M1: An Architecture for a Living Android's Mind

by Patrick M. Rael (pmrael@aol.com)

Abstract:  The goal of the M1-Project is to create a physical and virtual Android Robot which can be programmed to perform the creative behaviors somewhat similar to those in people.  The system will have a physical presence.  The system must be able to adapt to new situations.   It must be able to re-program itself.  The Android's Mind must be designed such that given enough time and education it can become an expert in fields of study and get advanced degrees.  A balance must be struck between too little initial programming, resulting in decades or centuries of learning, and too much initial programming resulting in too much change required for new learning to occur.
 
 

    The M1 Android Robot Mind Architecture as outlined below is a work in progress.  Ten areas have been identified as being significant or helpful in order to program a sentient robot.  These ten components will form the foundation of the android's "Mind".  It is critical to note that the android's mind is no mere I/O function.  The M1 Mind will not be a simplistic "if A then 1, elseif B then 2, ...".   Nor will M1 be a simplistic word pattern-matcher.  Rather, M1 will consist of a physical presence in a defined dimensional universe and will be heavily oriented toward learning about itself, others, the world and the universe.


Table of Contents.
    The ordering is arbitrary in that no particular item is considered of greater importance than any other item.

            Preface
            Introduction
            1. Spatial-Extents
            2. Species-Recognition
            3. Survival
            4. Temporal-Extents
            5. Analogy-Processor
            6. Mind-Casts
            7. Brain Store
            8. Sensory I/O
            9. Focus
            10. Language
            M2
            M3


Preface

    The design of the Temporal Architecture is very simplistic at this point.  It will be further refined and detailed in the coming weeks and months.  However, there is enough detail even at this early stage to see that the M1 Architecture is no mere I/O function.  The Feedback Loop provides a potentially continuous processing mechanism regardless of Sensor Input and Output.  ie;  The M1 Mind can be functioning and active regardless of any current input and output.

      1. Mind-Casts

1. Scientific
2. Religious
3. Warrior
4. Political
5. Business
6. Artistic
7. Bios
8. Herd
9. Self
10. Partner


    These Mind-Casts represent a high-level overview of the categories of behaviour and information processing.  I think it's critical to identify and differentiate between these categories of behavior because focusing on only one should prove to be a major complexity reduction when it comes to implementing these.  I propose that the Scientific Mind-Cast may be the easiest to implement because there is less randomess to code than in the other mind-casts.  The Scientific mind is the epitomy of objectivity, like a finely tuned machine.


Introduction

    Sometimes when trying to solve a complex system, it helps to have a focused, short-term goal.  Otherwise, the complexity may be overwhelming.  A good short-term goal which encompasses several of the components of the M1 Mind Architecture is the following:
 

As a first milestone of the M1 Mind Architecture, I'd like to be
able to place the computerized Android Robot head in any room
and let the robot identify 8 things with it's sensors:
  1. Ceiling
  2. Floor
  3. Walls
  4. Doors/doorways
  5. Windows
  6. Tables
  7. Chairs
  8. People
While the Robot is doing this, it will be building a fully 3-D
representation of what it perceives around it.  This will be called
the Robot's Minds-Eye.  This 3-D representation
is to be visible on the computer screen so that anyone can see
what the Robot perceives.

For example, if the Robot thinks a table is in the room, then it will
create a 3-D table object in it's Mind's Eye.  If the Robot thinks a
person is in the room, it will create an Avatar to represent that
person.  Note that the Robot can be wrong.  It may think a person
is in the room when only lifesize picture of a person is present.

The Robot's Mind's Eye operates under a system of Perceptual
Assumptions.  It is not necessary for the Robot' Mind's Eye to be
100% accurate; only that the Robot act on it's perceived reality.
The Robot may ask for help in identifying things and people.

1. Spatial-Extents (SE) [code=Fundamental].

    The bounds of self in space.   aka "I" in the physical
    sense.
        1. The physical "I".
        2. The immediate neighborhood which is not I.
        3. The universe.

