Are plants perceptual control systems?

[From Kent McClelland (2015.11.07 13.20]

Yes, plants are very definitely control systems, at least as I understand it. There’s a new branch of plant biology that has begun to document plant behavior (it’s controversial, of course). Plants move their stems and leaves to control access to sunlight, and they change the directionality of growth of their roots to control access to water, nutrients, and other chemicals in the ground. Here are a couple of relevant URL sources.

http://video.pbs.org/video/2338524490/

http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0029759#pone-0029759-g005

You may not be able to get the PBS video outside the US, but it gives interviews with botanists who study plant behavior and shows some of their experiments.

Kent

···

On Nov 7, 2015, at 11:26 AM, Rupert Young <rupert@perceptualrobots.com> wrote:

[From Rupert Young (2015.11.07 17.30)]

I couldn't find a discussion on this from a search of CSG. We always talk of living systems being perceptual control systems, but does that include plants? If so, what perceptions do they control and how do they do it?

Regards,
Rupert

[From Rupert Young (2015.11.07 17.30)]

I couldn't find a discussion on this from a search of CSG. We always talk of living systems being perceptual control systems, but does that include plants? If so, what perceptions do they control and how do they do it?

Regards,
Rupert

[Martin Taylor 2015.11.07.14.53]

[From Rupert Young (2015.11.07 17.30)]

I couldn't find a discussion on this from a search of CSG. We always talk of living systems being perceptual control systems, but does that include plants? If so, what perceptions do they control and how do they do it?

It was discussed a few times some years ago. The answer is yes, they are control systems, but their control loops (with some notable exceptions) are mostly slow, and depend on chemistry rather than electrical impulses. Watch a stop-motion movie of a field of sunflowers over a day. Lots of flowers do that, but it's pretty dramatic to see a whole field seeming to look straight at the sun as it moves across the sky, and it's hard not to see it as control, though I know of no TCV type studies to test exactly what about the sunlight is being perceived and controlled. (I try, unsuccessfully, to get our clematis to allow its flowers to face the house to its north, but all that test shows is that the plant actively controls a perception of something about the sun direction). Over a period of years, watch what a tree does after it loses a major branch in a way that opens a gap in the canopy. Again, without the TCV, we can't be sure just what perception the tree controls.

In the case of animals, one can postulate that there might be a place in the neural system where we could localize the specific functions and signals involved in controlling a particular perception. That's harder in the case of plants that don't have a centralized hub corresponding to a brain, but there's nothing in PCT that says control depends on localization. Lack of localization simply makes it harder to visualize what happens around the control loop and map that onto the organism's structure.

Martin

[Martin Taylor 2015.11.08.10.18]

[From: Richard Pfau (2015.11.08 8:30 EST)]

Rupert and Kent,

      Yes, plants are control systems, but not "perceptual

control systems" since Bill’s description of such systems is
based on organisms that have nervous systems based on neurons,
and plants do not have such nervous systems. (At least, that
is my understanding of PCT).

With Regards,

Richard Pfau

    Not my

understanding pf PCT. In my understanding, nervous systems are
just one possible method of conveying the variable values that
are involved in control. For example, chemical concentration is
a variable that could be a perceptual value or a reference
value. Or (in a nervous system) the perceptual value might exist
nowhere localizable, but could be the effect of multitudes of
fibres firing individually – come to think of it, isn’t that
where Powers gets his “neural current” value for his different
variables? Neural currents don’t exist anywhere in a nervous
system, do they?

    All PCT needs is some kind of way of comparing two variable

values and a way for the difference to influence one of them so
as to reduce the its value. That one is called a “perception”.
It doesn’t matter how the variables are represented within the
control system.

    Martin

[Tracy Harms 2015.11.08.11.11 EST]

In reply to Martin Taylor 2015.11.08.10.18

I agree with this, Martin. It is, to my mind, important to associate PCT with the systemics of behavior, with recognition that many mediums for such systems are possible.

I like it whenever I see us work with examples of perceptual systems in plants. There are advantages to examining behavior in its simpler manifestations. I know Bill was motivated by his abiding interest in psychology, but I find PCT strengthened by emphasis on basics that fall more within the specializations we call biology.

