Beliefs, factual and symbolic

Phenomena Collective phenomena
1

Dan Sperber found that the Dorze people in southern Ethiopia made assertions that they seemed both to believe and not to believe. For example, they said that “the leopard is a Christian animal who observes the fasts of the Ethiopian Orthodox Church.” Nevertheless, the average Dorze man guarded his livestock on fast days just as much as on other days. “Not because he suspects some leopards of being bad Christians, but because he takes it as true both that leopards fast and that they are always dangerous.”

Neil Van Leeuwen proposes that humans represent and use factual beliefs differently from symbolic beliefs, which he terms “credences.” Factual beliefs are for modelling reality and behaving optimally within it. Because of their function in guiding action, they exhibit features like “involuntariness” (you can’t decide to adopt them) and “evidential vulnerability” (they respond to evidence). Symbolic beliefs, meanwhile, largely serve social ends, not epistemic ones, so we can hold them even in the face of contradictory evidence.

Van Leeuwen’s ‘credences’, the beliefs in his ‘symbolic’ category are collectively controlled beliefs. They “largely serve social ends.” Their evidentiary basis is in successful control of social confirmation (“Yes, I agree with you, the earth is really flat. All those schoolteachers and NASA people are gaslighting us.”).

Evans-Pritchard, living with the Azande, studied how their belief in witches worked. “I always kept a supply of poison for the use of myself and neighbors and we regulated our affairs in accordance with the oracles’ decisions. I may remark that I found this as satisfactory a method of running my home and affairs as any other I know of.” Yet he also says “witches, as the Azande conceive them, clearly cannot exist.” The systematic ‘credences’ about witchcraft were eminently functional means for controlling collectively controlled variables together with other participants in them, irrespective of any physical correlation (or not) between (a) whatever CVs are the subject of oracular inquiry and (b) the biology chickens and whether they die or recover when poisoned.

The Azande case is discussed in Mattick, Paul Jr. 1986. Social Knowledge: An Essay in the Nature and Limits of Social Science.)

The bits about Sperber and van Leeuwen are partly quoted, partly paraphrased from Manvir Singh, “Don’t Believe What They’re Telling You About Misinformation” (New Yorker 4/15/24) If you can’t get past the paywall there, I’ve put a PDF image here.

2

In PCT, these “beliefs” would be implemented as the output functions of control systems. In a tracking task, for example, my “model” of the reality that connects me to the perception I’m controlling is embodied in the output functions that transform error into an amount of force applied to the mouse.

Could you please explain how the collective controlled of believe works, using a concrete example so that we can tell the difference between variables that are collectively controlled from those that aren’t.

3

For starters, Rick, what are you controlling by having typed the word “please” in your reply and how do you specify the environmental feedback function for that control loop? (Or if more than one control loop is involved what are the environmental feedback functions for those control loops?)

4

I think I was controlling for asking, without sounding confrontational, how you would explain the collective control of belief. Of course, I could only tell if adding “please” made my request sound less confrontational to me; I could only hope that it would also sound non-confrontational to you.

This is a great question. In trying to answer it I realized that the environmental feedback function is just a component of the complete feedback connection between a system’s output and the variable it controls, the controlled variable (CV). Here’s a little diagram to show you what I mean:

Output → Environment —> Sensory Input → Perceptual Hierarchy – > CV
{---- Environmental Feedback Function ----}
{--------------------------------- Feedback Function --------------------------------------}

The Environmental Feedback Function (EFF) is the part of the feedback loop that goes through the environment from system Output to Sensory Input. In the case of my writing “please”, the EFF goes from my key presses (my Output) to the sensory effect of the letters that appear on the screen (Sensory Input). The Environment in this case consists of the keyboard keys, the mechanism that closes the electrical contacts sending a pattern of voltages to the computer which are transformed by the computer into the pixel pattern of a letter on the screen. So the EFF in this case could be written as S = f(P), where S is the sensory effect of the pattern on the screen, P is the key press and f() maps key presses to sensory effects. For example, if P = B key then S = B, if P = H key then S = H, etc.

The Feedback Function (FF) involved in typing “please” includes the EFF as well as the neural processing performed by the Perceptual Hierarchy that relates Output to Controlled Variable (CV), which is a perceptual representation of the Sensory Input. In the case of my writing “please”, I was using my Output key presses to control several levels of CV. At the lowest level I was controlling for the configuration perceptions – letters – that make up the word “please”. At the higher levels I was using the same Outputs to construct words and sentences that took the form of what I perceived at even higher levels as polite questions about how collective control of beliefs works. Typing the word “please” was just part of the process of constructing questions that I perceived as being polite.

In most of our research we don’t pay much attention to the distinction between the EFF and the FF in a control loop. We usually just assume that these two terms describe the same thing – the connection between Output and CV. But I did write a paper where it was important to understand that I was studying the entire FF and not just the EFF. In order to study the entire FF I had to make assumptions about how the perceptual functions determine the nature of the CV. These assumptions were necessary to see whether the observed functional relationship between an environmental variable and the outputs that were apparently made in response to that variable was a behavioral illusion.

