Brown and Herrnstein

CSGnet Archive CSG2000
1

[From Bruce Abbott (2000.10.03.0830 EST)]

Sketch showing, from right to left and in order of decreasing size, a man
walking, a mother duck, a toy duck, and a duckling. The caption reads: "A
duckling imprinted on a person follows at a greater distance than one
imprinted on a mother duck, but it follows a small toy duck more closely.
Thus it keeps the visual angle of the imprinted stimulus approximately
constant." From Brown and Herrnstein's (1975) _Psychology_ (an introductory
psychology textbook), p. 58.

This reminds me of Bill P.'s "Crowd" demo where one "leader" circle moves
across the screen to a goal-point while being followed by a string of
"follower" circles. Bill will have to remind us what variables the follower
circles were controlling in order to keep a specific distance from each
other and the leader, but it would appear that with imprinted ducklings the
visual angle or something closely related (such as the size of the image of
the object being followed as projected onto the duckling's retinas.

At the time of its publication, Brown and Herrnstein's book provided a
radical departure from the typical introductory psych text. Consider the
title of the first chapter: Motivation: Forced Movement, Instinct, and
Imprinting. In the typical intro text, motivation does not receive explicit
treatment until well into the middle chapters, and then probably will
include only a brief mention of "physiological drives" in conjunction with
the concept of homeostasis. Brown and Herrnstein's first chapter outlines
some very basic behavioral systems that, while described without using the
term, represent a number of ways in which simple control systems are
implemented in the simpler forms of life. We find the humidity-seeking
kinesis of the sow bug, the flatworm's light-avoiding kinesis, the somewhat
more sophisticated light-avoiding taxis of the maggot, the meal-worm's
ability to steer a path directly away from light by balancing the brightness
of light between its two eye-spots, the eating mechanism of the blowfly.
Each mechanism is briefly described and here and there Brown and Herrnstein
note that such systems are purposive.

The book was too radical for its time, and would still be so today. It
failed to survive to second edition.

Bruce A.

2

[From Bill Powers (2000.10.03.1321 MDT)]

Bruce Abbott (2000.10.03.0830 EST)--

Interesting, that stuff about Herrnstein. Doesn't fit my prejudices at all.

Best,

Bill P.

3

[From Rick Marken (2000.10.03.1420)]

Bruce Abbott (2000.10.03.0830 EST) --

Sketch showing, from right to left and in order of decreasing
size, a man walking, a mother duck, a toy duck, and a duckling.
The caption reads: "A duckling imprinted on a person follows
at a greater distance than one imprinted on a mother duck, but
it follows a small toy duck more closely Thus it keeps the
visual angle of the imprinted stimulus approximately constant."
From Brown and Herrnstein's (1975) _Psychology_ (an introductory
psychology textbook), p. 58.

This is a great example of seeing behavior through control
theory glasses. Good find, Bruce. It suggests some interesting
tests. For example, if visual angle really is a controlled
variable then the ducklings' distance from the "imprinted
stimulus" should be directly proportional to the magnification
of lenses placed over the ducklings' eyes.

Possible controlled variables seem to be more noticeable in
some situations than in others. For example, a possible
controlled variable seems to have been noticed in studies
of "personal space": the distance between oneself and others.
But possible controlled variables have not been noticed in most
psychological research. I wonder why it is easier to notice
(or speculate about) controlled variables in some situations
(like the duck imprinting study described above) but not others.

By the way, Bruce. Thanks for not joining in the recent Rick-
bashing. You and I have been arguing about PCT for quite some
time and I know that you disagree with me about some things
very deeply. But I feel that our disagreements have always
been based on substance, not personal enmity. You have
typically argued against _what_ I said rather than against
_how_ I said it. I have never felt personally attacked by
you, even when you turned out to be right;-) You are a mensch.

Best

Rick

···

---
Richard S. Marken Phone or Fax: 310 474-0313
MindReadings.com mailto: marken@mindreadings.com
www.mindreadings.com

4

[From Bill Powers (2000.110.04.0144 MDT)]

Rick Marken (2000.10.03.1420)--

Bruce Abbott (2000.10.03.0830 EST) --

Piggy-backing on Rick's reply. Bruce A asked how the Crowd program
"Oneline" setup works. The lead "person" seeks a destination. The first
duckling maintains a fixed proximity to the lead person, at an angle of
zero (straight ahead). The second duckling does the same thing for the
first duckling, the third ducking for the second duckling, and so on.

Note that Lorenz (and everybody since) missed an obvious fact: only the
first duckling follows the mother duck. If all the ducklings independently
followed the mother duck, you'd have a result like the setup called "guru",
in which each follower independently tries to maintain proximity to the
lead person. The followers cluster around the lead person when they catch
up. Instead, you get a line of ducks, each following the one ahead of it.

The "proximity" that is controlled is proportional to the inverse square of
the distance to the object followed. Proximity is proportional to the
_area_ or _solid angle_ subtended by an object (all objects are the same
size in the program). If each object had its own weighting that multiplied
the proximity, the result would be the same as if the objects had different
sizes and solid angle were being sensed. The effects you noted would then
appear.

Best,

Bill P.

5

[From Bruce Abbott (2000.10.04.1130 EST)]

Bill Powers (2000.10.03.1321 MDT)]

Interesting, that stuff about Herrnstein. Doesn't fit my prejudices at all.

Nor mine. I happened to acquire the Brown and Herrnstein book two days ago
and have not had time to give it a thorough going over, but it is clear
enough already that Herrnstein was moving toward a conception of the
behaving organism that was quite different from Skinner's -- indeed I had
heard at one time that there was some friction between the two men, both of
whom were by then at Harvard University and working in the same department.

In his book, Herrnstein proposes a theory of motivation based on negative
feedback, but his the words he uses make it difficult to know just how much
he actually understood about control systems. Here's a slightly edited
description Herrnstein provides on pp. 109-110:

"Some endogenous or exogenous source knocks an internal regulator out of
kilter. Exogenous sources tend to be stimuli--like electric shock--that
threaten the physical integrity of the body although they may not. . . . But
most drives arise from a mixture of both external and internal sources, in
varying proportions across species and among the individuals in a species.

