Fundamentals of HPCT, Version 0

[From Bill Powers (920926.2200)]

RE: Organismism. This is my first attempt to lay out basic principles
of HPCT in a brief and systematic way. It undoubtedly needs thorough
examination for non-sequiturs, unwarranted (by HPCT) assumptions, and
gratuitous conclusions. I've tried to avoid saying things that are
obviously false in terms of the basic theory. I haven't defined
everything as if for a newcomer. While I haven't restricted my
imagination, I've tried to remain noncontroversial in my assumptions.
Have at it.

ยทยทยท

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

A. The basic architecture of human organization, according to HPCT.

A-1. The behavioral outputs of thge human organism consist of forces
applied by effectors to the environment (which includes the physics of
the body). All actions of the CNS on the environment take place
through these effectors.

A-2. The environment can affect the CNS through the sensory nerves.

A-3. The environment can affect the human organism physically and
chemically, through the food, water, and air it consumes and through
direct impingement of forces and energies on its body.

A-4. The basic principle of organization of the organism is that of a
control system. This organization is repeated at all levels from DNA
to the highest levels of the CNS.

A-5. In addition to being composed of control systems, the organism
can control through the acquisition of new control systems. The
process of acquisition is called, broadly, reorganization. The adult
CNS consists primarily of acquired control systems, brought into being
through reorganization.

A-6. Reorganization of the behavioral systems (the neuromotor systems)
is driven by a control system, perhaps of complex nature, that is
concerned only with maintaining the physical organism in a viable
state. This is called the reorganizing system.

A-7. This "viable state" is defined in terms of a set of critical
variables and a corresponding set of reference signals. The critical
variables are a sampling of biochemical (and possibly other) states in
the organism. The reference signals are genetically set -- quite
possibly, in many instances, simply as fixed thresholds of response in
chemical or neural sensory mechanisms. Such an arrangement could
combine sensing and comparing in a single neuron.

A-8. Changes of behavioral organization, including the acquisition of
the first behavioral control systems and all subsequent systems, are
driven by the difference between the values of the critical variables
and their respective reference signals.

A-9. A minimal model of reorganization adds the absolute values of all
these difference-signals together, the sum determining the frequency
with which random reorganizations are applied to the CNS.

A-10. The process of building new control systems works entirely
without a "teacher." The sole criterion for starting and stopping
reorganizations is the state of critical variables relative to their
reference levels. The sensors that detect critical variables and the
built-in reference signals and comparison process are inherited,
designed and built by evolutionary processes expressed through DNA.
The state of the critical variables in which reorganization will cease
is thus defined independently of the present-time environment.

A-11. There is no direct connection between behavioral control systems
and the reorganizing system. The acquired control systems become
organized to vary their actions on the environment in a way that keeps
perceptual signals matching CNS reference signals. For each controlled
perceptual signal, there is a state of the environment corresponding
to the state of the signal. Reorganization will cease only when this
variable state of the environment is in a state that keeps the
critical variables at their respective reference levels or prevents
disturbing them.

A-12. This method of acquiring behavioral organization requires no _a
priori_ knowledge of physical laws, entities, forces, relationships,
and so on in the environment. It makes no assumptions about the
environment other than the minimal ones concerning continuity and
consistency required to allow random reorganization to work.

A-13. It is probable, however, that the CNS is preorganized to make
the building of control systems efficient. There are strong
suggestions that the raw nervous system has a modular organization.
Perceptual computations occur in sensory nuclei and specialized layers
of the brain; output computations occur in motor nuclei and other
specialized layers. Comparison functions appear to occur within
restricted physical regions. There are, furthermore, layers of
organization in which different types of neural components exist,
presumably adapted to make possible certain type of computations. This
arrangement into layers may reflect evolutionary "knowledge" of the
environment, in that certain types of functions are important in
controlling all real environments regardless of their details (and
independent of our ways of describing physical laws).

