Bolles etc.

[from Jeff Vancouver 950918.1710]

Regarding the thread about Bolles, etc (Bruce A., Bill P. etc). I think
it is dangerous to lump all dynamic equilibruim process as either control
processes, as Bill P. _seems_ to be doing, or not, as Bolles seems to be
doing. A thermostat is a control process; a jostled glass of water that
returns to a level state is not. There has to be something that
functions as a reference signal (though there need not be some entity
that _is_ the set point). See Richardson's (1991) book on feedback
thought for a mathematic distinction (actually, the one subsumes the
other). I do not know whether it behooves the non-reference thinker to
demonstrate no goal or the reference thinker to demonstrate a goal, but
it behooves the scientist to discover which it is for a particular
variable. This is the hypothesis testing I am speaking about these days.

What I am not clear about is where biology, etc. generally stand on this
issue. I was under the impression the homeostatic processes and set
points were an accepted concept (although each must be empirically
demonstrated). Is this not the case?

Later

Jeff

···

_________________________________________________________________________
                           Jeffrey B. Vancouver
Assistant Professor Phone: (212)998-7816
Department of Psychology Fax: (212)995-4018
New York University e-mail: jeffv@xp.psych.nyu.edu
6 Washington Pl., Rm 572
New York, NY 10003

[From Bruce Abbott (950919.1235 EST)]

Jeff Vancouver (950919.0845) --

[Hans Blom, 950919c]
I'm not clear about the situation in biology, but I know that human
and non-human physiology is heavily control oriented. Guyton, in his
well-known textbook, refers to control again and again as the expla-
natory concept. In fact, he says more or less that physiology IS the
study of regulatory (control) processes.

Good, that was my impression. One that I often spew in my classrooms.
Thanks for the confirmation. So, what is the nature of these Bolles'
comments. Are they questioning the status quo?

The anthology in which Bolles's chapter appears (Toates and Halliday, 1980)
derived from the proceedings of a conference held at The Open University,
Milton Keynes, England, in September 1979 on the analysis of motivational
processes. What one gets from reading many of the chapters is an apparent
concensus that control theory, which had been enthusiastically pursued as
the key to understanding the classic "homeostatic" drives and the behavior
related to them, had failed in significant ways under empirical test.
Bolles's paper develops a simple argument based on evolutionary/functional
considerations to explain why one should have expected this result all along.

Bolles starts by imagining a simple system in which a "gustatory stimulus,"
a sensory quality correlated with nourishment, turns on a consummatory
mechanism, which produces eating. Fullness of the gut then produces a
satiety signal, and when the satiety signal becomes large enough, it
inhibits further eating. Bolles presents a simple diagram of this system in
which the gustatory stimulus (positive sign) and satiety signal (negative
sign) meet at a comparator, which produces a signal that turns eating on and
off; there is an arrow running back from eating to satiety and, as
mentioned, from satiety to the comparator. Bolles then reasons as follows:

    The first question is whether such a feeding system would be a regulatory
    system. The answer, curiously, does not seem to be so clear. The satiety
    mechanism does act something like a regulator in that it limits meal size.
    And certainly there is an obvious negative feedback loop that accomplishes
    stability. On the other hand, there is surely nothing in the system that
    is being held constant; there is no semblance of homeostasis. The
    limitation on meal size appears to do nothing toward maintaining a
    constant internal milieu. The system strikes us as far too simple to be
    regulatory, even though it does contain a control device.
        (Bolles, p. 65)

In the system Bolles describes, the comparator simply emits a go/no-go
signal to the feeding mechanism when the comparison is sufficiently positive
to exceed a threshold value. Given that there is a gustatory stimulus
present, the animal will feed so long as the gut is not too full. In the
absence of a gustatory stimulus, the animal will simply starve. It will not
alter its behavior so as to make the discovery of food more likely.

    The second question is whether such a primitive system could possibly
    serve the nutritional needs of a real animal. The answer appears to be
    no; the system is much too simple to work in the real world. There is
    nothing in the system that makes contact with motivation. Such an
    animal would never get hungy in any psychological sense, it would simply
    eat (for a while) in the presence of food. . . . The feeding system
    does not even make contact with the physiology of the animal. The
    nutritional state of the beast is not a factor in the feeding system.
    All the feeding system does is to insure that the animal will not eat
    too much at any one time. Such a system could not be viable.

    But the truth is that it is viable; it is, in fact, the system used by a
    great multitude of animals. The most familiar example is the fly
    (Dethier, 1976). . . . There are several aspects of the fly's situation
    that make such a simple feeding system workable. One is that the fly
    has great mobility and is, in some sense, always looking for food.
    Another factor is that the fly has distal chemoreceptors that direct
    taxic behaviors in addition to the proximal "G" receptors that actually
    trigger feeding. The most important consideration, however, is simply
    that there are lots of flies. Flies reproduce with an r-selection
    strategy (Pianka, 1970); they come from big families. The fly's
    feeding system is well suited to an environment where there is abundant
    fly food but it is not very useful in an environment where food is in
    short supply. With a serious shortage the fly just shrivels up and
    dies. but that is not a serious matter in the overall scheme of things
    because the fly has so many siblings and cousins that can reproduce its
    genes. There is no shortage of flies because, like everything else in
    the fly's world, its feeding system is designed to capitalize on good
    times. The fly does not have to defend against adversity because its
    reproductive strategy does that well enough already. It can, therefore,
    prosper with a primitive, one-sided, nonhomeostatic feeding system.
        (Bolles, p. 64)