A sufficient shift in the regulator of a drive activates the appetitive
responses, which are both learned and inborn. . . . A shift in the regulator
also causes certain stimuli to become rewards. If the regulator happens to
correspond to thirst, then water becomes rewarding, and so on. . . . a
source is defined by its effect on a regulator; a regulator is defined by
its effect on appetitive responses and rewards.
. . .
By definition, the consumatory event brings the regulator back to normal,
terminating the motivational sequence."

I've omitted quite a bit in the above, but I think that it conveys the
general tenor of Herrnstein's conception. Perhaps implicit but never
explicitly stated are the concepts of controlled variable and reference,
although it is clear in other passages in the book that Herrnstein is aware
of these. It seems to have been written in a way that would make it seem
compatible with the concepts of stimulus and response.

Something that seems to be missing from PCT but which I find worth
considering is the notion that certain errors (Herrnstein equates them with
drives), may produce actions indirectly, by making certain sensory inputs
"rewarding," or as I would put it, pleasurable (within some range of
intensity). One then controls for experiencing some level of that input.

Bill Powers (2000.110.04.0144 MDT)]

Note that Lorenz (and everybody since) missed an obvious fact: only the
first duckling follows the mother duck. If all the ducklings independently
followed the mother duck, you'd have a result like the setup called "guru",
in which each follower independently tries to maintain proximity to the
lead person. The followers cluster around the lead person when they catch
up. Instead, you get a line of ducks, each following the one ahead of it.

I rather suspect that the ducklings are doing something a bit more
sophisticated than just following the duck in front of them. Some years ago
I happened to walk out into my front yard just as a small flock of goslings
came running across the street into a corner of the yard. They had become
separated from their mother (who was nowhere in sight) and were running
rapidly. They definitely were not following each other in the usual string,
but were moving more like a school of fish -- one gosling would begin to run
in a given direction and all would then head off in the same direction in a
disorganzed gaggle, staying close to one another. This "leaderless"
behavior is of course very different from the following behavior they
exhibit when there is a mother present to head up the line, but I can only
speculate on what all the controlled variables might be.

Bruce A.

6

[From Bill Powers (2000.10.04.1138 MDT)]

Bruce Abbott (2000.10.04.1130 EST)--

Something that seems to be missing from PCT but which I find worth
considering is the notion that certain errors (Herrnstein equates them with
drives), may produce actions indirectly, by making certain sensory inputs
"rewarding," or as I would put it, pleasurable (within some range of
intensity). One then controls for experiencing some level of that input

All that's necessary is to raise the reference level from zero by adjusting
the reference signal for some specific controlled variable. The resulting
error, if there is no existing system for correcting it under current
circumstances, will lead to reorganization. Otherwise, the error will
simply be corrected. In either case, the appearance is that the particular
variable has suddenly become "rewarding," because the organism now will
behave in such a way as to produce a nonzero amount of the associated input
(apparent "reinforcement"). In PCT, of course, there is no such phenomenon
as "reward" or "reinforcement" per se; we explain the same phenomenon in
terms of different mechanisms.

Note that Lorenz (and everybody since) missed an obvious fact: only the
first duckling follows the mother duck. If all the ducklings independently
followed the mother duck, you'd have a result like the setup called "guru",
in which each follower independently tries to maintain proximity to the
lead person. The followers cluster around the lead person when they catch
up. Instead, you get a line of ducks, each following the one ahead of it.

I rather suspect that the ducklings are doing something a bit more
sophisticated than just following the duck in front of them. Some years ago
I happened to walk out into my front yard just as a small flock of goslings
came running across the street into a corner of the yard. They had become
separated from their mother (who was nowhere in sight) and were running
rapidly. They definitely were not following each other in the usual string,
but were moving more like a school of fish -- one gosling would begin to run
in a given direction and all would then head off in the same direction in a
disorganzed gaggle, staying close to one another. This "leaderless"
behavior is of course very different from the following behavior they
exhibit when there is a mother present to head up the line, but I can only
speculate on what all the controlled variables might be.

Why speculate? Set up a model to produce the same behavior. Of course the
situation you're describing, where the mother goose was not in sight, is
different from the following-in-line scenario. Some of the goslings did,
according to your description, follow another gosling rather than the
mother. If they all followed the lead gosling, or some other gosling
following the lead gosling, and if (in the absence of the mother) they set
their proximity reference levels very high, they would indeed "stay close
to one another." You'd have to do some experimenting to see what they were
controlling. Or perhaps the high proximity reference level does not change,
but the subtended angle of the target, being smaller for a gosling than for
a mother goose, requires the distance to be much smaller to achieve the
same sensed proximity (in the context of the previous post on this subject).

Or maybe it works like this: the lead duckling gets all the goose-proximity
it wants from the mother goose which looms directly ahead. The next
gosling, avoiding a collision with the first one, is farther back from the
mother goose, and thus senses less proximity from her, but the nearby
smaller gosling just in front supplies the rest of the required
goose-proximity. And so on down the line. The hypothesis here would be that
there is no particular ordering in the line (so that one gosling does not
always follow one specific other gosling). But once the mother goose starts
moving and the goslings start to follow, one gosling will be the front one
and the others will sort themselves out according to who is closest to the
next-in-line position when motion starts. On different occasions, under
this model, we would see different orderings of individual goslings after
the mother goose started moving.

If the ordering is always the same while in motion, the "oneline" setup is
sufficient. If not, the "total goose proximity" may do the trick.

Best,

Bill P.

7

[From Bruce Abbott (2000.10.04.1710 EST)]

Bill Powers (2000.10.04.1138 MDT) --

Bruce Abbott (2000.10.04.1130 EST)

Something that seems to be missing from PCT but which I find worth
considering is the notion that certain errors (Herrnstein equates them with
drives), may produce actions indirectly, by making certain sensory inputs
"rewarding," or as I would put it, pleasurable (within some range of
intensity). One then controls for experiencing some level of that input

All that's necessary is to raise the reference level from zero by adjusting
the reference signal for some specific controlled variable. The resulting
error, if there is no existing system for correcting it under current
circumstances, will lead to reorganization. Otherwise, the error will
simply be corrected. In either case, the appearance is that the particular
variable has suddenly become "rewarding," because the organism now will
behave in such a way as to produce a nonzero amount of the associated input
(apparent "reinforcement"). In PCT, of course, there is no such phenomenon
as "reward" or "reinforcement" per se; we explain the same phenomenon in
terms of different mechanisms.