B. The growth of behavioral organization.

B-1. The first aspect of behavior that must become organized after
birth is concerned with the use of muscles to control the perceptions
critical to survival. The neonate is helpless; it does not have enough
organization to take care of the needs represented by its critical
variables. It has just enough organization so that when a critical
error occurs, the reorganizing system can activate the motor equipment
to produce movement and noise in a global and unorganized way.

B-2. If caretakers did not interpret these output effects as a signal
to provide food or other aid, the baby would simply die of starvation,
dehydration, cold, heat, injury, illness, or suffocation. Its critical
variables would depart too far from their built-in reference levels to
allow life to continue.

B-3. This initial random-output process quickly becomes organized.
Sensory signals indicating various states of the environment (and
body) become connected to motor outputs. The sense of the connection
and the level of sensory signal at which an output will cease are
arrived at through random reorganization.

B-4. Further reorganization will be prevented when a sensory input
indicates a condition before it has effects on critical variables, and
when this sensed condition leads to action that removes the condition
before reorganization can begin. If the initially effective action
amounts to a signal to caretakers, the caretakers will correct the
condition, completing the control loop. This has the side-effect of
preventing critical errors, so reorganization does not further change
the neuromuscular connection.

B-5. Thus the CNS acquires the first control systems for controlling
the _sensory_ representations of body states related to the critical
variables. The particular motor actions that are used to control these
sensory signals depend on the caretakers -- what they regard as a
signal of distress, and how they interpret the signal in terms of the
remedial action they take.

B-6. These first control systems are crude and unreliable. They depend
on the continuous presence of caretakers, and on their ability to
intepret basically meaningless signals as indicating specific
problems. While major problems having to do with food and comfort are
usually handled properly, the critical variables inside the baby are
connected only loosely to events in the external world, and represent
far more detailed aspects of the organism than can be controlled
through a few crude gestures.

B-7. The result is that while major critical errors are prevented,
there must be a continual state of critical error that produces
continual reorganization. For example, even though the baby can signal
hunger sensations before they indicate a severe deviation from a
reference state, and even though the mother responds by picking up the
baby and nursing it, the baby still has to find the nipple and suck.
The baby may be put down in an uncomfortable position; it may have gas
bubbles that are painful; there may be a light shining too brightly in
its eyes; it may find its limbs cramped or immobilized, or its skin
itchy or inflamed. The mother can't be present every instant, and the
mother does not know how to interpret the random-looking struggles of
the baby in any but the broadest terms.

B-8. The baby therefore is continually required to learn to control
aspects of its experiences known only to it, as a means of preventing
critical errors that only its own reorganizing system can detect. It
must continue to refine its signals to the caretaker to make the
consequences more specific to the need, but it must also learn to
control other parts of the sensory world by moving its limbs, its
body, its eyes in ways that lessen the critical errors without the
help of the caretaker.

B-9. Through processes like these, the baby learns how to communicate
its needs to the caretakers in terms the caretakers are prepared to
(or learn to) recognize, and at the same time how to affect other
sensed aspects of its world that directly or indirectly bear on the
states of critical variables inside it.

B-10. Thus the foundations are laid for a process of ever-increasing
control of perceptions of the outside world and the body, with the
underlying reorganizing system always providing the ultimate judgment
of whether the resulting organization achieves results that are good
for the organism or bad for it. That judgment rests finally on
evolutionary experience.

C: Relationships of the behavioral systems to the environment.

C-1. The outside world can affect the organism's body directly, or
affect it through its sensory organs.

C-2. Direct effects on the body that are not represented as sensory
information (interoceptive or exteroceptive) are not controlled by the
acquired behavioral control systems. Their only possible influence on
behavior is through direct effects on critical variables that can
result in reorganization. Other effects on the body that do not affect
critical variables are either resisted by the biochemical systems or
alter the state of the organism without resistance and without
affecting behavior.

