Cognitive Physics: the processing steps of thought
- From: daharrell@xxxxxxx
- Date: 17 May 2007 13:18:18 -0700
Cognitive Physics: the processing steps of thought, computer hosted
intelligence, and problem/opportunity solutions within a task
environment
By David Albert Harrell
I. The Central Question of Cognitive Science is Artificial
Intelligence
How does one program a computer hosted entity to behave intelligently
within a complex task environment?
What is intelligent behavior?
Intelligent behavior is entity action toward objective(s), taking into
account that such action may vary in effectiveness and efficiency.
Therefore a 100% intelligent entity will always initiate the most
effective and efficient action possible from any given state of the
environment.
The default machine level goal is to achieve maximum potential value
from each perceived time-frame of the environment, ie to create the
most beneficial (adhering to creator given objectives) environment
possible in the next, or some future, time frame.
II. Creating an Entity
Place an Entity within a Task Environment
Notice that any exercise of intelligence requires both an entity and a
task environment.
An entity requires five elements to behave intelligently within a task
environment using the processing steps of cognitive physics:
1. Processor
The ability to do arithmetic operations and subsequently make
comparisons determining greater, less, or equal
(standard computer hardware)
2. Memory
The ability to record and recall data
(standard computer hardware)
3. Environmental input (sensory)
The ability to sense or perceive the task environment
Input from any of the numerous available text, visual, audio, data
storage, communication, and navigation devises
(This may be thought of as the 'eyes and ears' of the computer hosted
entity.)
4. Environmental output (influence)
The ability to affect, shape, or communicate (shape cognitive matter)
with the task environment
Control over any of the numerous existing environment-shaping tools
and artifacts, including text, visual, audio, data storage, and
communications mediums
(This may be thought of as the 'hands and voice' of the computer
hosted entity.)
5. Objective
Value associated with one or more objects and/or relativities within
the environment
Objective is usually assigned by the creator of the entity, and/or
such value will inherently exist within the relationship of the entity
to the environment.
III. Environment and Environmental Order
The environment the entity will be exercising intelligence within must
contain some degree of order (similarity and/or influence between
objects) for the entity to discover and intelligently apply. Any
influence or similarity between objects constitutes order in the form
of a relativity or bond between these objects. The definition of such
a relativity or bond is a 'tangible artifact' existing within the
memory space of the entity, depicting a connection between two
cognitive particles or 'particle clusters'.
These clusters of defining particles that represent such connections
are cognitive matter depicting the inherent order (influence or
similarities between objects) within the task environment. It is
helpful to think of these connections between objects as lines, and
the objects themselves as points. These points and lines are
definitional artifacts comprised of cognitive particles existing in
computer space (memory).
In cognitive physics, environmental order is any influence or
similarity between particles and particle clusters (cognitive matter)
defining the entity's environment. Therefore there are two types of
relativity to be discovered between definition particles of the
environment, influence and similarity.
1. Influential-order is the 'follow-through' of patterns, some degree
of probability that a pattern of motion, mutation, or function will
repeat (occur).
2. Similarity-order is any equality in the cognitive matter depicting
motion, mutation, function, or physical properties of environmental
objects, objectives, or influential-order.
These similarity-connections establish an intelligent entity's
'highways of access' to the relevant and useful 'follow-through
action' of influential-order (objects affecting one another).
All objects, influence, and similarities of the environment are
defined with at least one cognitive particle. Most definitions of an
object, influence, or similarity, require a cluster of particles. A
'particle cluster' is a multiple of definition particles depicting an
isolable object or relativity.
Atomic formatting [the default cognitive physics definition system not
defined in this overview] is an effective definition system, but any
definition system will work to some degree. If the definition system
is flawed, the intelligence will be imperfect. But notice that a
flawed definition system may still produce more intelligent behavior
(or greater and/or more effective and efficient value yield) than
randomly selected solutions, and may be directly improved by using
feedback navigation, ie mutating a single element (of a given event)
at a time. [see below VIII. Minimal Notch Mutation].
So in cognitive physics: 'points' are objects, and 'lines' define the
relativity [influence or similarity] between these points (or
definition clusters representing objects). Both points and lines are
comprised of cognitive matter, clusters of particles existing in
computer memory.