Yes, I know. But then, one has to ask why we have evolved the capacity to
experience pleasure (or displeasure). Clearly we could just have control
systems linked up so as to generate the appropriate behaviors without
involving these subjective feelings, yet we do often have them. The thirsty
rat may be drinking, not merely because there is a deficit in its water
balance, but because the rat, when thirsty, finds the experience of drinking
water to be pleasurable. As the water balance is restored, the pleasure of
drinking diminishes.

This conception is more complex than the simple control system that merely
raises the reference for drinking, but it may have a powerful advantage --
the single variable of pleasure can serve to motivate a large variety of
otherwise different activities, potentially involving many different
controlled variables. The rat will drink water because its taste evokes
pleasurable feelings (even in the absence of a bodily deficit), because it
has consumed salt and thus dried out its mouth and throat (drinking water
evokes pleasure because it reduces the unpleasant feelings of a dry mouth),
and it will seek out and consume food because consuming food evokes
pleasurable feelings, whether because the food is especially tasty (in the
absence of any error in nutritional state and even with a partially full
stomach) or because the control systems monitoring nutrient stores (or
related variables) are registering a deficit. Certain sensory experiences
will evoke, through connections in the brain, feelings of pleasure or
displeasure; usually these will have been associated through evolution or
learning with physical states that in the history of the species or organism
have fostered continued survival. By controlling for pleasure (or
elimination of displeasure), the rat also controls for being properly
nourished, hydrated, groomed, and so on.

You have suggested, I believe, that pleasure arises as an error in some CV
is reduced, and displeasure/discomfort when the error is increased.
Pleasure/displeasure are then mere byproducts of the operation of a control
system and have no real biological function. The conception I have proposed
above (somewhat badly, I am afraid), would give these affective qualities a
central function.

I like your suggestions on the gosling problem.

By the way, I did some checking on Roger Brown, Herrnstein's coauthor on the
textbook, as I was curious as to how much influence he may have had on the
writing of those chapters that touch on control theory. Brown's research
interests centered around language and language development, and according
to the book's preface, Herrnstein was the primary author on those chapters
I've been discussing. So it looks like those views I've been mentioning are
indeed Herrnstein's. I would like to have had the opportunity to have
talked with Herrnstein about it, or to ask Brown for his opinion of the
ideas Herrnstein has expressed in their book, but alas, both Herrnstein and
Brown have passed away, before their time.

Bruce A.

8

[Bill Curry (2000.10.04.1100 EDT)]

Bruce Abbott (2000.10.04.1130 EST)--

Something that seems to be missing from PCT but which I find worth
considering is the notion that certain errors (Herrnstein equates them with
drives), may produce actions indirectly, by making certain sensory inputs
"rewarding," or as I would put it, pleasurable (within some range of
intensity). One then controls for experiencing some level of that input.

If error reduction, as in satiating the intrinsic variable of hunger, is
somatically experienced as pleasurable sensations, why is there a need
hypothecate a missing pleasure mechanism? Given this emotion substrate, an
organism could control for such pleasurable feelings by creating perceptions
of consuming food. Indeed, isn't this the root of most hyperphagia--reproducing
the pleasurable feelings of eating to temporarily reduce other errors such as
threats to self-concepts, etc.? It all seems to be control to me.

Regards,

Bill

···

--
William J. Curry
Capticom, Inc.
capticom@landmarknet.net

9

[From Bruce Gregory (2000.1005.0513)]

[Bill Curry (2000.10.04.1100 EDT)]

If error reduction, as in satiating the intrinsic variable of hunger, is
somatically experienced as pleasurable sensations, why is there a need
hypothecate a missing pleasure mechanism? Given this emotion substrate, an
organism could control for such pleasurable feelings by creating perceptions
of consuming food. Indeed, isn't this the root of most
hyperphagia--reproducing
the pleasurable feelings of eating to temporarily reduce other errors such as
threats to self-concepts, etc.? It all seems to be control to me.

You seem to be proposing a new neural pathway from the comparator. Where
does this pathway go and what role does it play?

BG

10

Bruce Gregory (2000.1005.0513)]

>[Bill Curry (2000.10.04.1100 EDT)]
>
>If error reduction, as in satiating the intrinsic variable of hunger, is
>somatically experienced as pleasurable sensations, why is there a need
>hypothecate a missing pleasure mechanism? Given this emotion substrate, an
>organism could control for such pleasurable feelings by creating perceptions
>of consuming food. Indeed, isn't this the root of most
>hyperphagia--reproducing
>the pleasurable feelings of eating to temporarily reduce other errors such as
>threats to self-concepts, etc.? It all seems to be control to me.

You seem to be proposing a new neural pathway from the comparator. Where
does this pathway go and what role does it play?

Bruce, please help me understand your meaning. I see a self-concept error (e.g.,
"I am lonely") being reduced by setting lower level references for pleasure
attainment, in this example by controlling for perceptions of eating.

Regards,

Bill

···

--
William J. Curry
Capticom, Inc.
capticom@landmarknet.net

11

[From Bill Powers (2000.10.05.0553 MDT)]

Bruce Abbott (2000.10.04.1710 EST)--

Yes, I know. But then, one has to ask why we have evolved the capacity to
experience pleasure (or displeasure). Clearly we could just have control
systems linked up so as to generate the appropriate behaviors without
involving these subjective feelings, yet we do often have them. The thirsty
rat may be drinking, not merely because there is a deficit in its water
balance, but because the rat, when thirsty, finds the experience of drinking
water to be pleasurable. As the water balance is restored, the pleasure of
drinking diminishes.

If the reason the rat drinks is that it brings pleasure to do so, why does
it need a second mechanism for detecting and correcting a water deficit?
And if it has a mechanism for detecting and correcting a water deficit, of
what use is a second mechanism that gives pleasure for doing so?

