[om-list] Analytical models of irrational behavior

[Thomas L. Packer at home]

Mark

    You should play around with the Soar cognitive modelling architecture
(or Act-R), and see if that system could work.

http://www.eecs.umich.edu/~soar/docs/Gentle.pdf

    I bet you would have fun with it.  I have a bunch of Soar links on my
web page.

http://www.ontolog.com/Links/Soar.html

    (I will hopefully add more about Soar and especially about Act-R in the
future.)

        Soar has been successful in simulating the decisions of fighter
pilots in combat, including the simulation of talking to each other in their
restricted natural language.

http://www.soartech.com/

    I am almost certainly going to use Soar to build my language learner for
my masters project.  I've heard that the government uses Soar to simulate
the behaviours of other countries, and I think one company uses Soar to
simulate economic forces.

http://www.ers.com/

tomp

~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Víðar sum quem nihil obstat.
Omnia apud me Mathesis fiunt.
www.Ontolog.Com
~~~~~~~~~~~~~~~~~~~~~~~~~~~~


----- Original Message ----- 
From: "Mark Butler" <butlerm at middle.net>
To: "One Model List" <om-list at onemodel.org>
Sent: Saturday, 04 October, 2003 09:19
Subject: [om-list] Analytical models of irrational behavior


Errors in logic are responsible for an enormous amount of human
suffering, probably far more than starting from poor premises.  Logical
errors sustain the life of irrational beliefs in individuals and nations
by providing them with a false sense of the internal consistency of
their beliefs.  Minor logical errors creep into every body of human
thought, usually to no great consequence.  However major logical errors
can be fatal, both to people and to civilizations.

Current analytical models of human behavior are largely based on a
simple conception of self interest.  Analytical models of  moral
considerations and the irrational beliefs that flow from faulty analysis
of those moral considerations are rare or non-existent.

Moral considerations must be represented in terms an analytical model of
natural language, one powerful enough to model irrational thought
processes. A good analytical model should be capable of representing
both extremely rational processes (such as that used by the system
itself ) with gradual degradation down to simple Aristotelian logic (or
worse) that can be adjusted depending on an estimate of the logical
sophistication of the appropriate party or people.  It should also be
capable of generating multiple likely conclusions of parties whose
beliefs are not internally coherent, or consistent with the reality of
the outside world.

A simple logic engine based on fuzzy logic and analytical semantics is
easily tweakable to get all sorts of mental pathologies - treating
universals as simple entities and bipolar (Manichean) thinking to name
two of the most prevalent logical errors throughout history.

If a party is inclined to think in Cartesian terms, for example, we
should get results quite different, than if they are inclined to think
in terms of simple relations between Platonic universals.  We should get
even more dramatic differences if we increase the tendency to think
about everything in black and white instead of in shades of gray.  While
many moral laws are nearly absolute, prudential and sympathetic
considerations never are.  Errors in logic feed on each other - thinking
that universals are simple, the Platonic fallacy, lends itself to a
Manichean view of the world and vice versa.

No one has the resources to make a complete model of anything
particularly complex (such as unrestricted natural languages) anytime
soon, of course.  Progress is most likely to be made by increasing the
sophistication we bring to bear on simple bodies of fact. Collecting and
representing a voluminous body of fact should be a secondary priority.

Indeed, analytical models are useful as scientific tools in and of
themselves - a demonstration of  the errors in logic caused by severe or
bipolar depression would be very impressive.  An inference engine that
demonstrated the likely difference Aristotelian and Cartesian modes of
thinking about the same set of basic facts would be an incredible
educational tool.

   - Mark





_______________________________________________
om-list mailing list
om-list at onemodel.org
http://www.pairlist.net/mailman/listinfo/om-list