Interesting journal article

[From Bruce Abbott (970401.0830 EST)]

The following paper appeared in this month's issue of _Behavioral
Researcher_. I thought it would be of interest to CSGnetters.

                        Rocks as Control Systems

                             Hiam J. King
                        Quaker State University

                               Abstract

Six rocks of various sizes and shapes were tested for their ability to
control their position on a 4 X 6 foot table. The tests revealed, contrary
to expectation, that rocks are excellent control systems. It is concluded
that perceptual control theory (Powers, 1973), which was designed to account
for human and animal behavior, may have important applications that extend
beyond the realm of the animate.

                             Introduction

Perceptual control theory (Powers, 1973) asserts that living organisms are
in fact living control systems, which are organized so as to maintain their
inputs (perceptions) close to internally-specified reference values. These
reference values may be relatively fixed (as in the homeostatic control of
blood-sugar levels; Cannon, 1932) or may vary (as in servomechanisms, in
which the input is made to "track" or follow the reference). In either
case, deviations of the input from its current reference level produce an
action on the part of the system that tends to push the input back toward
the reference; thus, a characteristic of a control system is its ability to
counteract or resist disturbances to the input variable that otherwise would
tend to move the input away from its reference value.

Although control has been demonstrated in organisms as diverse as bacteria
and human beings, it has been thought that inert, non-living objects do not
exhibit this phenomenon, unless specifically designed to do so by human
engineers. However, this belief represents more an assertion of faith than
good science, as it has not to our knowledge been submitted to rigorous
test. The present study corrects this deficiency by examining the ability
of inanimate objects to control. To make this test as rigorous as possible,
I chose as subjects objects that seemed entirely unlikely to exhibit any
evidence of control, namely a set of six ordinary rocks of varying sizes and
shapes. Because one of the most notewothy characteristics of rocks is their
tendency to stay put, I chose to test for the rocks' ability to control
their position on the table.

                                  Method

_Subjects_

The subjects were six rocks of varying shapes and sizes that were acquired
from a stream-bed in Clarion, PA. They weighed between 131 and 972 grams,
and had average diameters of between 2.3 and 11.6 cm. Whether the rocks
were sedimentary, metamorphic, or igneous was not determined.

_Apparatus_

The apparatus consisted of an ordinary 4 X 6 foot (121.9 X 182.9 cm) table,
which was carefully leveled using an accurate carpenter's level (Stanley
Co.), to prevent gravitational effects from biasing the results.

_Procedure_

Each rock was tested individually, in a randomly-determined order. The rock
to be tested was placed at the approximate center of the table and observed
for five minutes in order to assess the rock's tendency for spontaneous
movement. Then the rock was pushed against (by the investigator's right
index finger) from several directions and with varying force in order to
assess the rock's ability to resist being displaced from its current
position. When this external disturbance was removed, the rock was again
observed to determine whether it would then return to its former position on
the table.

                               Results

A surprisingly simple and reliable pattern of results emerged from testing.
None of the six rocks was observed to move during the initial period of
observation. When disturbances were applied, each of the rocks eventually
began sliding across the table in the direction of the applied force
(although the force required to start this action varied directly with the
size of the rock). Finally, when the disturbance was removed, each rock
remained where it had come to rest.

                              Discussion

Our initial impression was that the rocks had failed to demonstrate any
evidence of control over their positions on the table. However, after
giving the problem some thought, I realized that in fact the results were
completely consistent with the view that the rocks were indeed controlling
their positions, but in an unexpected way. The entire pattern of results
can be accounted for within a control-system framework if it is assumed that
rocks are at least two-level control systems. In this view, the lower-level
system controls the position of the rock as previously hypothesized.
However, the reference value for this system is determined by the output of
the upper-level system, which is controlling for _pleasing the
experimenter_! This upper-level system senses the force being exerted on
the rock by the experimenter's push, and varies the lower-level system's
reference level for position in such a way that the rock then moves in the
direction the rock senses that the experimenter wants it to go. This
accounts for why the rock stays where it is put, and moves in the direction
it is pushed.

