The missionary and the Scientist (was WIRED:...)

[Martin Taylor 2019.01.07.14.21]

[Rick Marken 2019-01-06_14:17:16]

[Frank Lenk 2019.01.05.10:29 CST]

              FL: Rick, I agree that the story

about the wood lice rejects they have a purpose to
seek shade. But the next paragraph, along with its
footnote, modifies that:

        RM: I don't think it's particularly productive to try

find what see like PCT compatible ideas in the work of
others. Indeed, I think it’s counter-productive. My
experience is that in order to see the ideas of others as
being compatible with PCT requires that the essential
elements of Powers work be ignored. In Friston’s case (and
virtually every other case that I’ve encountered) you have
to ignore the fact that a central insight of PCT is that
behavior is control, that it is organized around the control
of a hierarchy of different types of perceptual
variables and that the reference states of these variables
are set autonomously by the system itself. I guarantee you
that there is no other theory of behavior contains any of
these insights.

  Different theories have different insights. For the Scientist,

the question is whether the theories are distinguishable in what
they predict should be observed under different conditions.
Friston’s insight doesn’t ignore a fact at the core of PCT,
rather, it uses a fact that PCT ignores. Do these mutual ignorings
mean that both are invalid, that both are valid, or that one or
the other is useful for different purposes? The Scientist asks
whether there are conditions in which both claim to make
predictions of what should be observed, and the claims are
different.

  Different theories have different insights. For the Missionary,

the question is whether one insight can be exalted and the other
suppressed. The validity of the insight is irrelevant to this.
Indeed, the nearer the “false” insight is to predicting the same
things as the “true” insight, the more important it is to persuade
people not to even consider the “false” one. The Missionary must,
above all, persuade people that the Scientist is to be ignored.
Examination of the “false” theory, for the Missionary, cannot help
illuminate what is true and good about the “true” theory, but runs
the risk that the examiner of the other theory might be misled
into apostasy. On examination, the other theory might turn out not
to be false even in light of the “true” theory, a potentially
disastrous result, so from the Missionary we get …

        RM: My recommendation is to ignore all other theories of

behavior besides PCT

However, the Scientist might agree with …

        and concentrate on the facts that these other theories

purport to account for. Then see if you can figure out how
PCT would handle these facts. That, rather than looking for
superficial similarities of other theories to PCT, would be
a major contribution to PCT science.

…if the last “PCT” was omitted.

basically correct theory and that when there are critical
experiments that could distinguish between PCT and any other
theory, PCT will win. But my personal objective is not to make
contributions to PCT science so much as to make contributions to
science through PCT. I would like all behavioural scientists
(psychologist, psycholinguists, human factors engineers,
sociologists, economists, political scientists) to know enough of
PCT to be able to see whether it works for their domain of
interest as well as does whatever theory they currently hold; and
I would like to know enough of their theories and data to
understand whether PCT does, as I expect it does, provide an
underlying theoretical basis. But I have no interest in narrowly
contributing to a hermetically isolated “PCT science”. I do not
think such a close-minded "science’ could be other than a
pseudo-science.

  In the case of Friston's all-too-similar "false" theory, if you

take the circuit I abstracted from the Seth and Friston paper in
[Martin Taylor 2017.07.11.10.35], relabelling “reference”
↔ “prediction” and “error”<->“surprise” as synonyms,
you find that they both are functionally the same circuit. The
connections differ, in that the S-F circuit sends up to the next
level the reference and the error separately, from which the next
level perceptual function can construct perception as their
difference, while the Powers circuit sends up perception alone.

  This does create a potential discriminative experiment. If you

could show that a reference value can be perceived and controlled,
that would favour Friston, but I’m not sure that it is even
possible in principle to show that, since all reference values are
already functions of higher-level outputs in the Powers circuit.
Powers assumed or asserted that reference values and error values
cannot be perceived, at least not in the unconscious hierarchy.
Both Powers and Friston say the error (surprise) can be
controlled or minimized in absolute value (Powers control of
perception to equal reference is equivalent to controlling error
to equal zero), so it is irrelevant whether error can be perceived
independently. Maybe the distinction comes down to choosing the
simpler circuit if there’s no neurological evidence either way and
both circuits perform exactly the same functions under different
labels.

