on a piece of
paper; when I control my perception of others understanding my
meanings, my behaviour is constructing an argument, not formulating
words, phrases and sentences, and not executing typing movements on
a keyboard.

Behaviour is part of any control loop. It is not the whole of the

control loop, for if it were, why have the two words “behaviour” and
“control”. But I have said all this before, for which I apologise if
you already saw it.

Martin

[Martin Taylor 2015.10.29.17.11]

Bruce, I find this whole discussion rather weird. Up to now, I have

always assumed that Bill wrote B:CP for people not already familiar
with PCT. If that was not the case, and he was preaching to the
choir ( which I seriously doubt), then I could accept the
interpretation you and Rick put on “behaviour”. But if he was trying
to grab the attention of a new audience, as I believe, then that
interpretation makes no sense. “Behaviour” in the title has to mean
what his intended audience would have understood it to mean, as
illustrated in your “sloppy” examples.
Here’s an expanded version of my “Behaviour is for the control of
perception”: “Behaviour is the observable outward manifestation of
what an organism does when controlling a perception”. That’s just as
much “in your face” as your conflation of behaviour with the entire
control loop, which gives a novel in-group-only reading to
“behaviour”.
Years ago, we had a discussion about “You can’t tell what someone is
doing by looking at what they are doing” and we used the example of
a man pushing a button set beside a door. I’m at the cottage and
have only my laptop, and on it I can’t find the multi-level diagram
that showed some of the possible behaviours the man might be doing,
or I would have included it at this point. Here are some
possibilities, anyway.
The behaviour might be:
Pressing a button
Testing whether a newly installed bell rings when the button is
pushed.
Visiting Aunt Maude.
Checking whether the house is empty.
Preparing to burgle the house.
Getting help after a traffic accident.

All of the above (though one has to imagine a rather strange nephew
and an odd coincidence for them all to be true simultaneously).
All of these are behaviours, possibly controlling perceptions of:
“the feel of the button pressure on the finger”
“the functioning of the bell”
“the pleasure Aunt Maude has when I visit”
“my uncertainty about whether the house is occupied”
“my uncertainty as to the vulnerability of the house to burglary”
“the presence of an ambulance at my car”
and so forth.
The behaviours, to me, are not the control loops. They are what
someone outside can see, and from which they might be able to infer
something about controlled perceptions, especially if they then
perform the Test for the Controlled Variable at the different
perceptual levels.
Why? Why not use the word exactly by analogy with “perception”, so
that it includes the everyday usage within the technical usage? Why
force it to mean something for which there is already a perfectly
good word, and thus require a new word to mean what the everyday
term means to everyone not on CSGnet, including people who might be
willing to learn PCT?
I have Rick’s reply, and my message is in my “Sent” box, but it
hasn’t appeared for me on CSGnet, as all my other postings have
done.
No, of course it isn’t. Those are lower-level behaviours used in
controlling lower-level perceptions.
No, nor is it the force applied to change that angle.
No, it’s part of a process, not the value of ANY variable or set of
variables. To equate “behaviour” with a variable is what is called
in engineering a “dimensional error”, of the same kind as equating
Force with Mass, on the grounds that when you apply force, the
amount of acceleration depends (in free space) only on the mass.
I would amend it slightly differently, for two reasons: the
dimensional error mentioned above, and that control is not always,
if ever, perfect. “Behaviour is what an organism can be observed to
do so as to bring a controlled variable to its reference state, as
manifest in the results on some observable property of the
observer’s environment.” If you observe someone turning a
doorhandle, pushing on the door, and when the door is open, passing
through the portal, you can reasonably infer that the person was
performing the behaviour of “going into another room” and to do so
performed the behaviour of “opening the door”, is support of which
he performed the behaviours of “turning the door handle” and
“pushing the door”. (Note all the "-ing"s).
Not to mention confusing anyone trying to learn about PCT.
Martin

···

(Extended quote to start, but
interspersed comments follow)

        [Bruce Nevin

(20151029.1310)]

              BN: Nice

post, Martin.

              BN: In the

title Behavior: the control of perceptio n, the
colon is equivalent to is in the sentence
"behavior is the control of perception.Â

MMT (2015.10.29.00.15 ): Not how I have ever read it. I
have always read it as “Behaviour is the way perception is
controlled” or, more shortly “Behaviour is for the control
of perception”.Â

        This identifies 'behavior'

with ‘behavioral outputs’ or ‘actions’. That is the usual
sense of the word. From
the Random House Unabridged Dictionary:

behavior

            —behavioral, adj.

—behaviorally, adv.

  /bi hayv"yeuhr/, n.

            Â  Â  Â  Â 1. manner of

behaving or acting.

            Â  Â  Â  Â 2. Psychol.,

Animal Behav.

            Â  Â  Â  Â  Â  Â  Â a.

observable activity in a human or animal.

            Â  Â  Â  Â  Â  Â  Â b. the

aggregate of responses to internal and external stimuli.

            Â  Â  Â  Â  Â  Â  Â c. a

stereotyped, species-specific activity, as a courtship
dance or startle reflex.

            Â  Â  Â  Â 3. Often,

behaviors. a behavior pattern.

            Â  Â  Â  Â 4. the action or

reaction of any material under given circumstances: the
behavior of tin under heat.

            Â  Â Also, esp. Brit.,

behaviour.

        Behavior (in the usual

sense) is " observable
activity ".
Observability extends probably inward as far as observable
muscle tensions and outward to proximal environmental
consequences of movements (e.g. bar presses), where it can
get rather sloppy with unacknowledged recognition of
intentions. (“He’s driving a nail. No, he’s adjusting the
position of that rafter. Oh … no , he’s killing ants. He just gets
really upset when he sees ants and the building’s not even
closed in yet.”)Â

        I hold to my 'in your face'

interpretation. In
support of my reading, there is abundant testimony that the
conventional use of the colon is to express equivalence,
e.g.:Â

        I understand that you would

prefer it to be otherwise in the title B:CP, but I hope that
you will at least acknowledge this widely accepted
convention and the legitimacy of reading it as “behavior is
the control of perception”.

        MMT:

That phrase, at least in CSGnet discussion, was often
qualified by the insertion of “intentional”, which
disqualifies nervous tics and side-effects as behaviour; but
it makes clear that Behaviour, in PCT, maps onto the
conventional usage of “behaviour” exactly by analogy to the
way Perception in PCT maps onto the everyday use of
“perception”. I can’t claim to know what Bill was thinking
when he created the book title, but I can claim that my
interpretation is natural, and makes logical sense in the
context of his use of “perception” to refer to the
controlled variable.

      I read that as an acknowledgement that the PCT sense of

behavior as control is at variance from the common usage (" observable activity in a human or
animal ") while communicating to the stream of
newcomers to PCT on CSG-net. I agree that ordinary usage is
natural, i.e. the conventional norm. I agree that the PCT use
of ‘perception’ (meaning a transform of the input quantity
derived from the observed CV) aligns with the ordinary usage
of that word; but the parallel term for the output quantity on
the other side of the loop is not ‘behavior’, it is
‘behavioral outputs’ or ‘actions’.