Tracy

[Martin Taylor 2015.11.08.14.06]

[From: Richard Pfau (2015.11.08 12:53 pm EST)]

Ref: [Martin Taylor 2015.11.08.10.18 10:29 am]

      We agree that control can occur in various ways, including

ways that don’t include nervous systems (neurons), as in
plants.

      However, PCT, as described by Bill Powers, focuses on the

nervous system (and neurons). Consider his definitions (from
Behavior: The Control of Perception, 1973 & 2005)

        "PERCEPTION: A perceptual signal (inside a system) that

is a continuous analog of a state of affairs outside the
system."

        "PERCEPTUAL SIGNAL: The signal emitted by the input

function of a system; an internal analog of some aspect of
the environment."

        "SIGNAL: A physical variable inside a system that gives

processes at one location a quantitative unidirectional
effect on processes at another location. ** A train
of neural impulses having a magnitude measured in units of
neural current, or impulses per second."** [Note:
bold type added by me/Pfau]

      And so, it seems clear that Bill is referring to

organisms having neurons and neural signals, not to all kinds
of control systems.

        That may be

your reading of the bolded sentence. Mine is that he describes
an example, because that example is what he wanted to pursue in
his book and most other writings. If we accept your
interpretation, we have to account for Bill’s acceptance of
perceptual control in organisms such as the e-coli bacterium.
(Which he used as an example when explaining how e-coli
reorganization worked).

    There's another point, quite independent of how to interpret the

bolded sentence, and that is whether science is best served by
restricting the application of a tool (concept, theory,
hypothesis) to the area for which the inventor intended it.
Fourier invented what we know as Fourier Analysis to study the
propagation of heat through a medium. Should we therefore ban
its use in electronics or acoustics? The world would be very
different if we did.

      It's your choice if you choose to generalize

beyond PCT as originated by BIll. However, rather than doing
so, you might consider Humberto Maturana’s thoughts. Maturama
states (in * From Being to Doing: The Origins of the
Biology of Cognition*,2004):

        "My own view is...that using a concept outside

its proper context of application means committing a double
fault: the concept will work properly neither in the
original nor in the new domain." (p. 106)

        This is a

general statement about the application of metaphor. We are not
using metaphors, unless you deny Bill’s main point, that the
biological system is subject to the same laws of physics as is
an engineered system. That’s a major reason why PCT is superior
to psychological or biological theories in which the key
concepts ARE metaphors – things work sorta like this, see! It’s
quite clear how to test whether perceptual control is occurring,
though the test may not always be practical.

    It's not clear whether ...
      For example, referring to Niklas Luhmann,

expanding upon use of the concept of “autopoiesis” to broad
social systems, he states:

        "Niclas Luhmann uses the concept of autopoiesis

as a principle in the explanation of social phenomena, which
does not illuminate the processes to be described nor the
social phenomena but tends to obscure them." (p. 107) [He
then goes on to explain further]

        ...

“autopoeisis” can be defined precisely either as a process or a
state, but if you could, you would be clear whether it applies
in a different situation. Use it as a sorta-like-that metaphor,
and you can get into trouble.

      The point is that you may choose to go beyond

Bill’s thinking in applying and linking PCT concepts to areas
such as the “perception” of plants. But doing so, if linked
to PCT, may dilute and damage PCT in the long run (as
Maturana’s thinking suggests).

        I disagree.

If there’s no evidence of control, then PCT is not applicable.
If control can be demonstrated, then other aspects of PCT can be
tested to see whether they are applicable. You can’t do that
with “autopoeisis”, can you? Nothing can dilute the concept of
perceptual control other than the use of it as a metaphor. If
you can hypothesise, or preferably determine, the different
components of a control loop, you are not using it as a
metaphor. “Neural Current” IS a metaphor, and one that might
cause trouble some day. Nobody expects to find any neural
currents in the brain, but that does not invalidate PCT, because
whether controlled perceptions are localizable somewhere or are
distributed over a region of the brain is irrelevant. The effect
is AS IF there is a neural current, so far as experiments can
show.

