From newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.csri.toronto.edu!rpi!usc!cs.utexas.edu!uwm.edu!ux1.cso.uiuc.edu!mp.cs.niu.edu!rickert Tue Jan 28 12:14:56 EST 1992
Article 2952 of comp.ai.philosophy:
Newsgroups: comp.ai.philosophy
Path: newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.csri.toronto.edu!rpi!usc!cs.utexas.edu!uwm.edu!ux1.cso.uiuc.edu!mp.cs.niu.edu!rickert
>From: rickert@mp.cs.niu.edu (Neil Rickert)
Subject: Adaptive intelligence.
Message-ID: <1992Jan21.155129.18807@mp.cs.niu.edu>
Organization: Northern Illinois University
References: <1992Jan20.185948.13056@mp.cs.niu.edu>
Date: Tue, 21 Jan 1992 15:51:29 GMT
Lines: 161

In article <1992Jan20.185948.13056@mp.cs.niu.edu> I wrote:

>    Adaptive intelligence.  The ability to adapt.  This provides clear
>         survival benefits to a species, so that there is an evident reason
>         for it to evolve.  Some degree of adaptive intelligence is present in
>         all living creatures, both plant and animal.
>
>         As might be expected, adaptive intelligence has evolved in more
>         advanced creatures.  The result of this evolution is an ever more
>         refined ability to recognize, to respond and to adapt, often with
>         great precision, to even quite subtle changes.  Such refinement
>         includes the evolution of memory systems, so that past experience can
>         be a guide in future adaptive reactions to similar experiences.
>
>         Common sense, instinctive reactions, intuition, are largely
>         consequences of a highly refined adaptive intelligence.

In private correspondence bill@nsma.arizona.edu (Bill Skaggs) writes:

>I don't understand this, but it sounds interesting.  It doesn't
>seem natural to me to speak of "adaptability" as a kind of
>intelligence.  Can you justify it?  Can you enlarge on this
>theme?
>
>Go ahead and post your response -- this topic is a lot more

In his book "Unified Theories of Cognition," Allen Newell lists his
requirements for cognition.  This is on page 19, in figure 1.7, with a title
"The multiple constraints that shape mind."  The first two of his thirteen
requirements are:

    1.  Behave flexibly as a function of the environment.

    2.  Exhibit adaptive behavior.

Notice that these requirements do not mention consciousness, although Newell
certainly lists consciousness amongst his other requirements.  It is these two
requirements that I am singling out, and referring to as adaptive
intelligence.  Adaptive intelligence does not require consciousness, but
consciousness may indeed contribute to the adaptive intelligence of more
advanced creatures.

 Let me begin with some examples:

1.  In cold climates, if they were without any heating devices, people would
    stay adequately warm by constructing shelters from the weather, by
    huddling close together to share body heat, and by eating an adequate diet
    to produce such body heat through metabolism.  If confined to the same
    shelter in summer, it might be considered desirable to use some kind of
    fan for a forced ventilation system, so as to not be overheated in the
    hot sun.

    Presumably you would consider such actions as exhibiting a degree of
    intelligence.  And they certainly are ways of adapting to the extremes of
    weather.  Yet bees too do this.  In winter they stay warm with body heat
    generated by consuming the honey stored over the summer.  In summer they
    keep their hive cool by using their wings as fans to provide ventilation.

    If these actions are considered intelligent when performed by humans, then
    should we not equally consider them intelligent when performed by bees?
    True, when humans do this it is learned behavior performed consciously.
    Because bees are so different from us, it is not at all clear whether they
    exhibit consciousness, not is it at all obvious how we could tell.  It is
    also not clear that it is learned behavior in bees; it might well be based
    on knowledge acquired genetically, rather than by experience.

2.  Sometime ago in an early agricultural society, some farmer discovered that
    even though the weather was warm and the soil moist, it was pointless to
    plant seeds for the days were short and the nights were long.