The observed functional relationship that was suspected of being a behavioral illusion was Steven’s Power Law, which is the consistently observed power relationship between the magnitude of an environmental stimulus, e, and numerical ratings of those magnitudes, o. The law can be written as

o = k*e^a

My hypothesis was that this power law is an example of a behavioral illusion inasmuch as it could be the inverse of a log relationship between o and e. My analysis is described in this paper , which was never published in a journal (I tried only once to get it published and it was rejected so I gave up) but I did publish it in my book Doing Research on Purpose, which is no longer in print but an electronic copy is available at archive.org.

The relevance of the paper to the distinction between the EFF and the FF is that the analysis took the perceptual component of the feedback connection between output and CV into account. The assumed that the relationship between numerical output and the CV is logarithmic rather than linear. This assumption is based on Fechner’s Law, which says that the relationship between stimulus and perceptual magnitude is logarithmic (rather than a power function).

In my analysis I assumed that the variable controlled when producing numerical estimates (o) of stimulus magnitude (e) was the difference between the perceptual values of e and one’s own numerical response, o. These perceptual values were assumed to be the logs of o and e. So the CV was log(o) - log(e). Thus the effect of o on this CV is log(o), so the FF in this case is CV = k log(o). The EFF in this analysis is absorbed into the FF since the means of producing the numerical estimates, o, are not considered in experiments that have found a power relationship between e and o.

5

I will be traveling the next four days but I will try to reply soon.

6

So to model what you were controlling by means of typing ‘please’ you would model two levels of control, motor control involved in making the letters appear on the screen and a higher level loop controlling ‘being polite’ by means of that lower level loop.

Both are within one individual person. A severely autistic person with no social skills.

To model asking the person being asked needs to be in the model

To model avoiding anything the model has to have a perception of that which is to be avoided and a low or zero amount specified as its reference value. When the unwanted input occurs, the avoidance system sends non-zero reference values to systems which carry out the actual avoidance. “I’m sorry, I didn’t mean to step on your toes!” In this case, the unwanted input is imagined, and as a consequence the avoidance system sends non-zero reference values to systems which carry out preventions from that input occurring. (Are anybody’s feet in the direction that I am about to step? Oops! Step where there are no feet, even though that may be otherwise inconvenient for me.)

Children learn to avoid tripping over obstacles to their feet by establishing memories of tripping and falling which are sources of a reference value for avoiding tripping and falling. The specific means of avoidance vary, depending upon what is perceived within the loop that is imagining tripping and falling. The reference value is present whether the perceptual input from the environment is present or not. That is a condition for controlling in imagination.

We’ve quite a ways to go to beliefs. First, the model has to acknowledge the existence of other people in the environment, and the environmental feedback function has to include not only the physical properties of the environment, it has to include the nonphysical control-system properties of other people. And it has to include the capacity to imagine what other people are doing.

7

I think “being polite” is a social skill by definition. Per the Dictionary:

polite: adj. having or showing behavior that is respectful and considerate of other people.

Perhaps. But my comment about what I was controlling for was just an answer to your question about what I thought I was controlling for when I used the word “please”.

I would just say the model has to have a reference that specifies a perception of a very polite, non-confrontational way of saying things.

I think we all – children and adults – learn to avoid tripping and falling over obstacles by reorganizing our hierarchy of control systems each time we have tripped or fallen. What’s remembered are the parameters of the control systems that get us through our environment with little of no tripping and falling (which had resulted in the errors that set off reorganization).

Of course. If people are part of the environmental component of the feedback connection between your output and the variable(s) you are controlling, then they have to be part of a model of your behavior. An example of people being in the feedback function between you and a controlled variable would be when you hire someone to do a job for you, like clean your house. But I think that, more often than not, people are in the role of disturbances to the variables we control. These disturbances can be helpful (as when someone helps you lift something) or harmful (as in the case of conflict). But since people are control systems themselves, there is always the possibility of conflict, even when they are trying to be helpful.

8

OK, before taking first steps in an explanation of how beliefs of the “credence” kind are collectively controlled perceptions, we have to get clear with one another about more fundamental matters.

You asked:

As a concrete example, I asked you what you were doing by means of putting the word ‘please’ in your request. You replied

I also asked you how you would specify the environmental feedback function for that.

You offered a model of motor control in one individual controlling the appearance of the word ‘please’ on your screen. I then said this model is inadequate because it does not include the person you are asking, and with whom you are avoiding being confrontational.

You also talked about a distinction between the environmental feedback function EFF and the feedback function FF, such that the EFF is included within the FF. This idea is new to me. You gave as illustration your contention about the power law and behavioral illusions. I couldn’t see any relevance to a model of what you were doing by adding “please” to your request for an explanation of how beliefs are collectively controlled perceptions. But I should consider it more carefully and ask you to explain its relevance.

You said that the part of the FF which is outside the EFF is “the neural processing performed by the Perceptual Hierarchy that relates Output to Controlled Variable (CV), which is a perceptual representation of the Sensory Input.” To explain this, you said:

It appears that you are saying that these higher levels of the perceptual hierarchy are a part of the feedback function. I do not understand this. Please explain.

