Perception and Reality

CSGnet Archive CSG2006
1

[From Rick Marken (2006.08.22.0900)]

Bill Powers (2006.08.21.1515 mdt)--

Rick Marken (2006.08.21.1600)]

I agree that "rights" (to the extent that they are perceptions and not references -- see below) are not objective, in the sense that there is nothing in our models of external reality that corresponds to what we perceive as "rights". But isn't that true of all perceptions?

I'm realizing that "rights," like "oughts", doesn't really refer to the things we have rights about, or ought to do. It's asserting something else about whatever it is...

So my conclusion is that a right exists only by agreement, and only if there is some means of enforcing it. The only actual rights are those we can persuade others to grant us, enough others so that the rights would be very hard to take away from us. It helps, of course, if they have reason to want the same rights.

Your discussion of rights is very interesting, and I agree with it. But it doesn't really address my question, which was about the relationship between perceptual variables and our models of external reality. I am wondering whether you think any perceptual variable can be considered objective in the sense that it corresponds to something in our models of external reality. I'm just interested in hearing a discussion of the relationship between perception and reality in PCT. I understand that many (most) perceptions, such as the taste of a milk shake, are constructions based on sensed aspects of external reality. There is no milk shake taste out there (according to our models); just the molecules that elicit various taste sensations, which are combined to produce the taste "milk shake".

In your earlier discussion you implied that some perceptions (like "rights", assuming they are perceptions) are less objective than others because these perceptions don't correspond to anything in our models of external reality. I was just asking whether some perceptions are more objective than others. For example, color could be considered an objective perception because changes in color correspond to changes in what we model as changes in wavelength of light. But we also know that a color perception that corresponds to a single wavelength can also result from the appropriate combination of 2 or 3 wavelengths. So it seems to me that color is no more objective, in terms of correspondence between perception and model of external reality, than the perception of a principle like "the right to bear arms". What do you think?

I had been arguing earlier for the idea that perception can be objective in the sense that people can agree (often after much discussion, pointing and whatever) that they are having the same perception (such as "the meter reading is 25 amps"). And I think in most of these cases they are having the same perception. So I think this kind of "inter-observer agreement" objectivity is exists but that the "correspondence of perception to external reality" type objectivity probably doesn't.. What do you think?

Best

Rick

···

---
Richard S. Marken Consulting
marken@mindreadings.com
Home 310 474-0313
Cell 310 729-1400

2

[From Rick Marken (2006.08.22.2220)]

Bill Powers (2006.08.22.1130 MDT)--

When the environment has more degrees of freedom than our perceptions have, as I think is the case, when you and I control "the same thing," our perceptions can be completely different from each other. You can see this in the ThreeSystemsC.exe program, which shows that a three-dimensional control system can be blind to rotations in the environment, while still keeping its perceptions matching its reference levels.

OK. I think I get your point (finally). I did an experiment that may speak to this point (it's described at the end of the "Degrees of freedom in behavior" paper in _Mind Readings_). In a two dimensional tracking task I suggested that there are at least two possibilities for the perceptual variables being controlled: the subject could be controlling a Cartesian or a polar representation of the situation. The Cartesian model controls just as well as a polar model. And the behavior of the two models is difficult distinguish, at least in terms of behavior of the controlled variable. But you can distinguish the two models when an abrupt disturbance is introduced in one (say, x) dimension only. The Cartesian model reacts only in the x dimension while the polar model reacts in both dimensions (because the disturbance to the x dimension is a disturbance to both perceptions -- angle and radial distance -- controlled by the polar model. Since people react only in the x dimension (at least, the few people I tested -- all my family members, so we may all be hereditary Cartesians) I concluded that the perceptions controlled in this two dimensional tracking task was a Cartesian rather than a polar representation of cursor position.

I suppose there are more than just these two ways -- Cartesian and polar -- of perceiving cursor position in a two dimensional tracking task. But is this Cartesian/polar difference what you are talking about when you say: When you and I control "the same thing," our perceptions can be completely different from each other? And, if so, doesn't the approach I describe (comparing human behavior to that of a model controlling different perceptions of the same situation) an approach to at least narrowing the possibilities regarding how we might be perceiving the same thing?

Best

Rick

···

---
Richard S. Marken Consulting
marken@mindreadings.com
Home 310 474-0313
Cell 310 729-1400

3

[From Bill Powers (2006.08.23.0030 MDT)]

Rick Marken (2006.08.22.2220)--

I suppose there are more than just these two ways -- Cartesian and polar -- of perceiving cursor position in a two dimensional tracking task. But is this Cartesian/polar difference what you are talking about when you say: When you and I control "the same thing," our perceptions can be completely different from each other? And, if so, doesn't the approach I describe (comparing human behavior to that of a model controlling different perceptions of the same situation) an approach to at least narrowing the possibilities regarding how we might be perceiving the same thing?

