[ACT-R-users] Spreading Activation and Retrieval Chunks

[Niels Taatgen]

I think I have to agree with Dan that one should be very careful to 
change something like spreading activation, innocent though it may 
sound. Arguments in the line that it is more logical from the 
perspective from the architecture are what you could call "computer 
science" arguments. But one of the typical things for ACT-R has always 
been that changes are driven by psychological considerations. So a 
reason for this particular change would be, for example, that you can 
model a certain phenomenon more parsimoniously, or even better, that 
you can model a phenomenon that you couldn't model before.
Apart from that, a change in the architecture will always have unwanted 
side-effects, and these side-effects may be hard to predict (actually, 
a manifestation of the good old Frame Problem for humans). A general 
change might be good for a particular type of model, but is it good for 
all models?
Niels
===============================================
Niels Taatgen - University of Groningen, Artificial Intelligence
web: http://www.ai.rug.nl/~niels     email: niels at ai.rug.nl
Telephone: +31 50 363 6435
===============================================