Black boxes again II

Rick Marken (961020.1700) sez:

Bill Powers (961020.1100 MDT) --

Anybody can "see order" in the [neurophysiological] literature if not
required to prove that the new picture is right.

Bill Benzon (961020) --

But "seeing order" is the first step to developing a picture you want
to test.

Can you describe the kinds of tests that have been or might be developed
based on the "order" you have found in the neurophysiological data. I'd be
particularly interested in how these tests relate to understanding
perceptual control.

But, as a prelude to tests, here's something which could be done by
reviewing the literature, which is to take a simple process used in the
cross-cultural literature and apply it to comparative neuropsych.

In the cross-cultural lit folks develop scales of cultural complexity by
making large lists of traits -- size of largest settlement, number of craft
specialties, levels of hierarchical social control, ritual specialists, how
food is obtained, etc. Such lists can run to several hundred. Then you
pull a random sample of known cultures, go to the literature on those
cultures in the sample, and score each culture on your traits. It almost
always turns out that you can place the traits in a order such that if a
culture has trait N, it will also have traits 1 though N-1. (This is called
a Guttman scale.)

As a very specialized example, if a culture has only 2 color-designatora,
they will mean, roughly, dark and light. If a culture has 3 designators,
the third will designate red. If 4 designators, the 4th will be either
green or blue. If 5 designators, then the 5th will be the one of the
green/blue pair which wasnt' the 4th designator. We can continue this to
about 11 color terms. If we then score these cultures on some general
measure of complexity, they will score in the same order.

This same technique could be applied to the comparative psych literature.
List all the various kinds of tests and measurements of animal perception,
behavior, and reorganization. Score each species for the items in the
list. And see whether or not you end up being able to arrange this tests
and measurments in a Guttman scale such that if a given species can do N,
it can do 1 through N-1. My guess is that you will get such a scale and
that it parallel phylogeny. (Though it's possible that birds will skew the

What's this get you? For one thing, it will reveal dependencies among a
whole range of perceptual and behavioral tasks. For another, if the order
turns out as I expect, then you've got a systematic way in to the nervous
system as it basically differentiates and adds new structures with
evolution. And this sort of regime has a natural explanation in terms of
the stack -- we evolve by adding new levels (either at the top or

A final meta-comment. Standard accounts of scientific method focus on
hypotheses and crucial tests. That's fine and dandy when you have
relatively simple models. But the nervous system really is complex. And
HPCT is a relatively complex model; 2 basic ideas (the stack and
reorganization) which imply some complex stuff. To validate models of such
complexity not only do you need many hypotheses & associated tests, but the
hypotheses and tests are inter-linked. I don't think standard accounts of
scientific method are adequate to this kind of investigation.


William L. Benzon 518.272.4733
161 2nd Street
Troy, NY 12180 Account Suspended
What color would you be if you didn't know what you was?
That's what color I am.