PCT and Behavioral Ecology

[_Gary_A_Cziko]

[from Gary Cziko 970308.2011 GMT]

I've been spending the last month or so studying ethology and behavioral
ecology and am quite impressed with what is now known about the behavior of
many animal species in natural settings even though this research is not
done from a PCT perspective.

Much of this research can be boiled down to finding animal preferences for
things like food, mates, and shelter. For example, northern European
honeybee colonies split into two groups, with the new one prefering large,
insulated locations for setting up a new hive that is far from the original
hive. In contrast, southern European honeybees prefer closer locations
that are smaller and not necessarily insulated. This makes adaptive sense
when one considers the climate and amount of available food sources. Of
course, this research doesn't tell us anything about the specific variables
that are controlled, but it seems useful nonetheless. There is also lots
of research on mate preference and food preference, much of it just from
observing, although occasionally these researchers resort to fiddling with
variables (see below).

I had lunch yesterday with Gene Robinson who studies honeybees and is
interested in how they divide labor within the hive. Young bees do
domestic work such as taking care of the the pupae and "housecleaning,"
while older bees forage for food. But if you remove the foragers, more of
the younger bees will more quickly turn into foragers. So it looks like
some controlled variable is involved here, but what is it? There are lots
of possibilities, but Robinson has fiddled with two in particular--the
amount of food available to the individual bees and the amount of stored
honey in the hive. When he made the bees hungry they progressed more
quickly to the forager stage. But if he removed honey from the honeycomb
(using a vaccum system to suck honey out of the cells) while keeping
another source of food available, they did not develop more quickly. So it
seems that keeping a certain amount of honey stored in the hive's honeycomb
is NOT a controlled variable but rather a side effect of having enough food
for individual consumption. Robinson was not guided in this research by
PCT, or The Test, or by the goal of finding a controlled perceptual
variable, but it seems to me that he is making good progress toward
understanding honeybee behavior just the same.

The moth _Catocala relicta_ has markings on its wings that nicely blend
into those on birch tree bark, but only when it is resting so that its
markings match the orientation of those on the tree, which usually means
with its long axis vertical--which is, as it turns out (thank you, Darwin)
is just the way it usually rests on the tree. But what is the moth
actually controlling? Does it just rest in this position as a side effect
of how it flies and lands? Uses gravity to keep its head up? Uses vision
to line up with the tree bark? Or feels the ridges in the bark with its
feet and lines up accordingly? I don't think the answer is known. But it
seems to me that a sharp researcher like Robinson would find out by doing
things like seeing how it rests in the dark or with its eyes covered, when
the tree trunk rotated 90 degrees, or when little rubber booties are placed
over its feet to find out what is going on. I don't see how explicit
knowledge of PCT would help much, other than preventing him from saying
that he found out what type of sensory information "guides behavior" and
what type does not. He'd be looking at independent variables and how they
influence behavior and coming up with, it seems to me, essentially the same
understanding a PCT entomologist would come up with.

Now here's a final example that doesn't seem to fit PCT at all. Some
reptiles and fish change their coloration in order to match their
background better. The adaptive significance of such behavior is clear
(thank you again, Mr. Darwin), but can this be exlained as the control of
some perceptual variable. The animal can't see its own color, so that
variable is out. Isn't this just a reaction to a perception rather than
the control of one?

Reactions appreciated, from Rick, or course, but others, too.

--Gary

[Richard_Marken]

[From Rick Marken (970308.1100 PST)]

Gary Cziko (970308.2011 GMT) --

I've been spending the last month or so studying ethology and
behavioral ecology and am quite impressed with what is now known
about the behavior of many animal species in natural settings even
though this research is not done from a PCT perspective.

OK. Let's see what is "known" about behavior based on non-PCT
research.

northern European honeybee colonies split into two groups, with the
new one preferring large, insulated locations for setting up a new hive
that is far from the original hive. In contrast, southern European
honeybees prefer closer locations that are smaller and not necessarily
insulated...

But what does this tell us about the variables that the honeybees are
controlling? As you correctly note:

this research doesn't tell us anything about the specific variables
that are controlled,

But then you say:

but it seems useful nonetheless.

I suppose it is useful if you want to herd northern European honeybees
into hives. But what does it tell us about the purposeful behavior of the
honeybees? As it sits, I think it tells us next to nothing. But it certainly
suggests hypotheses about the variables these bees are controlling for;
sensed temperature seems like one possibility. Making these honeybee
observations is what one does _before_ one starts Testing for Controlled
Variables. Now the hard part begins: coming up with hypotheses about
the perceptions which, if controlled by the honeybees, would result in the
behaviors we observe.

