[ACT-R-users] PM-Question

Wolfgang Schoppek Wolfgang.Schoppek at uni-bayreuth.de
Fri Oct 11 09:21:49 EDT 2002


> I'va a question concerning buffer stuffing: Does buffer stuffing
> autonomously command a movement of attention and subsequent encoding?
> I always thought it wouldn't do that, but now I encountered a
> phenomenon that looks like autonomous attention movement: The model
> does central cognition processing when the visual environment changes.
> The change makes the visual module "busy" which triggers the
> production "wait-for-free". When visual modle is free again, it
> returns a new encoding of a visual object, which was never commanded.
> The new visual object disturbs the normal proceeding of the model. The
> described effect occurs only sometimes and I haven't found any
> contingencies yet. Please, see attachment for a trace and the relevant
> production rules.
> -- WS

In the meantime, I've found the reason for the described behavior: When the
place where PM is currently looking happens to be the place where a change
is occurring, a new visual-object is created without need to move attention
(in this case the trace "<> MOVE-ATTENTION generated DMO VISUAL-OBJECT..."
is a bit misleading). When this place is the *only* place with changes,
after the encoding there's no new feature left and the production
"found-new-feature" cannot fire. Therefore, the model doesn't realize that
there's a new state in the world and halts.
So my question has changed: What's the best way of making an ACT-R/PM model
recognize that the state of the world has changed?

-- WS
-----------------------------------------------------------------
Dr. Wolfgang Schoppek          Universitaet Bayreuth
 Tel.: +49 921 554140
 http://www.uni-bayreuth.de/departments/psychologie/wolfgang.html
-----------------------------------------------------------------






More information about the ACT-R-users mailing list