[From Bruce Abbott (2017.12.30.0900 EST)]
Martin, I really like your âexperimentalistâ? alternative (or supplement to) the âconstructivistâ? position. Not only is perception âtop downâ? as well as âbottom up,â? it depends on taking advantage of correlations among our perceptions, particularly those that suggest causal relations. Thus, thunder generally follows lightning but rarely the other way around, and because causes must precede their effects, we infer that lightning causes thunder and not the other way around. Such relationships experienced passively can be learned and taken advantage of. But as you note, we also interact with the world of our perceptions, experience the correlations that emerge in those interactions, and infer our own abilities to exert causal influences on that world. We learn how the visual appearance of objects change as we manipulate them, and how those visual aspects correlate with the tactile impressions that arise during those same actions; we experience an objectâs weight and inertia as we lift and move it, how its color and shape appear to change, and in so doing build a multi-sensory perception of the object and its various experienced properties. In these ways the brain constructs and updates a model of world of our perceptions.
Bruce
[Martin Taylor 2017.12.29.11.28]
[From Bruce Nevin (2017.12.29.10:14 ET)]
Rick Marken (2017.12.28.1500) –
RM: In what way is science “constructivist”?
Rick, what you said following this question amounts almost to a definition of constructivism.
So let’s make sure we know how the term is defined.
For a constructivist mathematician, what’s valid are constructive proofs and entities that those proofs demonstrate, with the implication that such entities have no independent existence apart from the proofs that demonstrate them.
In education, particularly science education, “constructivism sees learning as a dynamic and social process in which learners actively construct meaning from their experiences in connection with their prior understandings and the social setting.” (Quoted from one of numerous discussions on the web.)
In science:
"Constructivist epistemology is a branch in philosophy of science maintaining that scientific knowledge is constructed by the scientific community, who seek to measure and construct models of the natural world. Natural science therefore consists of mental constructs that aim to explain sensory experience and measurements.
"According to constructivists, the world is independent of human minds, but knowledge of the world is always a human and social construction. Constructivism opposes the philosophy of objectivism, [which embraces] the belief that a human can come to know the truth about the natural world not mediated by scientific approximations with different degrees of validity and accuracy.
“According to constructivists there is no single valid methodology in science, but rather a diversity of useful methods.”
[From Wikipedia “Constructivist epistemology” (footnotes suppressed).]
“The terms “objectivity” and “objectivism” are not synonymous, with objectivism being an ontological theory that incorporates a commitment to the objectivity of objects.” [Wikipedia again.]
I don’t see how one could understand and accept PCT without having or adopting a constructivist epistemology.
/Bruce
All of these definitions are one-sided, looking only at the input side of our interactions with our environment. PCT has a different take on the question, which I might call “experimentalist” for lack of a better term (“enactivist” nearly does it, but I think it also may not be adequate). Here’s a short passage from the introductory chapter on “The Environment of Control” of “Powers of Perceptual Control.” Earlier in the book I introduced “Oliver”, who is weighing something on a classical balance scale that has a pointer that he sees to be on one side or the other of a neutral point where the two scale pans have equal weight.
=========begin quote======
At a higher level than we have yet considered, Oliver may perhaps see the pans, their contents, and the pointer all together as a configuration, but why should he be justified in assuming that in some outer world there is a real pan with real weights, and that his perception of the âheavy-or-lightâ? location of the scale pointer represents a property of a real scale pointer that indicates something about a property of another scale pan and its contents? Is he so justified?
There are two opposed facile answers to this. One is âYou have access only to your perceptions, and can know nothing about the world that appears to be âout thereâ since your perceptions might be created by something entirely different, such as a manipulative super-intelligence.â? The other is âThe world is whatever it happens to be, and what you perceive is what it is for you, but perhaps for nobody else.â? Neither answer is really helpful, though both may be true. Let us contemplate a different kind of answer, perhaps no more true, but perhaps more useful.
Although we could never prove it, we must act as though our perceptions are not entirely self-referential (solipsistic) or created at need by some super-being, and as though we are not merely inhabitants of some grand super-software simulation project. We must also assume that there is a distinction between what we perceive and what there is to be perceived thaat there is a âreal worldâ? of which we are a distinct part, separate from the rest of the world. What happens in the real world sometimes alters the tiny part of it that affects our perceptual apparatus, and how we act sometimes influences a little of what happens in the part outside of us, our âenvironmentâ?. If these assumptions are wrong, we cannot know that they are wrong, so there is no point in discussing that possibility. We must act as though the assumptions are correct.
Given the basic reality and separability assumptions, we can say a surprising amount about what is in our real-world environment. For example, Oliver can say that no matter what the âweightsâ? and âscale pointerâ? really are, if he perceives himself to be adding or subtracting âweightsâ?, the âscale pointerâ? will change its angle consistently from one side of vertical to the other depending on whether he is adding or subtracting weights. It rarely changes from âtoo heavyâ? to âtoo lightâ? when he adds a weight, or the reverse when he subtracts a weight.
More generally, we often find consistencies between what we do and what changes in what we perceive. Few of these consistencies are always observed, and some happen in contexts we rarely encounter, but many happen almost always in frequently encountered circumstances. We perceive something being put onto our outstretched hand and feel weight on that hand. A stage magician might be able to arrange conditions in which this didnât happen, but usually it does. We throw something into the air and perceive it to rise and then fall. We put it on a table and perceive that it stays there, while the perception of weight in the hand is reduced. Assuming that there is a real world out there, something about the interactions between it and our senses creates these normally observed consistencies. When they donât happen, we seek explanations for why they donât rather than thinking we were wrong to expect them to happen.
Perceptual Control Theory (PCT) is based in the philosophically obvious fact that if there is a âreal world out thereâ?, we can know of it only what we obtain through our senses. As we said above, we assume that the âreal actionâ? is in the âreal worldâ? outside our bodies, and thereafter simply treat it as a given that the real world has effects on our sensory systems and is acted on by our muscular and chemical (and for some species electrical) outputs. What happens in the real world, whether we can perceive it or not, determines whether we live or die, are sickened by radiation poisoning, bruised by hits from hard objects, are well nourished by our food, enjoy good social relations, and so forth.
Whatever is in the real world, our interactions with it have allowed us to stay alive long enough to be able to perceive these little consistencies and our ancestors to stay alive long enough to propagate their genes into their descendants, including us. We can perform certain actions and expect certain things to happen, which would not be the case if our perceptions were entirely divorced from what is âreallyâ? out there. In particular, if we can control a perception such as the angle of Oliverâs scale pointer by acting on the environment, that in itself is evidence (under our basic assumption) that our perception corresponds to something in the real world. We do what in a laboratory would be called âexperimentingâ?, influencing the environment in different ways and seeing what happens.
==========end quote=========
PCT takes not so much a constructivist view of the world as an experimentalist view. What we perceive depends in large part on the consistency of relationships among the parts of the world nearby in space and time, and on how they interact when we act on them. When I calculated the likelihood of prespecified small static patterns occurring in a tiny toy universe of black and white squares on a double-sized chessboard, the numbers were astonishing, verging on one in the number of atoms in the visible Universe for simple patterns of say 25 black square cross on a white 15x15 field.
Since the perceived Universe is much more complex than my little chessboard array, it would be beyond belief (literally) that the structures we produce in our perceptual functions are not either constructed for us by control systems (like the Ames Room) or are really out there in the real environment, to be used by successive levels of perceptual functions of increasing complexity, and tested by control.
Martin