            MMT: a message a couple

of days ago that I haven’t yet seen appear on CSGnet …

          I see it in  in [Martin

Taylor 2015.10.26.16.23]. I agree with Rick’s reply the
next day, [Rick Marken (2015.10.27.1000)]:

            RM: Behavior -- any

behavior – is not just the visible Means we see a
person using to get a Variable to the Reference state;
“Opening the door” is not just the grasping/pulling
forces exerted on the door.

Nor is behavior just
the varying state of the controlled Variable; “Opening
the door” is not just the changing angle of the door as
it is opened.

            Nor is behavior just the

Reference state of the controlled variable; Â “Opening
the door” is not just the angle of the door when it has
been open.

            "Behavior" is all three

variables [as the controlled variable is] brought to
[its] reference [state] by appropriate means – Â which
turn out to be the the three observable components of
control. That’s why I say that “behavior is
control”. “Behavior” is all three variables brought to
reference states by appropriate means – Â which turn out
to be the the three observable components of control.
That’s why I say that “behavior is control”.

      Rick, I was puzzled that

you said that all three variables–means, CV, and reference
state of CV–are brought to reference states by appropriate
means. I offer this friendly amendment:

            "Behavior" is all three

variables [as the controlled variable is] brought to
[its] reference [state] by appropriate means – Â which
turn out to be the the three observable components of
control. That’s why I say that “behavior is control”.

        Of course, the means of controlling CV = "door open"

(grasping, pulling) themselves involve subordinate CVs
controlled to reference values by subordinate means, and
there are superordinate purposes for which opening the door
is the means. Maybe that’ was confusing the picture as you
wrote that sentence.

/Bruce

        On Thu, Oct 29, 2015 at 1:12 AM,

Martin Taylor mmt-csg@mmtaylor.net
wrote:

            [Martin Taylor

2015.10.29.00.15]

On 2015/10/28 11:34 PM, Bruce Nevin wrote:

Nice post, Martin.

In the title * Behavior: the control of
perception*, the colon is equivalent to is  in
the sentence "behavior is the control of
perception.

            Not how I have ever read it. I have always read it as

“Behaviour is the way perception is controlled” or, more
shortly “Behaviour is for the control of perception”.
That phrase, at least in CSGnet discussion, was often
qualified by the insertion of “intentional”, which
disqualifies nervous tics and side-effects as behaviour;
but it makes clear that Behaviour, in PCT, maps onto the
conventional usage of “behaviour” exactly by analogy to
the way Perception in PCT maps onto the everyday use of
“perception”. I can’t claim to know what Bill was
thinking when he created the book title, but I can claim
that my interpretation is natural, and makes logical
sense in the context of his use of “perception” to refer
to the controlled variable.

                  Consequently, in order to talk about "behavior"

in this PCT sense we have to talk about the entire
loop, because it is the entire loop that controls
perception. The reference value doesn’t control
perception. The behavioral outputs don’t control
perception. Circular causation around the entire
closed loop controls the perceptual input in
accord with the reference value.

            Exactly! Isn't that what I tried to get across in my

messages to Boris? The behavioural output influences the
Complex Environmental Variable, and thereby influences
the perception. The whole control loop ensures (if all
the parameters are suitable) that the influence of the
behaviour on the perception is in the direction of
reducing the error.

            In the passage that Rick keeps trying to persuade people

to read (LCS I, 171-76) Bill restates the old point that
there is never one cause for anything that happens, but
he restates it in the context of perceptual control.
It’s a general truth that there is no one cause for
anything, and the fact that, with different reference
values and different disturbance values, the control
system needs different output values to achieve the same
result is just one more example of that general truth.

            If I trip over a misplaced toy and fall, the cause is

the toy, but I wouldn’t have fallen if I had been in
another room, so the cause of my fall is my being in the
same room as the toy. OK, that’s nonsense, but maybe it
makes the point. The cause of any observable event is a
confluence of many variables. When Bill talks about the
behaviour of opening the door, he lists a lot of things
that could vary, such as the force needed to cause the
door to open under the circumstances. That’s an error of
kind. Opening a door is not a force, and is irrelevant
to the control loop for which the controlled peception
is the door angle. The applied force is a behaviour
relevant to a much lower-level controlled perception.
Opening the door is sui generis, the observable
behaviour that influences the (believed to exist)
physical door-angle so as to bring the door-angle
perception to its reference value.

            Bill lists "Behaviours" on p172 just as I would, despite

Rick’s assertion to the contrary and despite Bill’s own
text on p 173. His “means” column lists behaviours used
in control loops for a few of the necessary lower-level
controlled perceptions. As I wrote (in a message a
couple of days ago that I haven’t yet seen appear on
CSGnet), when I control for producing the image of a
certain letter on my screen, my behaviour is hitting the
right key, not moving my finger to the right place,
bending it appropriately, lowering it, and then removing
it; when I control my perception of the party in
Government, my behaviour is voting, not walking to the
polling station, not picking up a pencil, not putting an
X on a piece of paper; when I control my perception of
others understanding my meanings, my behaviour is
constructing an argument, not formulating words, phrases
and sentences, and not executing typing movements on a
keyboard.

            Behaviour is part of any control loop. It is not the

whole of the control loop, for if it were, why have the
two words “behaviour” and “control”. But I have said all
this before, for which I apologise if you already saw
it.

                Martin

https://guinlist.wordpress.com/tag/equivalence/

[Martin Taylor 2014.10.30.10.57]

[Bruce Nevin (2015.10.29.23:15)]

        RM: Behavior -- any behavior

– is not just the visible Means we see a person using to
get a Variable to the Reference state; “Opening the door” is
not just the grasping/pulling forces exerted on the door.

        MMT: No, of course it isn't. Those are lower-level

behaviours used in controlling lower-level perceptions.

      Behavior has a visible public aspect and a private aspect

that can only be inferred from the model. The private aspect
includes the perception by the actor and the reference for
that perception…

I don't agree that this is a good use for the word "behaviour".

“Control” is a perfectly good word for what you describe. “Control”,
like other words with technical meanings in PCT, also means
something to a PCT-naive person. As with “perception”, the everyday
meaning of “control” is closely related to the technical meaning.
The way I prefer to use the word “behaviour”, it also is closely
related to the everyday meaning.

Nothing in this discussion has persuaded me that it is a good idea

(from the viewpoint of trying to propagate PCT) to (1) destroy the
relation between the technical and the everyday meaning, and (2) use
the word to mean something different for which a perfectly good
technical word already is in ordinary usage. Even eliminating use of
the word “behaviour” from PCT discourse (which I do not favour)
would be preferable.

I see little chance of reconciliation, but little does not mean

none. I think it’s a bad idea to have PCT-lingo split into mutually
unintelligible dialects, and a worse idea to widen unnecessarily the
inevitable split between PCT-lingo and its parent language –
English as understood by those we hope will learn and develop PCT.

Martin

[Martin Taylor 2015.10.30.11.21]

···

PS. In one of the messages Bruce
referred to “You can’t tell what someone is doing by observing
what they are doing” as evidence that “behaviour” should include
the whole of control. I would read the mantra differently, without
using the word “behaviour” or any PCT-technical words: “You can’t
tell what someone wants to achieve by observing their actions.” In
that, to make it technical, I would make these substitutions:
“what someone want to achieve” -> “what perception someone is
controlling with what reference value”, and “actions” ->
“behaviour”. In everyday speech, those two elements are quite
distinct, and indeed the ordinary observer might say something
like “If he wants to get what he wants, that kind of behaviour
won’t work”. Translated the way I would like to use the term:
“That behaviour is ineffective for control of this perception”, as
in “Riding a bike in a snowstorm at night isn’t a good way to stay
safe.”

