[ACT-R-users] Production satiation (UNCLASSIFIED)
Ion Juvina
ijuvina at andrew.cmu.edu
Tue May 15 11:47:07 EDT 2007
here is my 2 cents: a production's utility decreases when that
production fires frequently in a sequence of productions that leads
to very small rewards. so, all we need to do is make sure we have a
"competitive" production there, that is, a production that
contributes to getting higher rewards.
example: a "ride-a-bike" production could lose the competition in
favor of a "drive" production if the environment favors speed and
safety. however, the "ride-a-bike" production can still win when the
environment favors health and losing-weight criteria.
ion
On May 15, 2007, at 10:15 AM, Kelley, Troy ((Civ,ARL/HRED)) wrote:
> Classification: UNCLASSIFIED
> Caveats: NONE
>
> ACT-R users,
>
> I was wondering, and I have never heard of this, if there a concept
> of production satiation in ACT-R, or someway to implement such a
> concept? This seems to be a common problem in ACT-R where a
> production will continue to fire over and over and it is difficult
> to stop. I understand that the solution is to change the
> activation of a chunk so that something else will match, but this
> is difficult to accomplish when the matching chunk keeps increasing
> in activation, thus causing more matches. I remember a few ACT-R
> workshops back that someone, I think Richard Young, presented the
> idea of a production refractory period, and I was wondering if this
> idea has ever gained any ground? Or are there other ideas of
> satiation within ACT-R? I am asking this because we are having
> trouble with our robot performing the same productions over and
> over. Again, I understand that we need to activate other chunks to
> cause a different match, but the question is - Where does this
> increased activation come from? Is this a meta-cognitive
> function? If there are no “new” stimuli from the outside, what
> triggers an increase in activation?
>
> Troy D. Kelley
> US Army Research Laboratory
> Human Research and Engineering Directorate
> AMSRD-ARL-HR-SE
> Aberdeen, MD, 21005-5425
> Ph: 410-278-5869
> FAX: 410-278-9523
>
>
>
> Classification: UNCLASSIFIED
> Caveats: NONE
>
> _______________________________________________
> ACT-R-users mailing list
> ACT-R-users at act-r.psy.cmu.edu
> http://act-r.psy.cmu.edu/mailman/listinfo/act-r-users
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/act-r-users/attachments/20070515/774c526e/attachment.html>
More information about the ACT-R-users
mailing list