Against gobbledegook; for plain talk

[From Bill Powers (930906.0915 MDT)]

Gary Cziko (930906.0315 UT) --

Rick says:

You can decode p if you are given f(). That is, you can
determine what sensory state any particular >value of p "coded"
by solving s = f-1(p).

And you say:
I am a bit surprised by this kind of talk since you seem to be

taking a perspective of an outside observer of the perceptual
system. Now put yourself back inside. How can the perceptual
system itself know what the function f is?

There are many ways, all inferential, of course. I agree with you
that in the proposed formulation, p = f(s), what we experience is
p, and neither f() nor s. However, if you treat f and s as two
more kinds of perceptions, you can propose that the brain makes
models in which p and s are related in a certain way that we can
perceive as f(). This is the "perspective of an outside
observer," which is not outside at all, but results from looking
at a model in which some elements are imagined. There is no way
in which one can actually take the perspective of an outside

I think the key is in answering your question "How can the
perceptual system itself know what the function f is?" If by this
you mean how can the perceptual system obtain justified true
objective knowledge of the function f (emphasis on "know"), I
think the proper answer is that it can't. But we can emphasize
"how." The perceptual system can look at a perception s and
another perception p and come to recognize a relationship between
them: p depends on s according to a certain relationship. This
relationship, symbolized and worked into a logical framework,
becomes what we know as a "function."

The basic relationship p = f(s) is thus a hierarchical
relationship in two senses. First, recognition of a function f in
the behavior of p and s entails the use of higher levels of human
perceptual operations. And second, if p depends on s, then p is
by definition a higher level of perception than s. In a
sufficient array of such functional relationships, we can
perceive a principle that can be applied, in imagination, to
situations where we can experience only p, and neither f nor s.

When we look at most perceptions, we can also see lower-level
perceptions on which the higher-level one is constructed. So we
can see both s (the lower-level perception) and p (a higher-level
perception) and from a still higher-level point of view perceive
the dependence of p on s. This allows us to perceive the
function. However, when we explore these dependencies in the
downward direction far enough, we find perceptions (intensities)
for which we can discover no further lower-level perceptions of
which they are a function. We experience only the perception p,
and there is no way to detect an underlying s so as to derive the
functional dependence of p on s. Extrapolating from the general
principle we find at higher levels, however, we can suppose that
there must be an s, and some functional relationship that makes
this lowest level of p depend on the s. Thus we arrive at the
inference that there is an independent world on which the lowest
level of experience depends.

That development relies strictly on an analysis of subjective
experience by an isolated person. There is another line of
development, which is to consider the perceptions we call "other
people" and their actions, embedded in a set of perceptions that
we call "the environment." Now we can see what appear to be
elements of the environment of the other person: elements that
are affected by the actions of the other and which appear to
influence those actions at the same time. We hypothesize that
these other people are constructed internally just as we are.
However, we can't experience the perceptions in the other person
that correspond to our own: we can see things we might identify
as s for the other person, but not f or p. This is the reverse of
our subjective experiences at the lowest level, where we can
experience p, but not f or s.

From invasive study of other people and animals, we find that

there is a perceivable relationship between the variables we see
as s in the other's environment and variables we see as "signals"
in the nervous system of the other. From our standpoint, those
signals do not look like our own perceptions; they look like
blips on an oscillogram. But we can discern functional
relationships between environmental s's and signals, and find
structures in which this relationship seems to be embodied (which
we call sensory nerves). This functional dependence is similar in
form to what we find in our own higher levels as p = f(s).
Furthermore, we can see that some signals inside the other's
nervous system depend functionally on other signals, showing that
there are higher levels of dependence in the other, one signal
being a function of other signals.

All that remains now to achieve a consistent system concept is to
put these two views, one of our own experiences and one of
relationships observed in our perceptions of other people,
together. What single unified picture would explain what we find
by both kinds of investigation?

If we suppose that the blips in the oscillogram of the other's
afferent nervous-system activities are identically what we
experience as perceptions, the two views become consistent with
each other. When we observe subjectively that a configuration is
a function of the way sensations behave, and that in the other
person there are signals low in the nervous system that depend on
other signals even lower, but still in the nervous system, we can
propose that our subjective analysis of configurations into
sensations reveals the same functional relationship that we see
outlined in the other's nervous system. We conclude that what we
experience is the way neural signals look to the experiencer in
the same brain, while our experience of blips in the other's
nervous system is how neural signals look to an observer seeing
them as detected by electronic sensors and transmitted through
some electronic device to our eyes.

This merging of the external and interval views is made possible
by the assumption that the perceived other person is a system
organized inside exactly as we, the observers, are organized
inside. As a bonus, we can now also understand how we, the
observers, can know of the other person's relationships to the
world. This comes about through proposing that certain
perceptions represent an outside world that is external both to
ourselves and to the other person. We see this relationship of
externality primarily in observing others, for we can see things
that appear to affect the senses of the other person yet which
are different in nature from the signals they evoke inside the
other person. From this we can deduce that the other person would
see a similar situation in looking at us: the other would see
environmental variables affecting us which we know about only in
the form of internal signals. The other would see our experiences
as neural signals.

And finally, we can understand that this entire picture is being
constructed inside ourselves as a way of explaining our own
experiences to ourselves.

I haven't discussed action at all, but action would be an
intrinsic aspect of discovering all these relationships among
perceptions. We organize perceptions by acting on the world to
find which perceptions are controllable by means of varying which
other perceptions. In the world of science, this amounts to
experimental tests of hypotheses. But that gets us unto a
different subject, and other models.



Bill P.