Reorganisation and Memory reconsolidation

[bnhpct]

@MartinT

[BN]:Martin, I am not sure what you mean
when you say that we can become aware of errors and reference
values.

Have you never known what you
wanted and that the world has not (yet) provided it for you? In a
classical tracking study, do you not know where the cursor should
be compared to where it actually is?

The error signal output from a comparator (the rate of firing which results from synapsing its inhibitory perceptual input signal together with its excitatory reference input signal) is not the same as the relationship perception which the comparator controls.

When I have a perception of what I want (e.g. cursor at target), I am controlling that relationship perception in imagination. The current model of imagination posits that a copy of the reference signal is taken up as perceptual input. (This may involve copying reference signals at lower levels to perceptual signals sent up from those levels, depending upon how detailed and vivid the imagining is.)

Is it possible to control the relationship through the environment at the same time as controlling the relationship perception in imagination? The answer is not immediately obvious to me. If this can occur, it seems to me that it would require parallel systems with the same reference signals controlling the same perceptual input.

It appears that I can observe perceptual input from the environment (e.g. that distance between cursor and target) while knowing that it is not the desired state but not while simultaneously imagining that desired state (controlling zero distance between cursor and target in imagination). The “knowing that it is not the desired state” is awareness at the level above, at the comparator or comparators which is/are employing that relationship (zero distance between cursor and target) for its/their purposes.

This is my subjective experience. If the Friston-derived proposal would permit an error signal to be copied to perceptual input, I think that would predict subjective experience that contradicts mine.

@MartinT

“reference and error values go to the higher level perceptual input function”,

I am sometimes swamped by other obligations and at those times do not keep up with CSGnet traffic. I did not delve into the Friston discussion. If this is mathematically equivalent to the perceptual signal going to the higher level perceptual input function, it is because they are synapsed together in that perceptual input function, correct? Specifically, it does not follow that the error signal is a distinct perceptual signal of which we can be aware.

[MartinT]

Bruce,

One or other of us is seriously confused.

I am confused as to why you bring imagination into a conversation

about control within the hierarchy.

I am confused about why you say: "*      If the Friston-derived proposal

would permit an error signal to be copied to perceptual input, I* think**that would predict subjective experience that contradicts mine ."
Could you explain how you could possibly tell the difference, given
that the two proposals give the same mathematical result for the
output of any perceptual function at the next higher level, while
the Seth-Friston connection offers extra possibilities that do
conform to my conscious perceptual experience.

I am confused about whether you are talking about one or two levels

of control when you say: “* The error signal output from a
comparator (the rate of firing which results from synapsing its
inhibitory perceptual input signal together with its excitatory
reference input signal) is not the same as the relationship
perception which the comparator controls* .” If you are talking
about a single level that controls a relationship, it is a
tautology. If you are using the word “relationship” to refer to the
relationship between the perceptual and the reference signal values,
I don’t know where “which the comparator controls” fits into any
form of perceptual control theory.

I am confused about why you treat conscious perceptions as though

they were perceptions in process of being controlled in the
reorganized hierarchy.

Are you confused about what the Seth-Friston connection makes

possible, or do you think there is some evidence against it that you
did not state? If you want to refresh your mind about it, it is
described and contrasted with the Powers connection in [Martin
Taylor 2017.07.11.10.35], which presumably you can find in the
CSGnet archive.

I just noticed that this message was in response to "On 2017/07/9

6:16 PM, Bruce Nevin wrote:

•    The B:CP view of imagination is that a copy of the reference
  

signal (here, the reference for a missing bit of input) branches
across to create a perceptual input signal returning to the
originator(s) of the reference signal. B:CP depicts this as a
switch actively making and breaking a neural connection. This
has seemed implausible to me, so I proposed that the imagination
signal is always present. When there is actual perceptual input,
the copy of the reference signal augments it; when there is not,
then the copy of the reference signal provides some (weak) input
of that perception* ." The Seth-Friston accords with this
comment without the need for a special switch.

Martin

[bnhpct]

Yes,my reply to that post, together with your diagram, is at An alternative hierarchic connection circuit (was Re: Perceptual Cartoon).

In brief, I liked this way of having imagination always ‘on’, constituting the perceptual signal in the absence of perceptual input from below, and having negligible influence on the perceptual signal when there is perceptual input from below. More sophisticated than the way of doing it that I suggested in my chapter of the Handbook. I thought it might take a while to suss out unforeseen complications.

But the question at hand is not how the imagination signal is generated, but rather, can you perceive an error signal. In the Friston/Taylor proposal the imagined perceptual input results from synapsing error and reference together. The error signal never by itself constitutes a perceptual signal. (And error is zero when the perception is entirely imaginary, with no perceptual input from below, right?)

The ‘classical’ view is, no, you can’t perceive an error signal and you can’t be aware of an error signal; that is still my view. So you asked

@MartinT

Have you never known what you wanted and that the world has not (yet) provided it for you? In a classical tracking study, do you not know where the cursor should be compared to where it actually is?