9

Great.

Actually, it was a PCT model and only 1/2 of one at that since it was only my description of the environmental feedback function that was involved in my putting the word ‘please’ into my poorly typed request for an explanation of how the “collective controlled of believe [sic] works”.

My distinction between the EFF and FF was incorrect. Here’s the diagram I posted:
Output → Environment —> Sensory Input → Perceptual Hierarchy – > CV
{---- Environmental Feedback Function ----}
{--------------------------------- Feedback Function --------------------------------------}

You caught me before I could post a correction to this because I realized that both the terms “Environmental Feedback Function” and “Feedback Function” refer to the same thing: the function relating Output (O) to controlled variable (CV), CV = f (O). So my diagram should have looked like this:

Output ----------→ Environment —-------- > CV
{----- Environmental Feedback Function ------}
{--------------- Feedback Function ----------------}

I was thrown off by thinking of the CV as a perception, which it is. So I included the Perceptual Hierarchy as part of the Feedback Function since the perceptual functions in the hierarchy determine what perceptual variable is being controlled. But I realized that it is actually unnecessary to include those perceptual functions in a description of the Feedback Function because the CV is an observable phenomenon, which is visible to anyone building a model of a control system – a model that must include a Feedback Function.

The controlled variable is a perception in the sense that it is a perceptual aspect of the environment. The Feedback Function in a model of a control system’s behavior, whether that function is called the EFF or just the FF, is always the functional connection between the system’s Output and the perceptual aspect of the environment that the system controls: the CV. Indeed, a basic goal of PCT research is to determine the perceptual aspects of the environment; that is, which CVs organisms control. And this determination is done using some version of the Test for the Controlled Variable (TCV), which starts with a guess (or hypothesis) about the perceptual aspect of the environment that is being controlled. The TCV is all about Testing hypotheses about the CV. A good way to think of the TCV is that it is a way for an observer to perceive the same variable that the control system under study (a person, for example) is perceiving and controlling.

This characteristic of the Environmental Feedback Function is nicely illustrated by my "What is Size? demo, where there are (at least) two different aspects of the same environment that a person could be controlling: the area (A) or perimeter(P) of a quadrangle. These aspects of the environment are different functions of the horizontal (h) and vertical lines (v) on the screen: A = h * v and P = h + v. In the demo, h is a variable disturbance, d, and v is proportional to the person’s Output, o. If a person (like you) controls A (area is the CV) then the Environmental Feedback Function is A = d * o, and if the person controls P (perimeter is the CV) then the Environmental Feedback Function is P = d + o.

Note that there are two different Environmental Feedback Functions for these two different CVs – area and perimeter – even though the environmental connection between your Output (mouse movement) and each CV is the same! This shows that the nature of the Environmental Feedback Function depends not only on the nature of the environment between Output and CV but also on the nature of the CV itself.

My discussion of Stevens’ Power Law (which has nothing to do with the Power Law of Movement, by the way) was only aimed at illustrating how the Environmental Feedback Function can depend on the nature of the perceptual function that determines the nature of the CV that is affected by the system’s Output.

The higher level perceptual functions define more complex CVs. For example, in the “What is Size” demo, my ability to control perimeter (P) is generally worse than my ability to control area (A). Since the disturbance amplitude and frequency is the same in both cases, I would say that perimeter is a higher level perception than area.

This is relevant to your question about what I was doing when I typed “please” in my request for information about how collective control of belief works. Using the same outputs (pressing the computer keys), I was controlling for typing configurations in a sequence that makes up a word that is part of a contingent sequence (program) that was a sentence aimed at finding something out, and doing it in a way that was in accord with my principles which include trying to be polite. So I was controlling a hierarchy of CVs using the same outputs (pressing on the keys of the computer keyboard). The highest level CV was to answer your questions about the feedback function between output and CV when I typed “please” and about what I was controlling for when I typed “please”.

I hope this helps clarify.

10

Yes, it’s good to take some time replying. This is a permanent record. I confess, I’ve now set aside four different essays at an explanation of collective control. As many a writer has said, it is often the case that I write to find out what I’m thinking. Best to find out what I’m thinking before I press the Reply button.

I’ll start here on common ground which you have offered.

Yes. This presupposes that the parties have developed perceptual input functions with the same properties. We assume this is the case for the lowest levels of the hierarchy. Evolution and genetics justify that assumption. At the lower levels, exceptions to commonality are considered to be pathologies, e.g. blindness or color-blindness, deafness or inability to hear above, say, 2000 Hz.

Perceptual input functions for the social phenomena of interest here are not genetically innate, they are the product of learning. An example from a discussion of ethics of CRISPR in The Code Breaker: Jennifer Doudna, Gene Editing, and the Future of the Human Race points at the difference. If we could remove the disruptive and non-functional kink in DNA that causes Huntington’s Disease from the human germ line, most folks say heck yeah. That’s a defect in the human genome. What about the bit of DNA that is expressed as dark skin? Eh, no. That’s a defect of society. What needs intervention there is the processes that create and maintain some of those learned perceptual input functions. The fear/alliance us/them aspects of racism almost certainly have an innate basis, but the higher-level systems for identifying and maintaining alliances and other social arrangements for personal, family, and ‘tribal’ security are learned.