  Martin

[Martin Taylor 2014.10.30.10.57]

[Bruce Nevin (2015.10.29.23:15)]

          RM: Behavior -- any behavior

– is not just the visible Means we see a person using to
get a Variable to the Reference state; “Opening the door”
is not just the grasping/pulling forces exerted on the
door.

          MMT: No, of course it isn't. Those are lower-level

behaviours used in controlling lower-level perceptions.

        Behavior has a visible public aspect and a private aspect

that can only be inferred from the model. The private aspect
includes the perception by the actor and the reference for
that perception…

  I don't agree that this is a good use for the word "behaviour".

“Control” is a perfectly good word for what you describe.
“Control”, like other words with technical meanings in PCT, also
means something to a PCT-naive person. As with “perception”, the
everyday meaning of “control” is closely related to the technical
meaning. The way I prefer to use the word “behaviour”, it also is
closely related to the everyday meaning.

  Nothing in this discussion has persuaded me that it is a good idea

(from the viewpoint of trying to propagate PCT) to (1) destroy the
relation between the technical and the everyday meaning, and (2)
use the word to mean something different for which a perfectly
good technical word already is in ordinary usage. Even eliminating
use of the word “behaviour” from PCT discourse (which I do not
favour) would be preferable.

  I see little chance of reconciliation, but little does not mean

none. I think it’s a bad idea to have PCT-lingo split into
mutually unintelligible dialects, and a worse idea to widen
unnecessarily the inevitable split between PCT-lingo and its
parent language – English as understood by those we hope will
learn and develop PCT.

  Martin

[From Bruce Abbott (2015.10.30.1930 EDT)]

Martin Taylor 2015.10.30.11.21

MT: PS. In one of the messages Bruce [Nevin] referred to “You can’t tell what someone is doing by observing what they are doing” as evidence that “behaviour” should include the whole of control. I would read the mantra differently, without using the word “behaviour” or any PCT-technical words: “You can’t tell what someone wants to achieve by observing their actions.” In that, to make it technical, I would make these substitutions: “what someone want to achieve” -> “what perception someone is controlling with what reference value”, and “actions” -> “behaviour”. In everyday speech, those two elements are quite distinct, and indeed the ordinary observer might say something like “If he wants to get what he wants, that kind of behaviour won’t work”. Translated the way I would like to use the term: “That behaviour is ineffective for control of this perception”, as in “Riding a bike in a snowstorm at night isn’t a good way to stay safe.”
BA: As the risk of getting stuck to this “tar baby,” my own reading of the title to Bill’s book matches yours, Martin: “Behavior: the Control of Perception” I understand to mean that the function of behavior is to control perception. Read in this way, the title constitutes a broadside against the prevailing open-loop view of behavior as mere output: the product of environmental contingencies acting upon the organism.

BA: Back in the 1950’s when Bill was developing his ideas, Behaviorism was in its heyday, and for behaviorists, a major task of psychology was to identify functional relationships between the environmental conditions to which an organism is exposed and its behavior. The appearance of the term “behavior” in Bill’s title was sure catch the attention of behaviorists, especially with its bold assertion that behavior, far from being the mere product of a set of environmental contingencies acting on the organism through its physiological mechanisms, exists for the purpose of controlling certain of the organism’s perceptual inputs.

BA: Unfortunately the term “behavior” in ordinary usage can mean different things, leading to such statements as “you can’t tell what a person is doing by observing what they are doing,” in which the first “doing” refers to what changes the person is trying to bring about, and the second “doing” refers to what activities the observer sees the person engaged in. Bill liked to distinguish these two meanings by substituting the term “actions” for “behavior” when referring to the behavioral output of a control system. Such actions (or more accurately the perceptual consequences of them) can themselves be under control at a lower level, although they need not be. Thus the action of flipping a light (what the person is observed to “do”) may be the means by which the person controls the intensity of the lighting (what the person is “doing” by manipulating the switch).

BA: In ordinary speech we often talk about behavior in terms of its perceptual effects, e.g., “I’m turning up the heat” means that I’m doing something that I believe will increase the magnitude of my perception of the temperature. This usage tends to mask the role of the behavior as actions – outputs of a control system that act to bring about an intended state of a perception. We’re stuck with these two ways of talking about behavior, so the best we can do is to be clear as to our intended meaning in any given case: behavior as control, or behavior as the means of control.

Bruce

···

[Martin Taylor 2014.10.30.10.57]

[Bruce Nevin (2015.10.29.23:15)]

RM: Behavior – any behavior – is not just the visible Means we see a person using to get a Variable to the Reference state; “Opening the door” is not just the grasping/pulling forces exerted on the door.

MMT: No, of course it isn’t. Those are lower-level behaviours used in controlling lower-level perceptions.

Behavior has a visible public aspect and a private aspect that can only be inferred from the model. The private aspect includes the perception by the actor and the reference for that perception…

I don’t agree that this is a good use for the word “behaviour”. “Control” is a perfectly good word for what you describe. “Control”, like other words with technical meanings in PCT, also means something to a PCT-naive person. As with “perception”, the everyday meaning of “control” is closely related to the technical meaning. The way I prefer to use the word “behaviour”, it also is closely related to the everyday meaning.

Nothing in this discussion has persuaded me that it is a good idea (from the viewpoint of trying to propagate PCT) to (1) destroy the relation between the technical and the everyday meaning, and (2) use the word to mean something different for which a perfectly good technical word already is in ordinary usage. Even eliminating use of the word “behaviour” from PCT discourse (which I do not favour) would be preferable.

I see little chance of reconciliation, but little does not mean none. I think it’s a bad idea to have PCT-lingo split into mutually unintelligible dialects, and a worse idea to widen unnecessarily the inevitable split between PCT-lingo and its parent language – English as understood by those we hope will learn and develop PCT.

Hi Bruce,

I’m sorry to »jump in« here, but it seems odd that you make a final conclussion based on so »ambigous« oppinions about »Behavior is control«. I hope you’ll wait on »final deccision« on the bases of many arguments. But what it is, it is. I’d like to remand you that you also »changed« your mind about »Behavior is control«, as Bill and Rick did some times.

BN :

The key insight is that we do not control our behavior. Rather, behavior is variable in just the manner and extent necessary to make our experience be the way we want it to be. The title of the locus classicus of this science of psychology is Behavior: The control of perception, published in 1973 by William T. Powers

HB : It seems that something is not yet clear about PCT, because there is so much »double« meaning.

BN :

That reason is that the nature of behavior has been misunderstood until the advent of PCT. This is why Bill said in 2011, and I quote him again, “behavior is control of perception”. Or as he wrote in the 1973 Preface to B:CP, “Behavior is the process by which organisms control their input sensory data” (that’s in the last paragraph, on p. xiv of the 2005 edition). Not just the observable activities; the entire process.

HB : Of course I can give you much more Bill’s statements in which he affirm that »Behavior« or »Output« or »actions« are not Control.

Bill P : It means that we produce actions that alter the world of perception, and that we do so specifically to make the state of that world conform to the reference conditions we ourselves have choosen (to the extent we change the perceptions of our actions).

