[ACT-R-users] New ACT-R 6.0 release

[db30 at andrew.cmu.edu]

There's a new version of ACT-R 6 available from the ACT-R web site.
Below I will highlight the most significant changes since the last
release (June of 2011).  The changes fall into three categories:
what is (and isn't) available on the web site, what is included with
the downloads, and changes and additions to the software itself.


What is different about what's available on the ACT-R software page:

- The source code is now only available in a .zip which includes both
the Mac and Windows versions of the Environment application.

- There is no longer a version of the Mac standalone available for
PPC based machines.

- A set of notes on teaching ACT-R modeling provided by Bill Kennedy
is now available.


What has changed in the distributed materials:

- Both the Windows and Mac standalones now have both 64 bit and 32 bit
versions of the Lisp application included and the startup programs will
run the appropriate one for the machine.

- The Windows standalone now uses CCL instead of ACL, and like the Mac
version, it makes the Lisp prompt available directly instead of through
a custom dialog as it had done in the past.

- The Mac standalone now requires Mac OS 10.5 or newer to run because
the current version of CCL (1.7) has that requirement.

- The tutorial unit texts and the primary documentation files (reference
manual, Environment manual, and AGI manual) in the distributions are now
PDFs instead of Microsoft Word files, but the separate PDF downloads are
also still available as they were before.

- There is a new tutorial unit.  Unit 8, titled "Advanced Production
Techniques", covers dynamic pattern matching and the procedural partial
matching mechanism.

- Unit 1 of the tutorial has an additional text and model (unit1_modeling
and broken-addition respectively).  The text is an introduction to writing
and debugging models in ACT-R and the new model is a non-functioning
version of the addition model from unit 1 which the text walks through
debugging and fixing.


Changes to the ACT-R software:

- A parameter has been added which allows the activation trace details to
be saved instead of being printed in the trace.  If the :sact parameter is
set to t then the declarative module will record all of the retrieval
request details during the run and they can be accessed afterwards using
the print-activation-trace and print-chunk-activation-trace commands.  The
retrieval history tool in the Environment has also been updated to show
the detailed activation calculations now.

- The warnings printed for problems encountered while parsing productions
have been reordered so that they make more sense when read from top to bottom.

- A new tool has been added for displaying the production history information
that is recorded when the :save-production-history parameter is set to true.
The new viewer in the Environment is called "Production Graph" and it will
draw a graph showing the transitions which occurred between productions when
the model was run.  There is also a command called production-transition-graph
which can be used to output a representation of the graph in the DOT language
for those that would like to work with the graph in something like Graphviz
instead of using the ACT-R Environment.

- Two new options have been added for the :nearest request parameter in a
visual-location request.  Those options are clockwise and counterclockwise,
and there is also an additional request parameter called :center which can be
used to provide the reference point for calculating those.

- The *actr-enabled-p* variable has been removed from the system.  The reason
for that is because it didn't really do much of anything, but it was used in
most of the tutorial units as a flag to switch between a person and model doing
the task.  That usage, combined with its name, made it seem like it was an
important thing to set.  However, the only thing it really did in ACT-R was
determine where the get-time command got its time information from (either
ACT-R's clock or from get-internal-real-time).  Get-time now takes an optional
parameter which can be used to indicate which time source to use, and if the
parameter is omitted then it uses the ACT-R time.  Thus, that change shouldn't
affect any existing models which use get-time to collect data, but it does
mean that if one is using get-time to collect data for a person doing a task
it will likely require changing the experiment code.  All of the tutorial
models have been updated to not use *actr-enabled-p* and instead to pass an
appropriate value to get-time when needed.


If you have any comments, questions, or problems with this update please let
me know.

Dan