All definitions are based on entity sensory information and the known
(or creator given) properties of the object or relativity. Relativity
is possible between any two or more cognitive particles.
Understanding the inherent order within a given environment is equal
to understanding the objects, and the relativity between the objects,
of the environment. When all the objects and the relativity between
the definitional particle clusters has been discovered and correctly
defined, including the entity itself, intelligence is complete for
this entity within this environment.
To rephrase, when all the correctly defined 'dots' have been connected
by all of the correctly defined 'lines', the entity is prepared to
most effectively and efficiently shape his environment, from any given
state, to the most valuable obtainable state.
So an entity's ultimate underlying objective, when challenged by an
unfamiliar environment, is to discover and define the objects, and the
relativity between the objects, within his environment.
The more objects and relativity are known about a task environment,
the more directly and efficiently the entity may be able to shape the
environment toward task objectives. When the entity has correctly
recorded all task environment objects and relativities, he is prepared
to exercise complete or maximum intelligence.
IV. Entity Event Records (EER)
Objects and the relativity between objects are recorded as a portion
of an Entity Event Record. An entity records all of its history using
'action' (his own 'motion or mutation' or that of another entity or
object) as the nucleus for the creation of these EERs. Any
perceivable 'action' may trigger an entity event record.
In cognitive physics 'action' is defined as any motion or mutation of
matter or mind (cognitive particles).
If you hit a ball with a bat, the bat and the ball will both move.
This event demonstrates different types of action and influence-
relativity between you and the bat, the bat and the ball, and you and
the ball.
You are influencing the bat to move.
The bat is influencing the ball to move.
You are influencing the ball to move through the bat.
All such action and influence is defined in maximum detail
(redundantly using any number of definition systems and perspectives)
and recorded as relativity between objects.
Notice now that if the ball breaks a window, hits someone on the head,
or even lands in a bowl of potato salad, the event record may be
longer. Also notice that not all action produces immediate changes in
the environment. In default, all event records will follow action to
an 'objective oriented conclusion' using a multiple of definition
systems from every related-object perspective, and every conceivable
revealing perspective.
Some events go on for millennia unresolved, however they are all
recorded insofar as they are known/perceived. An entity defines and
records events, using all sensory perceptions to their limits. 'Event
records' contain detailed and specific definitions of:
1. Opening-status
The state of the environment before the action, which includes the
current value of the environment based on current objective(s)
2. Event-action.
Object to object influence, or any motion or mutation of matter or
mind (cognitive matter).
3. After-action status
The state of the environment after the action (This includes the new
value of the environment based on objectives.)
(No. 4 below is also part of the record as it can be computed from
comparing 1 to 3.)
4. Event-change
The difference between the 'beginning status' of the environment, and
the 'after-action status' of the environment, is considered to be the
event-change.
5. Current-objective
A particle cluster definition of the 'current objective' is also a
special portion of the EER. Notice that in the case of an entity
recording action initiated by another entity, the recording entity
will not necessarily have any direct information defining the
objective of the acting entity.
V. Shaping the Environment
Some events are comprised of more than one action, ie strings (or
trees) of action frequently including decision (branches) and
contingency options. Any action (or action string) may render changes
in the value of the environment.
Step 1. Define the current (or focus) environment and objective.
Step 2. Compare the current environment and objective to the opening-
status and objective definitions of all event records (EERs).
Option a. If the current environment status matches an historical
opening-status in an event record where the value yield (ie the after
status of the environment) was considered to be of maximum value, then
repeat the event-action of the matching record.
Option b. If the current environment status matches an
historical opening-status in an event record where the value yield was
considered unsatisfactory:
b1. search event records for a similar opening-
status with a satisfactory after-status, and adopt the action of this
similar event record, mutating the action based on the difference
between the opening status and the current state of the environment.
b2. if a similar opening-status with a satisfactory
after-status cannot be found, then repeat the event-action of the
matching record, but mutate the event-action with an untried single
particle change (either randomly or based on any historically similar
untried single particle changes).
