(Message 4) The Simple Observer
Links to Message 1, 2, and 3, so that I don’t have to repeat what is already there.
1: Real and Perceived Realities
2: Real and Perceived Realities
3. Real and Perceived Realities
By using the term Simple (or Pure) Observer, I am contrasting one who is observing and perceiving, but not controlling the perceptions of interest.
Sometimes we talk as though every perception that exists in the control hierarchy is necessarily controlled and built as an aspet of the reorganization of the control hierarchy. This segment of my contributions to this thread offers arguments that suggest that reorganization incorporates into the control hierarchy not only perceptual functions that produce controlled perceptions, but also perceptual functions that produce uncontrolled perceptions and even uncontrollable perceptions.
Here I am asking what an observer who cannot influence what is perceived — such as a newborn baby, a paralyzed person or an astronomer studying distant galaxies — can learn about Real Reality from the properties of the Observer’s sensory input alone. How can such an Observer build new perceptual functions when the perceptions cannot be tested by being used in control?
The answer is that our Observer, who we might as well name “Oona”, can do as Norbert Wiener’s builder of White Boxes did: look for correlations in the data from two or more unitary “terminal” signals (our sensor inputs), when the input terminals were fed with uncorrelated noise signals. If the Black Box output terminals output could be observed to exhibit positive or negative correlations of any form, linear or non-linear, Weiner’s engineer knew that he had to include a White Box or series of WBs that would produce such correlations when its inputs were uncorrelated.
Wiener introduced random signals at the input terminals to seek his correlations. Oona cannot do that. She, however, is not observing what comes out of a Black Box with well-defined input and output terminals. Oona is observing what Real Reality’s “output terminals” (Oona’s sensory inputs) present when she has no knowledge of the sources of the signals that they produce. From her viewpoint, they are quite random, and any correlations she observed in her sensory data are created by something in RR (which could be an “external input” to Real Reality (one might ask what that could possibly be).
Oona can do what Wiener did, even though she cannot influence the “input terminals” of RR. Since Wiener did not fee back the outputs from RR back to its inputs, his engineer-observer has the same kind of information as does Oona. The Engineer, like Oona, does not take advantge of any feedback, as we have assumed reorganization on the control hierarchy would do.
Oona can build a WB that would produce from uncorrelated inputs the correlation she observes. Since, however, Oona is not emulating a Black Box, but Real Reality, we should not use the term “White Box” to describe what she perceives (in PCT terms: “builds as a Perceptual Input Function” in her brain. For Oona, the equivalent to Wiener’s WB is an entity in her Perceptual Reality (PR) that performs the same correlation-producing functions as does an entity in RR, though the way it performs those functions may be entirely different.
Wiener built as many WB’s as he could find correlational patterns among the BB’s output signals. Oona can do the equivalent (just as Powers described in his Perceptual Control Hierarchy), apart from the fact that none of Oona’s PR entities (which we could call PREs) produce controlled perceptions. She could build Perceptual Functions from correlations among the outputs of the Perceptual Functions she has already built from her sensor signal correlations. She constructs Layers, or Levels of Perceptual functions, each of which exists as a PRE, an entity in her Perceptual reality.
If Oona is an astronomer, she can do no more than observe the starts, noting patterns among them that she calls by various names, such as globular clusters, planetary nebulae, galaxies — or, in times long gone, constellations. Using tools such as telescopes and spectroscopes, she can observe more about the stars and planets, but she is still observing. She can never manipulate what she observes.
On the other hand, if Oona is a newborn baby, the movements, sounds, and scents she produces do influence parts of Real Reality. She can perceive her own movements and correlate those perceptions with others she gets through her various sense organs, to create new perceived entities in her Perceptual Reality.
These new entities are different from those created by correlations among the world facing sense organs, in that they incorporate parts of herself into a single perceptual entity. If she feels “thus” from her arm, what she receives from some other sense organ will change “so” more probably than if she did not feel “thus” from her arm. Oona has perceived a cause-effect relationship, but not a simple one, because it depends on what she can perceive from kinaesthetic input from the arm, which is not in the world outside her skin. Nevertheless, there will be correlations that she might detect in the same way that she might detect the correlations among the world-facing sense organs. We are talking about the beginnings of control.
One of Powers’s talking points was that there’s no reson to take a model seriously unless the model has a plausible mechanism. In this Section on the Observer, I used “probability” in the form of “correlation” as a fundamental determiner of what patterns become manifest as Perceptual Functions, and thus as PRE’s (entities in the Perceptual Reality of the observer). But what calculates a “probability” or a “correlation” in the brain of a newborn? The word “calculates” presupposes the presence of a calculator, which we presume the newborn does not have. So what mechanism might implement a model that says a newborn builds perceptual functions from correlations and probabilities. One possible answer is Hebbian and anti-Hebbian synaptic modification.
Hebbian learning is encapsulated in the motto: “Nerves tha fire together wire together,” and anti-Hebbian by a similar mantra: “Nerves that don’t fire together wire separately”.In practice “wire together” implies an increase in the synaptic strength of interconnecting synapses, while “wire separately” implies s reduction of the strength of interconnecting synapses. Both effects have been observed in many parts of the brain. If nerves from two separate sensors tend to increase and decrease their firing rates together more often than not, the strengthening effect will predominate. If they tend to fire at unrelated times, the interconnecting synapses will weaken.
The effect is to produce strong outputs for correlated events in Real Reality, and less than average output if either of two anti-correlated events occurs. It is as though some calculator computed correlations and adjusted the synapses appropriately, to produce a perceptual function that reports the co-existence when it occurs.
We usually think of reorganization as being based on control of perceptions, and just as an experimental science can be more assured of a pattern that is the observed result of an experiment than of a pattern detected only by observation, so also is reorganization by way of perceptual control a more reliable way of understanding Real Reality than is reorganization by observation. Nevertheless, what I have just described is a plausible but by no means proven mechanism for a constructive form of reorganization that can be performed by a pure observer.
We address reorganization as the matching of Perceptual Reality to Real Reality in my next numbered contribution to the discussion.
Martin