    Discussion:
        I don't think it's absolutely necessary for there to be a physical component for life.
        I can imagine non-physical forces such as gravity, magnetism, etc, which I wouldn't
        want to rule out as being potential domains of life forms.   However, at the moment
        it will be exceedingly difficult to build such a non-physical based life form with current
        technology.  Therefore, a physical-based life form will be attempted.
 

2. Species-Recognition (SR) [code=Fundamental].

    Others like "I".  The Social-Context is derived from SR.
        1. Social Context.
        2. Remote Minds modeler [codename REMIND].  It will be necessary for the robot to maintain
            dynamic models of external remote minds.  From these models it will be possible for the robot
            to predict the reaction of remote minds to stimuli.
        3. Imitatation.

        Discussion:
            Most of the life forms we can currently study have the ability to recognize others
            of their own species.  This enables reproduction.  While the android robot we're
            building is not going to have it's own reproductive capabilities, it will none-the-less be
            useful for it to recognize it's own species, and it's creator species.

3. Survival (Su) [code=Fundamental].

        1. Food-Energy Input.
        2. Fear of  Dismemberment.

        Discussion:
            The Android robot will need an energy source.  It's not going to be much use if
            it has a constant umbilical cord  connected into the wall.  It will be sufficient for
            the robot to periodically replenish it's power sources.  Three times daily is acceptable.
            We can call that a Power-Breakfast, Power-Lunch, and Power-Dinner.  Snacks
            are ok too.

            The robot isn't going to be of much practical value if it readily disassembles itself.
            An aversion (fear-of) to dismemberment will have to be encoded.

4. Temporal-Extents (TE) [code=Fundamental].

    The passage of time.  The framework
    for conceptualizing time and the Temporal-Driver.  The Speed of Thought dilemma is addressed
    fully here in the Temporal-Driver.
        1. Temporal-Driver (TD) - Responsible for the passage of time.
        2. Temporal-Concept - Crystallizing Chronoleum Temporal Dimension prototype-1.
        3. Temporal-Cache - Recent events Cache.
        4. Temporal-ReCycler [TREC] - Addresses the speed of thought, the retirement of
            concepts, and the continuity of Focus.
                1. Speed of thought - v(t)
                2. Retiring of concepts
                3. Continuity of Focus over time.

        Discussion:
            Many early prototypes of intelligence were in the form of programs which
            pretended to carry on a discussion with people.  The person types input into the
            program and the discussion ensues.  Most of these programs suffer from the
            Black-Box-Syndrome: the thinking is that one MUST model intelligence as if it
            could live in a black box.  A very brilliant person invented this black-box system.
            Unfortunately, this person didn't himself provide an implementation using this system.

            Many of these programs lack the Temporal Extents component: they will wait
            indefinetly for input.  The ones which won't wait indefinitely usually have a simple
            chronometer-based timeout.  I don't think M1 should have a simple chronometer
            based temporal system.  I think M1's Temporal Extents system is at least 1 level
            of complexity more complex than a simple chronometer.

5. Analogy-Processor (AP) [code=Fundamental].

    The system to process analogies.
    ALPINE - AnaLogy ProcessINg Environment prototype to consist of 3 fundamental modules:
        1. Concept - Anything, including:
                    1. Discrete objects and activities, both physical and temporal.
                    2. Continuous objects and activities, both physical and temporal.
                    3. Collections of Discrete and Continuous entities.
        2. Gluon - the connection feature between concepts.  [The debate continues as to whether
            a generic Gluon will suffice, or if PV should be applied here.  Simplicity has merit and may
            win.]
                   1. Temporal Gluon
                    2. Relational Gluon
                    3. Derivational Gluon
        3. Trigger-Algorithms (TAs)
                    1. Neighbor-Trigger, default.
                    2. Near-Trigger - As opposed to Far.
                    3. Far-Trigger - As opposed to Near.
                   4. Range-Trigger - Depth Trigger.
                    5. N-Trigger - Multi-Trigger.
                    6. Leaf - Find the leaves.
                    7. T(x) - Arbitrary trigger functions.

        Discussion:
            Analogies are useful for understanding something new in terms of it's relation to
            something already known.