Bill P : Using the internal point of view, we can understand many aspects of behavior by seeing control as control of perception rather than of an objective world. We can make sense not only of other people’s behavior, but of our own, using the same concept of perceptual control.

Bill P : It means that we produce actions that alter the world of perception,

Bill P : If you change perception you change the world arround as it appears to be.

Bill P : Half of the jokes in the world are about one person assuming that everyone else sees the world the same way.

Bill P : Human beings do not plan actions and then carry them out; they do not respond to stimuli according to the way they have been reinforced. They control. They never produce any beahvior except for the purposes of making what they are experiencing become more like what they intend or want to experience, and then keeping it that way even in a changing world. If they plan they plan perceptions, not actions.

Bill P :

The output function … represents the means thiss system has for causing changes in it’s environment.

HB : And also other authors…

RM: Actually the other player (like everyone else) is controlling their own perceptions, not their actions.

RM : The basic tenet of the model is that organisms control perceptual input, not motor output. This is the fact of motor system operation. Control systems acts to keep their perceptions matching referene of what those perceptions should be. They do this by acting on the environment, producing effects which, when combined with prevailing environemntal disturbances, produce the desired perceptions….

RM :

To understand the behavior of a living control system, the observer must learn what perceptions the system is controlling : what reference images the system is trying to match. Living control systems produce many results, some of which may be contolled and others not. The observer must learn which results correspond to the percetpual variables that the system is actually controling.

Kent M : People’s actions are merely a by-product of their attempts to stabilize their perceptions in conformity with their own desires and preferences.

Kent M :

Experientially, then, controlling perceptions means controlling the world around us, and in practice controlling a perception can be equated with controlling some aspect of our physical or social environment.

Kent M :

In any event, we can only control our perceptions of our own actions, not the actions directly, because control depends upon perception.

Kent M :

Even when directing attention most closely to our actions and doing our best to act in a “controlled” manner … we end up controlling perceptions of our actions, not the actioons themselves, because our perceptions are our only means for controlling anything.

MT :

My “behaviour” is not control, but is the means by which the controller acts on the environment to influence the perception

BN :

The key insight is that we do not control our behavior. Rather, behavior is variable in just the manner and extent necessary to make our experience be the way we want it to be. The title of the locus classicus of this science of psychology is Behavior: The control of perception, published in 1973 by William T. Powers.

HB : So I think that labeling whether »Behavior« is control or not, can’t be done on the bases of frequency of statements. Everyone could have said also something contrary. But mostly, maybe I could estimate in 80 % of texts in PCT in not about »Behavior is Control«. So from this point, this hypothesis is unacceptable.

BN :It is central to the success of PCT to communicate that behavior is not what we have been led to believe. “Behavior is the process by which organisms control their input sensory data.” “Behavior is control of perception.”

HB :  If most members will decide that »Behavior is control«, then I have no reason to stay on CSGnet. I could never prove with physiological means that »Behavior is Control« so I have nothing to do hear any more. But I stronlgy advice that you get YIn’s oppinion (if he is physiologist) or any oppinion of any physiologist to confirm  that »Behavior can be control«.

I think that Bill was explicit about how much we can control our behavior despite his »changing of mind« :

Bill P :

Our only view of the real world is our view of the neural signals that represent it inside our own brains. When we act to make a perception change to our more desireble state … we have no direct knowledge of what wee are doing to the reality that is the origin of our neural signal; we know only the final result, how the result looks, feels, smells, sounds, tastes, and so forth…

Bill P.

If the driver had to execute any of the behaviors in Table 1 blindly, with no visual, auditory, kinestethic, or other sensory information to tell him the current status of the variable, it would be impossible for him to vary his actions so as to oppose unexpected disturbances. In Fact, we would find through continuing experiment that the only reliable consequences of the driver’s actions are those the driver can sense. This is a crucial hint about how this sort of phenomena is created.

HB : I think that in both cases Bill emphasized that »Behavior can’t be Control«. Both are clear cases at least to me, that »Behavior« is not possible without perception. It’s just »blind« action, not knowing what it’s doing to it’s environment. With perception it becomes action, which is just changing perception.

My other arguments why it’s not good to keep the definiton »Behavior is Control«  are :

  1.  First I think that introducing »Behavior is Control« into PCT, PCT becomes cheap »self-regulation« theory, where »Behavior is Control« is mantra. Please read Mary Powers about PCT and »self-regulation«
    

.

  1.  I see also  a problem in interpreting everyday activites. Do we eat, drink, dirve, walk, etc… with »Controllling our hands, legs…« ?
    
  2.  I think that »Behavior« can't be "Control", because we don't know what is "behavior". At the best shot we know what is "Perception of behavior (ouput)". All is perception. In this case "Perception of behavior can be controlled", as Martin mentioned many times. It depends from observer in whatever view we take perception (internal or external view). It's subjective. That's how individuals make their own interpretation of behavior, what depends on their references. If you and Rick observe behavior you will say that it is "control". If I observe it I would say that "Behavior" just affect environment, makes changes to my perception. People who do not know anything about Control Theory, would probably make interpretations of their own perceptions in whatever they beleive what is happening.
    
  3.  In any of Bill's "definitions" I exposed in my previous post, I can't see that "behavior" could be any kind of control or that "behavior" is involving control. So if you put in definition that "Behavior is Control" or at least that "involves control", you'll have to change at least some "definitions".  But would Powers ladies agree with this ? Wouldn't it be better that we left "definitions" as they are, but you and Rick make your own theory based on "Behavior is Control".
    
  4.  Considering your and Rick's statements also Bill's generic diagram has to be changed, because you'll want to put in the generic diagram "CV" in the external environment which is "controlled" by behavior.  But then diagram would not represent "general theory" anymore but special theory of something called "Behavior is Control". And even Rick based on my suggestion about »turning our heads« admitted that "actions" has double meaning.
    

RM :

This makes it clear that your actions (outputs) don’t necessarily have to have a direct effect on the environment… For example, when I move my head…

HB : So putting it all together my oppinion is that “BEHAVIOR IS NOT CONTROL” and it would be dangerous to introduce it into PCT. The most important problem I see, is that it can’t be proved with physiological means. So again I appeal that you provide physiological evidences for »Behavior is Control«. You mentioned Yin, or you can get anyone else or your can provide your own evidences. I think this is the most important evidence that is coming from the highest arbiter : nature.

Bill P : »If the effects of the model are just as hypothetical as the model, we don’t have a model, because we can’t check it against direct experience. The ultimate authority is always direct experience, the real reality we are incapable of doubting…« (LCS II, p.185)

Best,

Boris

···

From: Bruce Nevin [mailto:bnhpct@gmail.com]
Sent: Friday, October 30, 2015 11:12 PM
To: CSG
Subject: Re: Behaviour and “Behaviour” (was Re: – a lot of things)

[Bruce Nevin (2015.10.30.18:04 ET)]

Martin, my friend, we may secretly be in violent agreement. That’s often the case in terminological quarrels. Your main concern is to communicate clearly without undue barriers. I share that concern. Now let us reconsider just what it is that must be communicated. It doesn’t matter how clearly understood our words are if our readers understand from them the wrong thing.

