Special-purpose systems

[From Bill Powers (940717.1640 MDT)]

General comments on "alerting."

If "alerting" is a special phenomenon that requires treatment different
from that given to any other control phenomenon, then it seems to me
that there are many other phenomena equally deserving of a special place
in the PCT model.

There are, for example, "foraging" signals which indicate to the
organism that it should allocate some resources from the limited output
degrees of freedom to executing search strategies likely to produce

There are also "mating" signals, which indicate the presence of a
potential mate and tell the systems to switch to controlling for

And how about "calming" signals, which tell the systems that there is no
longer any need to be alert because all important variables are under
control and lower-priority activities can now safely be resumed.

Then we have "aggression" signals, which tell the organism that it
should switch from a defensive mode of action to a more proactive mode.

The list could probably be extended considerably by going through old
psychology books and finding all the traits and tendencies and
proclivities and instincts that have been observed and named from the
time of Aristotle onward. The classification and naming of patterns of
overt behavior is an ancient and honorable pursuit, as is the attempt to
find generalizations describing the circumstances under which they are

This, however, is not how modeling works. Modeling is concerned with the
organization of a system that can show all the classes and types of
behavior that are seen, under all possible circumstances. Taxonomy
creates the appearance of a huge number of unrelated patterns; modeling
attempts to show that behind all these patterns there is a single system
which behaves in the same way no matter what is going on. Once we start
trying to install specific architectures aimed at explaining how
specific behaviors occur under special circumstances, there will be no
end to it: the model, instead of being a general-purpose system, will
become a large set of special-purpose sub-organizations. This, it seems
to me, leads away from the goal I am pursuing.



Bill P.

<[Bill Leach 940717.21:19 EST(EDT)]

[From Bill Powers (940717.1640 MDT)]

I think that I find myself in full agreement with what you have said. I
think that PCT (or at least HPCT) should have the ability to explain all
such phenomenon (and it sure looks as though it can).

I also think that the 'rest of the world' (which in this case happens to
include myself) could well benefit from PCT/HPCT descriptions of such
phenomenon (as opposed to descriptions from some other theory of

If for no other reason than such descriptions would help people to
understand that there really is only one phenomenon that is significant
-- feedback control of perception.

OTOH, it is also easy to see where trying to say that such definitions or
descriptions are a "part" of PCT would indeed turn PCT in the same
encyclopedia of "observed phenomenon" as other behavioural "sciences".


From Tom Bourbon [940719.1202]

Bill P., how about the Jovian show? Amazing, isn't it?

[From Bill Powers (940717.1640 MDT)]

General comments on "alerting."

If "alerting" is a special phenomenon that requires treatment different
from that given to any other control phenomenon, then it seems to me
that there are many other phenomena equally deserving of a special place
in the PCT model.

And you continued with a nice list of "signals" that psychologists and other
behavioral scientists have advanced over the years. I am running late on
catching up with the mail and for a variety of reasons I have not jumped
into the discussion on "alerting signals" or "alerting stimuli." But from
time to time recently I have glanced through the indices of a few texts on
learning and conditioning, looking for other "stimuli" (aka "signals") that
are just as deserving of consideration for special status as are alerting
signals-stimuli. (Just as deserving, meaning not at all deserving.) Here
are a few that we would need to consider, were we to begin looking for ways
to model isolated "phenomena" carried over directly from traditional

If anyone is inclined to continue the attempt to port these traditional
concepts into PCT uncritically, have at it -- and keep us posted on the
progress of your modeling. You could begin by telling us the criteria you
use to identify the occurrence of the phenomenon associated with each of
these special stimuli. Remember, somewhere in behavioral science behavior
there are people who love each of these stimuli and believe they cause

Signals-stimuli from traditional psychology -- a partial listing:

fear-eliciting stimulus / fear stimulus
love s. / affectional s.
imprinting s.
eliciting s.
covert s.
discriminative s.
habituating s.
cognitive s.
primary s.
secondary s.
conditioning s. / conditioned s.
unconditioned s.
reinforcing s.
controlling s. / control s. (the s. that controls behavior)
drive s.
movement-produced s. / response-dependent s. / response-correlated s.
hedonic s.
aversive s. / appetative s. / neutral s. (think about that set for a while)
internal s. / external s.
intrinsic s. / extrinsic s.
potential s. / effective s. / functional s.
sensitizing s.
derived s.
private s.
textual s.
negative s. / positive s.
incidental s.
releasing s.
species-specific s.
disruptive s. (is it like an alerting s.?)
chained stimuli (something for people with more exotic sexual preferences?)

and one of my very favorite kinds of stimuli, the supraordinate s.

This list could go on forever, but I'm starting to get a little silly. I
think the idea is clear. Each of these "kinds" of stimulus has been
postulated to "explain" an allegedly different kind of behavioral
phenomenon. This list, and any other like it, is a museum piece -- a relic
from the day when behavioral scientists took seriously the notion that
behavior occurs in neat packets called "responses," with each "kind" of
response controlled by its own class of independent antecedent stimulus.
That's all history now -- why don't our editors and reviewers wake up to
that fact?

Alerting stimuli, anyone?