Instrumentalism versus realism?

CSGnet Archive CSG2017
1

[Martin Taylor 2017.02.24.06.48]

This diverges so far from the original thread title (CSG authorship 2017) that I though a new subject line without the "was" to be appropriate.

[Chad Green (2017.23.02.1045 EST)]

...

On 2/9/2017 Eetu mentioned that there may be a double bind between Rick and Martin who reflect two competing schools of thought: instrumentalism and realism. If true, then I think it would be fascinating if we found someone who could serve as a mediator of sorts by leveraging both perspectives to show how they are relational (e.g., the self and environment as fundamentally bound together).

Funny. I appear to be repeating myself.

I think you are simply repeating PCT. I believe PCT is the mediator you are seeking. PCT demands that both realism and instrumentalism are correct. They are not in opposition except in their extreme, exclusive, forms where one is only correct if the other is wrong. In the light of PCT, the opposition is between their proponents, not their concepts.

Why? Assuming there is a real world and this message is not a solipsistic exercise, enough organisms of a type must survive in it long enough to propagate their genes to descendants. Looking backwards in time, our own personal ancestors have succeeded in doing so for some 3 or 4 billion years, in really existing environments that have changed greatly over that time. Anyone reading this has personally survived in the real world long enough to be able to read it, despite having been exposed many times to situations in which they might not have done (and many like them did not) if chance had gone the other way.

One of Powers's insights was the concept of "reorganization". Reorganization does two things. One is to improve control of existing perceptions so that by controlling them the intrinsic variables are kept in the organism's "survival zone". the other is the arrange so that the control of multiple perceptions minimally interfere with each other -- in other words to "orthogonalize" our controls. I concentrate here on orthogonalization.

As Powers demonstrated in his "Arm" demo in LCS III Chapter 8, reorganization can orthogonalize the output degrees of freedom for control of several indeepndent perceptions (the arm with 14 different joint motions in the Demo). To "orthogonalize" them means to make them independent in the sense that changing one does not influence the others. In the actual demo, the perceptual functions are initialized to be orthogonal; changing one joint angle does not affect perception of the others. In the real world we are not guaranteed to be so lucky. Perceptual functions have to be developed through evolution or individual reorganization to operate together in ways that do not conflict, and at the same time work with the environment in ways that keep our intrinsic variables in good condition.

Look around you. Regardless of what actually is in the real world, what do you see? I think it is not edges, corners, and patches of colout, but books, chairs, walls, and at this moment a display screen, among other things. Are they in your environment? Perhaps not, if you just observe. But PCT says you can test, by acting to control something about them. Maybe you can take a book off the shelf and put it back. A whole lot of edges and colour patches moved in your visual field while kinaesthetic and touch sensations joined them in your (probably unconscious) set of perceptions. But you perceived an object, a book, moving with very little effect on the edges and colour patches of other objects you could see.

Your perception of the book was almost orthogonal to your perception of the rest of your world. The effects of your actions on the book had very little effect on the rest of your perceptual world. Evolution and reorganization have constructed perceptual functions and output hierarchies that work on the world as though there are really lots of touchable things that move coherently without affecting other objects (unless they bang into them).

Evolution and reorganization have made it so because when it is so, your ancestors' intrinsic variables (and yours) have been kept in survivable limits while you act in and are acted on by the real environment. We may not perceive the environment as it really is, and we are subject to illusions, but if we act as though the world is the way we perceive it to be, things usually work out better than if we act as though everything is illusion, and better than they would if we separately controlled every property of every perceived object and relationship.

Whereas the perceptual field is capable of containing many independent objects and their relationships and configurations at the same time, so is fairly easily reorganized into a reasonably orthogonal system, the output system is more constrained. In order to be able to control all those perceptions without much mutual interference between the individual perceptual controls, reorganization has to find another way to orthogonalize control. In engineering practice, this is solved by controlling one thing and then another (time multiplexing). In the Powers hierarchy, evolution and reorganization have (according to Powers) created a control level called "sequence control" that performs not only the orthogonalization, but also arranges that the effects of controlling one perception and then another make the inevitable carry-over effects of the earlier beneficial to the later, rather than impeding the later one.