That's one possibility, yes. But I think there's even more to it. Both Cartesian and polar coordinate systems (as you used them) have two dimensions. In the 2-D tracking task, all the degrees of freedom are used up, so it's a determinate system. But suppose that what you're looking at in both cases is really a 3-D world projected onto a 2-D plane, as in Flatland. Disturbances in the third dimension would be unresisted. While two people doing this might agree that they're controlling the same thing, in fact they could be looking at very different environments which don't coincide at all when the 2-D representations are the same.

In the ThreeSystemsC program, the situation is a little different -- the set of three control systems is looking at an environment with, it turns out, six degrees of freedom, and each time the input weights are shuffled it sees a different 3-D projection of that environmment. But because its output functions reorganize, it continues to control the same 3-D perceptual representation. Two of these systems sharing the same environment could experience 3-D worlds and control them, yet they would come into conflict with each other because they are controlling different 3D representations of the same world -- different Cartesian representations.

However, even if these systems resolved the conflict so they actually were using the same input weightings, and reorganized to use the same output weightings, they would still not be controlling exactly what is in the external world. There are three angles of rotation that neither system can sense, and therefore the behavior of the environment is not congruent with the behavior of the perceptions in either system.

I hang my hopes on the fact that when we introduce different viewing angles and more control systems, it may prove to be that the multiple control systems, after reorganization, will converge not only on self-consistent perceptions, but on mutually-consistent perceptions that are also true representations of the external variables. That would be a wondrous discovery, but I don't know how to get there from here, or even if it's possible. It is still possible that each of us lives in a unique world, and that we simply work out translations back and forth between us that in principle can eliminate all the differences we can see, but by no means all the differences there are.

Best,

Bill P.

.

4

[Martin Taylor 2006.08.23.10.11]

[From Rick Marken (2006.08.22.2220)]

Bill Powers (2006.08.22.1130 MDT)--

When the environment has more degrees of freedom than our perceptions have, as I think is the case, when you and I control "the same thing," our perceptions can be completely different from each other. You can see this in the ThreeSystemsC.exe program, which shows that a three-dimensional control system can be blind to rotations in the environment, while still keeping its perceptions matching its reference levels.

OK. I think I get your point (finally). I did an experiment that may speak to this point (it's described at the end of the "Degrees of freedom in behavior" paper in _Mind Readings_). In a two dimensional tracking task I suggested that there are at least two possibilities for the perceptual variables being controlled: the subject could be controlling a Cartesian or a polar representation of the situation. The Cartesian model controls just as well as a polar model. And the behavior of the two models is difficult distinguish, at least in terms of behavior of the controlled variable. But you can distinguish the two models when an abrupt disturbance is introduced in one (say, x) dimension only. The Cartesian model reacts only in the x dimension while the polar model reacts in both dimensions (because the disturbance to the x dimension is a disturbance to both perceptions -- angle and radial distance -- controlled by the polar model. Since people react only in the x dimension (at least, the few people I tested -- all my family members, so we may all be hereditary Cartesians) I concluded that the perceptions controlled in this two dimensional tracking task was a Cartesian rather than a polar representation of cursor position.

There's another possible interpretation of this result. This is based on thinking about the "Etch-a-sketch" which you are probably old enough to remember. It was a toy that allowed you to draw on an erasable surface, using two knobs, one each for the x and y positions of the drawing point.
Even if you were controlling for perceiving yourself to have drawn a circle, it was very hard not to draw staircases or straightish diagonal lines.

Imagine if someone had jogged the drawing point's location in x; correction would have been easy. But if the disturbance had been in r in a diagonal direction (r being your main controlled variable), I think you nevertheless would have made a corrective move in x and then y, rater than in r at the proper theta.

  So an interpretation of your result could be that even if the controlled perceptions were represented as polar coordiantes, the output and environmental feedback paths might at some point have been separated into individual paths for x and for y.

I'm not saying that this was the case in the actual experiment; mouse movements are pretty equally easy in all directions. But it's a consideration when using results like that to argue that they are necessarily due to the perceptual representation.

