Ethogram comment on Bruce's Researchgate PCT project

CSGnet Archive CSG2017
1

[Martin Taylor 2017.09.18.11.38]

This is a "testing" message to see whether I can post a shorter message that nevertheless has some content.

I hope some of you are "following" Bruce Nevin's PCT project on Researchgate <https://www.researchgate.net/project/Perceptual-Control-Theory-PCT>. If you are, you may have noticed comments by a person called Brad Jesness, advertising his "Ethogram" ideas for a general theory of psychology. Rick has responded to them, but this message is unrelated to the content of those comments and responses.

Jesness claims to be supportive of PCT, but the scientific method used is opposite to that of PCT. He asks that we read his 1985 draft of a book. I have started to do so, from both ends, as he in fact recommends, and I think I have an idea about where he is coming from.

PCT is what I would call a "functional" theory. It starts with the concept that there exists a mechanism (an Elementary Control Unit that can be organized into a hierarchy of similar units), and then asks how well that mechanism accounts for data. Probing (disturbing) hypothesized controlled variables is a primary experimental technique.

The word "experiment" seems antithetical to what Jesness proposes. His ideal procedure is purely observational. As is true in astronomy, he would not influence the subjects of his research at all, if he could avoid doing so. He would infer what happens inside the subject's skin, where observation is impossible -- or was in 1985, perhaps -- from correlations and contradictions that are or are not predicted and are or are not observed.

There's nothing wrong in principle with an observational science. It's just a lot harder to discover the functional underlay to what you observe than it is if you can probe and do experiments. So one often winds up with a purely descriptive statement of what you observe, but in more concise terms. That's just what the network nodes do in "Deep Learning". A complaint often raised about "Deep Learning" is that you can seldom find "Why" this description is as it is. PCT starts out the other way. It presents an answer to "Why" and then asks "What". The method is more powerful and the result more intelligible, but the results of the two approaches should not conflict.

Martin

2

[From Rick Marken (2017.09.18.1208)]

···

Martin Taylor (2017.09.18.11.38)

MT: This is a “testing” message to see whether I can post a shorter message that nevertheless has some content.

RM: As I told you privately, I got it and it posted.Â

MT: I hope some of you are “following” Bruce Nevin’s PCT project on Researchgate <https://www.researchgate.net/project/Perceptual-Control-Theory-PCT>. If you are, you may have noticed comments by a person called Brad Jesness, advertising his “Ethogram” ideas for a general theory of psychology. Rick has responded to them, but this message is unrelated to the content of those comments and responses.

MT: Jesness claims to be supportive of PCT, but the scientific method used is opposite to that of PCT. He asks that we read his 1985 draft of a book. I have started to do so, from both ends, as he in fact recommends, and I think I have an idea about where he is coming from.

MT: PCT is what I would call a “functional” theory. It starts with the concept that there exists a mechanism (an Elementary Control Unit that can be organized into a hierarchy of similar units), and then asks how well that mechanism accounts for data. Probing (disturbing) hypothesized controlled variables is a primary experimental technique.

RM: Good point. I think that’s the essential problem we have in discussions with Jesness and other  fans (and non-fans) of PCT. PCT is a functional (dare I say it, mechanistic) theory of behavior the essence of which is trying to explain behavior using working models whose behavior can be be compared to the behavior to be explained, just like in real sciences. It’s difficult, if not impossible, to have productive discussions about the comparative merits of theories of behavior with people (like Jesness and many of those in the SlateStarCodex discussion) who don’t take the modeling approach to understanding behavior.

Best

Rick

The word “experiment” seems antithetical to what Jesness proposes. His ideal procedure is purely observational. As is true in astronomy, he would not influence the subjects of his research at all, if he could avoid doing so. He would infer what happens inside the subject’s skin, where observation is impossible – or was in 1985, perhaps – from correlations and contradictions that are or are not predicted and are or are not observed.