  The functional equivalence of the two circuits says nothing about

which insight (minimizing perceptual surprise or controlling
perceptual value) is more valid than the other, even if some
future experiment were to demonstrate that the S-F circuit
represented human brain function better than the Powers circuit.

Martin

···

From my point of view, I have long thought that PCT must be a

[Rick Marken 2019-01-07_16:14:52]

[Martin Taylor 2019.01.07.14.21]

        RM: I don't think it's particularly productive to try

find what see like PCT compatible ideas in the work of
others…Â

MT: Different theories have different insights.

RM: I think the theory is the insight. And it’s a good insight if it predicts the data; and it’s a bad insight if it doesn’t.

  MT: For the Scientist,

the question is whether the theories are distinguishable in what
they predict should be observed under different conditions.

RM: It seems to me that the scientific question is whether one theory accounts for the data better than another. If theories are indistinguishable in terms of their ability to account for data (observations) then they are, for all intents and purposes, the same theory and you just go with the one that is most parsimonious (Occam’s razor).

  MT: Friston's insight doesn't ignore a fact at the core of PCT,

rather, it uses a fact that PCT ignores.

RM: What is Friston’s insight? What is the fact that Friston’s insight uses that PCT ignores? What data does Friston’s theory explain that PCT doesn’t?Â

rsm

···
  Do these mutual ignorings

mean that both are invalid, that both are valid, or that one or
the other is useful for different purposes? The Scientist asks
whether there are conditions in which both claim to make
predictions of what should be observed, and the claims are
different.

  Different theories have different insights. For the Missionary,

the question is whether one insight can be exalted and the other
suppressed. The validity of the insight is irrelevant to this.
Indeed, the nearer the “false” insight is to predicting the same
things as the “true” insight, the more important it is to persuade
people not to even consider the “false” one. The Missionary must,
above all, persuade people that the Scientist is to be ignored.
Examination of the “false” theory, for the Missionary, cannot help
illuminate what is true and good about the “true” theory, but runs
the risk that the examiner of the other theory might be misled
into apostasy. On examination, the other theory might turn out not
to be false even in light of the “true” theory, a potentially
disastrous result, so from the Missionary we get …

        RM: My recommendation is to ignore all other theories of

behavior besides PCT

However, the Scientist might agree with …

        and concentrate on the facts that these other theories

purport to account for. Then see if you can figure out how
PCT would handle these facts. That, rather than looking for
superficial similarities of other theories to PCT, would be
a major contribution to PCT science.

…if the last “PCT” was omitted.

  From my point of view, I have long thought that PCT must be a

basically correct theory and that when there are critical
experiments that could distinguish between PCT and any other
theory, PCT will win. But my personal objective is not to make
contributions to PCT science so much as to make contributions to
science through PCT. I would like all behavioural scientists
(psychologist, psycholinguists, human factors engineers,
sociologists, economists, political scientists) to know enough of
PCT to be able to see whether it works for their domain of
interest as well as does whatever theory they currently hold; and
I would like to know enough of their theories and data to
understand whether PCT does, as I expect it does, provide an
underlying theoretical basis. But I have no interest in narrowly
contributing to a hermetically isolated “PCT science”. I do not
think such a close-minded "science’ could be other than a
pseudo-science.

  In the case of Friston's all-too-similar "false" theory, if you

take the circuit I abstracted from the Seth and Friston paper in
[Martin Taylor 2017.07.11.10.35], relabelling “reference”
↔ “prediction” and “error”<->“surprise” as synonyms,
you find that they both are functionally the same circuit. The
connections differ, in that the S-F circuit sends up to the next
level the reference and the error separately, from which the next
level perceptual function can construct perception as their
difference, while the Powers circuit sends up perception alone.

