From newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.csri.toronto.edu!bonnie.concordia.ca!garrot.DMI.USherb.CA!uxa.ecn.bgu.edu!mp.cs.niu.edu!rickert Wed Apr 22 12:04:07 EDT 1992
Article 5152 of comp.ai.philosophy:
Path: newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.csri.toronto.edu!bonnie.concordia.ca!garrot.DMI.USherb.CA!uxa.ecn.bgu.edu!mp.cs.niu.edu!rickert
>From: rickert@mp.cs.niu.edu (Neil Rickert)
Newsgroups: comp.ai.philosophy
Subject: Knowledge acquisition.
Message-ID: <1992Apr20.173841.651@mp.cs.niu.edu>
Date: 20 Apr 92 17:38:41 GMT
References: <1992Apr16.142423.10650@oracorp.com>
Organization: Northern Illinois University
Lines: 86

In article <1992Apr16.142423.10650@oracorp.com> daryl@oracorp.com (Daryl McCullough) writes:

>I would say that you know things by a combination of (1) receiving
>data, and (2) interpreting that data according to some conceptual
>scheme (which might be innate or learned).

 Much of an individual's knowledge does not seem to fit this model.  I have in
mind the knowledge of how to walk, how to coordinate the various components of
the vocal apparatus to speak, etc.  This type of knowledge is gained by
receiving data, but there need be no "interpretating ... according to some
conceptual scheme" involved in the knowledge acquisition.

  I am of course quite aware that this type of knowledge is often dismissed as
being different in nature from what is usually referred to as "knowledge".
But it seems to me that such dismissal is not easily justified.  A great deal
of intuitive knowledge, even in dealing with quite abstract topics, seems to
be acquired in a comparable manner.  Perhaps of greater importance, much of
the knowledge base needed to convey semantic content via language is of this
form.

  To put this in perspective, it is useful to look at man-made mechanical
methodologies for knowledge acquisition, and see how they might relate to
mental processes:

    Passive observation: this more-or-less corresponds to the term "receiving
    data" as you used the term.  For mechanical knowledge acquisition this
    could include use of thermometers, barometers, cameras, etc.  Passive
    observation is time consuming, since it requires waiting for nature to
    take its course.

    Active observation, or testing: this is where you observe the effects of
    interacting with your environment.  For a person, it might include gently
    putting down you foot to sense if the ground is slippery before you
    transfer your weight.  Testing is generally superior to passive
    observation, since the type of data available is far broader.  It is often
    faster, since the tester has some degree of control over the events being
    observed.

    Simulation:  For mechanical knowledge acquisition this would include both
    simulation with models (such as the use of wind tunnels to test scale
    models of aircraft) and simulation on a digital computer (often referred
    to as numeric simulation).  For humans, thought appears to be a form of
    simulation.  Simulation is the most versatile and often the most
    economical method of knowledge acquisition.  It is, however, always
    slightly risky, for if your simulation does not use a sufficiently
    accurate model, inaccurate knowledge may be acquired.

 With respect to this subdivision, I would place your "interpreting that
data according to some conceptual scheme" in the category of simulation,
achieved in this case with the aid of a mental model (the "conceptual scheme"
you refer to).

>                                           I don't think introspection
>is any different from other ways of getting data about the world,
>except that it isn't independently verifiable.

  There is one important way in which introspection differs.  When we
technologically or mechanically do a simulation we build an external model,
or we find an available computer on which to perform that simulation.  When
the human does a simulation via the processes of thought, he does the
simulation on himself, since that is the only hardware available.  This
makes such a simulation highly reflexive in nature.  It means that the
simulations effect various parts of the body, such as when we wave our
hands while thinking.  Sometimes the physical effects on the body are
incidental, but they are often a central part of the simulation.  For
example a simulation via thought may bring the body to a near state of
preparation for a physical fight and thus trigger the emotional responses
of anger.  This reflexive nature of self simulations may be the underlying
mechanism behind many features of the mind.

  Imagine, as a near approximation, that we have an automated factory set up.
To simulate some possible changes in the assembly line we run a simulation.
But instead of running the simulation on a separate machine we do it on the
assembly line machine.  During the simulation some (but not all) of the wheels
of the factory begin to turn simply due to the fact that the simulation
changes states of the assembly line machine and these states trigger
mechanical action on the assembly line as an incidental side effect.  Some
feedback is returned from sensors, due to the fact that this assembly line
controller has built into its system the monitoring of the state of the
factory.  Perhaps this produces the equivalent of pain (when the feedback from
the sensors indicate something wrong) or elation (when everything seems to
coordinate and fit just right).  This interaction between the simulation and
the production line is in part a nuisance, but in part it is also very
valuable since it provides a degree of feed back for evaluation of the
simulation which might not be available if the simulation were done on an
independent system.