There’s nothing wrong in principle with an observational science. It’s just a lot harder to discover the functional underlay to what you observe than it is if you can probe and do experiments. So one often winds up with a purely descriptive statement of what you observe, but in more concise terms. That’s just what the network nodes do in “Deep Learning”. A complaint often raised about “Deep Learning” is that you can seldom find “Why” this description is as it is. PCT starts out the other way. It presents an answer to “Why” and then asks “What”. The method is more powerful and the result more intelligible, but the results of the two approaches should not conflict.

Martin

–
Richard S. MarkenÂ

"Perfection is achieved not when you have nothing more to add, but when you
have nothing left to take away.�
                --Antoine de Saint-Exupery

3

[From Bruce Nevin (2017.09.18.20:53 ET)]

Rick, I saw your comment on ResearchGate (RG) asking Jesness to clarify, and I saw where he responded “just read all this wonderful stuff I’ve written.” Fair enough on the surface, we say pretty much the same to PCT newbies. But It seems pretty clear to me (my thoroughly subjective perceptions :wink:  that he sees himself as an elder in a supportive role to up-and-coming students and researchers. Will he be willing to relinquish the pleasures of that self-image to become an up-and-coming learner of PCT?

Martin, I’m glad you’ve taken him up on it and looked into his book.

“Ethogram” was a new one on me. Wikipedia to the rescue:

An ethogram is a catalogue or inventory of behaviours or actions exhibited by an animal used in ethology.

The behaviours in an ethogram are usually defined to be mutually exclusive and objective, avoiding subjectivity and functional inference as to their possible purpose.

On this basis I, too, replied to him with a comment, but I don’t see it on RG. Probably just as well, as I was quite short with him, saying that it would be difficult for a conception to be more orthogonally different from PCT. Not very diplomatic at all, at all.Â

···

On Mon, Sep 18, 2017 at 3:07 PM, Richard Marken rsmarken@gmail.com wrote:

[From Rick Marken (2017.09.18.1208)]

Martin Taylor (2017.09.18.11.38)

MT: This is a “testing” message to see whether I can post a shorter message that nevertheless has some content.

RM: As I told you privately, I got it and it posted.Â

MT: I hope some of you are “following” Bruce Nevin’s PCT project on Researchgate <https://www.researchgate.net/project/Perceptual-Control-Theory-PCT>. If you are, you may have noticed comments by a person called Brad Jesness, advertising his “Ethogram” ideas for a general theory of psychology. Rick has responded to them, but this message is unrelated to the content of those comments and responses.

MT: Jesness claims to be supportive of PCT, but the scientific method used is opposite to that of PCT. He asks that we read his 1985 draft of a book. I have started to do so, from both ends, as he in fact recommends, and I think I have an idea about where he is coming from.

MT: PCT is what I would call a “functional” theory. It starts with the concept that there exists a mechanism (an Elementary Control Unit that can be organized into a hierarchy of similar units), and then asks how well that mechanism accounts for data. Probing (disturbing) hypothesized controlled variables is a primary experimental technique.

RM: Good point. I think that’s the essential problem we have in discussions with Jesness and other  fans (and non-fans) of PCT. PCT is a functional (dare I say it, mechanistic) theory of behavior the essence of which is trying to explain behavior using working models whose behavior can be be compared to the behavior to be explained, just like in real sciences. It’s difficult, if not impossible, to have productive discussions about the comparative merits of theories of behavior with people (like Jesness and many of those in the SlateStarCodex discussion) who don’t take the modeling approach to understanding behavior.

Best

Rick

The word “experiment” seems antithetical to what Jesness proposes. His ideal procedure is purely observational. As is true in astronomy, he would not influence the subjects of his research at all, if he could avoid doing so. He would infer what happens inside the subject’s skin, where observation is impossible – or was in 1985, perhaps – from correlations and contradictions that are or are not predicted and are or are not observed.

