[From Bill Powers (950225.1105 MST)]
Richard Thurman (950224.1630)--
RE: memory
I hope you will treat what BCP says as proposals concerning memory, not
Gospel.
I do think that memory has to be an inherent part of the system, if only
to explain how you can "learn" someone's telephone number.
I also do not think that everything we include under the label of
"learning" is one phenomenon. There are several phenomena which seem to
involve entirely different principles: recording and playback,
reorganization, "tuning" of the parameters of an existing system, and
systematically acquiring and organizing information according to some
learned algorithm ("learning to learn"). And maybe others. To classify
these disparate phenomena under one word is a little like using
"financial transaction" to refer to everything from being robbed to
winning the lottery to buying something to counterfeiting ten-dollar
bills.
···
----------------------------------------------------------------------
Martin Taylor (950224 13:30) --
You are right that in our agreements, there may well remain
disagreements that depend on how the words are being understood.
I assert: (1) There is no way I can determine whether anything
exists outside of my own perceptions--including myself. (1a)
Position (1) is no fun, and should be disregarded for the sake of
further argument. Therefore substitute (2) there is a reality
outside my perceptions, and (2a) all I know about it is what my
perceptions tell me--and that includes my perceptions of myself, as
well as all rules and regularities that I "discover" about that
reality. They are all perceptions.
The danger in (2) is that in saying that all I know about reality is
what my perceptions tell me, I may tacitly assume that my perceptions
are telling me correctly, without any need for interpretation on my
part. I think it is much safer to stay with (1), even though it's not as
much fun, provided it's reworded a little.
Actually, I think we can determine in the usual way (by considering
evidence) whether something _exists_ outside of us. We do have to learn
what actions are needed to control our perceptions, and we do quite
often experience changes in our perceptions that we didn't intend to
happen. That is evidence of something existing independently of us. But
we have to recognize that the evidence does not tell us WHAT exists.
We can assume an existent reality, I think, without too much fear of
contradiction or disagreement. But as we currently understand how
perception works, we must realize that whatever is out there is being
projected by our perceptual apparatus onto a coordinate system of our
own making, through transformations of which we currently know very
little but which are part of our human nature (innate and acquired) and
not necessarily the nature of the external world. If we all saw the
world through anamorphic lenses, and saw our effects on the world the
same way, we would never suspect the existence of those lenses. So it is
for all input transformations that lie between the reality we assume to
exist and the reality we experience in the form of perceptions -- all
transformations, that is, that human beings share in common. These can
be not only physical transformations, but intellectual ones.
I think it is important to stay with your choice (1), modified to admit
that we can infer existence without knowing form, because doing so
forces us to be clear about the distinction between perception and
imagination, between observation and inference. I observe that an apple
is red on the outside, white on the inside, sweet, and crunchy; I infer
that there is an object in the real world corresponding to these
experiences. I can't be mistaken about observations; I can easily be
mistaken about inferences, or not realize that more than one inference
from the same observations is possible.
In assuming that our perceptions are telling us about reality, we can
easily forget just how much human work has been put into reaching a
consensus about reality. We can too easily assume that the accepted view
is the correct version, and judge your view and mine in terms of
deviations from what "really is." This is why physics, so often, is
taken as the final arbiter in disuputes about theories; we easily forget
(and in many instances, probably, never realized) that physics is about
human perceptions, not about the reality they are assumed to represent.
(3) When I act, my actions are in "reality" but I know nothing of
them except through my perceptions. However, (still in 3), my
actions affect my perceptions THROUGH reality, at least in part.
(Here I allow for hallucination, the admixture of hope with
sensation...)
I'm trying to establish that we agree on the implications of this. If my
actions affect my perceptions _through_ reality, it follows that I must
find those actions that will have the correct effects; the fact that I
can do this tells us that reality has properties of its own. But there
may be a transformation between what I perceive as my actions and the
reality, and between the reality and the perceptions I am controlling,
of which we know nothing. If the same transformation is involved, the
transformation and its inverse are being applied between our perceived
outputs and the perceived consequences, making the transformation
invisible to us. So even though we can claim that there is some regular
relationship between the properties of the world and the properties we
infer by analyzing our own perceptions (of actions and their
consequences), there is no direct way of knowing the actual relationship
between the real properties and the inferred ones.
