Behavior IS Control?

Continuing the conversation:

RM: what we call “behavior” IS controlled perception seen from the point of view of the observer.

BN: Yes. We’ve noticed many times that there are two senses of ‘behavior’, behavior from the control system’s point of view and behavioral actions or outputs as seen from an observer’s point of view. “Behavior (from the organism’s point of view) is the control of perception” is consistent with “behavior (from the observer’s point of view) is the means by which we affect and, hopefully, control our perceptions.”

RM: There is only one sense of “behavior” in PCT because behavior is the phenomenon explained by PCT. PCT starts with the observation that behavior IS control. What this means is that most of the things we see organisms doing that we give names to, such as “walking”, “talking”, “begging”, “flying”, “nesting”, “courting”, “reading”, “pursuing”, “evading”, “analyzing data”, sewing, “biking”, etc, etc – are consistent results that are produced in the face of disturbances.

RM: “By paying careful, even compulsive, attention to detail…”(Powers, LCS I, p173) Powers was able to see that these behaviors are the reference states of controlled variables. And that “in these reference states…we have the heart of the problem to which control theory is addressed.” (Powers, LCS I, p. 175). These reference states are the data which PCT is designed to explain!!

RM: As Bill says “The existence of these reference states is not conjectural; once behavior has been defined in terms of an appropriate variable [or variables–RM], such reference states always exist. They can be discovered experimentally and defined in terms of observable relationships. Whether or not they should exist according to anyone’s theory, they do exist… To dismiss the existence of reference states as an illusion is simply to discard data” (Powers, LCS I, p175).

RM: So the “behavior” that PCT explains – the “behavior” in the title of Bill’s book Behavior: The control of perception – are the objectively observable reference state of controlled variables. The theory that explains the existence these reference states posits that controlled variables correspond to perceptual signals that are kept matching constant or varying reference signals by the actions of the behaving system.

RM: Because the word “behavior” is not a technical term it is often used to describe the means by which a controlled variable is brought to and maintained in a reference state. Bill uses it that way right in the middle of the very same chapter of LCS I where he is explaining that what we call “behaviors” are the reference states of controlled variables: “What kind of a system can behave in such a way that a variable will, under a variety of unpredictable conditions, always approach the same state?” (LCS I, p. 175, emphasis mine).

RM: I think there is really no way to avoid this usage of “behave” (as a synonym for the more appropriate “action”) and I don’t think it’s necessary to try to avoid it. Once you understand that the lay meaning of “behavior” corresponds to observed reference states of controlled variables you’ve got your PCT glasses on correctly. And the fact is that virtually all of the observable “actions” that keep a controlled variable in a (constant or varying) reference state are themselves “behaviors” inasmuch as they are controlled variables being kept in reference states by still lower level actions. Indeed, the only “actions” that are not controlled variables are probably the firing rates of neurons – such as motor neurons – that are at the lowest level of the nervous system. And only neurophysiologists using special equipment can see the action of those neurons when looking at some overt “behavior”, such as turning the head towards a food reward.

RM: What Bill is describing (and recommending) in the quotes from the “Cybernetic Model for Research” chapter of LCS I (which I highly recommend!) is what I call the “Phenomena Phirst” approach to understanding the PCT model. I think it’s the only approach to understanding PCT that will produce any progress in our understanding of human nature (and the nature of living control systems in general).

Best, Rick

Got it. How about “behaving is controlling”?

Fred

Hi Fred

FN: Got it. How about “behaving is controlling”?

RM: Sounds great. As Jerry Seinfeld might have said “Love the gerund”:wink:

Best, Rick

Indeed it is.

Trouble is, everybody in the world who does not understand PCT has this misunderstanding. Over the years, to help them understand PCT we have sometimes said that outputs, actions, control actions, behavioral outputs, etc. are the observable means for affecting the state of the controlled variable. I agree that we should reserve “behavior” to refer to what the entire control loop does.

The process of coming to understand PCT develops new perceptual input functions. At first, these are abstract concepts with mostly imagined inputs. Looking at an activity and saying that we are observing control at this stage is something of a hopeful (and instructive) pretense. Over time applying PCT our perception of control in a given situation is more richly supplied with lower-level perceptual inputs, and we can more legitimately make the claim.

