[ACT-R-users] Perception and Motor

[Dan Bothell]

--On Wednesday, July 12, 2006 4:37 PM -0300 Fabio Gutiyama 
<fgutiyama at gmail.com> wrote:

> Greetings, ACT-R users,
>
> I'm new in this group and I am starting to use ACT-R on a research about
> human error in Brazil (Escola Politécnica, São Paulo University - USP).
> Some aspect I'd like to know is about the perception and motor modules.
> Is there any implementation that open possibility to ACT-R interact not
> only with the listener or the experiment window but with the entire
> enviroment of the operational system, seeing what is presented in other
> programs and generatig keyboards' in/out(s) to windows or any program,
> for example?
>

>From the model's perspective the world is represented by what's called
a device.  The device provides all of the ins and outs for the current
perceptual and motor modules of ACT-R.  You can find some general
information on the device for ACT-R 6 in the framework-API.doc document
in the ACT-R 6 docs directory and more detailed information at:

<http://chil.rice.edu/projects/RPM/docs/index.html>

Note however that the web site describes the ACT-R 5 code and may
not always match with ACT-R 6 (work is currently ongoing to update
the documentation for ACT-R 6).

So, your question comes down to basically whether or not there is
a device that supports the type of access you desire, and basically
the answer is no.  The devices included with ACT-R don't do that, but
there are a couple of possibilities.

First, it is possible to add new devices.  So, it's not impossible
to have that type of interaction, but one would have to do the
work necessary to create the appropriate device for ACT-R.

Also, the device that's provided for use with ACL (Allegro Common
Lisp) under Windows does actually generate system-level mouse and
keyboard actions.  They will be sent to whatever application has the
current focus.  So, if you are using that Lisp and OS combo you can
have the model send actions to any application, but it can't "see"
them.

As for seeing, there was a project called SegMan being developed
by Robert St. Amant which was looking to provide a general image
processing system that would allow for visual information to come
from "any" window, but I don't know too many details or the current
status of that project at this time.

Hope that helps,
Dan