This conception is more complex than the simple control system that merely
raises the reference for drinking, but it may have a powerful advantage --
the single variable of pleasure can serve to motivate a large variety of
otherwise different activities, potentially involving many different
controlled variables. The rat will drink water because its taste evokes
pleasurable feelings (even in the absence of a bodily deficit), because it
has consumed salt and thus dried out its mouth and throat (drinking water
evokes pleasure because it reduces the unpleasant feelings of a dry mouth),
and it will seek out and consume food because consuming food evokes
pleasurable feelings, whether because the food is especially tasty (in the
absence of any error in nutritional state and even with a partially full
stomach) or because the control systems monitoring nutrient stores (or
related variables) are registering a deficit. Certain sensory experiences
will evoke, through connections in the brain, feelings of pleasure or
displeasure; usually these will have been associated through evolution or
learning with physical states that in the history of the species or organism
have fostered continued survival. By controlling for pleasure (or
elimination of displeasure), the rat also controls for being properly
nourished, hydrated, groomed, and so on.

Yes, but _which_? If the very same sensation results from doing all these
different things, how is it that the rat or person knows which of them
needs to be done to get pleasure, since any of them gives pleasure? If the
rat is dehydrated, why doesn't it have sex? If it lacks salt, why doesn't
it groom itself? A single all-purpose pleasure sensation is incapable of
directing the organism to what it needs, particularly when it needs many
things at the same time.

And if a certain act is done to get pleasure, why does it ever stop? Why
doesn't the rat go on eating until it bursts? As soon as you posit this
explanation, you must invent another one to explain why it stops working:
the rat stops eating when eating more would cause it more pain than
pleasure. But again, why is it that pain would cause it to stop _eating_
rather than, say, drinking? It's the very same sensation that would arise
from being too hot or too deficient in salt or too dehydrated -- if you
refuse to consider the concept of error signals in specific different
control systems, you can't explain the fortuitous specificity of pleasure
and pain.

To make these explanations work you must invent still more complex
mechanisms. Now the rat or person must be able to distinguish which action
goes with pleasure in the current circumstances, so as not to perform the
wrong action. It must have some sort of spatio-temporal correlation
capabilities, so it can tell that the feeling of pleasure that exists now
arose because of the appropriate immediately-preceding action and not from
an action somewhat earlier or of a different kind. "Just what did I do that
I am being reinforced for (or punished for) this time?" In an organism that
must simultaneously maintain dozens of variables close to specific levels
in order to survive and propagate its species, trying to answer this
question would tax a brain much bigger than a rat's. "If I am feeling pain,
which of these pleasurable activities needs to be stopped? And if I am
feeling pleasure, which aspect of my activities should I continue
producing, or produce more of?"

Again, you must invoke more complexity to handle these problems. Now you
must invent something called "saliency," meaning that among all the
activities, one of them somehow stands out from the others and indicates
itself to the rat or person as the one that needs to be increased or
stopped. The longer you go on with this, the more tangled the logical
threads get and the more entities you have to invent to explain why the
basic assumption goes wrong in one or another special case. Thus has
psychology grown, inventing and reifying concepts one after another simply
to handle failures of the basic ideas.

The control model does away with all these problems. There are reference
signals which by their settings (high or low) define whether a given
perception is to be sought or avoided. There are error signals which
indicate whether action needs to be taken and which, at the lower levels,
are converted into the required actions. Imagining and even modeling many
such systems working at the same time is no problem at all; each control
process occurs independently of and simultaneously with many other control
processes at the same level. There is no danger of one kind of error being
confused with another.

You have suggested, I believe, that pleasure arises as an error in some CV
is reduced, and displeasure/discomfort when the error is increased.
Pleasure/displeasure are then mere byproducts of the operation of a control
system and have no real biological function. The conception I have proposed
above (somewhat badly, I am afraid), would give these affective qualities a
central function.

No, that's not what I suggest. You're still treating pleasure and
displeasure/discomfort as if they were separate phenomena, signals that
arise somewhere else when error appears or disappears, as if these signals
existed separately from the signals in control systems. What I am proposing
is that pleasure and pain are simply words we use for the states of control
systems as we experience them. When we experience the results of control
systems having large error signals, we call that experience "bad" or
"painful" or "unpleasant," and we try not to experience it any more, or
again. When we experience a marked diminution in or disappearance of the
effects of errors, the name we give to that constellation of perceptions is
"pleasurable" or "good". We seek or try to prolong that kind of experience.
"Pleasure" and "pain" are _classifications_ in which we find many different
kinds of experiences, and what is common to each category is the state of
error and change of error in many different control systems, as well as
what we do when such errors exist, increase, decrease, or vanish. Pleasure
and pain do not exist as single specific sensation-signals in the brain. Or
to put that differently, pleasure or pain associated with one experience is
not the same signal as pleasure or pain associated with a different
experience. You can have one system in a "pleasant" state at the same time
another system is in an "unpleasant" state.

Best,

Bill P.

12

[From Bruce Gregory (2000.1005.0945)]

Bill Powers (2000.10.05.0553 MDT)

The control model does away with all these problems.

Perhaps it is _too_ successful.

What I am proposing
is that pleasure and pain are simply words we use for the states of control
systems as we experience them.

Here the world "simply" covers a great deal that seems to me at least to be
far from simple. Am I correct that PCT does not model "as we experience
them"? In fact, as you point, out PCT "solves" the problem of experiences
by ignoring them. A commendable economy, in my view, but one that not
everyone embraces.

When we experience the results of control

systems having large error signals, we call that experience "bad" or
"painful" or "unpleasant," and we try not to experience it any more, or
again.

Exactly how do we "experience the results of control systems having large
error signals"? Once again, I was under that impression that this is not
modeled in PCT. Am I incorrect?

When we experience a marked diminution in or disappearance of the
effects of errors, the name we give to that constellation of perceptions is
"pleasurable" or "good".

What are these "effects of errors"? What "constellation of perceptions" are
you referring to?

We seek or try to prolong that kind of experience.

Why? Is this a control process? I assume it must be, but I am at a loss as
to how to diagram it.

"Pleasure" and "pain" are _classifications_ in which we find many different
kinds of experiences, and what is common to each category is the state of
error and change of error in many different control systems, as well as
what we do when such errors exist, increase, decrease, or vanish. Pleasure
and pain do not exist as single specific sensation-signals in the brain. Or
to put that differently, pleasure or pain associated with one experience is
not the same signal as pleasure or pain associated with a different
experience. You can have one system in a "pleasant" state at the same time
another system is in an "unpleasant" state.