The implications of this finding, if confirmed, are enormous. For one
thing, it suggests that even inanimate objects have desires and purposes and
will attempt to achieve them under proper conditions. The rocks tested
evidently were positively disposed toward the experimenter (perhaps because
he saved them from further wear by the action of water in the streambed),
but it would be premature to conclude that rocks in general hold this
attitude. For example, the falling rocks encountered along highways cut
through the sides of mountains may be attempting to get even with drivers
for disturbing their environment. Additional research will be required to
determine whether this is the case.

In our laboratory, we are already engaged in examining the range of this
phenomenon. Currently we are getting set up to determine whether ice cubes,
by sensing their surface temperatures, can infer the desire of the
experimenter to melt them, and act accordingly.

                                References

Cannon, W. B. (1932). _The wisdom of the body. New York: Norton.

Powers, W. T. (1973). _Behavior: The control of perception._ Chicago:
     Aldine.

[From Bill Powers (970401.0719 MST)]

Bruce Abbott (970401.0830 EST)--

The following paper appeared in this month's issue of _Behavioral
Researcher_. I thought it would be of interest to CSGnetters.

Thank you for finding this important article, Bruce. Here is a reply which I
hope to offer to JEAB; it would be helpful if you would write a cover letter
introducing me to the editors.

            A commentary on King's 'Rocks as Control Systems'
                          William T. Powers

Hiam J. King's stimulating analysis of rocks as control systems calls for a
careful response. As the primary cited author I feel compelled to say that
in the years since the cited work was published in 1973, I have repeatedly
encountered a different interpretation of the kinds of phenomena that King
has investigated, and have realized that it has an appeal that is difficult
to resist. I find that in recent months I have come to offer this new
interpretation more and more frequently, to the point where the class of
utterances I have previously produced has all but ceased to occur.

King's report on the behavior of rocks led to a situation in which
observations of a set of similarly-derived rocks on a table that resembled
the one described were occurring. These rocks, however, were first observed
from the banks of a stream where the same class of rocks could be found. The
water was abundant and moving very fast, and some of the rocks could be seen
moving about in the more energetic parts of the flow. The rocks gave the
appearance of moving about purposively, as if to escape from the more
turbulent parts of the flow and seek positions of rest in calmer waters, in
apparent confirmation of King's report of purposive behavior.

However, it became clear that no rock moved except that rushing water was
exerting a strong force upon it. The rocks' movements were clearly a
response to the rushing of the water, which served to discriminate the
situation in which movement was appropriate from that in which it was not. A
survey of many rocks showed that there were two classes of rocks: those
which moved in response to rushing water, and those whose movements ceased
and did not again occur. It was also observed that every rock which was
initially observed to move eventually became one of the class of rocks the
movement of which had ceased.

While a number of interpretations suggested themselves, one gradually came
to be more likely than the others: the only explanation that coincided with
all the observations was that the rocks were being punished for moving in
response to the class of situations typified by being in fast-moving water,
and that the behavior of moving into quiet water and ceasing to move was
being negatively reinforced by the cessation of the collisions with other
rocks (contingent on being in fast-flowing water).

While rocks were being collected, it was observed that when the hand of the
experimenter grasped various rocks and lifted them, those rocks which moved
most readily in the rushing water also moved most readily when in the
experimenter's hand, while those rocks that were least stimulated into
movement by the water were also the least responsive to lifting. Clearly,
the larger the spatial dimension of the rock, the more susceptible it was to
negative reinforcement, and the less readily it could be conditioned to move
in the presence of the class of stimuli associated with applied forces.
These apparently contradictory findings actually reveal two distinct
phenomena: punishment for moving, and negative reinforcement for avoiding
the discriminative stimulus of rushing water. The behavior of the rocks
reflects the balance between these two contingencies, as in [Bruce, perhaps
you can supply me with an appropriate reference to the literature on choice
behavior].