I suppose there are more than just these two ways -- Cartesian and polar -- of perceiving cursor position in a two dimensional tracking task. But is this Cartesian/polar difference what you are talking about when you say: When you and I control "the same thing," our perceptions can be completely different from each other? And, if so, doesn't the approach I describe (comparing human behavior to that of a model controlling different perceptions of the same situation) an approach to at least narrowing the possibilities regarding how we might be perceiving the same thing?

I think I would say that the experiment biases the likelihoods (not probabilities) of different kinds of explanation. But that's all that evidence can ever do. [Why likelihood? Because probabilities must sum to unity, and if you include the class "all others not yet taken into account", that class has probability within epsilon of 1.0 in most situations. Likelihoods don't neet to sum to unity.]

Martin

5

[From Rick Marken (2006.08.23.1120)]

Bill Powers (2006.08.23.0030 MDT)--

That's one possibility, yes. But I think there's even more to it. Both Cartesian and polar coordinate systems (as you used them) have two dimensions. In the 2-D tracking task, all the degrees of freedom are used up, so it's a determinate system. But suppose that what you're looking at in both cases is really a 3-D world projected onto a 2-D plane, as in Flatland. Disturbances in the third dimension would be unresisted. While two people doing this might agree that they're controlling the same thing, in fact they could be looking at very different environments which don't coincide at all when the 2-D representations are the same.

OK, I think I get it. Is this it: both people could be controlling for "square", for example, but in one case the environment is a cube and in the other it's a pyramid? Or is it more like the 3 D environment is the same for both people but each person is controlling a different 2 D aspect of that environment?

In the ThreeSystemsC program, the situation is a little different -- the set of three control systems is looking at an environment with, it turns out, six degrees of freedom, and each time the input weights are shuffled it sees a different 3-D projection of that environmment. But because its output functions reorganize, it continues to control the same 3-D perceptual representation.

Now I'm lost. So three control systems control three different weightings of the six input degrees of freedom? These three different perceptions are then the 3 dimensions (3 D projection) being controlled? Is that right? And when the input weightings of each system are changed (so that three new perceptual variables -- dimensions -- are created) the outputs reorganize so that these three variables are kept under control? Is that it?

Two of these systems sharing the same environment could experience 3-D worlds and control them, yet they would come into conflict with each other because they are controlling different 3D representations of the same world -- different Cartesian representations.

Now I'm lost. Are the two systems you refer to individual control systems of systems consisting of three individual control systems, the variables controlled by each system being the 3 dimensions controlled? I'm not understanding how systems would come into conflict because they are controlling _different_ 3 D representations of the same world. Wouldn't there be conflict only if these systems were controlling the _same_ (or similar) 3 D representations of the same environmental degrees of freedom?

However, even if these systems resolved the conflict so they actually were using the same input weightings, and reorganized to use the same output weightings,

How can the systems resolve a conflict by using the _same_ input weightings?

they would still not be controlling exactly what is in the external world. There are three angles of rotation that neither system can sense, and therefore the behavior of the environment is not congruent with the behavior of the perceptions in either system.

Now I'm completely lost. This may be beyond my meager intellectual capabilities. I don't quite get the details but the overall message of your comments seems to conflict with some of my general understandings of a couple features of PCT:

1. I thought one way conflict occurs is when two (or more) control systems control the same (or a similar) perceptual representation of the same environment. Conflict can occur when control systems control different perceptual representations of the same environment. But in that case the conflict does not occur _because_ the systems are controlling different perceptions.

2. The test for the controlled variable is a way to determine what perceptual variable a system is controlling. so it can be seen as a way for one person (the experimenter) to have a perception that is equivalent to what another (the subject) is perceiving and controlling.

Are these two points wrong in some fundamental way? If so, I know how I'm going to spend my retirement: learning PCT!

Best

Rick

PS. I agree with Fred Nickols: your posts on "rights" and looking for a "better way" to control for "rights" other than through the use of force were wonderful. I rarely wear a hat (to Linda's chagrin) but I put one on this morning so I could take it off to you, too.

I hang my hopes on the fact that when we introduce different viewing angles and more control systems, it may prove to be that the multiple control systems, after reorganization, will converge not only on self-consistent perceptions, but on mutually-consistent perceptions that are also true representations of the external variables. That would be a wondrous discovery, but I don't know how to get there from here, or even if it's possible. It is still possible that each of us lives in a unique world, and that we simply work out translations back and forth between us that in principle can eliminate all the differences we can see, but by no means all the differences there are.

Best,

Bill P.