I had lunch yesterday with Gene Robinson who studies honeybees and
is interested in how they divide labor within the hive.

Already I can see that Gene is looking at the problem from the wrong
perspective: that of the observer rather than that of the honeybee. I think
it's highly unlikely that honeybees are able to control for anything like
"division of labor". Division of labor is almost certainly a side effect
(seen by an observer) of the fact that honeybees are controlling
perceptions of variables related to chemical concentrations, relative
temperatures, etc.

if you remove the foragers, more of the younger bees will more
quickly turn into foragers.

My guess is that the removal of foragers leads to an increase (for all bees)
in the perceived level of a variable that is being controlled by all the bees.
The young bees with higher loops gains with respect to this variable
would be the first to act on it when the foragers are taken out. Eventually,
enough young bees join in so that this variable is brought back to its
reference level _for all the bees_. But, as you ask "what is this controlled
variable" (or variables)? You say:

There are lots of possibilities, but Robinson has fiddled with two in
particular--the amount of food available to the individual bees and the
amount of stored honey in the hive.

These are, indeed, two possible controlled variables. They are described
from the observer's point of view, but it's a good first step.

When he made the bees hungry they progressed more quickly to the
forager stage.

I presume this means that he disturbed (by removal of food?) the amount
of food available to the bees. Unfortunately, he only monitored the
response to this disturbance, not the state of the (presumed) controlled
variable (amount of food available). This is like monitoring only the
relationship between disturbance and response in the coin game. If the
subject is controlling _something_ about the coins almost anything the
experimenter does to disturb the coins will result in a response. But this
response doesn't tell you much about what it is about the coins that is
being controlled. You should show Robinson the "coin game" and
explain it to him carefully in terms of what the bees might be up to.

But if he removed honey from the honeycomb (using a vacuum system
to suck honey out of the cells) while keeping another source of food
available, they did not develop more quickly.

So he has found that foraging is not related to a variable disturbed by
removing honey from the honeycomb. This is like finding that the subject
in the coin game does nothing (there is no "response") when you open
the window. You are disturbing a variable (sensed temperature) that is
not controlled by moving the coins. It doesn't tell you that sensed
temperature is not controlled.

So it seems that keeping a certain amount of honey stored in the hive's
honeycomb is NOT a controlled variable but rather a side effect of
having enough food for individual consumption.

Not at all. The amount of honey in the honeycomb MAY be controlled;
but all we know from Robinson's research is that foraging is not related
to the variable disturbed when we remove honey from the comb. I would
be very surprised if some perceptual variable related to the amount of
honey in the comb were not controlled by the bees. But Robinson's data
certainly don't speak to whether such a variable is controlled or not.

Robinson was not guided in this research by PCT, or The Test, or by
the goal of finding a controlled perceptual variable,

Yes. And, lo and behold, he found no controlled variables;-)

but it seems to me that he is making good progress toward
understanding honeybee behavior just the same.

I agree that he's just a millimeter away. But not many have been able to
go that extra millimeter. It _might_ help if you would explain to him what
a controlled variable is and why he can't determine what variable is
controlled by simply observing stimulus-response relationships (like
the relationship between access to food and foraging). Use the demos to
show him why he should _monitor_ the hypothetical controlled variable
when applying disturbances in order to determine whether the suspected
controlled variable is (or is not) under control. If he's still paying
attention you might also discuss the importance of considering the
importance of considering the possible _speed_ of control system operation
when applying disturbances and monitoring their effects on possible
controlled variables.

I don't see how explicit knowledge of PCT would help much

It would help him understand what a controlled variable IS and why S-R
relationships cannot be used to infer what these relationships are. PCT
would keep him from coming to foolish conclusions like the one above
( that "honey stored in the hive's honeycomb is NOT a controlled
variable but rather a side effect of having enough food for individual
consumption".)

He'd be looking at independent variables and how they influence
behavior and coming up with, it seems to me, essentially the same
understanding a PCT entomologist would come up with.

Oh, come on Gary. Get real;-)

Now here's a final example that doesn't seem to fit PCT at all. Some
reptiles and fish change their coloration in order to match their
background better.

This implies that you know that obtaining such a match is under control.
How do you know this?

but can this be explained as the control of some perceptual variable.

Only if some perceptual variable is under control.

Isn't this just a reaction to a perception rather than the control of one?

Maybe. Maybe not. But you can't tell by just looking. You have to Test
for a controlled variable.

Reactions appreciated, from Rick, or course, but others, too.

My reaction is that I think you should spend more time doing the PCT
demos and experimnents and less time talking to Bruce Abbott;-)

Best

Rick