Here’s another illustration of the difference:

collective-or-not
collective-or-not765×303 39.5 KB

All control is intentional but only collective control can be performative.

Consider a parallel to the TCV. To control being confrontational (or avoiding it) requires the confrontational one to perceive and control a variable that the ‘mark’ (the person confronted or not) is controlling, and to perceive ways of disturbing that CV. Their perceptual input functions have to be ‘the same’, just as in the TCV. (‘Same’ means sufficiently congruent or homologous for the purpose, sameness cannot be known exhaustively and is hugely improbable at the synaptic level.) Another parallel: to be performative the performer and the members of the audience must have the same perceptual input functions for some variables. A performance often draws attention to the arbitrariness of norms. Lenny Bruce was transgressive. Hippies were transgressive. Abbie Hoffman wrote a book titled Steal This Book. Trump is transgressive.

It’s not a difference between kinds of variables, those that are collective and those that are not. It’s a difference in how perceptual input functions are developed, and often but not always it’s a difference in how the collectively normalized variables which they create are controlled. We recognize a stop sign and know what we’re supposed to do when approaching it in a car. Our stopping the car (or at least slowing) is controlled by the individual driver. As I approach a stop sign and come to a complete stop before proceeding, the driver behind me may be annoyed at what she perceives as over-punctilious obedience, and anyway she’s late to work. Or as I slow down, glance both ways, and cruise through without stopping observers outside or in the passenger seat may go tsk tsk, and a bored cop may stop me and issue a traffic ticket.

  • The driving behavior is my idiosyncratic control process.
  • Its conformity to legal and informal social norms (which disagree with each other) is something that I control as an individual, with variable gain.
  • My observed conformity to or violation of a norm is collectively controlled. Every time someone else transgresses a norm in a way that is a disturbance to my control, the memory strengthens the reference signal for that norm, and vice versa for other people when I transgress a norm.
  • I control my (imagined) perception of how others perceive how I navigate the stop sign. I remember that others’ transgressions have disturbed me, I remember how others who perceived a transgression retaliated.

And so on.

You had a simple request: “explain how the collective control of belief works”. You hedged this with a deeply conventional “please”. More elaborately, “Could you please”. A brief detour now, if you please, into traces of the social processes (see above) by which generations of English users forged these conventional politeness idioms which we have inherited from them.

I said deeply conventional. Deeply, as in the early French origins of s’il vous plaît (still au courant in France), translated directly to English “If it please you”. From there, the it referring to the request disappeared, leaving if you please. Then the conditional if went away, leaving please you as a two-word gauze of counterfeit deference. Finally, the recipient of the request is completely effaced in the mid 1600s, leaving only please.

The secondary hedge could you is obviously not to be understood literally as a question about my abilities. It’s a similar bleaching of the semantic bones in would you, a literal request which seems to put the conditional if back into the transaction, except that there’s no if-then conditional. Another semantically bleached convention suggesting social deference which may only be pretended. Both of these politeness tags are often used with aggressive irony. Would you please shut that door!

I was reminded of the ’90s thriller Basic Instinct. When the character Catherine Tramell tells visiting detectives to “get the fuck out of here, please,” she sums it up: The word can brilliantly convey anger, irony, passive aggression, condescension, formality, or desperation—all without a hint of true politeness.
— “How Please Stopped Being Polite” (The Atlantic)

We can still retrieve if you please if you please, with its own social ambiguities. I hope that brief historical detour pleases you.

Perceptual input functions create these non-literal interpretations of words and phrases, their more literal meanings, and the words themselves. These perceptual input functions are in remarkable conformity from one native speaker of English to another. The processes by which these perceptual input functions are created, refined, and maintained, are examples of collective control. Their usage to transmit an intended meaning from one person to another also requires collective control.

For collective control, the participants need to perceive the same variables as the same. As above, ‘same’ means sufficiently congruent or homologous for the purpose.

Take your time. It’s a big ask.

[Next day, a brief addendum:

image
image903×300 153 KB

11

The TCV is a research method so one of the parties to the Test is typically a researcher who would have to develop ways to perceive the variables that are suspected of being under control by the system under study. If that system is a human, like the researcher, then it’s possible that the researcher’s own perceptual equipment can be used to perceive suspected controlled variables. If, however, the system is not human then the researcher will probably have to develop devices that can perceive what the system can perceive, as is done, for example, in the study of the variables controlled by bats when they navigate to prey.