Let’s back up. I believe the origin was in discussion of Rick’s spreadsheet. This spreadsheet is organized according to the table on p. 172 of LCS I and the discussion on pp. 171-176. I note that on p. 171 Bill called it “a partial listing of activities”. The first column is labeled “Behavior” and contains “phrases of a kind used both in ordinary discourse and in scientific psychology to denote what an organism is doing.” For lay people and for most scientists “No analysis takes place.”

A partial analysis is laid out in the other three columns. Bill says (p. 173, 4th paragraph) “We can see immediately that the real actions of the driver … are all in the Means column. What is so casually called behavior results from the conjunction of many forces, only one of which is contributed by the driver. The Behavior column really lists consequences of the driver’s actions, consequences that are not determined by the driver’s actions, but are only influenced by them.” As an example, “Grasp, pull” gives more detail to “open [the] door”. Of course, yet more detail could be given, e.g. for “grasp”: extend fingers, reduce distance between fingers and handle, flex fingers around handle, etc. This is a partial analysis.

If the real actions are in the Means column, the intended consequences of the actions are in the last two columns, the controlled variable (“angle of door”) and the reference state for that variable (“80 degrees”). Because these consequences of actions are not just any consequences.

I apologize for belaboring this. This is the stuff of a PCT 101 class, or even a Pre-PCT cass. (Wouldn’t that be an excellent thing for our children to attend!) I know you have no quarrel with anything that Bill wrote here. You just feel that we should use the word “behavior” to mean “observable activity”, because it is the usage of the word that is most familiar to everyone. Why erect barriers to understanding, you ask. I laud your desire to be easily understood. I recently quoted a bit of your writing as an example of good communication over difficulties a lot of people have with the word “control”.

But just as you helped your readers understand that control is not what they might think, we also must help people understand that behavior is not what they might think. Behavior is not just observable activity. Behavior is always purposeful. (Setting aside the metaphoric usage in e.g. “the behavior of sodium chlorite in the presence of an acid” or “the behavior of trees in a high wind”.) Behavior has consequences and (unless it is thwarted) those consequences are intended.

You say we have a perfectly good word for this, namely, control. Well and good, but we also have perfectly good words for that to which you want to limit “behavior”, namely, observable actions, behavioral outputs, etc. There are good reasons why Bill and others have always employed such words to be clear, rather than referring to them as “behavior”, which the naive reader would naturally understand in the old way.

That reason is that the nature of behavior has been misunderstood until the advent of PCT. This is why Bill said in 2011, and I quote him again, “behavior is control of perception”. Or as he wrote in the 1973 Preface to B:CP, “Behavior is the process by which organisms control their input sensory data” (that’s in the last paragraph, on p. xiv of the 2005 edition). Not just the observable activities; the entire process.

It is central to the success of PCT to communicate that behavior is not what we have been led to believe. “Behavior is the process by which organisms control their input sensory data.” “Behavior is control of perception.”

/Bruce

On Fri, Oct 30, 2015 at 11:12 AM, Martin Taylor mmt-csg@mmtaylor.net wrote:

[Martin Taylor 2014.10.30.10.57]

[Bruce Nevin (2015.10.29.23:15)]

RM: Behavior – any behavior – is not just the visible Means we see a person using to get a Variable to the Reference state; “Opening the door” is not just the grasping/pulling forces exerted on the door.

MMT: No, of course it isn’t. Those are lower-level behaviours used in controlling lower-level perceptions.

Behavior has a visible public aspect and a private aspect that can only be inferred from the model. The private aspect includes the perception by the actor and the reference for that perception…

I don’t agree that this is a good use for the word “behaviour”. “Control” is a perfectly good word for what you describe. “Control”, like other words with technical meanings in PCT, also means something to a PCT-naive person. As with “perception”, the everyday meaning of “control” is closely related to the technical meaning. The way I prefer to use the word “behaviour”, it also is closely related to the everyday meaning.

Nothing in this discussion has persuaded me that it is a good idea (from the viewpoint of trying to propagate PCT) to (1) destroy the relation between the technical and the everyday meaning, and (2) use the word to mean something different for which a perfectly good technical word already is in ordinary usage. Even eliminating use of the word “behaviour” from PCT discourse (which I do not favour) would be preferable.

I see little chance of reconciliation, but little does not mean none. I think it’s a bad idea to have PCT-lingo split into mutually unintelligible dialects, and a worse idea to widen unnecessarily the inevitable split between PCT-lingo and its parent language – English as understood by those we hope will learn and develop PCT.

Martin

[Martin Taylor 2015.11/-1/08/09]