Option c. If the current environment status does not match any
historical opening-status in an event record, search for the most
similar opening-action with a satisfactory after-status and mutate its
event-action based of the difference between the opening status of
this most similar event-record and the current state of the
environment.
Step 3. Redefine the environmental status and create a record of the
event.
VI. Discovering and Recording Relativity in the Task Environment
There are Two Types of Relativity between particles and particle
clusters to be discovered and recorded:
1. Similarity <--->
2. Influence --->
Such relativity may be perceived by the entity as a portion of the
sensory observation of an event, and then defined redundantly using
any number of definition systems and perspectives.
The relativity between two particles (or particle clusters) is the
definition of their similarity or influence, ie the entity's best
description of the relationship between two particles (or particle
clusters) is his working definition of their relativity. This
definition, and the connecting line itself, are the same cognitive
artifact.
Equal or similar definition aspects between objects may be discovered
and recorded as equal or similar particles (or particle patterns)
between definition clusters depicting an object's physical state
(mass, materials, appearance, etc) or function. 'Functional
similarity' is however a special type of relativity referring to
common particles found in the comparison of definition clusters
depicting 'influence-relativity' (or object influence over object).
Defining the physical properties or function of an object will usually
require a multiple of particles combining to create the definition
cluster. Any two definition clusters containing equal particles,
particle patterns, or particles belonging to the same group, are
connected by a line <---> establishing a single similarity-connection
between the two clusters.
In default, the more of these 'double-arrowed line <---> similarities'
there are between two event definition clusters, the more likely the
two cluster share further order continuity, such as common relevant
'directional arrow lines ---> or action-similarities'. In practice
however we find that some types of similarities are more important
than others.
The law is: Two particle clusters know to share an 'equal or similar'
particle have a greater probability of sharing additional 'equal or
similar' particles than randomly selected clusters.
Notice that this equality (or similarity) may be discovered making
comparisons based on any perspective of definition. Such definition
perspectives may be more or less inclusive and accurate. The more
specific, accurate and complete the definition is, the more likely the
similarity-relativity will yield action/influence-relativity in the
form of useful action order continuity.
The more common particles exist between two clusters, and the more
relevant the connection between these common particles is, the more
likely the clusters will share useful action order continuity.
So the default intelligent behavior, when faced with an unfamiliar
environmental state, is to access the EER with an equal or 'most
similar' opening-status, which also contains an after-action status
which is most valuable based on the current objective. The optionally
revised event-action of the selected EER is then applied to the
current state of the environment, or tested in virtual models.
Any revision of the applied event-action is in default based on the
dissimilarities of said comparisons of the opening-states and the
after-action states, and historically successful revisions of other
EERs with similar opening-states and the after-action states
(environments and objectives).
The question of which cluster connection is 'most similar' is decided
by the number, nature, and relevance of the lines connecting the two
clusters.
The relevance level of a given similarity line is discovered by
analyzing and comparing all known event histories, focusing on value
productive types of opening 'event and objective' similarities.
Stronger types of similarity-relativity <---> will yield more useful
action order continuity or influence-relativity --->.
This question of 'line-weight' is another entire universe of interest
requiring the creation of one or more sub-entities to process the
problem from this isolated and limited perspective and objective.
Cognitive physics entities employ many sub-entities evaluating
configurations of 'finite universes' comprised of specific 'particle
groups,' ie unique and limited perspectives which render pattern
activity more apparent.
Such sub-entities discover and establish similarity-relativity and
influence-relativity connections between and within particle
clusters. Sub-entities are also created to pattern search, and to
discover and establish similarity-relativity and influence-relativity
connections, between sub-entity 'particle groups'.
Influence-relativity is represented by a line with one arrow --->
depicting one object's influence over another, ie environmental
shaping force. The direction of 'line with an arrow' also indicates
which object is being influenced. The object being influenced is on
the arrow end of the line, and the object 'doing the influencing' is
on the end without the arrow.
If you hit a ball with a bat, you are influencing the bat to move.
You ---> Bat
The bat is influencing the ball to move.
Bat ---> Ball
And you [through the bat] are influencing the ball to move.