I think there is no reason to believe that the perceptual functions for social phenomena (phenomena involving other people) differ from those for non-social phenomena or that they are not innate. We perceive social phenomena in terms of the same types of perceptual variables that are presumed to be the basis of all perceptual experience. These are the types of perceptual variables hypothesized by the PCT model: intensities, sensations, transitions, configurations, relationships,… programs, principles and system concepts. For example, we perceive the distance relationship between people in the same way as we perceive that relationship between inanimate objects; we perceive the program involved in bidding for the proper contract for a bridge partnership game in the same way as we perceive the program involved when an individual solves a math problem.

I think the evidence (the fact that all humans seem to perceive the world in terms of the same types of perceptual variables) is that all perceptual functions – of all types – are innate, though, per the work of the Plooijs, it seems that the higher level perceptual functions come “on-line” later than the lower level ones. It’s the arguments to these functions – the aspects of the environment that are the inputs to these functions – that are not innate. These arguments are inputs to the perceptual functions that are the outputs of lower level perceptual functions. So, for example, we learn to perceive words by learning which outputs from lower level configuration (phoneme) perceiving functions should be included as the inputs to the innate sequence perceiving input functions.

According to Chat-GPT “performative behavior refers to actions that are primarily intended to convey a certain image or message to others”. If “collective control” includes controlling that is done for the sake of other people then performative behavior is an example of collective control by definition. One question I have is: “Is it still collective control when you are just practicing the performance with no people present”?

Since you start by describing the behavior under consideration as “to control being confrontational” then it “requires” nothing more than the ability to control a perception of being confrontational. It certainly doesn’t require controlling a variable that the ‘mark’ considers confrontational. The perception of being confrontational is likely to be based on a guess about what the ‘mark’ is controlling for, but that guess doesn’t have to be correct. As soon as you described the behavior of the confrontational person as “To control being confrontational (or avoiding it)” you have produced a general model the behavior of the confrontational person. The behavior of the person being confronted exists in the model only as a possible disturbance to the confrontation perception being controlled by the person doing the confronting.

It’s that second thing that I’m interested in understanding. What is the difference between collectively controlled and non-collectively controlled variables in how they are controlled? I think you try to answer that question here:

This model of collective control does differ considerably from how control works in individuals, according to PCT. For one thing, “strengthened” reference signals are not part of PCT, nor is the idea of disturbances leading to strengthen reference signals. This sounds to me more like reinforcement theory dressed up in PCT terminology.

Yes, and if the explanation is anything like the above “observed disturbances, created by other people, leading to memories that strengthen reference signals” then I’m satisfied that collective control is not part of my approach to understanding social phenomena.

Well, I don’t understand how I was “hedging” my request for you to explain how the collective control of belief works, but I enjoyed your detour. Here’s an abbreviated quote of it:

This was quite interesting and, as I note below, consistent with my model of the emergence of differences in pronunciation observed by Labov.

I think these processes are control processes – which operate on the inputs to the perceptual functions and not on the functions themselves – like those that I used to explain the regional differences in pronunciation observed by Labov. The basic process involved in that model was imitation: controlling for pronouncing a diphthong in the same way as the people you interact with.

I think imitation is the oft neglected control process that corresponds to the “meme” that is responsible for many important social phenomena, not least of which being cultural differences in language, cuisine, music, literature, science, etc.

12

You are saying that your perceptual input function for the word “please” is genetically innate. To be consistent, you must also claim that a French person’s perceptual input function for «s’il vous plaît» is also innate.

13

I am saying that the perceptual functions that make it possible for you to perceive words, like “please”, or phrases, like “s’il vous plait”, are innate. Indeed, I believe that all perceptual functions are innate. I base this conclusion on several different observations: 1) The Nobel prize winning findings of Hubel & Wiesel (1962), which demonstrated the existence of “built in” perceptual functions called receptive fields that perceive increasingly complex aspects of the visual environment of the cat 2) the complexity of the neural implementation of a receptive field, which makes it seem implausible that such perceptual functions – especially the complex ones that perceive words and phrases – could be built in the short time it takes for humans to learn to perceive and control these aspects of the environment 3) the implausibility of the idea that all people would end up developing the same complex neural functions that produce the perception that all people can control, such as the neural perceptual functions that make possible the perceptions of, and, therefore, the control of, words 4) the fact that all humans seem to perceive the world in terms of the same types of perceptual variable and 5) the fact that the PCT hypothesis of a hierarchy of perceptual variables applies to all people, which implies that the same neural hierarchy of different types of perceptual functions is an innate characteristic of the nervous systems of all humans.

As I said in my previous post, I believe it is not the perceptual input function for the word “please” that is learned but, rather, the arguments (inputs) to that function that are learned. And once learned, the output of that perceptual function is stored as a reference for the perception “please”.

This might be easier to understand by looking at Bill’s model of a perceptual function for perceiving the word “juice” (Figures 11.2 and 11.3 in both editions of B:CP). I see the “built in” or innate perceptual function as the series of looped, thresholded neural connections (with a reset switch) that is the “raw material” for a sequence perceiving perceptual function. Figures 11.2 and 11.3 show a sequence perceiving perceptual function that has learned that the appropriate arguments (inputs) of this function are the phonemes “J”, “OO”, “S”, occurring in that order. When these inputs occur, the perceptual function will produce maximum output; the word “juice” will be clearly perceived. Other inputs, such as “J”, “OO”, “S”, “K”, that differ from “J”, “OO”, “S”, would produce far less than maximum output; what you would perceive is a muddled version of “juice”.