      So be careful!  In either case, as Bill's

definition and books indicate, he was describing PCT as
applicable to organisms as systems that have neurons and
nervous systems, not to all types of control systems.

        Yes, that's

where most of his thinking and writing were directed. By no
means does that imply that other people should not see how the
mathematics and the underlying concepts and theories apply in
other domains. Never did I notice (that I remember) Bill raising
a complaint when it was suggested that perceptual control was
THE indicator of whether something was alive. If he wanted to
proscribe the application of PCT to organisms without neural
systems, don’t you think he would have raised a teeny objection?

    As an aside, Bill's version of PCT (the strict hierarchy of

eleven intuitively invented levels) is only one of an indefinite
number of different possible forms of Perceptual Control Theory.
The hierarchy has many nice properties, but the core, the genius
of the theory, is that simple scalar perceptual control units
connected in complex structures CAN do an awful lot more than
can any theory built to fit specific data in some small domain.
For PCT, the data fitting is parameter value choice, not changes
in the structure of the theory to model each new situation. And
that includes everything from simple tracking to the madness of
crowds to the ecology of forests or gut microbiota. We just have
to be careful not to rely on metaphor when we (eventually) study
its application in such disparate domains of inquiry.

    Martin
···

mmt-csg@mmtaylor.net
csgnet@lists.illinois.edu

[Martin Taylor 2015.11.08.10.18]

[From: Richard Pfau (2015.11.08 8:30 EST)]

Rupert and Kent,

                Yes, plants are control systems, but not

“perceptual control systems” since Bill’s
description of such systems is based on organisms
that have nervous systems based on neurons, and
plants do not have such nervous systems. (At least,
that is my understanding of PCT).

With Regards,

Richard Pfau

              Not

my understanding pf PCT. In my understanding, nervous
systems are just one possible method of conveying the
variable values that are involved in control. For
example, chemical concentration is a variable that
could be a perceptual value or a reference value. Or
(in a nervous system) the perceptual value might exist
nowhere localizable, but could be the effect of
multitudes of fibres firing individually – come to
think of it, isn’t that where Powers gets his “neural
current” value for his different variables? Neural
currents don’t exist anywhere in a nervous system, do
they?

              All PCT needs is some kind of way of comparing two

variable values and a way for the difference to
influence one of them so as to reduce the its value.
That one is called a “perception”. It doesn’t matter
how the variables are represented within the control
system.

              Martin

[From Rick Marken (2015.11.08.1305)]

···

Kent McClelland (2015.11.07 13.20)_-

KM: Yes, plants are very definitely control systems, at least as I understand it. … Here are a couple of relevant URL sources.

http://video.pbs.org/video/2338524490/

RM: Yes, indeed. Plants are definitely control systems because they can be seen to control, as shown in the wonderful “Nature” video that you posted. Note that the big question at the beginning is “Do plants behave?”. It’s pretty clear to me that what they are trying to figure out is whether plants control! And its pretty obvious from the marvelous time-lapse photography that they do – though usually very slowly. The most obvious examples of plant control, perhaps, are the photo-tropisms that Martin Taylor mentioned (and are shown in the Nature film), where the time-lapse photography shows the plant’s\ flowers tracking the sun as it moves slowly across the sky.

RM: There are many other wonderful examples of plant behavior as control in the nature film. Indeed, it might be nice to include some plant behavior in the “Behavior as control” spreadsheet. The photo-tropisms would be easy. How about someone add it to the spreadsheet!

RM: Once it’s clear that plant behavior is control then we know that the only explanation is control theory (PCT). But, of course, the functions and variables in the model – input and output functions, perceptual, reference and error variables – would have to be implemented by means other than neural nets and neural firing rates. But it would still be a control model, so it would have to work by controlling a perceptual representation of the controlled variable. It would be PCT, for plants.

Best

Rick

Richard S. Marken

www.mindreadings.com
Author of Doing Research on Purpose.
Now available from Amazon or Barnes & Noble