[ACT-R-users] ACT-R users Digest, emotional Stroop

[Roman Belavkin]

Hello Ann,

> I am new to the group, and I hope I've accessed it properly.  I am doing
> a study of emotion in ACT-R and am wondering if there is any literature
> on how ACT-R might reflect emotional Stroop data.  It seems from this
> newbie's perspective that making G part of the condition of every
> production might prevent the distractibility that needs to be modeled
> for this task.  If one chooses not to do that, how might one reflect the
> increase in activation due to G (goal)?

First, let me pass you couple of references of some work we've done on
modelling the effects of emotion using ACT-R.

Belavkin, R. V. (2003). On Emotion, Learning and Uncertainty: A Cognitive
Modelling Approach. PhD Thesis
http://gold.mdx.ac.uk/~rvb/publications/rvb-thesis.pdf

Belavkin, R. V. (2001). The role of emotion in problem solving. In
Proceedings of the AISB'01 symposium on Emotion, Cognition and Affective
Computing (pp. 49--57). Heslington, York, England
http://gold.mdx.ac.uk/~rvb/publications/rvb-aisb1.pdf

Belavkin, R. V. & Ritter, F. E. (2003). The Use of Entropy for Analysis and
Control of Cognitive Models. In F. Detje, D. D\"orner, & H. Schaub (Eds.),
Proceedings of the Fifth International Conference on Cognitive Modelling
(pp. 21--26). Bamberg, Germany: Universit\"ats--Verlag Bamberg
http://gold.mdx.ac.uk/~rvb/publications/rvb-fer-iccm03.pdf

You may find some answers or hints there.  Basically, yes, G is probably the
only way in ACT-R at the moment to represent motivational `strength' (or
urgency you may call it), and it affects the choice of rules quite
dramattically (there is even some asymtotic analysis of that in the above
papers).  There are some other mechanisms and parameters, however, that you
may find useful.  Fot example, some of us believe (Dorner, myself) that
noise varaince is related to appreciation of uncertainty, and it changes the
behaviour from random to more focused and vica versa.  In neural networks
you could simmulate this by changing the bias (or threshold of activation)
in neurons, and there is some evidence that certain neurotransmitters have
such an effect.  So, what you might simulate in ACT-R by changing noise
variance, in the brain may be the result of activity of some brain areas
(e.g. amigdala) that interfere with the neocotex activity to control its
operation.

Now, regarding the memory and retrieval of chunks, remember that in ACT-R
the lower the activation of a chunk is the longer it takes to retrieve it
(see the retrieval time equation).  If G is high, however, this may
represent a situation of a great urgency, that is there is not much time to
retrieve the chunks.  You may simulate this in ACT-R by rising the retrieval
threshold (:rt), and also possibly activation noise variance.  This may lead
to retrieving `worng' chunks or a retrieval failure.

hope this helps a bit,

Cheers!
Roman