variabilization of slots

[Wolfgang Schoppek]

Tina Schorr wrote:

> I try to model some activation-based proceeding in the working on a task
> where the solution of the task should be held in one single chunk. The
> mentioned activation should result from external cues (reading text files).
> Depending on the kind of the last notified cue there will be a binding of a
> part of the answer on a certain slot. Now the problem is that I am looking
> for some way to variabilize slots in the production specification. This
> would allow to use one production for this mechanism.
> Is there any way to do so? The only way I can see at the moment is to write
> some nearly identical productions where only the explicitly given slot
> varies. So, the response on the kind of the cue will be the use of
> different productions instead of only different bindings. But that doesn't
> seem to be a really elegant solution.

I'm not sure I've understood you right. Do you mean the problem that you want
to check for a certain chunk in any of the goal slots which means a disjunction
in the lhs of the production?

For example:

(p prod-1
  =goal>
   isa dingsbums
   slot-a cue1
   <or>
   slot-b cue1
...
)


Then the only way to do that is actually writing two (or more) productions.
If you want to have that kind of check in a retrieval, then you can try to omit
the symbolic constraints at all and rely on the spread of activation.

For example:

(p prod-2
  =goal>
   isa dingsbums
   slot-a cue1
   slot-b cue2
   result nil
 =chunk-x>
   isa solution
==>
 =goal>
   result =chunk-x
)


The probability of retrieving the correct chunk if it has cue1 and cue2 in any
of its slots is quite high and your model produces quite plausible errors
(without the assumption of partial matching).

-- Wolfgang

P.S. To learn more about ACT-R, it would be a good idea to attend the ICCM on
March 23-25 in Groningen, The Netherlands.
--------------------------------------------------------------------
 Dr. Wolfgang Schoppek                <<< Tel.: +1 703-993-4663 <<<
 HUMAN FACTORS & APPLIED COGNITION PROGRAM, George Mason University,
 Fairfax, VA 22030-4444
  http://www.uni-bayreuth.de/departments/psychologie/wolfgang.htm
--------------------------------------------------------------------