(4) (Assumption, rather than assertion) By some means or other
(according to PCT, it's "reorganization") I come to connect some
actions with some perceptions, and these links persist only so long
as reality has some regularity that "validates" them.
What is validated is that the apparent effect of perceived action on
perceived consequence is regular and reliable, and that whatever is
happening between action and consequence in the external world is
regular (although perhaps not of the same form as the regularity we
infer) at least as projected into human perceptual space. If the reality
has more degrees of freedom than the human perceptual system does (and
that is almost certainly true), then we are simply blind to all the
changes that take place which don't happen to alter the relationship
between our perceived actions and their perceive consequences. As far as
human perception is concerned, such changes are "don't-care" conditions,
even though they may in fact have important consequences in the real
world.
The reason I am so picky on this subject is that I think there is a
great reward in being constantly aware of the difference between
perception and inference. To be aware of inference is to be aware that
inference can be mistaken, and that there is more to be known than we
know. In fact, I begin to get an inkling that because in general we fail
to make this distinction clearly, we have yet to realize how much more
may be contained in the reality around us than we have so far been able
to infer even with the most sophisticated of our sciences and methods of
analysis.
There's another side to this, too. The import of all the above is not
that experienced reality has been proven to be different from boss
reality. It is that so far we have not developed any way to know whether
it is or not. If such a way is possible, we will never find it by trying
to force our understanding onto the external world just because we would
be a lot happier if doing that were valid. We must first appreciate
fully and in detail what the problem is, and that is what I think I am
trying to do. If there is a way of bootstrapping our understanding so it
actually says something about the real world, discovering that way will
depend on knowing exactly where we stand now -- and, I should add, on
the advent of intellects considerably more powerful than mine.
(5) If (and it's a big IF) I am perceiving some actions under 4,
together with the related perceptions (I do mean "perception of a
set of perceptions" which may be the same as a higher-level
perception), then I may perceive the regularity that persists in 4
as "a rule." Otherwise, there's no perception of a rule, though an
outside observer might be able to describe my behaviour in terms of
a rule.
Here we have a problem that shows up in other contexts as well. Consider
looking at a room to perceive regularities in the form of spatial
relationships. The problem here is not that it is difficult to find and
verify spatial relationships, but that it is too easy. Once you start
looking for spatial relationships, you can continue enumerating them for
the rest of your life. Any one of them, such as the fact that your nose
is higher than the floor, or than a bit of lint on the floor, can be
verified and be shown to be reproducible under proper conditions. Yet
the longer we look for such relationships, and find them, the clearer it
becomes that we are simply using our capacity to construct
relationships, and are saying nothing of import about the room.
The same happens with rules. It's been known for quite some time that in
any finite sequence of numbers, such as 1, 2, 4, 8, 11 ... not only can
we find a rule that will generate the sequence to date, but we can find
an infinity of rules that will generate the same sequence. The problem
is not one of finding a rule; it is one of finding as many rules as we
like, all generated by our own capacity to formulate rules and behave
according to rules.
The regularities of nature are thus chosen by us from an infinity of
regularities that could be drawn from the same phenomena. Your words
imply that in any perceived relationship of action to consequence there
is just one regularity to be discovered. In fact, there are many
possible regularities, and which one we see depends on the perceptual
interpretation we bring to the experience.
One regularity in isolation therefore means nothing. What counts is the
whole network of regularities that we formulate and apply, with the goal
of making them all consistent with each other. There may be no unique
network, and there may be no knowable relationship between one network
of regularities, another equivalent network, and whatever network
actually exists outside us. So we come back, as always, to the same
fundamental problem: is there a unique "right" view of the world?.
------------------------------------------------------------------------
Best,
Bill P.