These observations, which occurred during the collection of rocks that
preceded the replication of the King experiment, have greatly influenced the
conclusions drawn after the replication was finished. The most probable
conclusion is now that the rocks had no magical ability to divine the
so-called "desires" of the experimenter, a response to the situation which
is obviously no more than a supersitious vestige of our timid, savage past.
The clearly indicated conclusion is that if the rocks did not spontaneously
move, this was a result of their _history of reinforcement_. This history,
which showed itself during the process of collecting the rocks, adequately
explains why the rocks moved under the stimulus of the experimenter's push
(the larger rocks, as previously observed, moving the least readily), and
why, when they reached a position where forces were no longer exerted upon
them, movement was extinguished, the most throughly in the case of the
larger rocks.

Subsequent experiments have reinforced this conclusion beyond doubt. Igneous
rocks were collected by chipping them out of ancient formations which had
clearly never been subject to aqueous influences, and they exhibited exactly
the same behavior. While they were being collected, it was observed that the
larger rocks showed a lower tendency to respond to chipping forces than the
smaller rocks, and that all rocks showed a strong tendency to cease their
movements when they escaped from the chipping action (Thorndyke, 1925).
Similar experiments with wood chips of various sizes showed that this
phenomenon occurs whether the material is organic or inorganic (ref). While
no experiment by itself has been conclusive, the preponderance of evidence
from many related studies shows that the basic phenomenon occurs reliably
and repeatably. There is even strong evidence that Isaac Newton observed a
similar relationship between size and tendency to respond to the class of
stimuli known as "forces" with a class of responses known as
"accelerations." Newton himself nearly discovered the operant over 350 years
ago!

It is unfortunate that King was influenced by a series of misleading
contingencies that produced what can now be seen as a superstitious pattern
of behavior. The conclusion that rocks are purposive is not supported by
observation. It is self-evident that rocks obey the same laws of behavior
that govern all other objects, as shown by the above experimental findings.

···

----------------------------------

Happy 970401,

Bill P.

                              Abstract

Six rocks of various sizes and shapes were tested for their ability to
control their position on a 4 X 6 foot table. The tests revealed, contrary
to expectation, that rocks are excellent control systems. It is concluded
that perceptual control theory (Powers, 1973), which was designed to account
for human and animal behavior, may have important applications that extend
beyond the realm of the animate.

                            Introduction

Perceptual control theory (Powers, 1973) asserts that living organisms are
in fact living control systems, which are organized so as to maintain their
inputs (perceptions) close to internally-specified reference values. These
reference values may be relatively fixed (as in the homeostatic control of
blood-sugar levels; Cannon, 1932) or may vary (as in servomechanisms, in
which the input is made to "track" or follow the reference). In either
case, deviations of the input from its current reference level produce an
action on the part of the system that tends to push the input back toward
the reference; thus, a characteristic of a control system is its ability to
counteract or resist disturbances to the input variable that otherwise would
tend to move the input away from its reference value.

Although control has been demonstrated in organisms as diverse as bacteria
and human beings, it has been thought that inert, non-living objects do not
exhibit this phenomenon, unless specifically designed to do so by human
engineers. However, this belief represents more an assertion of faith than
good science, as it has not to our knowledge been submitted to rigorous
test. The present study corrects this deficiency by examining the ability
of inanimate objects to control. To make this test as rigorous as possible,
I chose as subjects objects that seemed entirely unlikely to exhibit any
evidence of control, namely a set of six ordinary rocks of varying sizes and
shapes. Because one of the most notewothy characteristics of rocks is their
tendency to stay put, I chose to test for the rocks' ability to control
their position on the table.

                                 Method

_Subjects_

The subjects were six rocks of varying shapes and sizes that were acquired
from a stream-bed in Clarion, PA. They weighed between 131 and 972 grams,
and had average diameters of between 2.3 and 11.6 cm. Whether the rocks
were sedimentary, metamorphic, or igneous was not determined.

_Apparatus_

The apparatus consisted of an ordinary 4 X 6 foot (121.9 X 182.9 cm) table,
which was carefully leveled using an accurate carpenter's level (Stanley
Co.), to prevent gravitational effects from biasing the results.