There’s nothing wrong in principle with an observational science. It’s just a lot harder to discover the functional underlay to what you observe than it is if you can probe and do experiments. So one often winds up with a purely descriptive statement of what you observe, but in more concise terms. That’s just what the network nodes do in “Deep Learning”. A complaint often raised about “Deep Learning” is that you can seldom find “Why” this description is as it is. PCT starts out the other way. It presents an answer to “Why” and then asks “What”. The method is more powerful and the result more intelligible, but the results of the two approaches should not conflict.

Martin

Richard S. MarkenÂ

"Perfection is achieved not when you have nothing more to add, but when you
have nothing left to take away.�
                --Antoine de Saint-Exupery

–

4

[Martin Taylor 2017.09.18.23.31]

[From Bruce Nevin (2017.09.18.20:53 ET)]

      Rick, I saw your comment on ResearchGate (RG) asking

Jesness to clarify, and I saw where he responded “just read
all this wonderful stuff I’ve written.” Fair enough on the
surface, we say pretty much the same to PCT newbies. But It
seems pretty clear to me (my thoroughly subjective perceptions
:wink: that he sees himself as an elder in a supportive role to
up-and-coming students and researchers. Will he be willing to
relinquish the pleasures of that self-image to become an
up-and-coming learner of PCT?

My subjective impression is a little different. But here we have an

issue that is relevant to what he says. We are both attempting to
perceive his reference for his self-image perception without having
(yet) applied much disturbance to it. We have his comments on
something Rick said and something (I can’t see what) that Rupert
said. From those and what little of pure observation (his preferred
approach) we can use, I see him as a member of a class of
“misunderstood geniuses who know how the world really works, if only
everybody would listen.” His emphasis on the number of pages and the
number of minor articles he has written rather than on the themes he
is trying to argue seem consistent with this notion. The result is
often to say “I’m right and I have no need to tell you why, because
somewhere in all those pages is the truth Ye seek.” I think most of
us who contribute appreciably to CSGnet have similar tendencies.

The problem with such people is that if they are sane in the

everyday sense of the word, their grand theories usually are built
on some reasonable foundation, and their results seem internally
coherent. But as with Aristotle, along the way there are usually
logical possibilities not seen but important, and they make serious
mistakes, such as that an object in motion needs continued
application of force to keep it in motion – which is true of
objects we see in everyday life, but for reasons Aristotle never
considered (so far as we know).

Aristotle had a big hold on serious thought for many centuries,

because he got so much right that people believed he got everything
right, and failed to examine critically his claims to test whether
they were right or wrong. But we don’t dismiss him as a crank, and
nor, I think, should we dismiss people like Jesness because of their
grandiloquent, dismissive, style. At least, not until we translate
what he is saying into some form more congenial to our own
backgrounds. At the moment, having read his preamble and Chapter 5
(as he suggests in the long preamble), my takeaway is that he will
infer from ethographic observation of consistencies of behaviour (at
some level of his perception, as we would understand it) to infer
what typically happens inside the studied type of organism. I see
nothing inherently wrong with that. If he observes in some community
typical behaviours at the level of, say, greeting strangers, and
those behaviours would be hard to predict using PCT, there’s a
pointer to something that needs investigating.

Anyway, I may change my mind entirely when I have read the rest of

his 156-page semi-book (it’s semi because chapters 3 and 4 haven’t
been written).

      Martin, I'm glad you've taken him up on it and looked into

his book.

“Ethogram” was a new one on me. Wikipedia to the rescue:

An ethogram is
a catalogue or inventory of behaviours or actions
exhibited by an animal used in ethology.

          The

behaviours in an ethogram are usually defined to be
mutually exclusive and objective, avoiding subjectivity and functional
inference as to their possible purpose.

      On this basis I, too, replied to him with a comment, but I

don’t see it on RG. Probably just as well, as I was quite
short with him, saying that it would be difficult for a
conception to be more orthogonally different from PCT. Not
very diplomatic at all, at all.

As yet, I don't see his conception to be orthogonal to, conforming

with, or opposed to PCT. He claims to see a relationship. His method
is certainly opposite, but how his results compare is (in my mind)
yet to be seen.

Martin