C-3. Environmental effects can influence the body directly at the same
time that they influence sensory information. Ingested food, for
example, affects the biochemical processes in the body, and
simultaneous affects taste, smell, tactile, and visual sensory organs.
What makes a substance into "food" is not its sensory effects, but its
effects on critical variables. Those effects have first priority; as a
consequence, the sensory aspects of the substance become controlled
(through reorganization of the behavioral systems) in favor of
ingesting the substances or in favor of doing something else with them
-- spitting them out or avoiding them. Sensory aspects of the food do
not determine which result will occur; only effects on the critical
variables can make that difference.

C-4. The basic capacity to sense the environment and body is built-in
in the form of sensory nerve-endings. Higher levels of perception,
derived from the basic level, are not built in but are acquired
through reorganization. All higher perceptions are acquired as a
component of a control system. The only purpose for acquiring any
control system is to maintain critical variables near their reference
levels. Therefore all higher perceptions are organized in order to
control them, and thus, indirectly, to control critical variables.

C-5. The sensory effects of the environment at levels higher than the
first thus always affect perceptions that are, or sometimes are, under
control.

C-6. If the effect of the environment is to alter a controlled
perception, the amount of alteration is normally kept small by the
actions of the control system. The sensitivity to error and the
capacity of the outputs of a control system are such that any
remaining effect of a normal external disturbance will not cause
critical error. Reorganization simply continues until that is the
case. Abnormal disturbances cause reorganization to begin again. If
the organism survives, these disturbances then become "normal," in
that the control systems change until the new disturbances can be
sufficiently resisted. Perceptions that are controlled well enough to
prevent critical error are said to be "unaffected" by disturbances.

C-7. Actions that control perceptions have side-effects on the body
and on other perceptions. Side-effects on the body are unimportant
unless they cause critical errors, in which case reorganization occurs
and persists until those side-effects are removed.

C-8. Side-effects of control actions on perceptions not directly
related to the control action are important only if they disturb other
controlled perceptions. If the disturbances are successfully countered
by other control systems, without inducing further disturbing side-
effects that can't be controlled, there is no important effect on the
behaving system as a whole.

C-9. Perceptions that are not currently under control can be affected
by the environment without any action that resists the disturbance.
These perceptions then simply change. If that change does not disturb
any other control systems to which those perceptions are inputs, there
is no important effect on the behaving systems. If the external
circumstance producing the disturbance does not also impinge on the
body so as to cause critical errors, there is no important effect on
the organism as a whole. If such errors are caused, reorganization
will commence and will not end until the perception being affected (as
well as the circumstance doing the affecting) is brought under active
control. So a perception that was not controlled under those
circumstances becomes controlled under the same circumstances. The
result is that uncontrolled perceptions remain uncontrolled only as
long as disturbances of them make no difference to the organism.

C-10. The external world can disturb perceptions in two ways: by
disturbing variables, and by changing parameters.

C-11. Disturbing a variable perceived by a control system means
applying the same kind of influence to the variable that the control
system's own outputs have. The aspect of the environment sensed by the
control system determines whether an affect in its environment is or
is not a disturbance. This also determines the kind of output that the
control system will have; the output must have an effect that can
change the sensed aspect of the environment. So disturbances are
always expressed in units commensurable with the units of output.

C-12. Parametric disturbances cause a change in the form of the
environmental function through which a control system's outputs affect
its own perceptions. Such changes have an effect on behavior only in
the dimension of the original effect of the output on the perception.
If the parametric change results in the output having new side-
effects, those new side-effects are irrelevant unless they result in
an autogenic disturbance of some other control system in the organism
-- if the output begins to have effects on other perceptions that it
did not have before -- or if they affect the body in a way that
induces reorganization. The only relevant effect on the original
control system that a parametric disturbance can have (without
involving reorganization) is to change the amount of effect a given
output has on the perception -- changing the loop gain of the control
system. The greatest effect a parametric disturbance can have is to
reduce the effect of output on input to zero, or to increase the
amount of that effect to the point where the control system becomes
unstable. Between those extremes, parametric disturbances change only
the amount of effect the system's output has on its input.