    The discovery that one could predict the onset of winter by the shortening
    of the days was surely something we would consider to be an example of
    intelligence.  And yet many plants have this same knowledge, and use the
    shortening of the days as the signal to stop active growth and ripen their
    wood ready for winter.

    It is certain that this knowledge in plants is acquired genetically rather
    than by experience.  And yet if the corresponding actions by humans are
    considered an intelligent adaptation to changing seasons, why should they
    not be considered intelligent when carried out by plants?

3.  At some time when you were a child, you discovered you could breath
    through your nose even while you kept your mouth open.  You did it by
    suitably positioning your tongue.  Yet you possessed this knowledge long
    before you discovered it.  Indeed, you consciously discovered it by
    observing yourself.  You unconsciously acquired the knowledge when quite
    young, because you needed it to avoid choking when you breathed during
    eating.

    It is not completely clear in this case whether the knowledge is acquired
    genetically or through experience.  But, either way, it is important
    knowledge which is regularly used, and of which there was no conscious
    awareness before the time of your discovery.

4.  You are able to make quite complex sounds with combined action by your
    mouth and vocal apparatus.  Indeed, your ability to speak depends on it.
    Clearly it requires some complex knowledge to coordinate all of the
    necessary motions.  Yet you are not conscious of much of that knowledge.
    Indeed, to the extent that you are consciously aware, you have obtained
    that knowledge by consciously observing yourself.

    The knowledge of how to make the various sounds of your language is
    clearly learned knowledge.  Different people use different ways to make
    similar sounds, which is quite different from what we would expect if the
    knowledge were genetic.

    We know that a young child learns to make these sounds with a great deal
    of practice, often known as babbling.  But when the child is acquiring
    this knowledge, it is not due to any intentionality to acquire language.
    Indeed at that age the child has no knowledge as to what language is, and
    no reason to believe there is any benefit in learning it.  The learning
    method the child uses is simple trial and error.  Nobody explains to the
    child what motions to make, and indeed such explanations would be
    difficult since there is as yet no suitable language for such
    communication.

    There is some strong genetic acquired predisposition toward imitation,
    which stimulates the child to learn to speak.  Thank goodness for this
    predisposition, sometimes referred to as a susceptibility to peer
    pressure, or else we should surely learn nothing.  But the knowledge
    itself, although acquired through conscious action in imitation, is itself
    knowledge in which we are not consciously aware of the details.

It is my contention that there is a great deal of knowledge held by humans, of
which we are not consciously aware, and that this unconscious knowledge is a
critical component of our cognition.  It may even be that the amount of
unconscious knowledge we hold far exceeds the knowledge we consciously hold.
Scientific discoveries are sometimes made by us becoming consciously aware of
knowledge that we formerly held at an unconscious level.

Please note that I do not claim that adaptive intelligence implies
consciousness.  Clearly it does not.

I do claim, however, that it is adaptive intelligence that has evolved.
Evolution does not care whether we are conscious.  There is no advantage to
natural selection in our having an understanding of what we are doing.  All
that matters for evolutionary processes is that the way we behave contributes
to the survival of the species.  If a gazelle sees a lion stalking it, whether
the gazelle is aware of the danger, or understands that the lion could kill it,
is of no importance.  What matters is that the gazelle run like hell,
regardless of whether there is any awareness of the reason.  Evolution
already agrees with the Turing test if we interpret that test as meaning
that it is only behavior that matters.

Certainly consciousness has evolved.  But it has done so only because it
substantially enhances adaptive intelligence.  It has not evolved on its own
right.  It stretches credulity beyond belief to imagine that something as
complex as human consciousness has evolved by random accident.  It must be,
then, that consciousness has evolved as a side effect of the evolution of
adaptive intelligence.

If higher levels of cognition have evolved from adaptive intelligence, and
have evolved primarily to enhance adaptive intelligence, then we cannot hope
to create a true artificial intelligence until we understand adaptive
intelligence.

-- 
=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=
  Neil W. Rickert, Computer Science               <rickert@cs.niu.edu>
  Northern Illinois Univ.
  DeKalb, IL 60115                                   +1-815-753-6940