The lay meaning of “behavior” corresponds to observed reference states of controlled variables when the lay meaning refers to the intent of the observed actions. He’s pushing the doorbell, he’s seeing if anyone is home, he’s selling vacuum cleaners, etc. It’s hard to describe an activity without implying intent. Does anybody except a conventional psychologist attempt that veneer of ‘objectivity’, with no attribution of purpose?

Everybody (except ‘scientific psychologists’, and even they most of the time) attributes intent to observed actions. With PCT, we describe intent as reference values in a hierarchy of controlled perceptions. There lie avenues of communication.

The common-parlance usage of ‘behavior’ meaning actions is more apt when perceptions of actions come under control. Example: the dancer before a mirror in a practice studio. Or in another way, perceptions of outputs for controlling A may be among the perceptual inputs for control of B at a higher level. Examples include threats, welcomes, and all kinds of more sophisticated communication-perceptions. In these situations, the perceptions of behavioral outputs may themselves also come under direct control. “Don’t do that, it can break!” “Do it upside down.” “Insert tab A into slot B this way.” The speaker is using language to control their perception of the listeners’ behavioral outputs, and the attentive, compliant listeners thereupon attend to and control their own perceptions of those outputs and their consequences.

Blast from the past, this is from another nice exposition of yours, Rick, that’s worth revisiting:

Hi Bruce

BN: We’ve noticed many times that there are two senses of ‘behavior’, behavior from the control system’s point of view and behavioral actions or outputs as seen from an observer’s point of view. […]

RM: This is … a fundamental misunderstanding of the scientific basis of PCT

BN: Indeed it is.

RM: Actually, you left out the part that I was referring to as a fundamental understanding:

You said: “Behavior (from the organism’s point of view) is the control of perception” is consistent with “behavior (from the observer’s point of view) is the means by which we affect and, hopefully, control our perceptions.”

RM: The misunderstanding here is that “behavior is the control of perception” (from any point of view) is consistent with “behavior is the means by which we affect and, hopefully, control our perceptions" (also from any point of view). In PCT, “behavior is the control of perception” means that what we see as behavior is controlled aspects of the environment (which are perceptual variables).

BN: Trouble is, everybody in the world who does not understand PCT has this misunderstanding. Over the years, to help them understand PCT we have sometimes said that outputs, actions, control actions, behavioral outputs, etc. are the observable means for affecting the state of the controlled variable. I agree that we should reserve “behavior” to refer to what the entire control loop does.

RM: I think it’s far more important to explain to those new (and old) to PCT what I explained at the beginning of this thread: that what we call an organism’s “behavior” is a controlled aspect of the organism’s environment; it is a controlled perceptual variable, and controlled variables are the basic data of the study of behavior from a PCT perspective.

BN: The lay meaning of “behavior” corresponds to observed reference states of controlled variables when the lay meaning refers to the intent of the observed actions.

RM: Actually, I think the lay meaning of “behavior” doesn’t confine itself just to intended results of action. It also often refer to the actions themselves, even when the result was unintended (she “knocked” over the cup by accident). Of course the actions that we see as the means of producing an intended or unintended result are also controlled variables (the arm movement that knocked over the cup was a controlled results, though poorly when overturning over the cup was not the intended result.

BN: He’s pushing the doorbell, he’s seeing if anyone is home, he’s selling vacuum cleaners, etc. It’s hard to describe an activity without implying intent.

RM: Right, and what PCT brings to the party is not just the idea that behavior has goals or purposes but also an explanation of what goals and purposes are: they are the reference states of controlled variables. So the first step in understanding these behaviors is determining, in detail, what variables are being controlled. My initial hypothesis of each of your examples would be 1) the sound of a doorbell 2) the sight of someone at the door 3) the receipt of payment for purchase.

BN: Does anybody except a conventional psychologist attempt that veneer of ‘objectivity’, with no attribution of purpose?

RM: I think behavioral scientists in general are prone to avoiding talk of purpose. Economists, for example, are big on “incentives” which are events that are seen as the cause of behaviors rather than the goals of purposeful action, which they actually are. I have found, by he way, that Gary Cziko’s The Things We Do is a treasure trove of examples of life scientists who were quite unabashed about describing behavior in terms of its possible purposes – that is, in terms of the variables that the organism is controlling.

BN: With PCT, we describe intent as reference values in a hierarchy of controlled perceptions. There lie avenues of communication.

RM: More importantly, I think, is that PCT describes what intended results are: they are controlled variables, which are the observable consequences of an organisms actions that are protected from the effects of disturbance. Controlled variables are not only an avenue of communication with psychologists; psychologists have to learn controlled variables are the data that are essential to the development of a science of purpose.