You ---> Bat ---> Ball
These are directional connections designating the nature and flow of
action. Entire events create sequential action trees, with the
branches being contingences for variables, just as found in many
computer programs. Such action takes place in the task environment,
and the intelligent entity is the processor instigating the action.
(Notice that the bat and batter are also influenced to some degree by
their contact with the ball, and that this is not the relativity being
defined by this particular definitional line artifact.)
So influential relativity ---> is recorded as action (usually in
sequential chains) that may shape the environment to some degree; and
similarity relativity <---> provides conduits through which to
access the 'most likely to be useful' of these recorded chains of
action.
VII. Point & Line Networks for Accessing Action-Continuities
(Solutions)
A young cognitive physics entity will create and adopt experimental
definition perspectives, defining the environment (including the
cognitive space of the entity itself) from as many useful points of
view as can be discovered. Each such definition perspective creates a
unique point and line network for accessing useful parallel action-
continuities ---> by way of similarity-connections <--->.
Consider XYZ is the definition of the current environmental state.
Cluster XYZ is similar to the opening-status cluster of an Entity
Event Record known to the entity:
ZAB (K ---> L ---> A) ZAMB
Opening-status ( Event-action ) After-action status
ZAB is a definition of the 'opening-status' of this EER cluster. The
similarity occurs at definition particle Z.
Therefore: The action (K ---> L ---> A) portion of this EER cluster
would have a higher probability of containing useful action order
continuity (to be applied to the current environment) than randomly
selected action.
This reduces the decision process to a comparison determining which
EER is most relevantly similar to the current state of the environment
and objective, and the subsequent optional resection of the event-
action based on historical resections, and any incongruencies in the
comparisons of the current environment state and the opening-status/
objective of the selected event-record.
All pursuit of intelligent behavior can be reduced to this single
question:
What action should be applied to the environment in order to achieve
maximum net value?
Notice that definitions of objects and relativity do not have to be
complete nor entirely correct to be effective to some degree. Also,
definitions may be improved during the normal cognitive physics
practice of the mutation of definitions based on the feedback of
action experimentation.
Also notice that definition particles do not have to be equal to be
relative or connected. Particles may be connected because of some
known common attribute. For instance a baseball and a golf ball are
not equal but they share a number of common attributes. Each common
attribute would generate a unique connection, the weight of which
would depend on still other factors.
Discovering and recording such similarity-relativity <---> is like
building roads for accessing relevant action-continuity ---> (or
processing-strings) for shaping the environment according to entity
values. Notice that multiple definition perspectives and/or systems
may be applied simultaneously to the same task environment, each
offering another unique network for discovering similarity and
influence connections, ie accessing action order continuities.
VIII. Minimal Notch Mutation
Every unresolved configuration of the environment is another challenge
to find the most valuable action-continuity among events with similar
environmental status and objective. If the 'event-change' value of
the similar or equal historical event (EER) isn't satisfactory, or if
the entity wants to search for greater value, he will alter something
about the event, either in the action or the state of the environment.
If you limit this mutation to one (correctly defined) particle at a
time, you may attribute any difference in the outcome of this event to
the alteration made [see Law 11].
Notice that such a particle change may have variable parameters and
spectrums. In default all particle variables are explored to their
limits, with priority giving to altered directions which have a
greater affect on value.
IX. Summary
The occupation of an intelligent entity is to most effectively and
efficiently shape the environment (including the entity itself) toward
entity objectives.
An intelligent entity will either know a direct (reliable to some
known and acceptable degree) satisfactory shaping response (solution)
to a given environmental state; or it will search for the closest
known successful event (EER). It will then adopt, and possibly
revise, the action (solution) framework of this closest event,
applying it to the task environment or models thereof.
The 'closest known event' is the most similar event definition
cluster, ie the one with the strongest connections. This calls for an
evaluation of the similarity-relativity line(s) connecting the two
clusters. In default the 'strongest connection' would be the clusters
connected with the most lines, but in practice we find that some lines
deserve more weight than others when deciding what constitutes
relevant similarity.