_Procedure_

Each rock was tested individually, in a randomly-determined order. The rock
to be tested was placed at the approximate center of the table and observed
for five minutes in order to assess the rock's tendency for spontaneous
movement. Then the rock was pushed against (by the investigator's right
index finger) from several directions and with varying force in order to
assess the rock's ability to resist being displaced from its current
position. When this external disturbance was removed, the rock was again
observed to determine whether it would then return to its former position on
the table.

                              Results

A surprisingly simple and reliable pattern of results emerged from testing.
None of the six rocks was observed to move during the initial period of
observation. When disturbances were applied, each of the rocks eventually
began sliding across the table in the direction of the applied force
(although the force required to start this action varied directly with the
size of the rock). Finally, when the disturbance was removed, each rock
remained where it had come to rest.

                             Discussion

Our initial impression was that the rocks had failed to demonstrate any
evidence of control over their positions on the table. However, after
giving the problem some thought, I realized that in fact the results were
completely consistent with the view that the rocks were indeed controlling
their positions, but in an unexpected way. The entire pattern of results
can be accounted for within a control-system framework if it is assumed that
rocks are at least two-level control systems. In this view, the lower-level
system controls the position of the rock as previously hypothesized.
However, the reference value for this system is determined by the output of
the upper-level system, which is controlling for _pleasing the
experimenter_! This upper-level system senses the force being exerted on
the rock by the experimenter's push, and varies the lower-level system's
reference level for position in such a way that the rock then moves in the
direction the rock senses that the experimenter wants it to go. This
accounts for why the rock stays where it is put, and moves in the direction
it is pushed.

The implications of this finding, if confirmed, are enormous. For one
thing, it suggests that even inanimate objects have desires and purposes and
will attempt to achieve them under proper conditions. The rocks tested
evidently were positively disposed toward the experimenter (perhaps because
he saved them from further wear by the action of water in the streambed),
but it would be premature to conclude that rocks in general hold this
attitude. For example, the falling rocks encountered along highways cut
through the sides of mountains may be attempting to get even with drivers
for disturbing their environment. Additional research will be required to
determine whether this is the case.

In our laboratory, we are already engaged in examining the range of this
phenomenon. Currently we are getting set up to determine whether ice cubes,
by sensing their surface temperatures, can infer the desire of the
experimenter to melt them, and act accordingly.

                               References

Cannon, W. B. (1932). _The wisdom of the body. New York: Norton.

Powers, W. T. (1973). _Behavior: The control of perception._ Chicago:
    Aldine.

[From Rick Marken (970401.0840)]

Bruce Abbott (970401.0830 EST) --

The following paper appeared in this month's issue of _Behavioral
Researcher_.

A most illuminating article.

It is concluded that perceptual control theory (Powers, 1973), which
was designed to account for human and animal behavior, may have
important applications that extend beyond the realm of the animate.

This is a very exciting discovery because it shows that PCT can account
for a phenonmenon that was previously accounted for only by reinforcment
theory and Christianity -- the behavior of dead people.

Of course, everyone is probably aware of the biblical descriptions of
the phenonenon of dead people behaving (Lazarus, Jesus) but I'm not sure
everyone is aware of the operant research that has been done on this
topic. One such research project was described in a monograph that I ran
across by chance many years ago while browsing through the stacks of the
library at the University of Minnesota. Unfortunatly, I do not remember
the author or title but I do remember the research itself.

Operant conditioning was used to shape the behavior of a corpse. The
corpse was that of a person who had just died so the body still made
the occasional twitch. These twitches provided the raw material for the
operant conditioning procedure. Reinforcements (I can't recall
what these were; possibly electrical stimulation to the reinforcement
centers of the brain) was initially delivered after ANY twitch. Once
this reinforcement procedure produced a high enough rate of twitching,
the shaping procedure began. Reinforcement was delivered only after
twitches that were a closer approximation to the desired behavior
(which, I believe, was playing a Chopin mazurka) than previous twitches.
This was, of course, a slow and tedious process but, eventually, the
researcher had the corpse playing the mazurka and
a couple of etudes to boot.

I don't know if PCT can explain the kind of behavior the operant
researcher was able to get out of a corpse. But I think King's
demonstration of PCT's ability to handle the behavior of rocks
is a good start.

Best

Rick