14

Yes, many of the perceptual input functions which create the lowest levels of the perceptual hierarchy are formed during embryonic development. Others are learned under genetic direction, e.g. functions for faces and for human vocalizations are pretty surely innate as a basis for learning the mother’s specific face and voice. Other perceptual input functions are created by learning: door, spoon, plate, book, car, truck, boat …. (My grandsons were all fans of cars and trucks at an astonishingly early age.) Yes, a capacity for perceiving configurations in general is innate, but the structures for perceiving specific kinds of configurations (or configurations of configurations, e.g. trucks have wheels) are learned.

The learned systems depend upon innate systems; the only tabula rasa is way back in something like RNA World at the beginning of evolutionary time, and that was bootstrapped too.

The topic here is not a distinction of innate vs. learned. Nor is the topic a distinction of ‘perceptual function’ vs. ‘perceptual input function’.

The subject matter of this Discourse topic concerns a distinction in the character of the environmental feedback function. The mathematics of physics suffices for the environmental feedback function in individual control of configurations, etc. in an asocial environment. In addition to the mathematics of physics, control of the ‘social’ variables that are of concern here requires the mathematics of control, because it must take into account the behavior of other control systems. It must do so because the control systems in a social environment take into account the behavior of the other control systems in their environment.

In simple cases, it’s basically a matter of the influences of other systems’ outputs on a controlled variable. Tug of war, and so on. In the more interesting cases, it is necessary for a control system to control a perception of another control system’s behavior, it’s intention, and it is necessary for a control system X to control a perception that another control system Y controls a perception of X’s intention.

Interaction games with small children ("pat a cake, pat a cake, baker’s man …), and interaction rituals (greetings, requests, introductions, …) impose mutually known limitations within which the other’s immediate intentions are known. Other kinds of interactions are structured to focus mutual attention on finding out what the other is controlling, as for example ask-answer, request-respond.

Certainly, the asocial control hierarchy adequate for a solipsistic universe is the basis which ‘makes possible’ control in a populated world. There may be innate systems for some of this. Heck, even nerve cells in vitro project and retract dendrites, make and release synaptic connections, as though reorganization were the default state. But surely most of it is learned.

Lying, deceit, trust, alliance, betrayal, intention itself, are not described with the mathematics of physics.

15

[quote=“bnhpct, post:14, topic:16110, full:true”]

As you can see, I offered my ideas about the innateness of perceptual functions because you asked if I really believed that the perceptual input function that produces the word “please” was innate. I gave you my reasons why I thought it was. Indeed, I also explained why I thought all perceptual input functions are innate. By the way, there is no distinction between the terms “perceptual function” and “perceptual input function”. In PCT, the terms are synonymous, the former being preferred to that latter because all perceptual functions are input functions so adding the word “input” to “perceptual function” is simply redundant.

In that case, the influence of the other system’s outputs on a controlled variable is simply a disturbance; the other system is not part of the feedback function. As I noted in my “Conflict” presentation at the last IAPCT meeting, a disturbance produced by another control system differs in important ways from a disturbance produces by an inanimate object. So you do have to take the disturbing system’s properties as a control system into account when analyzing its interaction with another control system.

Yes, when you try to control the behavior of another control system you are placing that control system in the feedback connection between your outputs and a variable you are controlling (which is something about that person’s behavior). This can get ugly, especially if you either ignore the fact that the system that makes up the feedback function is a control system or (just as bad or possibly even worse) or think that you know what that system wants.

Unless you are purposely trying to get the child to behave in a particular way, the its behavior is a disturbance to the variable you are controlling (patting hands). Same is true of the interaction rituals, unless you are purposely controlling for the person making the “ritual” response that you want them to make.

Yes, in this case the person is in the feedback connection between you and what you want (an answer to your question). This can get really ugly if the questioner is controlling with high gain for getting particular answers, in which case the questioner’s outputs can get rather torturous.

Of course learning is involved in developing the ability to control. And some of that learning surely involves learning to perceive the variable(s) to be controlled. I just think that the neural perceptual functions that compute the perceptual variables that an organism can and, often, must be able to control, are “built in” and what is learned (in the case of the higher level perceptual variables) are the inputs upon which these functions should operate.

16

In B:CP things like the Red Sox, the US Constitution, PCT, and the word “please” are perceptions. They aren’t innate. If all perceptions are innate, and these are not innate, then in your view they aren’t perceptions after all. What are they, if they aren’t perceptions?

In my view, a handshake is a controlled event perception. How would you describe the event called “a handshake”? Is it an innate perception? If it is not, then in your view it is not a perception, because in your view all perceptions are innate. What is a handshake, then?

Rather than imagine this in the abstract, observe examples of the phenomenon, if you can. For example, it’s possible to find or arrange a situation where it is natural and comfortable to face someone you don’t know or know only by sight, say “Hi, I’m Rick Marken” and stick out your hand.