Too complex for me, I fear. Perhaps a diagram would help.

BG

13

[From Bruce Gregory (2000.1005.0950)]

[Bill Curry (2000.10.04.1100 EDT)]

Bruce, please help me understand your meaning. I see a self-concept error
(e.g.,
"I am lonely") being reduced by setting lower level references for pleasure
attainment, in this example by controlling for perceptions of eating.

That's fine, but it involves no "experience", any more that my thermostat
experiences angst when it turns the furnace on. In the case of my
thermostat, it has no "angst experiencing" system. Human beings, at least
some of them, seem to have such systems. In the same way, we experience
making painful choices, but the model, like a computer program's IF...THEN
logic, makes "choices" without experiencing anything.

BG

14

[From Bruce Gregory (2000.1005.1003)]

I think it might be helpful to keep the following in mind:

"Control theory is about controlling. It is not about responding to
stimuli, or planning actions and carrying them out; it's not about the
effects of traumatic incidents on later behavior; it's not about particular
things people do under particular circumstances. It's not about attitudes
or habits or beliefs or tendencies. It's not about predicting. It's just
about one kind of behavior that we can see people carrying out, called
controlling."

-- William Powers

BG

15

[From Bruce Abbott (2000.10.05.1025 EST)]

Bill Powers (2000.10.05.0553 MDT) --

Bruce Abbott (2000.10.04.1710 EST)

Yes, I know. But then, one has to ask why we have evolved the capacity to
experience pleasure (or displeasure). Clearly we could just have control
systems linked up so as to generate the appropriate behaviors without
involving these subjective feelings, yet we do often have them. The thirsty
rat may be drinking, not merely because there is a deficit in its water
balance, but because the rat, when thirsty, finds the experience of drinking
water to be pleasurable. As the water balance is restored, the pleasure of
drinking diminishes.

If the reason the rat drinks is that it brings pleasure to do so, why does
it need a second mechanism for detecting and correcting a water deficit?

The second mechanism establishes the level of pleasure (and/or relief from
displeasure) to be had by consuming water.

And if it has a mechanism for detecting and correcting a water deficit, of
what use is a second mechanism that gives pleasure for doing so?

Drinking water may serve to correct errors in many controlled variables --
level of hydration of the cells, blood volume, wetness of the throat, body
temperature, level of irritation of the membranes of the mouth (as disturbed
by hot peppers for example). On the other hand, drinking to correct errors
in some variables sometimes will disturb others (drinking to cool the
burning sensation of hot peppers may result in too much water in the
bloodstream). The pleasure experienced when drinking the water provides a
universal code signifying immediately to the organism that drinking water is
the right thing to be doing. (It _feels_ right!) The organism will thus
drink even when the error that arouses drinking may not be immediately
corrected by the act. (I am thinking here of the need for the water
consumed to be absorbed in the digestive tract and transported to the body's
cells, all of which takes time. If pleasure results from the reduction of
error, then drinking motivated by this error would produce no immediate
pleasure.)

. . . Certain sensory experiences
will evoke, through connections in the brain, feelings of pleasure or
displeasure; usually these will have been associated through evolution or
learning with physical states that in the history of the species or organism
have fostered continued survival. By controlling for pleasure (or
elimination of displeasure), the rat also controls for being properly
nourished, hydrated, groomed, and so on.

Yes, but _which_? If the very same sensation results from doing all these
different things, how is it that the rat or person knows which of them
needs to be done to get pleasure, since any of them gives pleasure? If the
rat is dehydrated, why doesn't it have sex? If it lacks salt, why doesn't
it groom itself? A single all-purpose pleasure sensation is incapable of
directing the organism to what it needs, particularly when it needs many
things at the same time.

Pleasure does not exist by itself; it is inevitably linked to a particular
sensory experience -- when thirsty, drinking water feels especially good,
and feeling thirsty feels bad. I am suggesting that, in higher organisms at
least, the brain includes a mechanism that attaches _values_ to these
sensations, which are experienced as pleasures and displeasures, likes and
dislikes, of varying degree. One learns that, when thirsty or when the
mouth is dry, or when the pepper sensation gets too intense, drinking water
produces pleasure, whereas having sex, to use one of your examples, does not.

And if a certain act is done to get pleasure, why does it ever stop? Why
doesn't the rat go on eating until it bursts? As soon as you posit this
explanation, you must invent another one to explain why it stops working:
the rat stops eating when eating more would cause it more pain than
pleasure. But again, why is it that pain would cause it to stop _eating_
rather than, say, drinking? It's the very same sensation that would arise
from being too hot or too deficient in salt or too dehydrated -- if you
refuse to consider the concept of error signals in specific different
control systems, you can't explain the fortuitous specificity of pleasure
and pain.

I'm afraid that you've misunderstood my proposal. If you go back and reread
it, you will find that I have indeed considered the concept of error signals
in specific different control systems. When error exists in a certain
specific control system, certain specific sensory experiences become
pleasurable (although they may be at least somewhat pleasurable even without
such error -- as witness the eating of sweets after consuming a full meal).

To make these explanations work you must invent still more complex
mechanisms. Now the rat or person must be able to distinguish which action
goes with pleasure in the current circumstances, so as not to perform the
wrong action. It must have some sort of spatio-temporal correlation
capabilities, so it can tell that the feeling of pleasure that exists now
arose because of the appropriate immediately-preceding action and not from
an action somewhat earlier or of a different kind. "Just what did I do that
I am being reinforced for (or punished for) this time?" In an organism that
must simultaneously maintain dozens of variables close to specific levels
in order to survive and propagate its species, trying to answer this
question would tax a brain much bigger than a rat's. "If I am feeling pain,
which of these pleasurable activities needs to be stopped? And if I am
feeling pleasure, which aspect of my activities should I continue
producing, or produce more of?"

When you are thirsty and drink, just how much trouble do you have in
determining that it is the water you are drinking that is producing all
those really good feelings? How much trouble do you think a rat would have?