Richard S. Marken Consulting
marken@mindreadings.com
Home 310 474-0313
Cell 310 729-1400

6

[From Rick Marken (2006.08.23.1335)]

Martin Taylor (2006.08.23.10.11)--

Rick Marken (2006.08.22.2220)]

Since people react only in the x dimension (at least, the few people I tested -- all my family members, so we may all be hereditary Cartesians) I concluded that the perceptions controlled in this two dimensional tracking task was a Cartesian rather than a polar representation of cursor position.

There's another possible interpretation of this result. This is based on thinking about the "Etch-a-sketch" which you are probably old enough to remember.

I knew "Etch-a-sketch". "Etch-a-sketch" was a good friend of mine. My results were no ""Etch-a-sketch" results;-)

Imagine if someone had jogged the drawing point's location in x; correction would have been easy. But if the disturbance had been in r in a diagonal direction (r being your main controlled variable), I think you nevertheless would have made a corrective move in x and then y, rater than in r at the proper theta.

Rather than go into an unnecessarily long explanation of why this "Etch-s-sketch" explanation is probably not correct, all I will say that my conclusions were based on the fact that subject behavior resembled the behavior of one model (Cartesian) rather than another (polar) . I'm sure that, if a disturbance had been made on the diagonal rather than in only the x dimension, the x and y handle movements of the person would have looked like the x and y handle movements for the Cartesian model. And these corrective handle movements, for both person and model, would occur in the x and y dimensions simultaneously.

Best

Rick

···

---
Richard S. Marken Consulting
marken@mindreadings.com
Home 310 474-0313
Cell 310 729-1400

7

[Martin Taylor 2006.08.23.16.55]

[From Rick Marken (2006.08.23.1335)]

Martin Taylor (2006.08.23.10.11)--

Rick Marken (2006.08.22.2220)]

Since people react only in the x dimension (at least, the few people I tested -- all my family members, so we may all be hereditary Cartesians) I concluded that the perceptions controlled in this two dimensional tracking task was a Cartesian rather than a polar representation of cursor position.

There's another possible interpretation of this result. This is based on thinking about the "Etch-a-sketch" which you are probably old enough to remember.

I knew "Etch-a-sketch". "Etch-a-sketch" was a good friend of mine. My results were no ""Etch-a-sketch" results;-)

Imagine if someone had jogged the drawing point's location in x; correction would have been easy. But if the disturbance had been in r in a diagonal direction (r being your main controlled variable), I think you nevertheless would have made a corrective move in x and then y, rater than in r at the proper theta.

Rather than go into an unnecessarily long explanation of why this "Etch-s-sketch" explanation is probably not correct, all I will say that my conclusions were based on the fact that subject behavior resembled the behavior of one model (Cartesian) rather than another (polar) .

I think you missed my point. I believe I said (and I apologise if I didn't) that I thought Etch-a-Sketch was not a correct explanation for_your_ experimental results, because you used a mouse as your action device, and a mouse has no preferred directions. I was pointing out that as a generic approach, you can run into interpretation problems if you don't ensure that the output and feedback paths are unbiased as to the coordinates (in general, the dimensions and functions) being compared.

I suggested that had you used Etch-a-Sketch knobs to influence the cursor, and the subject were really controlling a polar-coordinate perception, you would have got the results you did get. I know that's silly in the case at hand, but it's not silly when you get to testing more complex percpeptions at higher levels.

I'm sure that, if a disturbance had been made on the diagonal rather than in only the x dimension, the x and y handle movements of the person would have looked like the x and y handle movements for the Cartesian model. And these corrective handle movements, for both person and model, would occur in the x and y dimensions simultaneously.

Yes, but I bet they wouldn't b a smooth diagonal if you did have Etch-a-Sketch controls. I suspect then that the corrective trace would look something like

···

__
   \
    \_
      >
      \ _
       \__|

Mind you, that's still Cartesian, but I think the result would be like that even if the controlled perceptions were polar.

The point is that you, the experimenter, perceive the effect on the environment caused by the subject's actions, and if the only bit of the environment you look at is what you presume to be the data input to the subject's sensors, it's ambiguous where in the backtrack any particular effects are happening. The subject might be controlling for any two perceptions that are functions of your (experimenter's) x and y variables, but if the output functions or the output apparatus or the environmental feeback paths somewhere include a place where the effects of the two error values interact, that interaction is going to affect what you (experimenter) see the subject as controlling for.

It's no big deal for your particular experiment, but it is another riff on what Bill P has been saying about it being essentially impossible to determine exactly what someone else is perceiving.