Seen across time, the sequential level of control has the same effect on reducing internal conflict as does the control of an object property as compared to the independent control of the same property across the different parts of the object. If one want to move a chair, it's easier to move the whole chair than it is to move each leg, the seat, and the back separately and hope that they all fit together again to make a chair in the new place. Likewise, if one wants to, say, perceive a shelter in the forest, it is easier to gather branches then arrange them as a supporting structure and then cover them with leaves than it is to put up the leaves and hope they stay up while you gather the branches. To perceive and control the sequence is a way at worst to orthogonalize (preventing you from trying to wash the bathroom floor while you are vacuuming the living room) and at best to ease the control of the parts of the sequence.

Is the perceived sequence in the environment? Its reference value is not, but I could (at this moment) argue for either answer to the question. On the one hand, the relationships among the properties of different objects that determine the mutual influences important to the sequence clearly are in the environment, while on the other one might say that what is in the environment at this instant is not extended in time, so the sequence cannot be in the environment at any instant during its execution. The same might be said of any kind of time-binding. A velocity is a measure of rate of change, and you can't have a change at a single instant. The same ambiguity arises for an event such as a car crash. The crash unfolds over time, or at least its perception does. And one perceives the crash as happening in the external environment. It has real-world consequences such as the death of an organism.

As I said, I don't have a personal answer to this one, though I lean toward an extended perception of "now" that covers at least the loop transport lag time of the perception being controlled. With that proviso, one can at any instant perceive the whole sequence, but it occurs in the real environment extended over time.

The take-away point of this message is that we cannot know to what extent our perceptions of the environment are constructed arbitrarily by our evolved and reorganized sensory and perceptual structure or are predetermined by what is really out there. Nevertheless, PCT provides an answer that both realism and constructivism (instrumentalism) are correct in part. I suppose that makes PCT a precise form of enactivism.

Martin
PS. Vyv, I'm sorry this is another essay. It's hard to balance accuracy and concision.

2

[Vyv Huddy 26.02.2017.2123)]

[Martin Taylor 2017.02.24.06.48]

MT: PS. Vyv, I’m sorry this is another essay. It’s hard to balance accuracy

and concision.

VH: This is a fascinating and very thought provoking post Martin. I hereby withdraw my call to limit the length of posts on CSGNet! Now my young children are in bed I’m reflecting that it’s my responsibility to prioritise what to read and put
aside time for learning. I was having a particularly busy week and my suggestion reflected an impatient state of mind.

3

[Martin Taylor 2017.02.28.10.49]

  Here's another "essay". For those who like a shorter read, here's

the bottom line. Shakespear said it best: "
There are more things in heaven and earth, Horatio,/Than are
dreamt of in our philosophy." (First Folio version)

··· 
I thought I might try a different approach, to see if we could

dispose of the bogeyman Rick introduced, about the necessary
existence of an environmental processor to create the environmental
Qi correlate of the output of a perceptual function in the organism.
I’m leaving my original essay as a postscript to this message,
ignoring subsequent messages in this and other threads. I’m going to
take Richard Dawkins’s “Selfish Gene” approach and start with
evolution.

First, some basic assumptions that I don't want to defend. I think

they are reasonable.

Assumptions:
(1) Reality assumption: There is a real world in which unspecified

things interact in consistent ways.
(2) Consistent pattern assumption: There are consistent patterns
in the way things interact.

(3) Context sensitivity assumption: One such pattern is that two

“things” may interact in different ways if different third, fourth,
… other things are present.

(4) Organism boundary assumption: Among the things that may interact

in this real world are organisms, and among the patterns of
interaction are patterns that tend to exist in all organisms
together with their immediate environments.

(5) Heredity assumption (i.e.Genes and learning): Patterns of

interaction (both internal and with the environment) are likely but
not guaranteed to be passed from ancestor organisms to their
descendants.

(6) Mutation and recombination assumption: Descendant organisms may

have patterns of interaction not inherited from their ancestors.

These assumptions say nothing about the number or type of patterns

of interaction, but it says that they are consistent and contextual,
which implies that an all-seeing observer who could see all the
“things” in the real world would not see perfect consistency if even
one contextual element was ignored when looking at a particular
pattern of interaction. All the same, it is unreasonable to assume
that everything that happens in the whole Universe is equally
important for the consistency of any particular interaction. Indeed,
the conceptual segregation of a pattern of interaction from a
context implies that patterns are somewhat modular, some things
being more important to the pattern consistency than others.

For example, we know (at least our current theories are) that both

Jupiter and the Earth interact gravitationally with a stone we hold
in a hand, but when we let go of it, it always approaches the Earth
more consistently than it approaches Jupiter. Furthermore, when we
measure as accurately as we can how fast it falls, we see no effect
of Jupiter’s position. So although we know there is an effect, we
usually ignore it.