The control model does away with all these problems. There are reference
signals which by their settings (high or low) define whether a given
perception is to be sought or avoided. There are error signals which
indicate whether action needs to be taken and which, at the lower levels,
are converted into the required actions. Imagining and even modeling many
such systems working at the same time is no problem at all; each control
process occurs independently of and simultaneously with many other control
processes at the same level. There is no danger of one kind of error being
confused with another.

Nor is there any reason why the organism should experience any subjective
feelings of pleasure.

You have suggested, I believe, that pleasure arises as an error in some CV
is reduced, and displeasure/discomfort when the error is increased.
Pleasure/displeasure are then mere byproducts of the operation of a control
system and have no real biological function. The conception I have proposed
above (somewhat badly, I am afraid), would give these affective qualities a
central function.

No, that's not what I suggest. You're still treating pleasure and
displeasure/discomfort as if they were separate phenomena, signals that
arise somewhere else when error appears or disappears, as if these signals
existed separately from the signals in control systems. What I am proposing
is that pleasure and pain are simply words we use for the states of control
systems as we experience them. When we experience the results of control
systems having large error signals, we call that experience "bad" or
"painful" or "unpleasant," and we try not to experience it any more, or
again. When we experience a marked diminution in or disappearance of the
effects of errors, the name we give to that constellation of perceptions is
"pleasurable" or "good". We seek or try to prolong that kind of experience.
"Pleasure" and "pain" are _classifications_ in which we find many different
kinds of experiences, and what is common to each category is the state of
error and change of error in many different control systems, as well as
what we do when such errors exist, increase, decrease, or vanish. Pleasure
and pain do not exist as single specific sensation-signals in the brain. Or
to put that differently, pleasure or pain associated with one experience is
not the same signal as pleasure or pain associated with a different
experience. You can have one system in a "pleasant" state at the same time
another system is in an "unpleasant" state.

Why should we experience these states at all? Why should we label these
experiences as good or bad, pleasant or unpleasant? What is doing the
evaluating? For what purpose? You suggest that pleasure/pain are
classifications based on error, but in your system the classification has no
functional role to play. Apparently it is simply a mental game we play for
our own amusement.

I say that the pleasantness or unpleasantness itself is a perceptual
experience, an integral part of the perception aroused by the sensory inputs
themselves, and dependent on the state of error in those control systems for
which those sensory inputs correlate strongly with physiological effects
that close the negative feedback loop. Without this label being attached to
the sensory experience, becoming part of it, we simply have no basis for
choosing among the various alternatives, as Damasio points out so
forcefully. A bland bowl of cold mush and a hot, juicy New York Strip may
both correct an error in our nutritional state equally well; given the
choice, we may choose based on which we expect to arouse the most pleasure.
And there is nothing in this proposal demanding, as you claim, that pain or
pleasure exist as some disembodied entity unassociated with specific sensory
experiences, or that different systems can't be in different states.

Bruce A.

16

[Bill Curry (2000.10.05.1050 EDT)]

Bruce Gregory (2000.1005.0950)--

>[Bill Curry (2000.10.04.1100 EDT)]
>
>Bruce, please help me understand your meaning. I see a self-concept error
(e.g.,
>"I am lonely") being reduced by setting lower level references for pleasure
>attainment, in this example by controlling for perceptions of eating.

That's fine, but it involves no "experience", any more that my thermostat
experiences angst when it turns the furnace on. In the case of my
thermostat, it has no "angst experiencing" system. Human beings, at least
some of them, seem to have such systems. In the same way, we experience
making painful choices, but the model, like a computer program's IF...THEN
logic, makes "choices" without experiencing anything.

I think I get your meaning now. Good point and better question! Where does the
feeling or angst component of experience come from since it is isn't an output
of the PCT signal processing model? Speaking in terms of my example, I didn't
mean to infer that the pleasurable somatic feelings we associate with satiation
are generated via electrically signaled comparator outputs from the brain.
Physiologically speaking, I see these somatosensations as an affective
expression of complex streams of neuropeptide-mediated chemical feedback
signaling from the enteric system acting on brain cell receptors (e.g., the
opiate receptors) as gut fill and digestion proceeds. Obviously, these
parasynaptic information processes would have to be interfaced or "plumbed in"
somehow with the model, but I don't have the foggiest idea how this could be
modeled. What's your take on the angst experiencing system vis a vis the
model?

Regards,

Bill C.

···

--
William J. Curry
Capticom, Inc.
capticom@landmarknet.net

17

[Bill Curry (2000.10.05.1205 EDT)]

Rick Marken (2000.10.03.1420)--

Bruce Abbott (2000.10.03.0830 EST)

> Sketch showing, from right to left and in order of decreasing
> size, a man walking, a mother duck, a toy duck, and a duckling.
> The caption reads: "A duckling imprinted on a person follows
> at a greater distance than one imprinted on a mother duck, but
> it follows a small toy duck more closely Thus it keeps the
> visual angle of the imprinted stimulus approximately constant."
> From Brown and Herrnstein's (1975) _Psychology_ (an introductory
> psychology textbook), p. 58.

This is a great example of seeing behavior through control
theory glasses. Good find, Bruce. It suggests some interesting
tests. For example, if visual angle really is a controlled
variable then the ducklings' distance from the "imprinted
stimulus" should be directly proportional to the magnification
of lenses placed over the ducklings' eyes.

Cool idea! Another simple test of the visual angle variable would be to
collect distance data with the imprinted person waddling his flock (like
a neighbor apparently caught Lorentz doing) and when leading them while
walking upright. Another might be to collect data from same aged
gaggles having imprints of differing heights (small children thru
basketball centers). I assume in all of these cases the imprint's
forward velocity would be have to be held constant to eliminate that
bias on the following distance.

Regards,

Bill C.

···

--
William J. Curry
Capticom, Inc.
capticom@landmarknet.net

18

[From Bill Powers (2000.10.05.1146 MDT)]

Bruce Gregory (2000.1005.1003)]

I think it might be helpful to keep the following in mind:

"Control theory is about controlling. It is not about responding to
stimuli, or planning actions and carrying them out; it's not about the
effects of traumatic incidents on later behavior; it's not about particular
things people do under particular circumstances. It's not about attitudes
or habits or beliefs or tendencies. It's not about predicting. It's just
about one kind of behavior that we can see people carrying out, called
controlling."