This actually echoes questions that I asked and which you quoted in that 2017 thread that you referenced.

Yes, I can direct my attention to an imagined perception of the desired state of affairs from the point of view of the comparator which sets that reference signal. In a tracking example, that comparator is above the relationship level, setting a reference for ‘cursor at target’ as means of controlling its purpose, ‘satisfy the experimenter,’ perhaps. And I can direct my attention to the actual spatial relationship between cursor and target. To do that, my attention is from the point of view of the relationship comparator. Can I attend to both perceptions at once? Perhaps, but more likely my attention shifts between them.

The disparity between the imagined perception (the reference value for the relationship) and the ‘actual’ perception could conceivably be perceived as some kind of relationship perception, but that is not the error signal.

It may be confusing sometimes that we do perceive the somatic consequences of error. I believe this ‘affect effect’ is extremely important for clinical applications, and have suggested that to Eva.

Over the years I have occasionally heard phrases like ‘experiencing error’ and it has seemed to me that these phrases express this confusion. At the heart of it is the very important and puzzling distinction between perception-as-experienced and perceptual signal.

The experience of perceiving is all we directly have. Indirectly, inferentially, we have theories and models. ‘Perceptual signal’ is an experienced perception for a neuroscientist under certain experimental conditions, but for the most part, ‘perceptual signal’ is a theoretical entity in a model, and though the relationship between the former and the latter has been demonstrated in certain experiments it is more often assumed. We have a good grasp of each end of the relationship between ‘perceptual signal’ and ‘perception-as-experienced’, but the presumed bridge between them is quite foggy. The reason it is foggy is that intellectual constructs and immediate experience seem to be incommensurate in character. The point of science is to use the former to explicate the latter, and use the latter to test and refine the former.

The error signal is firmly on the theoretical construct side, with some few glimpses perhaps by neuroscientists, and our awareness of feelings and emotions is firmly on the experiential side. Feelings are perceptions in the somatic branch of the hierarchy and emotions are perceptions interpreting those feelings at levels above the divergence of the behavioral and somatic branches.

I refer to Bill's PCT model of emotion (version 2).

I elaborated a bit on An opinion of the brain about sensations in the body.

[MartinT]

Bruce,

bnhpct

              August 28

…
But the question at hand
is not how the imagination signal is generated, but
rather, can you perceive an error signal. In the
Friston/Taylor proposal the imagined perceptual input
results from synapsing error and reference together. The
error signal never by itself constitutes a perceptual
signal. (And error is zero when the perception is entirely
imaginary, with no perceptual input from below, right?)

No, it’s the actual perceptual signal that is synthesized (not
synapsed) from the reference value and the error value, not the
imagined perceptual signal. That synthesis is presumed to be a part
of the processing done by the next higher level perceptual functions
to which the perceptual signal would contribute in the Powers
version.

And error is zero when the perception is
entirely imaginary, with no perceptual input from below, right?

I don’t see why that would be necessary, or even usual. Imagined
error is certainly not incorporated into my diagram, but I don’t see
any reason why it couldn’t be.

I the connection circuit that I based on the Seth-Friston proposal,

the error signal and the reference signal are values that can go
anywhere, not just one place. A function that takes both as inputs
can create from them what would be the perceptual value in the
Powers hierarchy (apart from caveats discussed in the old CSGnet
message).

What might the error signal do by itself? For one example, it might

contribute to variation in the rate of reorganization in that part
of the hierarchy, which has sometimes been supposed to depend in
part on the absolute value of the error and its derivative summed
over a larger or smaller part of the hierarchy. It might function as
a perception that could be made conscious. I expect that modellers
might well find other uses for it, but I think that’s enough to make
the point.

The ‘classical’ view is,
no, you can’t perceive an error signal and you can’t be
aware of an error signal; that is still my view. …

The error signal is
firmly on the theoretical construct side,

As is the entire perceptual control hierarchy, in any specific form,
and perhaps in general, though the construct of perceptual control
as the source of behaviour seems to be necessarily implied if one
believes in the general correctness of classical mechanics as an
approximation appropriate to the scales of human perception and
output forces and chemical emissions. Since you can’t have
perceptual control without controlling perceptions to some intended
(reference) values, I don’t see any form of perceptual control that
does not include an error signal.

Everything is theoretical construct except one's conscious

perceptions. You can’t get away from the truth that all you know or
can know depends on whatever you are conscious of. That includes all
theories of perception and of behaviour, as well as of physics and
neuroscience. So to describe something as just a theoretical
construct is hardly a valid argument for its lack of value within
the context of a theory that is accepted for the sake of argument.

I cannot experience your experiences. For myself, I can't think of

an instance in which I am consciously aware of a perception and of
its optimum value to me when I am not also conscious of any
discrepancy between those two values.

I do lean toward the Friston inter-level circuitry in preference to

the Powers version, but that is because it seems to explain
observations that the “classic” connection ignores entirely. But I
have no expectation that it actually describes what goes on in the
brain, even at the coarse level of “neural bundles” and “neural
currents”.

Martin