Martin

Martin

8

[Jim Dundon 08.25.2006.1200edst]

Rick, You said

Your discussion of rights is very interesting, and I agree with it. But
it doesn't really address my question, which was about the relationship
between perceptual variables and our models of external reality. I am
wondering whether you think any perceptual variable can be considered
objective in the sense that it corresponds to something in our models
of external reality. I'm just interested in hearing a discussion of
the relationship between perception and reality in PCT

Jim

Rick, when you "wonder whether or not something can be considered objective" are you not asking whether or not it can be considered an object? That is, can we objectivise it? In the English-language we think of a great many intangable concepts as things. Our first experience as children with words taught us that we apply names to things in our environment. Having learned that process we learn words for feelings or mental processes, words like love, thought, in the English language they are utilized as things in the sentence and so take on the aspect of an object. There are great many words not thinged which could be thinged if we wanted to take the liberty, such as the word be. If we reason that we shall use the word be to represent a characteristic of existence of conditions and then stipulate that all existing conditions be noted as bes we can now say about certain conditions that they constitute a be. A be is now an abstract objectivised concept. And we could proceed to carry on an intelligent conversation about various bes that each of us knows about. If we were to speak about the process of naming something a be we could speak of the act of being [that is naming our act], that is we could say let us be that process with the word in each of the foregoing meaning a naming act. In other words we be'd them, we named them bes. Having done all this work we could then proceed to talk this way by agreement, with comfort, confident that we would understand each other which is pretty much the purpose of language. So are you not asking for agreement and seeking common terminology? And yet you ask the question as though it could yield an absolute nature of its own.

There is probably no concept which we cannot think of as an object and no object that we have not named and therefore no concept which we cannot think of as a product of our naming of our experiencing. We create much of our experiencing with our application of namings. Can we completely sort out external reality from the internal, I doubt it. The use of words requires a certain measure of predictability and is predictability that we seek in languaging. Our existence depends on it.

Your investigation into the relationship between perception and reality in PCT uses the words relationship, perception, reality and PCT as objects. You have already made them objects you have already objectivised them.

You said:

I was just asking whether some perceptions are more
objective than others. For example, color could be considered an
objective perception because changes in color correspond to changes in
what we model as changes in wavelength of light.

Jim:

Are you not asking here whether or not a perception should be considered less valid if we cannot measure something in scientific terms, scientific here meaning narrowed parameters and purified components?

I have a more liberal view of the word scientific and so a more liberal view of the word valid. I believe science begins with the child's first words. I view each word as the naming of an experiencing, the naming of a unit of experience,[thanks Bill] the naming of a percept. But then as a child not narrowed the parameters, as he not culled from among the variables in his environment a few with which to tag, make an object of, a particular phonetic tag?

There are several realities here one of which is the nature of science another is at what level do we name and objectivise. Is PCT not a perception? You are then asking what is the relationship between perception [generaly] and_the_perception named PCT. The answer to that of course is PCT is in the set of perceptions. It looks to me therefore like PCT is a reality. The question is, "is it the all of reality?". Does it encompass all realities?

Best

Jim D

9

[From Rick Marken (2006.08.25.1220)]

Jim Dundon (08.25.2006.1200edst)

Rick, when you "wonder whether or not something can be considered objective" are you not asking whether or not it can be considered an object?

No, I was asking (or, at least, trying to ask, about objectivity in the sense of the correspondence of perception to external reality. I myself don't think any perception is more objective in this sense than any other.

You said:

I was just asking whether some perceptions are more objective than others. For example, color could be considered an objective perception because changes in color correspond to changes in what we model as changes in wavelength of light.

Jim:

Are you not asking here whether or not a perception should be considered less valid if we cannot measure something in scientific terms, scientific here meaning narrowed parameters and purified components?

No. I was asking whether some perceptions correspond to external reality (as we know it via our models, which, of course, is the only way we can know external reality) better than others.

Is PCT not a perception?

Yes. Everything is perception.

You are then asking what is the relationship between perception [generaly] and_the_perception named PCT.

No. I was asking about the relationship between perception and the presumed external reality on which our perceptual experience is presumably based.

It looks to me therefore like PCT is a reality.

I think you're using thee term "reality" in a different way than I tend to use it. For me, reality is an assumption; it's a mental model of what is on the "other side" of my perceptual experience. I think of PCT as a set of perceptions -- a set of imaginings -- that aim to explain the reality that underlies our perceptions of the behavior of living systems, just as physics is a set of imaginings that aims to explain the reality that underlies our perceptions of non-living systems.

Best

RIck

···

---
Richard S. Marken Consulting
marken@mindreadings.com
Home 310 474-0313
Cell 310 729-1400