Why is it comfortable and natural in some situations but not in others?

17

Actually, those are words that point to various states of perceptual variables. For example “Red Sox” could refer to the state of a perceptual variable that is a system concept called “baseball teams”. It could also refer to the states of lower level perceptions from which the system concept perception is derived, such as the event perception of Carlton Fisk coaxing his home run ball to stay fair in the 1976 World Series. (I experienced the Carlton Fisk event on TV in 1976 but I was on the first base line with my son in Dodger Stadium when Kirk Gibson hobbled to the plate to hit the game-winning home run in the bottom of the ninth in Game 1 of the 1988 World Series. That was one heck of an event perception!!)

Of course not!! What is innate are the neural networks that are the perceptual functions that can produce those perceptions. Actually, what the innate perceptual functions produce are perceptual VARIABLES. Perceptual functions produce the possibility of perceiving what impinges on our sensory receptors as remarkable events produced by members of system concepts called teams.

I agree. It is made up of configuration, transition and relationship perceptions.

I would describe it as grasping one hand with another at varying levels of tightness and moving the grasp up and down at varying amplitudes, frequencies and durations. It is an event perception that you learn to perceive by monitoring the outputs of your innate perceptual functions that are capable of perceiving that set of configurations, transitions and relationships as an event that can vary , from “firm” to Ko-Ko’s (of the Mikado) “All persons who in shaking hands, shake hands with you like that” to zero (no handshake).

There are many different things that have affected my comfort or lack therefore with handshaking. These different “things” are the higher level perceptual variables I am controlling. For example, during the pandemic I was controlling for not touching other people. So when I was in a situation were a handshake was expected, I set my reference for the handshake perception to zero; no handshake. In other situations where I saw the handshake was appropriate but would be done with someone who I disliked, I would set my reference for the “handshake” perception to “limp”. And when I met Tom Hanks at the intermission of a play that his daughter was in, I set my reference for the handshake perception to “firm” (but hopefully not too firm).

We vary our references for lower level perceptual variables as the means of controlling higher level perceptual variables. This has been demonstrated empirically many times. But to get a good feel for hierarchical perceptual control you have to start looking a perception in terms of perceptual variables, which are the outputs of perceptual functions that produce these variables based on variable inputs to these functions from lower level perceptual functions.

18

This began with the distinction between ‘factual beliefs’, which derive from and are verifiable by sensory input, and ‘credences’, which depend in significant measure on imagination. Never underestimate the ability of higher cortical functions to rationalize inconsistencies between these two kinds of ‘beliefs’. If pressed on the matter of his guarding his fields on Sunday, that Ethiopian farmer might say, “We are all capable of sinning. Why should I tempt my fellow Christian by leaving my animals unguarded on Sunday?”

In our conversation here, you made a request, “could you please explain” how some beliefs (‘credences’) are collectively controlled. We stepped back to your more basic request for any example of collective control. I offered your word “please” as an example. I asked you how you would model what you were controlling by adding “could you please” to the simple request, “explain”. You sketched a diagram indicating several levels of control in one person. I said a model of requesting or asking has to include the person being asked. You have not yet sketched a model including more than one person, although you have acknowledged that there exist cases where

This describes one characteristic of collective control.

The situation of asking and of “avoiding being confrontational” while asking and answering seems to be too complex to consider at this stage of our conversation. In a recent reply I asked how you would model a handshake. You replied:

This describes my right and left hands grasping each other and moving up and down. Nope. Your model has to include the other person.

Usually, one person initiates a handshake by putting their right hand out (not their left), and then the other responds in kind. Those are the first two steps in the event. It continues from there. The difference from a sequence like taking a sip of tea is that each successive step must be performed by another person. For example, after the move putting out your hand, if the other person does not move his hand toward yours there is no handshake.

To include in your model of a handshake higher-level purposes being controlled by means of shaking hands, take a simple case, like congratulating the daughter of a friend for receiving an MD diploma. No need to get into the motor control weeds, just the distinct steps in the event.

A somewhat similar example that makes the need for coordination between individuals even more obvious is the ‘high five’ slap of palms. Many a pair of people intending this have experienced colliding wrists, crossed arms, or whiffing misses. Skill at this can’t be gained by practice in solitude, since in its nature it requires coordination with another person. There is a key reference perception that helps a lot, and it’s not the perceived alignment of the palms. Instruction speeds the acquisition of this insight, and that of course requires other people.

Handshakes and high fives are means of controlling the perception “we can coordinate our actions to control the same variable.” On that basis the participants can direct their attention to other variables which are best controlled by more than one person in coordination.

Some of these, at very low levels of the hierarchy, are innate. Learned perceptual functions at higher levels make use of innate functions as means of controlling their inputs.