Yes, I agree that it would be very helpful to keep this in mind. Anytime
you're tempted to talk about responding to stimuli, planning actions,
effects of traumatic incidents, and so on, you should remind yourself that
control theory explains the phenomena underlying these artificial
categories or concepts in a way completely different from the traditional
or common-sense way. These conceptual artifacts, for the most part,
disappear under control theory, and are replace by other, and I think
better, constructs.

Best,

Bill P.

19

[From Bill Powers (2000.10.05.1153 MDT)]

Bruce Abbott (2000.10.05.1025 EST)--

Drinking water may serve to correct errors in many controlled variables --
level of hydration of the cells, blood volume, wetness of the throat, body
temperature, level of irritation of the membranes of the mouth (as disturbed
by hot peppers for example). On the other hand, drinking to correct errors
in some variables sometimes will disturb others (drinking to cool the
burning sensation of hot peppers may result in too much water in the
bloodstream). The pleasure experienced when drinking the water provides a
universal code signifying immediately to the organism that drinking water is
the right thing to be doing. (It _feels_ right!) The organism will thus
drink even when the error that arouses drinking may not be immediately
corrected by the act. (I am thinking here of the need for the water
consumed to be absorbed in the digestive tract and transported to the body's
cells, all of which takes time. If pleasure results from the reduction of
error, then drinking motivated by this error would produce no immediate
pleasure.)

There is absolutely no function you propose for either pleasure or pain
that would not be fulfilled by a higher level of control system. The
problem here is that you're trying to give pleasure and pain an existence
separate from _what_ is pleasant or painful, the perception itself. This
reification is unnecessary, and anyway you fail to propose any mechanism
that could accomplish it.

. . . Certain sensory experiences
will evoke, through connections in the brain, feelings of pleasure or
displeasure; usually these will have been associated through evolution or
learning with physical states that in the history of the species or

organism

have fostered continued survival. By controlling for pleasure (or
elimination of displeasure), the rat also controls for being properly
nourished, hydrated, groomed, and so on.

Wrong way around. By controlling for being properly nourished, hydrated,
groomed, etc., the rat controls those things it desires -- which it the
same as saying it controls what is pleasant for it. "Pleasant" means no
error; "unpleasant" means error -- and, I claim, nothing else.

Yes, but _which_? If the very same sensation results from doing all these
different things, how is it that the rat or person knows which of them
needs to be done to get pleasure, since any of them gives pleasure?

Pleasure does not exist by itself; it is inevitably linked to a particular
sensory experience -- when thirsty, drinking water feels especially good,
and feeling thirsty feels bad. I am suggesting that, in higher organisms at
least, the brain includes a mechanism that attaches _values_ to these
sensations, which are experienced as pleasures and displeasures, likes and
dislikes, of varying degree.

All right, what exactly is your model of that "mechanism?" I say the model
is a control system. To "like" something is to set a nonzero reference
level for it; to dislike it is to set a zero reference level for it. That
_is_ the mechanism; no other is needed.

One learns that, when thirsty or when the
mouth is dry, or when the pepper sensation gets too intense, drinking water
produces pleasure, whereas having sex, to use one of your examples, does not.

You are describing nothing more than control processes. Pleasure is not
drinking water when the pepper sensation exceeds its reference level; it is
the restoration of that sensation to its desired level. A higher-level
system may conclude that water is good because of that effect, but that is
a different level of control.

I'm afraid that you've misunderstood my proposal. If you go back and reread
it, you will find that I have indeed considered the concept of error signals
in specific different control systems. When error exists in a certain
specific control system, certain specific sensory experiences become
pleasurable (although they may be at least somewhat pleasurable even without
such error -- as witness the eating of sweets after consuming a full meal).

And you have misunderstood mine. I am saying that the concept of pleasure
as a signal different from any in the control systems is superfluous.

To make these explanations work you must invent still more complex
mechanisms. Now the rat or person must be able to distinguish which action
goes with pleasure in the current circumstances, so as not to perform the
wrong action.

When you are thirsty and drink, just how much trouble do you have in
determining that it is the water you are drinking that is producing all
those really good feelings? How much trouble do you think a rat would have?

None at all, under my proposal. "All those [hypothetical] good feelings"
are the thirst decreasing (as well as any other errors that may decrease as
a consequence, such as the fear of dying from thirst), and nothing else.
That is _why_ it's so easy. But in saying it is easy, and no trouble, you
do not explain why it is so easy and so little trouble. You say nothing at
all about what the mechanism is.

You have suggested, I believe, that pleasure arises as an error in some CV
is reduced, and displeasure/discomfort when the error is increased.

No, that's not what I suggest. You're still treating pleasure and
displeasure/discomfort as if they were separate phenomena, signals that
arise somewhere else when error appears or disappears, as if these signals
existed separately from the signals in control systems.

Why should we experience these states at all? Why should we label these
experiences as good or bad, pleasant or unpleasant? What is doing the
evaluating? For what purpose?

The control system is doing the "evaluating" such as it is. But the
evaluation is embodied in the fact that we have nonzero reference levels
for some things and zero reference levels for others. At the same time, our
thinking levels can form cognitive opinions about our experiences which
_reflect_ the reference settings, and can influence them in the future. The
taste of cool clear water when we are thirsty is made of many perceptions,
and because of the way we feel when we drink it we set the reference values
quite high when we are thirsty -- but not when we are not thirsty. Why
should setting the reference level to zero change the associations with the
perception from pleasant to unpleasant? The answer is that it doesn't,
because there are no "associations."

My point is that we don't need anything outside the perceptual hierarchy
and the control hierarchy to explain what we mean by pleasant and unpleasant.

You suggest that pleasure/pain are
classifications based on error, but in your system the classification has no
functional role to play. Apparently it is simply a mental game we play for
our own amusement.

Not at all. If we perceive something as pleasant, we may plan on
experiencing it again, and the categorization thus plays a part in our
control processes.

I say that the pleasantness or unpleasantness itself is a perceptual
experience, an integral part of the perception aroused by the sensory inputs
themselves, and dependent on the state of error in those control systems for
which those sensory inputs correlate strongly with physiological effects
that close the negative feedback loop. Without this label being attached to
the sensory experience, becoming part of it, we simply have no basis for
choosing among the various alternatives, as Damasio points out so
forcefully.