  This does create a potential discriminative experiment. If you

could show that a reference value can be perceived and controlled,
that would favour Friston, but I’m not sure that it is even
possible in principle to show that, since all reference values are
already functions of higher-level outputs in the Powers circuit.
Powers assumed or asserted that reference values and error values
cannot be perceived, at least not in the unconscious hierarchy.
Both Powers and Friston say the error (surprise) can be
controlled or minimized in absolute value (Powers control of
perception to equal reference is equivalent to controlling error
to equal zero), so it is irrelevant whether error can be perceived
independently. Maybe the distinction comes down to choosing the
simpler circuit if there’s no neurological evidence either way and
both circuits perform exactly the same functions under different
labels.

  The functional equivalence of the two circuits says nothing about

which insight (minimizing perceptual surprise or controlling
perceptual value) is more valid than the other, even if some
future experiment were to demonstrate that the S-F circuit
represented human brain function better than the Powers circuit.

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

[Martin Taylor 2019.01.08.23.09]

[Rick Marken 2019-01-07_16:14:52]

[Martin Taylor 2019.01.07.14.21]

                  RM: I don't think it's particularly productive

to try find what see like PCT compatible ideas in
the work of others…

MT: Different theories have different insights.

        RM: I think the theory is the insight. And it's a good

insight if it predicts the data; and it’s a bad insight if
it doesn’t.

            MT: For the Scientist, the question is whether the

theories are distinguishable in what they predict should
be observed under different conditions.

        RM: It seems to me that the scientific question is

whether one theory accounts for the data better than
another. If theories are indistinguishable in terms of their
ability to account for data (observations) then they are,
for all intents and purposes, the same theory and you just
go with the one that is most parsimonious (Occam’s razor).

We agree 100% to this point, at least if you actually understand

Occam’s razor, and apart from your dazzling insight that a theory
consists only of an insight. Jut how would an insight predict any
data, by the way?

            MT: Friston's insight doesn't ignore a fact at the core

of PCT, rather, it uses a fact that PCT ignores.

        RM: What is Friston's insight? What is the fact that

Friston’s insight uses that PCT ignores? What data does
Friston’s theory explain that PCT doesn’t?

  That has been explained in the WIRED thread, and if that is

insufficient for you, you can look it up in Wikipedia or in
Friston’s writings.

Martin

[Rick Marken 2019-01-09_10:39:27]

[Martin Taylor 2019.01.08.23.09]

            MT: Friston's insight doesn't ignore a fact at the core

of PCT, rather, it uses a fact that PCT ignores.

        RM: What is Friston's insight? What is the fact that

Friston’s insight uses that PCT ignores? What data does
Friston’s theory explain that PCT doesn’t?Â

  MT: That has been explained in the WIRED thread, and if that is

insufficient for you, you can look it up in Wikipedia or in
Friston’s writings.

RM: The closest I get to anything that sounds like PCT from the wiki is that the free energy principle says that a system “tries to minimize the difference between its model of the world and the perception of its sensors”. But things kind of goes off the rails when it says that the difference between perception and world model is “minimized by constantly updating the world model”. This seems analogous to PCT saying that an error signal is reduced by changing the reference for the perception. This is a way to “control” the error, keeping it at 0, but it certainly isn’t a way to control perception. Perception would actually just be causing the world-model to match it. Doesn’t seem very PCT like to me.

RM: The only thing I see that might be considered a “fact” that PCT ignores is that biological systems “maintain their order (non-equilibrium steady-state) by restricting themselves to a limited number of states.” I can’t find anything in the article that gives concrete examples of a non-equilibrium steady state and how you measure it. So I don’t know how the free energy principle accounts for that “fact”. Indeed, I don’t see how the free energy principle connects to any measurable aspect of the behavior of living things.Â

RM: The thing I liked about Powers theoretical work from the start was that it was always tied to experimental demonstrations of the phenomena to be explained. There was no hand-waving or attempts to impress with mathematics heavy laden with Greek symbols. It was always clear how the math related to the phenomena to be explained and the math itself was always presented clearly, like Bill actually wanted you to understand it. Friston’s approach would have impressed me when I was a grad student. But I can now tell when math is being used to make an impression rather than provide understanding. And I am not impressed by Friston’s free energy principle and lovely mathematics.

rsm

···


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