There are lots of contextual effects we cannot see, but that do

matter to the patterns of the way the real world interacts with us.
They matter in the sense that in different contexts, a pattern that
is in one context innocuous may in another be fatal. For example, if
we walk on a sidewalk beside a tall city building, most of the time
there is no problem, but if the context includes a stone dropped
from a high balcony, or a pane of glass blown out by the wind, we
may die. The basic interaction is between our forces on the sidewalk
and on us, but the context includes the possibility that other
undetected forces may also have effects on that simple interaction.
That kind of context matters a lot but doesn’t matter often, whereas
the Jupiter context doesn’t matter much but exists all the time.

Now let's go back to early times in evolution, not so early as the

beginning of life, but early enough to allow for organisms as
complex as bacteria. These organism might be affected by such things
as chemical concentrations in their environments, and maybe, like
the e-coli famous in PCT, were sometimes able to move in directions
of more salubrious chemical climates. Any one of them might have
myriads of descendants, most of which would die without reproducing,
because if each one had, say only a thousand descendants and it
weighed a nanogram, in six generations the descendants would weigh a
gram, in twelve generations a ton, and in six more they would weigh
more than the Earth. At some point, survival gets to be so difficult
that on average every one of those organisms could have no more than
one descendant. The other 999 die early because of the way they
interact with their environment.

Such simple organisms do not sense a lot of patterns that more

complex organisms can sense. They have to take a lot of chances with
environmental patterns that affect them undetected, along the lines
of being hit by a rock tossed from a high balcony as you walk past.
But already the kind organisms I hypothesise have by evolutionary
chance developed a pattern that involves both their interior and
their environment. Do they perceive their environment? In Powers’s
sense, yes they do, because some property in their interior is
influenced by some property of the environment in which they live.
Do they control their perception of the chemical property of their
environment? That’s hard to say.

The reason it's hard to say whether they control their perception of

a chemical property of the environment is that we have as yet no
criterion for the perception. In usual PCT-speak, they have no
reference value. They do die if the chemical environment has some
property values, and survive if they live in a more benign chemical
environment. Survival is more probable if evolution has arranged
that they can move independently than if they do not, because if
they do not, they will soon exhaust the local energy resources
(food) and be in a cloud of their own waste. Those organisms would
soon die, leaving only the ones that move to pass on their genes.
This is pure Darwinism.

Does this organism perceive its environment. I think not. All it

does is passively use a through flow of energy to maintain its
internal structure by exporting the entropy introduced by its
contact with its environment. Nothing it does is affected by changes
in the environment.

  [Martin Taylor 2017.02.24.06.48]




  This diverges so far from the original thread title (CSG

authorship 2017) that I though a new subject line without the
“was” to be appropriate.

    [Chad Green (2017.23.02.1045 EST)]




    ...




    On 2/9/2017 Eetu mentioned that there may be a double bind

between Rick and Martin who reflect two competing schools of
thought: instrumentalism and realism. If true, then I think it
would be fascinating if we found someone who could serve as a
mediator of sorts by leveraging both perspectives to show how
they are relational (e.g., the self and environment as
fundamentally bound together).

    Funny.  I appear to be repeating myself.

  I think you are simply repeating PCT. I believe PCT is the

mediator you are seeking. PCT demands that both realism and
instrumentalism are correct. They are not in opposition except in
their extreme, exclusive, forms where one is only correct if the
other is wrong. In the light of PCT, the opposition is between
their proponents, not their concepts.

  Why? Assuming there is a real world and this message is not a

solipsistic exercise, enough organisms of a type must survive in
it long enough to propagate their genes to descendants. Looking
backwards in time, our own personal ancestors have succeeded in
doing so for some 3 or 4 billion years, in really existing
environments that have changed greatly over that time. Anyone
reading this has personally survived in the real world long enough
to be able to read it, despite having been exposed many times to
situations in which they might not have done (and many like them
did not) if chance had gone the other way.

  One of Powers's insights was the concept of "reorganization".

Reorganization does two things. One is to improve control of
existing perceptions so that by controlling them the intrinsic
variables are kept in the organism’s “survival zone”. the other is
the arrange so that the control of multiple perceptions minimally
interfere with each other – in other words to “orthogonalize” our
controls. I concentrate here on orthogonalization.

