[Martin Taylor 2019.05.03.10.01]
Over the nearly three decades I have been on CSGnet and its predecessor, one thing has been almost universal -- especially after Bill passed away. Bill's design for the control hierarchy within an individual has been largely ignored. The current twists in the RREV/CEV thread have brought this issue to the fore.
There's nothing sacred about the precise form of hierarchy proposed by Bill, as he acknowledged both explicitly and implicitly (by using different forms in some of his simulations). But there are characteristics that are important, and should not be thrown away without serious thought. What important aspects of the hierarchy are being thrown in the trash can in that thread?
I think the two most important ideas that usually get thrown away work together. They are that whatever happens in the hierarchy is not conscious, and that every perception in the hierarchy, controlled or uncontrolled, is a scalar variable. A scalar variable is one that can be described by a single number. Our conscious perceptions are not. We perceive a wildly complicated field all at once, with contents that shift as we do what we call "paying attention to different things". Bill thought that everything in conscious perception could be found as a perception somewhere in the reorganized control hierarchy. If that is the case, then conscious perception is a collation of many perceptions in the hierarchy.
Conscious perception contains what we perceive as "objects" such as tables, furniture arrangements, social structures, and the like. Scalar-valued perception are of properties of these objects, not of the whole object. Conscious perceptions are of the objects in the context of other perceived objects, whether or not they are tangible.
You may say "but each perceptual functions that has been reorganized into the control hierarchy is the same. It takes as input a set of perceptions and generates a scalar value that indicates how well its input structure of properties matches its own structure as a filter function." Or you may put it more succinctly and say that both the reorganized perceptual function and consciousness collate lots of perceptual values from the sensory and imagined/remembered environment, so what's the difference? The difference is that in consciousness there is a very large number of ways this collection might be put together, whereas the perceptual function defines just one way.
Consciousness, to use Eetu's metaphor, is a mapmaker mapping the Real Reality terrain, whereas a perceptual function is an already drawn map of a small aspect of the terrain -- perhaps of the proportion of households with running water on tap in a village street. The value of the perception changes as people install faucets linked to the water supply as the supply pipe breaks, but what is mapped does not change. The "mapper" that is conscious perception can choose to map the household water supply, but it can also choose to map the proportion of residents on the street younger than 18, or the surface quality of the roads to nearby villages.
What has all this got to do with RREV and CEV? A CEV is a perception, which in the Powers hierarchy, a scalar variable. But if it were just a perception as output from a perceptual function, would there be any point in using a different term for it? No, there would not. "CEV" stands for "Corresponding Environmental Variable", in which the key word is "Environmental". The CEV is consciously perceived as being "out there", in a complex world of other CEVs. Some of those "other CEVs" are perceptions that serve as inputs to the first one, some are perceptions to which this one provides input. Accordingly, the conscious environment in which all these CEVs live is an environment in which these relationships exist.
We are not conscious of all these CEVs all the time. Way back, someone (maybe Bill P) used the metaphor of an attentional searchlight that can play around in the tangled thicket of inter-related CEVs. We are conscious only of the parts of the "Perceived Environment" lit by this searchlight, whose beam is not very well focussed. But all of the perceptual hierarchy exists all the time, going about the business of keeping our intrinsic variables in good condition -- keeping us alive an well. And the effectiveness with which it does that job depends ONLY on how well the structure of the hierarchy fits with the structure of Real Reality.
What, then, is the function of the conscious perception of an external environment that consists of the related perceptions of the hierarchy in the form of CEVs? One possible answer is that it provides a test-bed in which the value of controlling arbitrary collections of perceptions put together in arbitrary ways and controlled by different methods can be tried out. If the tested method works in consciousness, maybe it will work in reality, but the proof of that pudding is, to mix the metaphor, in a series of test flights. If Real Reality incorporates variables that more or less match the CEVs in the ways they influence one another, then the test flights will not end in crashes, and we can say that RR contains structures that interact in the ways our CEVs do, at least insofar as our actions influence our sensory inputs. Those variables are RREVs.
If the test flight are wobbly or end in crashes, consciousness can change the design of the tested aircraft (perceptual function plus action output) to correct problems, and try out the revised version. If it works well, the new controller (Perceptual function plus comparator plus output function) can be basted into the hierarchy and firmed up as it continues to work well under new conditions. The conscious perceptual function that created a novel CEV from its Perceived Reality context merges into the hierarchy until it eventually becomes as non-consciously controllable as any other in the hierarchy.
This can happen only if the CEV the temporary conscious perceptual function produces is somewhere near an RREV in respect of what other structures in Real Reality a corresponding RREV influences and is influenced by. The structure of those influences is what reorganization depends on, and every "object" in our conscious perception is such a structure.
I think it is important to distinguish between the temporary "scaffolding" that is the tentative proposal for a perceptual function constructed in consciousness and perhaps discarded after one trial from the perceptual functions baked into the hierarchy and controlled non-consciously. Much CSGnet discussion treats them as though they were the same. Sometimes this is appropriate, because the structure of "temporary perceptual function - temporary reference input - temporary output distribution to cause action" is the same as an ECU (perceptual function, comparator, output function) in the hierarchy. But at other times to treat them as the same is misleading, and that is especially the case when we are dealing with contextual effects.
Bill Powers showed that a perceptual control hierarchy can function as specified, and that in principle it can account for a lot of observable behaviour of arbitrary actions of any kind of living thing, whether mobile (animals, fish, insects) or sessile (vegetation). The hierarchy operates without consciousness, and when we discuss the operations of the hierarchy we need not invoke Real Reality or RREVs. But when we deal with how the hierarchy came to be what it is, then the RREVs are essential, and the matching of their relationships with those among the CEVs determines the likelihood that the organism will live a long and healthy life. As the old saying goes: "We propose, but Nature disposes".
As usual, this is much longer than I had initially intended. I hope it makes some sense and isn't too much like bafflegab.
Martin