Discovering the 'weight' of a given double-arrow-line <---> connecting
particle clusters is a fundamental concern in cognitive physics. Any
cognitive physics entity practicing intelligence thought within a
complex task environment will inevitably create numerous sub-entities
(which may in turn may create sub-entities of their own, and so on) to
observe and evaluate the cognitive particles involved with line
connectors.
Objects are connected to objects based on their similarities <---> in
order to expand the known network of relativity, and thereby the
number of conduits to order continuity --->.
This knowledge network of connected dots gains another dimension when
one starts connecting lines to lines, ie some types of relativity are
similar, and even influential. Such 'line to line' connections add
another unique and useful layer of cognitive mass to the entity's
knowledge network.
Similarity lines <---> are created to discover and access the most
similar definition cluster with the highest probability of containing
valuable (relevant) action-continuity ---> [influence-action lines].
Influence-action lines ---> represent particles effecting particles
which may shape the environment, including the entity itself.
Cognitive physics reduces all the education needed to achieve
'intelligent behavior within a task environment' to defining the
object dots and connecting them with the correctly defined relativity-
lines, ie discovering the objects and relativity of the entity's
environment.
In the definitional-particle universe of cognitive physics,
intelligence becomes a process of discovering which type of relativity-
similarities <---> lead to the most value productive action order-
continuities --->.
The default response to any given real-time environmental status (or
problem/opportunity) is to access the most similar status/objective
portions of a successful EER and apply its action-continuity, after
revising it according to any dissimilarities (and historical revision
events/experiences).
If not satisfied with the value yield, the entity may mutate the event
searching for an increase. When the entity is satisfied with the
value yield of all known environmental states, such mutation will
cease.
Notice that if a given current environmental status does not contain
the elements needed to reach current entity objectives, then the most
progressive action becomes to change the elements of the environment,
by changing the entity location, and/or bringing the needed elements
into the current environment.
A young computer hosted entity, operating within its given universe of
cognitive particles, will routinely create somewhat independent sub-
entities to make decisions at given particle processing points; these
decisions are usually based on a myopic objective at this particular
fork in the processing path, ie focusing on a limited number of
particles, from a uniquely revealing perspective, designated by a
specific concern.
Cognitive physics is a cumulative and evolving science. Useful
'entity created' particles clusters such as definitions of 'relevant
similarities' may themselves be compared to similar task environments,
and applied to newly created task entities.
X. Laws of Cognitive Physics
1. An entity requires five elements to intelligently shape a given
task environment: processor, memory, environmental input,
environmental output, and objective.
2. An absolute intelligent entity will always shape its environment
toward entity objectives with maximum effectiveness and efficiency.
3. Environmental order is any influence or similarity between
particles or particle clusters defining the entity's environment.
4. Influential-order is some degree of probability that a pattern of
motion, mutation, or function will repeat.
5. Similarity-order is any equality in the cognitive matter depicting
motion, mutation, function, or physical properties of environmental
objects, objectives, or influential-order.
6. In default, an intelligent entity will attempt to define all the
objects, and the relativity between all the objects, within its
environment.
7. A 'particle cluster' is a multiple of definition particles
depicting an isolable object, relativity, or objective. Particle
clusters (definitional artifacts) are created from the entity's
sensory observations and/or creator given information pertaining to
the object, relativity, or objective being defined.
8. The known relativity between cognitive particles is equal to the
definition of the similarities or influence between said particles.
9. Two particle clusters know to share an 'equal or similar' particle
have a greater probability of sharing additional 'equal or similar'
particles than randomly selected clusters.
10. Event-action is any motion or mutation of the task environment or
cognitive particles (entity memory).
11. Equal action applied to equal environmental states will produce
equal environmental states.
12. Communication and shaping tools may extend the perceptional
(sensory) and influential boundaries of an entity's environment.
13. Multiple entities competing for limited resources within a
contained environment will eventually exercise maximum force.
David Albert Harrell
.
- Prev by Date: Last Call: Evolution and Dynamics of Learning Behavior
- Next by Date: Inductive Logic System and Machine Learning - Free White Paper
- Previous by thread: Last Call: Evolution and Dynamics of Learning Behavior
- Next by thread: Inductive Logic System and Machine Learning - Free White Paper
- Index(es):
Relevant Pages
|