This statement and many of your selected quotes make me think you

have misunderstood Bruce. Or I have. So far as I understand Bruce,
he and I have the same model of perceptual control in mind, but
differ on how to use words to describe it. The main reason for my
disagreement with him is in the clarity of expression when talking
with people only vaguely familiar with PCT or hearing about it for
the first time.
You write to Bruce as though he used “Behaviour”, “Output” and
“Actions” interchangeably. Neither he nor I would do that, but for
different reasons. Bruce made his distinction explicit when he said:
"
."
As far as I understand Bruce’s reasoning (and I fear that I don’t
understand it very well), he interprets the passage of LCS 1 pp171ff
differently from the way I interpret it, and chooses to include the
entire control loop as “the process by which perceptions are
controlled”. That’s a choice, and if all his readers understood the
word the same way, nobody would have a problem. Problems arise when
the writer and reader infer different things from way a word is
used. What I don’t understand is the reason for his choice. What is
clear is that he does distinguish between “behaviour” and “actions”.
One is control, the other isn’t.
Here’s another look at my preferred usage of the words. It is hard
to make clear the distinction when we are dealing with only one
control loop at one level. As I would use the terms, for a single
isolated control loop, “behaviour”, “behavioural output”, “output”,
and “actions” all mean the same thing, diagrammed as the signal that
is passed along path between the output function and the CEV. Let’s
start there and consider the words, because at least we can separate
the Bruce-Rick usage of “behaviour” and the other three at this
stage. Let’s stick to the “Opening a door” example, and forget about
all the muscular actions involved.
From the Observer’s viewpoint, there is a door, initially closed,
and an actor who appears to influence the position of the door until
it comes to a state that the Observer would call “open”. At this
point the actor stops influencing the position of the door. If, by
some chance, the wind or another person changes the state of the
door to what the Observer calls “closed”, the actor again starts to
influence the position of the door until the Observer sees it as
“open”. The Observer sees a process called “opening the door”, and
calls that process the “actor’s behaviour” or the “actor’s action”.
The observer may make other inferences, such as that the actor
wanted the door to be open, or equivalently if the Observer knows
PCT, that the actor was controlling a perception of the state of the
door with a reference value of “open”. The Observer would have
inferred from the fact that the actor stopped influencing the door
position that the observed actions or behaviour had served this
control process effectively.
Now consider the Analyst’s viewpoint, which differs from the
Observer’s viewpoint because, God-like, the Analyst accurately
perceives everything that happens in the control loop, whereas the
Observer can only infer it according to his previous understanding
or model of the way the world works. The Analyst might start by
noting that the actor perceived the door to be closed and had a
reference to perceive it to be open. As a consequence, the error
signal was non-zero, which resulted in the output function producing
a change in the output that created forces on the door, eventually
resulting in the error decreasing to zero. In the Bruce-Rick
terminology, all of this is “behaviour”, but only the part between
the output function and the door is “behavioural output”, “output”,
or “action”.
Now let’s consider “Grasping the handle”, which I would like to call
another behaviour. Now the Analyst sees two control loops, one
controlling a perception of the position of the door, another
controlling a perception of the relationship of the hand to the
handle. At this point, my terminology separates “behaviour” from
“actions”. The Observer sees exactly the same actions, whether the
Analyst sees one control loop or two. However, the Observer can
choose to separate the actions into separate behaviours – “Opening
the door” and “grasping the handle”. All the actions are part of
“Opening the door”, but only some are part of “Grasping the handle”.
The two behaviours, however, are entirely separate, one serving the
other, just as a salesperson in a shop serves a customer, both being
part of a transaction, but each playing separate and different parts
of that transaction. Similarly, the behaviour “Pulling the door”
serves the “Opening the Door” behaviour without being part of either
it or of the “Grasping the handle” behaviour, which it does not
serve.
The Analyst sees my “serve” in the foregoing as “controlling its
perception to a reference value set eventually by”, but the Observer
sees only that “Grasping the handle” and “Pulling the door” both
stop when the door is open. The analyst sees that the reference
values for both have changed so that their errors have gone to zero
and no output change is needed. The observer sees a cessation of the
two behaviours, and also of the behaviour they both serve, the door
now being open.
Now we can segregate the words a little better, at least as I like
to use them. The “output” of what the Analyst sees as an Elementary
Control Unit (ECU) is a simple variable, perhaps a neural firing
rate. The Observer never sees this. The “actions” used in the
behaviour “Opening the Door” include all the actions used in the
behaviours “Grasping the handle” and “Pulling the door”, but the
actions used in those two behaviours are separate and independent,
although one cannot pull the door without grasping the handle. The
handle-grasping actions are completed before the door-pulling
actions begin.
I’m not sure how I would characterize the other term Bruce uses:
“behavioural output”. My terminology doesn’t seem to need it, but I
suppose it could refer to the way the output of the “Opening the
Door” control loop is distributed to the other two behaviours. The
words seem to suggest a kind of variation different from the simple
temporal variation of the “output”, but I don’t find any clear need
for the term. Bruce equates it with “actions”, so perhaps it could
be used of actions that form part of a behaviour exclusive of
actions that form part of a supporting behaviour. But are there any?
Perhaps the transition between “Grasping” and “Pulling while
grasping” might be an action that is part of neither, and could be
something to which “behavioural output” might be applied.
I’ve concentrated on my usage of the terms “output”, “actions”, and
“behaviour” because I think Bruce and Rick are better placed to
describe better why they choose to use those words the way they do.
And I’ve concentrated on the Observer’s viewpoint, because that is
the viewpoint a PCT-naive reader has to take. Only someone who has a
reasonable grasp of PCT can take a PCT-Analyst’s view. Since the
people unable to take a PCT-Analyst’s view are the people we try to
reach, the Observer’s view seems the most suitable to use in
defining the terminology,
Martin

···

On 2015/11/1 1:40 AM, Boris Hartman
wrote:

        Hi

Bruce,

        I'm

sorry to »jump in« here, but it seems odd that you make a
final conclussion based on so »ambigous« oppinions about
»Behavior is control«. I hope you’ll wait on »final
deccision« on the bases of many arguments. But what it is,
it is. I’d like to remand you that you also »changed« your
mind about »Behavior is control«, as Bill and Rick did some
times.

      BN :
      The key insight is that we do not control

our behavior. Rather, behavior is variable in just the manner
and extent necessary to make our experience be the way we want
it to be. The title of the locus classicus of this
science of psychology is * Behavior: The control of
perception*, published in 1973 by William T. Powers

        HB

: It seems that something is not yet clear about PCT,
because there is so much »double« meaning.

BN :

      That reason is that the nature of behavior

has been misunderstood until the advent of PCT. This is why
Bill said in 2011, and I quote him again, “behavior is control
of perception”. Or as he wrote in the 1973 Preface to B:CP,
“Behavior is the process by which organisms control their
input sensory data” (that’s in the last paragraph, on p. xiv
of the 2005 edition). Not just the observable activities; the
entire process.

        HB

: Of course I can give you much more Bill’s statements in
which he affirm that »Behavior« or »Output« or »actions« are
not Control.

  •  I agree that the PCT use of 'perception' (meaning a transform of
    

the input quantity derived from the observed CV) aligns with the
ordinary usage of that word; but the parallel term for the output
quantity on the other side of the loop is not ‘behavior’, it is
‘behavioral outputs’ or ‘actions’*

Boris Hartman (2015/11/1 1:40 AM)

I’m sorry to »jump in« here, but it seems odd that you make a final conclussion based on so »ambigous« oppinions about »Behavior is control«. I hope you’ll wait on »final deccision« on the bases of many arguments. But what it is, it is. I’d like to remand you that you also »changed« your mind about »Behavior is control«, as Bill and Rick did some times.

We have an ambiguous term, “behaviour”. The ambiguity is imperceptible to people who do not understand control. If they have studied behavior, they may be troubled by “Brunswik’s lens” (B:CP p. 6) or the equivalent, but behavior is behavior, darn it. The ambiguity is introduced as soon as we introduce someone to PCT. Yes, we can start out talking about “behaviour” as observable actions, but immediately we have to say “there’s more to it than that”.

What we witness as behavior only makes sense when we understand that what the person is really doing, what those actions are really doing, is controlling perceptions. And that is why, eventually, we have to bring them around to the realization that behavior is control of perceptions.

Bill came down quite equivocally on both sides of this, as evidenced by the various quotations we have assembled.

···

On Sun, Nov 1, 2015 at 12:07 PM, Martin Taylor mmt-csg@mmtaylor.net wrote:

[Martin Taylor 2015.11/-1/08/09]

  On 2015/11/1 1:40 AM, Boris Hartman

wrote:

        Hi

Bruce,

        I'm

sorry to »jump in« here, but it seems odd that you make a
final conclussion based on so »ambigous« oppinions about
»Behavior is control«. I hope you’ll wait on »final
deccision« on the bases of many arguments. But what it is,
it is. I’d like to remand you that you also »changed« your
mind about »Behavior is control«, as Bill and Rick did some
times.

      BN :
      The key insight is that we do not control

our behavior. Rather, behavior is variable in just the manner
and extent necessary to make our experience be the way we want
it to be. The title of the locus classicus of this
science of psychology is * Behavior: The control of
perception*, published in 1973 by William T. Powers

        HB

: It seems that something is not yet clear about PCT,
because there is so much »double« meaning.

BN :

      That reason is that the nature of behavior

has been misunderstood until the advent of PCT. This is why
Bill said in 2011, and I quote him again, “behavior is control
of perception”. Or as he wrote in the 1973 Preface to B:CP,
“Behavior is the process by which organisms control their
input sensory data” (that’s in the last paragraph, on p. xiv
of the 2005 edition). Not just the observable activities; the
entire process.

        HB

: Of course I can give you much more Bill’s statements in
which he affirm that »Behavior« or »Output« or »actions« are
not Control.

This statement and many of your selected quotes make me think you

have misunderstood Bruce. Or I have. So far as I understand Bruce,
he and I have the same model of perceptual control in mind, but
differ on how to use words to describe it. The main reason for my
disagreement with him is in the clarity of expression when talking
with people only vaguely familiar with PCT or hearing about it for
the first time.