BN: Blast from the past, this is from another nice exposition of yours, Rick, that’s worth revisiting:

RM: So I believe Bill used the word “perception” in the title of his book because he knew that psychologists would see it as referring to the “input” of an organism, while the word “behavior” would be seen as referring to the organism’s “output”…It’s the first claim – that output controls input – that is the tough one (for psychologists), so tough that it has been nearly impossible to get psychologists to pay attention to PCT for more than a few seconds.

RM: Thanks. It’s nice to know that some people are reading my stuff.

Best, Rick

Eh. Reaching for the salt was her behavior. Knocking over the cup was not, it was an accident. Accidental byproducts of control are not ‘behavior’ in any sense, PCT or lay.

Yes, that’s good: not just to name the intention, but to specify it in terms of the desired outcome. What management-talk calls measurable objectives. But you will grant that there is no substantive contradiction here, you are saying the same thing with greater specificity. In lay usage, the ‘measurable objective’ may not be called for; in PCT research it is.

Good reference to an excellent book. I’ll go back and look at it again. (Also without Miracles.) But I’m a little unclear what point you’re making here. Are you making a contrast between behavioral scientists and life scientists? Saying that behavioral scientists don’t talk of purpose, but life scientists do?

Discourse suggests other similar topics when we create a new one. It’s always interesting and valuable to follow these up in new tabs. And the Search function turns up great stuff in the archive when you search on a word or phrase from the current discussion. I recommend it.

Hi Bruce

RM: …Of course the actions that we see as the means of producing an intended or unintended result are also controlled variables (the arm movement that knocked over the cup was a controlled results, though poorly when overturning over the cup was not the intended result.

BN: Eh. Reaching for the salt was her behavior. Knocking over the cup was not, it was an accident. Accidental byproducts of control are not ‘behavior’ in any sense, PCT or lay.

RM: My point was just that the means used to produce results, controlled or accidental, are typically controlled variables themselves. Reaching is a controlled variable; knocking over the cup is probably not, but it certainly could be.

RM: Right, and what PCT brings to the party is not just the idea that behavior has goals or purposes but also an explanation of what goals and purposes are: they are the reference states of controlled variables. So the first step in understanding these behaviors is determining, in detail, what variables are being controlled. My initial hypothesis of each of your examples would be 1) the sound of a doorbell 2) the sight of someone at the door 3) the receipt of payment for purchase.

BN: Yes, that’s good: not just to name the intention, but to specify it in terms of the desired outcome. What management-talk calls measurable objectives. But you will grant that there is no substantive contradiction here, you are saying the same thing with greater specificity. In lay usage, the ‘measurable objective’ may not be called for; in PCT research it is.

RM: Yes. And great point about measurable objectives. They are very similar to reference states of controlled variables. Both are descriptions of an intended perceptual result (eg. the sound of a doorbell, the sight of someone at the door, the receipt of payment for purchase). The only difference is that measurable objectives are descriptions of the perceptual results that should be produced when you are asked to carry out a particular behavior (task); the reference states of controlled variables are descriptions (derived from the test for the controlled variable) of the perceptual results that are being produced when you see an organism carrying out a particular behavior.

RM: … I have found, by he way, that Gary Cziko’s The Things We Do is a treasure trove of examples of life scientists who were quite unabashed about describing behavior in terms of its possible purposes – that is, in terms of the variables that the organism is controlling.

BN: Good reference to an excellent book. I’ll go back and look at it again. (Also without Miracles.) But I’m a little unclear what point you’re making here. Are you making a contrast between behavioral scientists and life scientists? Saying that behavioral scientists don’t talk of purpose, but life scientists do?

RM: The only point I am making (or trying to make) is that controlled variables are readily observable phenomena that are the basic data of PCT. According to PCT the behavior of organisms is organized around the control of these variables. Controlled variables are the observable analogs of the theoretical perceptual variables (signals) of the PCT model. Failure to take the existence of controlled variables into account in behavioral research results in studies of behavioral phenomena that are essentially illusions in the sense that they seem to tell us something about how the organism “works” but actually don’t.

BN: Discourse suggests other similar topics when we create a new one. It’s always interesting and valuable to follow these up in new tabs. And the Search function turns up great stuff in the archive when you search on a word or phrase from the current discussion. I recommend it.

RM: Thanks. I’ll give it a try.

Best, Rick

Got it.