C-13. Parametric disturbances change a system's loop gain. They can
also increase or reduce the range of effective control and affect the
stability of control. They do not directly alter perceptions, as
disturbances of variables tend to do. Their effects are visible only
during active control behavior.

C-14. These special effects of parametric disturbances are otherwise
like those of disturbances of variables, with similar consequences
when they are not successfully resisted by the organism.

D. Relationships among organisms

D-1. The control actions of one organism can affect other organisms.
These can be direct effects on the bodies of other organisms, or they
can be disturbances of controlled and uncontrolled variables or
parametric disturbances of other control systems.

D-2. All considerations under C above hold when the effects of the
environment on an organism are produced by the actions of another
organism.

D-3. There are two added considerations when the source of an effect
is the action of an independent organism. The stability of control in
both systems can be affected because of feedback effects; this is
mutual parametric disturbance. Also conflict can develop; this is
mutual disturbance of variables. The two cases go together; unless the
two systems are tightly coupled to each other, neither conflict nor
instability will result.

D-4. Tight coupling between independent organisms exists when either
the input or the output variables are so nearly identical that one
system cannot act on its own perceptions without having an equal
effect on the other's perceptions of the same kind.

D-5. When the input variables are tightly coupled, one system's input
variable must necessarily be the same as the other system's input
variable. But the other system's action must be tightly coupled to its
own perception if control is to exist, and the two systems will not
necessarily control the input variable relative to the same reference
level. Therefore tight coupling of input variables leads to conflict.

D-6. When the output variables are tightly coupled, one system's
action must necessarily be the same as the other system's action. As
the same action can't in general affect both inputs in the ways each
system requires, again conflict occurs.

D-8. The effects of conflict depend on the loop gains in the two
control systems.

D-9. If both loop gains are low, the systems will produce outputs
within their normal range of output, opposing each other. Both will
experience large error, but with low loop gain will do little to
correct it. This implies that the errors are not important -- that is,
that critical error does not depend much on keeping those perceptual
errors small. If such errors are important in that sense, both systems
will begin to reorganize and increase their loop gains.

D-10. If one loop gain is high and the other low, the system with the
high loop gain will keep its error small while the other system
experiences large error -- in the same range it would experience if
had no control. The system with lower loop gain will begin to
reorganize if that amount of error results in critical error.

D-11. If both loop gains are high, both systems will produce maximum
output and experience maximum error. Both systems will lose control of
their perceptions. These perceptions became controlled as a way of
preventing critical error. Loss of control therefore implies that
reorganization will start.

D-12. Conflict, if important in the sense of affecting the critical
variables in the systems that have lost control, will lead to
reorganizations that escalate the conflict, bringing both loop gains
into the high region and increasing the outputs to their maximum. This
will continue until one or both organisms dies, or until one organism
reorganizes in a way that eliminates the conflict altogether.

D-13. Therefore the normal steady-state interaction between organisms
will be one that involves no conflict.

D-14. Interactions without conflict require that independent organisms
in close contact control perceptual variables that are linearly
independent of each other. This is the solution to conflict that will
occur, through reorganization, if both organisms survive. It is the
only solution, other than breaking contact, that can persist over
time.

D-15. When there are no conflicts, each organism controls variables
that are independent of the other's controlled variables. The actions
of each organism can amount to disturbances of variables or parametric
disturbance of the other's control systems. These disturbances,
however, remain within the range where each system can continue to
control all its perceptions by varying its own output actions. Each
organism continues to maintain its own critical variables near their
reference levels, independently of the other.

D-16. The shared environment will then come to a state in which it
simultaneously satisfies all the reference conditions in both
organisms. This can involve a close intertwining of physical
relationships, because for the behaving systems, "environment" is all
that is not CNS or reorganizing system. It includes, therefore, the
bodies of both systems, not just the world between the systems.
-------------------------------------------------------------------
Best,

Bill P.