You write to Bruce as though he used "Behaviour", "Output" and

“Actions” interchangeably. Neither he nor I would do that, but for
different reasons. Bruce made his distinction explicit when he said:
"

  •  I agree that the PCT use of 'perception' (meaning a transform of
    

the input quantity derived from the observed CV) aligns with the
ordinary usage of that word; but the parallel term for the output
quantity on the other side of the loop is not ‘behavior’, it is
‘behavioral outputs’ or ‘actions’*."

As far as I understand Bruce's reasoning (and I fear that I don't

understand it very well), he interprets the passage of LCS 1 pp171ff
differently from the way I interpret it, and chooses to include the
entire control loop as “the process by which perceptions are
controlled”. That’s a choice, and if all his readers understood the
word the same way, nobody would have a problem. Problems arise when
the writer and reader infer different things from way a word is
used. What I don’t understand is the reason for his choice. What is
clear is that he does distinguish between “behaviour” and “actions”.
One is control, the other isn’t.

Here's another look at my preferred usage of the words. It is hard

to make clear the distinction when we are dealing with only one
control loop at one level. As I would use the terms, for a single
isolated control loop, “behaviour”, “behavioural output”, “output”,
and “actions” all mean the same thing, diagrammed as the signal that
is passed along path between the output function and the CEV. Let’s
start there and consider the words, because at least we can separate
the Bruce-Rick usage of “behaviour” and the other three at this
stage. Let’s stick to the “Opening a door” example, and forget about
all the muscular actions involved.

From the Observer's viewpoint, there is a door, initially closed,

and an actor who appears to influence the position of the door until
it comes to a state that the Observer would call “open”. At this
point the actor stops influencing the position of the door. If, by
some chance, the wind or another person changes the state of the
door to what the Observer calls “closed”, the actor again starts to
influence the position of the door until the Observer sees it as
“open”. The Observer sees a process called “opening the door”, and
calls that process the “actor’s behaviour” or the “actor’s action”.
The observer may make other inferences, such as that the actor
wanted the door to be open, or equivalently if the Observer knows
PCT, that the actor was controlling a perception of the state of the
door with a reference value of “open”. The Observer would have
inferred from the fact that the actor stopped influencing the door
position that the observed actions or behaviour had served this
control process effectively.

Now consider the Analyst's viewpoint, which differs from the

Observer’s viewpoint because, God-like, the Analyst accurately
perceives everything that happens in the control loop, whereas the
Observer can only infer it according to his previous understanding
or model of the way the world works. The Analyst might start by
noting that the actor perceived the door to be closed and had a
reference to perceive it to be open. As a consequence, the error
signal was non-zero, which resulted in the output function producing
a change in the output that created forces on the door, eventually
resulting in the error decreasing to zero. In the Bruce-Rick
terminology, all of this is “behaviour”, but only the part between
the output function and the door is “behavioural output”, “output”,
or “action”.

Now let's consider "Grasping the handle", which I would like to call

another behaviour. Now the Analyst sees two control loops, one
controlling a perception of the position of the door, another
controlling a perception of the relationship of the hand to the
handle. At this point, my terminology separates “behaviour” from
“actions”. The Observer sees exactly the same actions, whether the
Analyst sees one control loop or two. However, the Observer can
choose to separate the actions into separate behaviours – “Opening
the door” and “grasping the handle”. All the actions are part of
“Opening the door”, but only some are part of “Grasping the handle”.
The two behaviours, however, are entirely separate, one serving the
other, just as a salesperson in a shop serves a customer, both being
part of a transaction, but each playing separate and different parts
of that transaction. Similarly, the behaviour “Pulling the door”
serves the “Opening the Door” behaviour without being part of either
it or of the “Grasping the handle” behaviour, which it does not
serve.

The Analyst sees my "serve" in the foregoing as "controlling its

perception to a reference value set eventually by", but the Observer
sees only that “Grasping the handle” and “Pulling the door” both
stop when the door is open. The analyst sees that the reference
values for both have changed so that their errors have gone to zero
and no output change is needed. The observer sees a cessation of the
two behaviours, and also of the behaviour they both serve, the door
now being open.

Now we can segregate the words a little better, at least as I like

to use them. The “output” of what the Analyst sees as an Elementary
Control Unit (ECU) is a simple variable, perhaps a neural firing
rate. The Observer never sees this. The “actions” used in the
behaviour “Opening the Door” include all the actions used in the
behaviours “Grasping the handle” and “Pulling the door”, but the
actions used in those two behaviours are separate and independent,
although one cannot pull the door without grasping the handle. The
handle-grasping actions are completed before the door-pulling
actions begin.

I'm not sure how I would characterize the other term Bruce uses:

“behavioural output”. My terminology doesn’t seem to need it, but I
suppose it could refer to the way the output of the “Opening the
Door” control loop is distributed to the other two behaviours. The
words seem to suggest a kind of variation different from the simple
temporal variation of the “output”, but I don’t find any clear need
for the term. Bruce equates it with “actions”, so perhaps it could
be used of actions that form part of a behaviour exclusive of
actions that form part of a supporting behaviour. But are there any?
Perhaps the transition between “Grasping” and “Pulling while
grasping” might be an action that is part of neither, and could be
something to which “behavioural output” might be applied.

I've concentrated on my usage of the terms "output", "actions", and

“behaviour” because I think Bruce and Rick are better placed to
describe better why they choose to use those words the way they do.
And I’ve concentrated on the Observer’s viewpoint, because that is
the viewpoint a PCT-naive reader has to take. Only someone who has a
reasonable grasp of PCT can take a PCT-Analyst’s view. Since the
people unable to take a PCT-Analyst’s view are the people we try to
reach, the Observer’s view seems the most suitable to use in
defining the terminology,

Martin

[Bruce Nevin (2015.11.01.2150 ET)]

Sorry that went out accidentally without clear indications who says what.

Boris Hartman (2015/11/1 1:40 AM)

I’m sorry to »jump in« here, but it seems odd that you make a final conclussion based on so »ambigous« oppinions about »Behavior is control«. I hope you’ll wait on »final deccision« on the bases of many arguments. But what it is, it is. I’d like to remand you that you also »changed« your mind about »Behavior is control«, as Bill and Rick did some times.

BN: Boris you’re looking for an unequivocal answer when the answer really is equivocal.

BN: We have an ambiguous term, “behaviour”. The ambiguity is imperceptible to people who do not understand control. If they have studied behavior, they may be troubled by “Brunswik’s lens” (B:CP p. 6) or the equivalent, but behavior is behavior, darn it. The ambiguity is introduced as soon as we introduce them to PCT. Yes, we can start out talking about “behaviour” as observable actions, but immediately we have to say “But there’s more to it than that”.

BN: What we witness as behavior only makes sense when we understand that what the person is really doing, what those actions are really doing, is controlling perceptions. And that is why, eventually, we have to bring them around to the realization that behavior is control of perceptions.

BN: Bill came down quite equivocally on both sides of this, as evidenced by the various quotations we have assembled.