Bolles goes on to argue in the balance of the paper that feeding systems in
general are non-homeostatic "because there really is no mechanism within the
system to hold anything close to some preset value," but do achieve some
degree of regulation "by protecting the system from a particular kind of
disturbance." This is the conception I talked about in my first post
describing Bolles's view of regulation. If it is important to prevent a
quantity from exceeding certain values, a mechanism will have evolved to
limit its excursion. In the case of the fly, only the excursion of gut
content above a certain volume is limited; excursions below a certain volume
are passively limited by the fact that the fly is always on the move and,
given an environment sufficiently rich in fly food, the fly will always
initiate feeding in time to save itself.

I really have no quarrel with most of Bolles's analysis. It comes down to
this: in general one should not expect to find simple control systems
directly regulating some fundamental quantity (e.g., energy balance, body
weight) via the classic mechanism embodied in the standard diagram.
Instead, the mechanisms to be found most likely will consist of systems in
which excursions of a variable are actively limited only when the
environment makes such limits necessary, and even then the variables under
direct control may not be those whose stability is thus usually guaranteed,
as when, by feeding whenever there is opportunity and the gut is
sufficiently empty, the fly guarentees that its nutrient reserves will be
maintained. Other mechanisms which limit the amount of nutrient the fly can
store or alter the metabolic rate may act in concert with this simple
feeding mechanism in such a way that circulating nutrient levels, body
weight, and so on are maintained with high precision at a nearly constant
value without there necessarily being any single, specific control system to
detect the levels of these quantities, compare them to a specific reference
value, and take action based on the error.

Bolles's argument goes farther than this; to do it justice I would have to
quote most of his article. Instead, I would encourage anyone who is
interested in the issues Bolles raises to read it. The writing is clear and
at times even funny. Yet it is easy to conclude from Bolles's article that
classical control principles simply do not apply to real organisms; that
regulation is accomplished in other ways, usually more aptly described as
equilibrium processes. What seems to be lacking in Bolles's exposition is
any recognition that the analytic tools of control theory apply just as well
to the kinds of systems he describes as they do to systems that physically
parallel the classic diagram. Even the fly's little feeding mechanism is a
perfectly competent control system that attemtps to keep the perception of
gut distension between certain limits despite the disturbances acting upon
it, in almost exactly the way that a flush-toilet maintains its perception
of water-level in the storage tank nearly constant. Yet for Bolles, only
the satiety mechanism, acting to shut off feeding, qualifies as a "control
device."

Regards,

Bruce

[Avery.Andrews (950920)]
(Bruce Abbott (950919.1235 EST))

Even the fly's little feeding mechanism is a
perfectly competent control system that attemtps to keep the perception of
gut distension between certain limits despite the disturbances acting upon
it, in almost exactly the way that a flush-toilet maintains its perception
of water-level in the storage tank nearly constant. Yet for Bolles, only
the satiety mechanism, acting to shut off feeding, qualifies as a "control
device."

But I only noticed you describing one side of it as an actual control
system, the side that Bolles describes that way. The story as I understood
it (from your account, not the original) was that a perception of too
much food in the gut inhibited feeding, but that there was no perception
of too little in the gut. What keeps the gut full is what might be called
`friendly dynamics', rather than control systems.

Avery.Andrews@anu.edu.au

[From Bruce Abbott (950920.0845 EST)]

Avery.Andrews (950920) --

(Bruce Abbott (950919.1235 EST))

Even the fly's little feeding mechanism is a
perfectly competent control system that attemtps to keep the perception of
gut distension between certain limits despite the disturbances acting upon
it, in almost exactly the way that a flush-toilet maintains its perception
of water-level in the storage tank nearly constant. Yet for Bolles, only
the satiety mechanism, acting to shut off feeding, qualifies as a "control
device."

But I only noticed you describing one side of it as an actual control
system, the side that Bolles describes that way. The story as I understood
it (from your account, not the original) was that a perception of too
much food in the gut inhibited feeding, but that there was no perception
of too little in the gut. What keeps the gut full is what might be called
`friendly dynamics', rather than control systems.

Perhaps I'm misunderstanding Bolles's intent, but after presenting a proper
diagram of this little control system, he described it as "containing" a
control system, the satiety mechanism. The "satiety mechanism" simply
provides the negative feedback; it is not the control mechanism itself,
which would include the whole loop plus the reference signal input. It is
possible that Bolles understands this, but is simply using the term "control
mechanism" to refer to the feedback mechanism through which control is
established.

As you suggest, it is a one-way control system, and that is one of Bolles's
main points. As he takes pains to point out, what keeps the fly from
starving is not a growing motivation to locate food as the gut empties, but
an environment in which food will almost certainly be encountered in the
fly's travels. In this regard the fly's gut is a bit like a rain barrel--if
it rains often enough the barrel will always contain water. If the
environment is friendly enough that a certain critical quantity never
reaches a critical value in the lives of most flies ("friendly dynamics"),
there will be no natural selection for mechanisms that impose a limit
through control action. In such cases we should not be surprised to find
that such a mechanism does not exist.

Regards,

Bruce