[ACT-R-announce] Change in Utility learning mechanism

[Dan Bothell]

The new Utility learning mechanism which John presented at
the last Workshop and which was taught during the last
Summer School is going to become the default mechanism
in ACT-R 6.

When the next version of the software is made available on
the ACT-R web site in mid-January it will have the new
mechanism active by default.  The old mechanism will still
be included as an option, but will not be loaded by default.

Information about the new mechanism can be found in John's
talk "A New Utility Learning Mechanism" from the last ACT-R
Workshop and in the paper:

Fu, W-T. & Anderson, J. R. (2006). From recurrent choice to
skill learning: A reinforcement-learning model. Journal of
Experimental Psychology: General, 135(2), 184-206.

The paper is available from the ACT-R web site:
<http://actr.psy.cmu.edu/publications/pubinfo.php?id=684>

It is also described in documents available with the currently
available version of ACT-R 6 in the extras/new-utility
directory.  There you will find a file called new-utility.doc
and updated ACT-R tutorial units and models which use the
new mechanism.

For those getting the ACT-R files through Subversion,
as of r332, the change has already taken place.  The new
utility module is in place as are the corresponding new
tutorial units and models.  The old Utility module and the
units/models that go with it are still available as an
alternative in the extras/old-utility directory.

If you have any questions or concerns about the change
please let me know.

Dan