···

On Sun, Nov 1, 2015 at 12:07 PM, Martin Taylor mmt-csg@mmtaylor.net wrote:

[Martin Taylor 2015.11/-1/08/09]

  On 2015/11/1 1:40 AM, Boris Hartman

wrote:

        Hi

Bruce,

        I'm

sorry to »jump in« here, but it seems odd that you make a
final conclussion based on so »ambigous« oppinions about
»Behavior is control«. I hope you’ll wait on »final
deccision« on the bases of many arguments. But what it is,
it is. I’d like to remand you that you also »changed« your
mind about »Behavior is control«, as Bill and Rick did some
times.

      BN :
      The key insight is that we do not control

our behavior. Rather, behavior is variable in just the manner
and extent necessary to make our experience be the way we want
it to be. The title of the locus classicus of this
science of psychology is * Behavior: The control of
perception*, published in 1973 by William T. Powers

        HB

: It seems that something is not yet clear about PCT,
because there is so much »double« meaning.

BN :

      That reason is that the nature of behavior

has been misunderstood until the advent of PCT. This is why
Bill said in 2011, and I quote him again, “behavior is control
of perception”. Or as he wrote in the 1973 Preface to B:CP,
“Behavior is the process by which organisms control their
input sensory data” (that’s in the last paragraph, on p. xiv
of the 2005 edition). Not just the observable activities; the
entire process.

        HB

: Of course I can give you much more Bill’s statements in
which he affirm that »Behavior« or »Output« or »actions« are
not Control.

This statement and many of your selected quotes make me think you

have misunderstood Bruce. Or I have. So far as I understand Bruce,
he and I have the same model of perceptual control in mind, but
differ on how to use words to describe it. The main reason for my
disagreement with him is in the clarity of expression when talking
with people only vaguely familiar with PCT or hearing about it for
the first time.

You write to Bruce as though he used "Behaviour", "Output" and

“Actions” interchangeably. Neither he nor I would do that, but for
different reasons. Bruce made his distinction explicit when he said:
"

  •  I agree that the PCT use of 'perception' (meaning a transform of
    

the input quantity derived from the observed CV) aligns with the
ordinary usage of that word; but the parallel term for the output
quantity on the other side of the loop is not ‘behavior’, it is
‘behavioral outputs’ or ‘actions’*."

As far as I understand Bruce's reasoning (and I fear that I don't

understand it very well), he interprets the passage of LCS 1 pp171ff
differently from the way I interpret it, and chooses to include the
entire control loop as “the process by which perceptions are
controlled”. That’s a choice, and if all his readers understood the
word the same way, nobody would have a problem. Problems arise when
the writer and reader infer different things from way a word is
used. What I don’t understand is the reason for his choice. What is
clear is that he does distinguish between “behaviour” and “actions”.
One is control, the other isn’t.

Here's another look at my preferred usage of the words. It is hard

to make clear the distinction when we are dealing with only one
control loop at one level. As I would use the terms, for a single
isolated control loop, “behaviour”, “behavioural output”, “output”,
and “actions” all mean the same thing, diagrammed as the signal that
is passed along path between the output function and the CEV. Let’s
start there and consider the words, because at least we can separate
the Bruce-Rick usage of “behaviour” and the other three at this
stage. Let’s stick to the “Opening a door” example, and forget about
all the muscular actions involved.

From the Observer's viewpoint, there is a door, initially closed,

and an actor who appears to influence the position of the door until
it comes to a state that the Observer would call “open”. At this
point the actor stops influencing the position of the door. If, by
some chance, the wind or another person changes the state of the
door to what the Observer calls “closed”, the actor again starts to
influence the position of the door until the Observer sees it as
“open”. The Observer sees a process called “opening the door”, and
calls that process the “actor’s behaviour” or the “actor’s action”.
The observer may make other inferences, such as that the actor
wanted the door to be open, or equivalently if the Observer knows
PCT, that the actor was controlling a perception of the state of the
door with a reference value of “open”. The Observer would have
inferred from the fact that the actor stopped influencing the door
position that the observed actions or behaviour had served this
control process effectively.

Now consider the Analyst's viewpoint, which differs from the

Observer’s viewpoint because, God-like, the Analyst accurately
perceives everything that happens in the control loop, whereas the
Observer can only infer it according to his previous understanding
or model of the way the world works. The Analyst might start by
noting that the actor perceived the door to be closed and had a
reference to perceive it to be open. As a consequence, the error
signal was non-zero, which resulted in the output function producing
a change in the output that created forces on the door, eventually
resulting in the error decreasing to zero. In the Bruce-Rick
terminology, all of this is “behaviour”, but only the part between
the output function and the door is “behavioural output”, “output”,
or “action”.

Now let's consider "Grasping the handle", which I would like to call

another behaviour. Now the Analyst sees two control loops, one
controlling a perception of the position of the door, another
controlling a perception of the relationship of the hand to the
handle. At this point, my terminology separates “behaviour” from
“actions”. The Observer sees exactly the same actions, whether the
Analyst sees one control loop or two. However, the Observer can
choose to separate the actions into separate behaviours – “Opening
the door” and “grasping the handle”. All the actions are part of
“Opening the door”, but only some are part of “Grasping the handle”.
The two behaviours, however, are entirely separate, one serving the
other, just as a salesperson in a shop serves a customer, both being
part of a transaction, but each playing separate and different parts
of that transaction. Similarly, the behaviour “Pulling the door”
serves the “Opening the Door” behaviour without being part of either
it or of the “Grasping the handle” behaviour, which it does not
serve.

The Analyst sees my "serve" in the foregoing as "controlling its

perception to a reference value set eventually by", but the Observer
sees only that “Grasping the handle” and “Pulling the door” both
stop when the door is open. The analyst sees that the reference
values for both have changed so that their errors have gone to zero
and no output change is needed. The observer sees a cessation of the
two behaviours, and also of the behaviour they both serve, the door
now being open.

Now we can segregate the words a little better, at least as I like

to use them. The “output” of what the Analyst sees as an Elementary
Control Unit (ECU) is a simple variable, perhaps a neural firing
rate. The Observer never sees this. The “actions” used in the
behaviour “Opening the Door” include all the actions used in the
behaviours “Grasping the handle” and “Pulling the door”, but the
actions used in those two behaviours are separate and independent,
although one cannot pull the door without grasping the handle. The
handle-grasping actions are completed before the door-pulling
actions begin.

I'm not sure how I would characterize the other term Bruce uses:

“behavioural output”. My terminology doesn’t seem to need it, but I
suppose it could refer to the way the output of the “Opening the
Door” control loop is distributed to the other two behaviours. The
words seem to suggest a kind of variation different from the simple
temporal variation of the “output”, but I don’t find any clear need
for the term. Bruce equates it with “actions”, so perhaps it could
be used of actions that form part of a behaviour exclusive of
actions that form part of a supporting behaviour. But are there any?
Perhaps the transition between “Grasping” and “Pulling while
grasping” might be an action that is part of neither, and could be
something to which “behavioural output” might be applied.

I've concentrated on my usage of the terms "output", "actions", and

“behaviour” because I think Bruce and Rick are better placed to
describe better why they choose to use those words the way they do.
And I’ve concentrated on the Observer’s viewpoint, because that is
the viewpoint a PCT-naive reader has to take. Only someone who has a
reasonable grasp of PCT can take a PCT-Analyst’s view. Since the
people unable to take a PCT-Analyst’s view are the people we try to
reach, the Observer’s view seems the most suitable to use in
defining the terminology,

Martin