They are all built in; that is, they are all built. The sensors at the periphery of the system are built by genetically directed developmental processes that produce specialized cells in functional relationships to one another. Likewise, genetically directed developmental processes organize control functions at the lowest levels which vary those inputs by interacting with the environment. The simplest way interactively to vary them is by moving, e.g. directing the gaze, moving surfaces of skin with their pressure sensors, turning the head and ears relative to apparent sources of sound, and so on. (This is why Bill affirmed that the expectation that the sun will rise in the east and not elsewhere is an example of control.) And all perceptual functions include input signals which can be traced down the hierarchy ultimately to intensity receptors, except where they depend on imagined input, e.g. QAnon believers perceiving what Q is telling them or Ethiopian farmers perceiving that the leopard is a Christian.

If interaction with the environment is restricted the development of neural and chemical control systems is incomplete and functionally defective (Ardiel & Rankin 2010), If the foundation of intensity inputs for a sensory modality is missing, even innate sensory input systems that would depend upon intensity signals diminish in size and function and may be reorganized to serve other functions (Merabet & Pascual-Leone 2010). And the plasticity of the mature brain, after the recognized stages of genetically directed development, is well established. No one would claim that DNA/RNA genetic and epigenetic functions have no role in this. So all perceptual functions are ‘built’ and the building in every case (except probably the intensity receptor cells) requires interaction with the environment. At every level, perceptual functions impose a many-one relation on perceptual signals. Even intensity receptors are affected by many physical variables (e.g. photons, sound waves) but emit just one signal.

Ardiel EL, Rankin CH. The importance of touch in development. Paediatr Child Health. 2010 Mar;15(3):153-6. doi: 10.1093/pch/15.3.153. PMID: 21358895; PMCID: PMC2865952.

Merabet LB, Pascual-Leone A. Neural reorganization following sensory loss: the opportunity of change. Nat Rev Neurosci. 2010 Jan;11(1):44-52. doi: 10.1038/nrn2758. Epub 2009 Nov 25. PMID: 19935836; PMCID: PMC3898172.

19

Yes. And I would argue that it is not very common characteristic of collective control. A far more common characteristic of collective control is when a person’s outputs are a disturbance to the variable(s) other people are controlling. This is the characteristic of collective control that is demonstrated in Kent’s model. In that model, the outputs of each person in the collective are a disturbance to the variable controlled by each of the others. No person in that collective is part of the environmental component of the feedback connection between another person’s output and the variable he or she is trying to control.

Similarly, after putting out your hand to lift the teacup, if the cup does not lift (because it’s stuck in some dried tea), there is no sip (just a lot of tea in your face when it finally becomes unstuck). The point here is simply that in both cases the failure to produce the intended result (handshake or sip) is a result of the same control phenomenon: an insuperable disturbance.

You only need to model the source of disturbance in each case if you want to know why the insuperable disturbance occurred (setting a reference for “no hand movement” in the handshake case and evaporation in the tea sip case.

Yes, higher level perceptual functions “make use” of perceptual variables that are the outputs lower-level perceptual functions. At least that’s the way it works in the PCT model. What makes me think these higher-level functions are innate is the fact all people learn to “make use” of those lower level perceptual variables in the same way!

For example, all people seem to be able to control a perceptual variable called “honesty”. They might control this variable relative to different reference specifications in different situations. But virtually all people seem to know what honesty is; they are able to perceive and control the principle of being honest.

Honesty is a perceptual variable that must be computed by a perceptual function that makes use of many lower-level perceptual variables. The function itself must be incredibly complex, which makes me think that it is unlikely that random reorganizations of neural networks, even if constrained by connections to all the relevant lower level perceptual variable, could all come up with a function that is able to perceive principles like “honesty”, “fairness”, “cooperativeness”, etc.

But, as you say, the innateness (or non-innateness) of perceptual function is not really germane to our discussion of collective control. What is germane (at least to me) is the answer to my original question “How are beliefs collectively controlled? I don’t believe I’ve ever gotten a clear answer to that.

But this discussion has not been a waste of time for me. Indeed, it led me to realize that there are two very different ways individual organisms (controllers) can be involved in “collective control” : 1) as disturbances to the variables controlled by other controllers and 2) as components of the feedback functions of other controllers. I’d be interested in what you (and others involved in the study of collective control) think about this distinction.

20

Here, you are mistaken, just as you are mistaken in your persistent belief that Kent’s 1993 demo of conflictive control is all there is to say about collective control and all that Kent has presented about collective control. A much larger proportion of collective control concerns conventionalized ways to avoid conflict so that the even larger proportion, more ad hoc forms of collective control, can proceed while minimizing or avoiding conflict.

You still haven’t indicated how you would model a handshake. Other than conflict, you have no models in which “other people are part of the environmental component of the feedback connection between your output and the variable(s) you are controlling”.

In LCS IV, the Handbook vol. 1, read Kent’s chapter and Martin’s chapter, and read Martin’s 4-volume book. Practically all of it is about the latter kind of collective control.

Even if it had been, it would not have been for me, because you are not the only reader.

Addendum, next day:
Bill’s 1995 conversation with Bruce Abbott about contrafactual beliefs is relevant. If factual data can’t be rationalized to fit then either denial or reorganization follows. I suppose denial is an extreme form of rationalization.