6. Mind-Casts (MC) [code=Fundamental]

    The more obvious patterns in human ways of information
    processing.  Many human behaviours correlate with certain MC's.  It will be possible to focus on a
    single MC and make the Android Mind emulate that.  This will vastly simplify the programming task
    by orders of magnitude.  The design of the MC architecture will be modular.  MC's will be run-time
    switchable. Several MC's may be running at the same time.  When a particular MC is the principle
    cast for the Android, that is called X-Major, X=MC.  It will not be necessary nor desireable to
    implement several Mind-Casts, further simplifying the task.

       1. Mind-Cast Categories with a sampling of some properties.
            1. Science* - Facts, proof, hypothesis, theory, objectivity, rigor, mathematical, logic, deduction,
                induction, axioms, education, cause-effect.
            2. Religion - Belief, faith, salvation, mysticism, happiness, perception, appearance, good/evil,
                super-coordinator, ways to live, subjective codes of conduct, multi/after-lives, balance with
                nature, angels, spirits, ghosts, super-natural, purposeful, creationism, fear of eternal fire/plasma
                and/or punishment.
            3. Warrior - Military, battle, superiority, conquer, fight, survival, defend, intimidation, force,
                fight or flight.
            4. Political - Authority, popularity, control of others, appearance, polls, election, coup,
                public service/control.
            5. Business - Profit, wealth, material, money, greed, transactions, marketing, commerce, labor.
            6. Art - Subjectivity, appearance, creativity, originality, imprecision, entertainment, authoring.
            7. Bios - Body, health, sports, pleasure, pain, reproduction, appearance of, organic, gut,
                maintenance.
            8. Herd - Mob, hive, collective, self-less, one-of-many, strength-in-numbers, the flow, what
                others think and are doing is of supreme importance.
            9. Self - I, me, yo, self-ish, the center of the universe is ME.
            10. Partner - Significant other, extended self, family.

         2. Mind-Cast System Analysis Methodologies (SAM) - Certain MCs are associated with
            methodologies in analyzing systems.  For example, in Science the scientific-method prevails
            in analyzing systems.  Certain SAMs below are generic and may apply to all MCs.   The
            information below is preliminary and subject to revision.
            1. Give Things A Name (GTAN) - Give things a name or find the name of things.
            2. Principle of Variables (PV)  - Find the variable (that which can vary) properties in
                things.  Consider the system's state when the values of the variables are changed.
            3. Hierarchies
                1. Generalize-Categorize (GC) - The more general category.
                2. Specialize-Instantiate (SI) - The more specific example.
            4. Domain-Analysis (DA) - A system for analyzing the legal and illegal domains
                of systems.
            5. Idealism (Id) - A system for decoupling what one wants or what should happen
                from how to achieve it or how it will come about.
            6. Ways of Choosing (WC) - A system for making choices.
                1. Does the choice have to be correct? Does a wrong choice exist?
                2. Is this a Defining choice?  Is any choice okay?
                3. Is there a set or range(es) to choose from?
            7. Cause and Effect modeling.
            8. Logic
            9. Scientific Method
            10. more ...

        Discussion:
            In any Android Robotic project, eventually the point is reached where behavioural
            modeling is addressed.  Unless all behaviours are fair game [not necessarily a bad
            choice] then the domain of human behaviours is usually identified next.  Then a
            categorization of the large number of behaviours may ensue, from which it can be
            observed that not all of the domain of behaviours are worth modeling.

7. Brain-Store (BS) [code=Fundamental]

    The system for storage of memory including features which
    enable or encourage the human way of thought.
        1. Conceptual-Net
                1. Concept links from 1:N
                2. Visitor - Visits concepts.
        2. Analogy-Enabler Architecture prototype-1.  System requirements are below.
                1. Simple, low complexity.
                2. Robust - Tried and proven technology.
                3. Browseable Offline and Online mode.
                4. Reconfigurable Online and Offline.
                5. Decoupled from hardware.
                6. Cloneable.
                7. Low-Tech.