  As Powers demonstrated in his "Arm" demo in LCS III Chapter 8,

reorganization can orthogonalize the output degrees of freedom for
control of several indeepndent perceptions (the arm with 14
different joint motions in the Demo). To “orthogonalize” them
means to make them independent in the sense that changing one does
not influence the others. In the actual demo, the perceptual
functions are initialized to be orthogonal; changing one joint
angle does not affect perception of the others. In the real world
we are not guaranteed to be so lucky. Perceptual functions have to
be developed through evolution or individual reorganization to
operate together in ways that do not conflict, and at the same
time work with the environment in ways that keep our intrinsic
variables in good condition.

  Look around you. Regardless of what actually is in the real world,

what do you see? I think it is not edges, corners, and patches of
colout, but books, chairs, walls, and at this moment a display
screen, among other things. Are they in your environment? Perhaps
not, if you just observe. But PCT says you can test, by acting to
control something about them. Maybe you can take a book off the
shelf and put it back. A whole lot of edges and colour patches
moved in your visual field while kinaesthetic and touch sensations
joined them in your (probably unconscious) set of perceptions. But
you perceived an object, a book, moving with very little effect on
the edges and colour patches of other objects you could see.

  Your perception of the book was almost orthogonal to your

perception of the rest of your world. The effects of your actions
on the book had very little effect on the rest of your perceptual
world. Evolution and reorganization have constructed perceptual
functions and output hierarchies that work on the world as though
there are really lots of touchable things that move coherently
without affecting other objects (unless they bang into them).

  Evolution and reorganization have made it so because when it is

so, your ancestors’ intrinsic variables (and yours) have been kept
in survivable limits while you act in and are acted on by the real
environment. We may not perceive the environment as it really is,
and we are subject to illusions, but if we act as though the world
is the way we perceive it to be, things usually work out better
than if we act as though everything is illusion, and better than
they would if we separately controlled every property of every
perceived object and relationship.

  Whereas the perceptual field is capable of containing many

independent objects and their relationships and configurations at
the same time, so is fairly easily reorganized into a reasonably
orthogonal system, the output system is more constrained. In order
to be able to control all those perceptions without much mutual
interference between the individual perceptual controls,
reorganization has to find another way to orthogonalize control.
In engineering practice, this is solved by controlling one thing
and then another (time multiplexing). In the Powers hierarchy,
evolution and reorganization have (according to Powers) created a
control level called “sequence control” that performs not only the
orthogonalization, but also arranges that the effects of
controlling one perception and then another make the inevitable
carry-over effects of the earlier beneficial to the later, rather
than impeding the later one.

  Seen across time, the sequential level of control has the same

effect on reducing internal conflict as does the control of an
object property as compared to the independent control of the same
property across the different parts of the object. If one want to
move a chair, it’s easier to move the whole chair than it is to
move each leg, the seat, and the back separately and hope that
they all fit together again to make a chair in the new place.
Likewise, if one wants to, say, perceive a shelter in the forest,
it is easier to gather branches then arrange them as a supporting
structure and then cover them with leaves than it is to put up the
leaves and hope they stay up while you gather the branches. To
perceive and control the sequence is a way at worst to
orthogonalize (preventing you from trying to wash the bathroom
floor while you are vacuuming the living room) and at best to ease
the control of the parts of the sequence.

  Is the perceived sequence in the environment? Its reference value

is not, but I could (at this moment) argue for either answer to
the question. On the one hand, the relationships among the
properties of different objects that determine the mutual
influences important to the sequence clearly are in the
environment, while on the other one might say that what is in the
environment at this instant is not extended in time, so the
sequence cannot be in the environment at any instant during its
execution. The same might be said of any kind of time-binding. A
velocity is a measure of rate of change, and you can’t have a
change at a single instant. The same ambiguity arises for an event
such as a car crash. The crash unfolds over time, or at least its
perception does. And one perceives the crash as happening in the
external environment. It has real-world consequences such as the
death of an organism.

  As I said, I don't have a personal answer to this one, though I

lean toward an extended perception of “now” that covers at least
the loop transport lag time of the perception being controlled.
With that proviso, one can at any instant perceive the whole
sequence, but it occurs in the real environment extended over
time.

  The take-away point of this message is that we cannot know to what

extent our perceptions of the environment are constructed
arbitrarily by our evolved and reorganized sensory and perceptual
structure or are predetermined by what is really out there.
Nevertheless, PCT provides an answer that both realism and
constructivism (instrumentalism) are correct in part. I suppose
that makes PCT a precise form of enactivism.

  Martin


  PS. Vyv, I'm sorry this is another essay. It's hard to balance

accuracy and concision.