Season Greetings

[Christian J Lebiere]

At last year's workshop the remark was made that there was little
opportunity to stay abreast of work happening in the ACT-R community
between workshops.  So we are venturing this message at the turn of the
year and the midway point between meetings.

The major project that is occupying us is getting the book out.  This is
a conceptual description of ACT-R 4.0 plus some demonstrations of its
applications.  We feel a bit like salesmen giving the specs on a new car
and then giving some test drives to show what it can do.  A Table of
Contents follows, with additional principal contributors listed.  The
book should be available by or before the 1998 Summer Workshop:

The Atomic Components of Thought.  Mahwah, NJ:  Erlbaum.
Chapter 1.  Introduction
Chapter 2.  Representation
Chapter 3.  Performance 
Chapter 4.  Learning 
Chapter 5.  Visual Interface 
Chapter 6.  Action and Perception (Mike Byrne)
Chapter 7.  List Memory 
Chapter 8.  Choice (Marsha Lovett)
Chapter 9.  Cognitive Arithmetic
Chapter 10. Analogy (Dario Salvucci)
Chapter 11. Scientific Discovery (Chris Schunn)
Chapter 12. Reflections

To review, since the workshop the major changes in ACT-R 4.0 have been:

A power-law time decay similar to that used in the base level and
strength formulas has been added to the computation of production
utility parameters.  The production compilation mechanism (nee analogy)
now applies when a dependency structure is popped rather than in
parallel with production conflict resolution.  Cosmetic improvements
have been made to the appearance of printed chunks and productions, the
format of the !output! command and the variable names of automatically
generated productions.  There will probably be another one or two beta
releases before the final release of ACT-R 4.0 next summer, but few if
any significant changes are expected.

A big push in our efforts is to increase the public accessibility of
ACT-R.  One step in this direction is the Published Models directory
which is available from the ACT-R research page.  The goal is to allow
not just experienced ACT-R users but any interested researcher to try
out models and inspect their code.  We encourage others to submit models
for this repository.   The Published Models directory is intended to
provide the public documentation for research papers.  There is no
requirement that these papers actually be "published" but they should be
available.  In fact, it might be nice to announce on the mailing list
papers for which ACT-R 4.0 models are available.  To help model that
practice we list our first two such papers meeting this criterion:

Anderson, J. R. & Reder, L. M. (in press).  The fan effect: New results
and new theories.  Journal of Experimental Psychology: General.

Anderson, J. R., Bothell, D., Lebiere, C. & Matessa, M. (in press).  An
integrated theory of list memory.  Journal of Memory and Language.

While these two papers are in the memory domain, there are many other
domains for which successful ACT-R models exist as the book will
illustrate.  It is important for the community to find traditional
outlets for publication of ACT-R research.  Traditional journals can be
receptive to properly presented papers.  We would be happy to advise on
how to prepare suitable papers.

On another dimension of public accessibility, it has become apparent to
us that a major limitation in the access of ACT-R is difficulties
researchers are having in wrapping code around their ACT-R models to
administer the experiment and collect statistics.  We are working on a
scripting extension which should help alleviate those problems.  We hope
to have it in beta release soon and officially unveil it at the 1998
Summer School and Workshop.  We also plan to hold a special session at
the workshop devoted to that topic.

On another dimension of public availability we intend to place our
instructional material on the Web for general access with the
publication of the book.  This will include the Interbook-based (Peter
Brusilovsky) tutorial and documentation and the ACT-R Environment (Jon
Fincham).

We welcome other suggestions about matters you perceive critical to the
conceptual or technical development of ACT-R.  We wish you all truly
happy holidays and a prosperous (i.e. lots of successful ACT-R modeling)
New Year. 

John Anderson
Christian Lebiere