        Discussion:
            It would be very useful if it were possible to debug the robot's mind while it's
            running.  Ideally, I'd like to implement the Brain-Store as a simple Unix File
            System.  This would allow the use of the advanced Shell tools.  Also, the idea
            of tar'ing the file system to tape is appealing.  I'm going to work hard to see
            this through.

8. Sensory I/O [code=Fundamental]

        1. Electro-Magnetic Spectrum - Video, Infrared, UV, ...  (aka vision)
        2. Molecular Vibration Waves in mediums - Audio (aka hearing, speech)
        3. Molecular/Atomic vibrations - Temperature
        4. Physical Bounds Intersections -
                1. Short-range extended proximity sensors (aka hair).
                2. Perimeter sensors - skin.
        5. Chemical Analysis
                1. Solid/Liquid-Phase Sampler - taste
                2. Gaseous-Phase molecular sampler - olfactory
                3. Ingestion
        6. Balance - Gravitic/Inertial UP-vector stabilizer.
        7. Motor Control - The control of the robot's moving parts and other hardware features.

        Discussion:
            Motor control is implied.  In the beginning, video-cameras, microphones, speakers,
            buttons for skin, and gyroscopes will have to suffice for Sensory I/O.

            There is a common misconception about Skin I'll clear up now.   Many people assume
            incorrectly that a skinned robot is more advanced than a metal or plastic framed robot.
            This is incorrectly interpolated from the Human Superiority Complex.   There are many
            reasons why Skin isn't superior to durable metal: durability for one.

9. Focus [code=Fundamental]

    What I'm doing.  The root of sentience. *****.
    This module will be implemented as the Minds-Eye (ME) prototype.
        1. Minds-Eye - Where am I, where did I come from, where am I going.  What I'm doing now,
            what I did earlier, what I want to do, what I'll do next, what will happen.  This is to be
            implemented using the brain-store.
                1. I +
                2. Activity +
                3. Temporality.

        Discussion:
            This module is going to be the tough one.  Nine other components have been removed
            from it and all that remains is the root core of how I'm going to define sentience.  This elusive
            component is arguably the holy grail of Artificial Intelligence.  Good luck Pat!  Thanks, I'll
            need it.

10. Language [code=Fundamental]

    A means of communicating.  Using the model below it will be
     possible to make specific languages modular so that the modules can be replaceable by other
      language modules.
        1. Internal language-independent representation of concepts.
        2. External-Representation Language Translation. eg, English, German, French, Japanese,
            Swahili, ....

        Discussion:
            I'm convinced that many people think in a stream of networked concepts.  Mixed into
            that is language idiosyncracies.  So I think it's not just language, and not just concepts, but
            it's a mixture of the two.   However, my first cut at this will be a highly modular pair of
            Concepts and Language.  It's not clear to me how to proceed any other way while still
            providing a system powerful enough to model any language.  Most systems lock-in to
            a specific language very early.  This system being designed for M1 will not lock-in to
            any particular language ever!
 
 
 

M2 Abstract Behavioral Complexes

    Extension format interface for Abstract Behaviorial Complexes (ABCs).  This layer is to
        be composed of Abstract Behavioral Complex objects.  These objects modify the M1
        architecture such that the M1 system will begin emulating the behaviour of the specific
        ABC.  Rough outline and examples below.  This layer is to be implemented in M2.

    1. Honor
    2. Loyalty
    3. Modesty
    4. Stubborn
    5. Dignity
    6. Pride
    ...

M3 - Positronic.

    Ethical codes of social conduct.  The three laws of Robotics as per Isaac Asimov
    are obvious first-cut codes.  There's only three for now.  This should be quite a handful to
    encode.  These are purposely left unspecific.  eg, what is harm and inaction.  For the time being
    we will leave out the zeroth law [0. A robot must not harm humanity, nor through inaction allow humanity
    to come to harm].

    1. A robot must not harm a human being, nor through inaction allow a human to come to harm.
    2. A robot must do what it's told by a human being, except where that would conflict with law 1.
    3. A robot must protect itself, except where that would conflict with laws 1 and 2.