But that just isn't true. There is _always_ a basis/

A bland bowl of cold mush and a hot, juicy New York Strip may

both correct an error in our nutritional state equally well; given the
choice, we may choose based on which we expect to arouse the most pleasure.
And there is nothing in this proposal demanding, as you claim, that pain or
pleasure exist as some disembodied entity unassociated with specific sensory
experiences, or that different systems can't be in different states.

But why do you confine your reference signals to "nutritional state?" How
about the taste itself? If you have come _for any reason_ to set a high
reference level for experiencing some taste, you will seek that taste
regardless of the nutritional effects of consuming whatever gives that
taste. In other words, you will act and feel in all respects as if the
taste is pleasurable. At a higher level, you may even have a positive
reference level for the idea that you are here having this sought-after
experience -- for example, you might get snobbish satisfaction out of
eating caviar even though at a lower level the desire for experiencing the
taste of caviar can be satisfied with a vanishingly small portion of it.

I think we're having an Occam's-Razor kind of argument. You want pleasure
and displeasure to be a separate phenomenon, while I am claiming that there
is no need for this to be the case. If we just drop the notion of a generic
kind of experience of pleasure and displeasure, and let the state of the
control systems in the brain substitute for them, we will lose nothing at
all in explanatory power, and we will shed two constructs that are not
needed. The _very same experiences_ to which you refer will still occur;
only the explanation will be different.

Best,

Bill P.

20

[From Bill Powers (2000.10.05.1102 MDT)]

Bruce Gregory (2000.1005.0945)]

Bill Powers (2000.10.05.0553 MDT)

The control model does away with all these problems.

Perhaps it is _too_ successful.

Are you suggesting that to improve the model, we should introduce some
features that predict incorrectly, just to play fair with other models?

What I am proposing
is that pleasure and pain are simply words we use for the states of control
systems as we experience them.

Here the world "simply" covers a great deal that seems to me at least to be
far from simple. Am I correct that PCT does not model "as we experience
them"? In fact, as you point, out PCT "solves" the problem of experiences
by ignoring them. A commendable economy, in my view, but one that not
everyone embraces.

That's a pretty distorted picture of the situation. PCT contains
explanations for how all experiences come into being as perceptions. Like
every other theory anyone has ever proposed about human existence, it does
not explain who or what is doing this experiencing of the activities in the
brain. The nature of consciousness remains mysterious no matter whose
theory you're talking about. PCT most certainly does not ignore the
_content_ of experience.

When we experience the results of control

systems having large error signals, we call that experience "bad" or
"painful" or "unpleasant," and we try not to experience it any more, or
again.

Exactly how do we "experience the results of control systems having large
error signals"? Once again, I was under that impression that this is not
modeled in PCT. Am I incorrect?

Yes. In PCT, "to experience" is synonymous with "to perceive," because the
theory can't explain what makes the difference between an unconscious and a
conscious perception. To experience the results of control systems having
large error signals, we must _at least_ contain perceptual signals
representing those results. PCT goes that far, but it does not add anything
about what is required for those signals to be objects of awareness. The
results will be things like unwanted body states or very large corrective
efforts.

When we experience a marked diminution in or disappearance of the
effects of errors, the name we give to that constellation of perceptions is
"pleasurable" or "good".

What are these "effects of errors"? What "constellation of perceptions" are
you referring to?

I'm wondering why you even have to ask, but I'll play along.

The main effect of error signals is to readjust lower-level reference
signals, and usually to create error-opposing actions. We (meaning some of
our higher-level subsystems) can perceive those actions, and also we can
perceive body states that may go with them, or that go with the existence
of error signals (for example, pain or fear, as somatic states). What we do
about those perceptions, if anything, depends on the reference signals that
specify their preferred states. The "constellation of perceptions" includes
many different perceptions and many levels of perception: what you think
and feel as the experience goes on. If we find ourselves strongly rejecting
an experience, we can take that as what we mean by saying it is an
unpleasant experience. We may perceive in imagination that we have a very
low or zero reference level for that experience, which is also part of our
evaluation.

We seek or try to prolong that kind of experience.

Why? Is this a control process? I assume it must be, but I am at a loss as
to how to diagram it.

This means only that we have a non-zero reference level (or reference
levels) for that kind of experience. We don't have the nonzero reference
level because the experience is pleasant; we call the experience (when at
its reference level) pleasant _because the reference level for it is
nonzero_. It is an experience we want to have (bring to some positive
level) rather than to avoid (bring to zero). If the perceptions involved
are below their reference levels, we will act to bring them up to those
levels: that is what "seeking" the experience means. If the perceptions
already nearly match the reference levels, we will act to maintain them in
the match: that is what "prolonging" the experience means. Both statements
refer to the way any control systems given non-zero reference signals will
act.

"Pleasure" and "pain" are _classifications_ in which we find many different
kinds of experiences, and what is common to each category is the state of
error and change of error in many different control systems, as well as
what we do when such errors exist, increase, decrease, or vanish. Pleasure
and pain do not exist as single specific sensation-signals in the brain. Or
to put that differently, pleasure or pain associated with one experience is
not the same signal as pleasure or pain associated with a different
experience. You can have one system in a "pleasant" state at the same time
another system is in an "unpleasant" state.

Too complex for me, I fear. Perhaps a diagram would help.

I don't think a diagram would help. All it would show would be a set of
perceptions from lower systems converging into two category-level
perceptual input functions. Out of one of them would come a perceptual
signal labelled "pleasant" and out of the other a signal labeled
"unpleasant." Any input connecting to these perceptual functions would give
rise to either the "pleasant" or the "unpleasant" perceptual signal (that
is how I conceive of category perception). Lower-level perceptions not
connecting to either PIF would not give rise to either signal. You can
postulate any kind of lower-level perception you please as being a
potential candidate for categorization. Of course the outcome should be the
kind of evaluation that people typically put on those perceptions, or
better, that a specific individual puts on them.

I wish you'd take a stab at answering some of these questions for yourself.
You certainly know enough about PCT and HPCT to do so.

Best,

Bill P.