[ACT-R-users] model validation

["Marc Halbrügge"]

Hi,

> I have implemented a few ACT-R 6 models of attentional networks simulating
> experimental studies. I have human data available to validate my model.
> the human study itself uses statistics like ANOVA to determine significane of
> the conditions and interactions between variables.
> 
> Is is sufficient (from the point of view of model validation ) for me to
> just use Correl and SD to validate the model results against the human
> data or is it better to repeat the statistical test to show the same
> significane? 
model validation is a complex (and still open) field, so don't expect a short and simple answer 

You could of course repeat the ANOVA on your model results. But: Model data is of a different quality than human data. For example, you don't really have independent observations. Therefore your ANOVA's df would be too high and your computed significances too low. I discussed this problem in more detail in a short paper last year (see refs below).

There's a good deal of information on model validation in Gluck and Pew (2005): "Modeling Human Behavior with Integrated Cognitive Architectures". You should find this book in your local library.

Greetings
Marc

Refs:
Halbrügge, M. (2007). Evaluating Cognitive Models and Architectures. In Kaminka, G. A. and Burghart, C. R., Eds. Evaluating Architectures for Intelligence. Papers from the 2007 AAAI Workshop, p. 27-31, Menlo Park, California: AAAI Press


-- 
GMX Kostenlose Spiele: Einfach online spielen und Spaß haben mit Pastry Passion!
http://games.entertainment.gmx.net/de/entertainment/games/free/puzzle/6169196