Working Memory

Marsha Lovett lovett at andrew.cmu.edu
Thu Apr 18 10:07:52 EDT 2002


Lynne Reder, Christian Lebiere, and I define working memory in 2 ways:
(1) content-wise: a subset of highly activated declarative memory elements 
(including elements in the goal)
(2) process-wise: the mechanism of source activation propagating from the 
goal to elements in declarative memory.

When you combine these two ideas (noting that the goal is actually the MOST 
active set of declarative items, cf. Neal Cowan's work), the sentence with 
"star" appearing twice does not cause a problem. As one reads that 
sentence, the visual processing of the words creates new chunks (one for 
each 'star') that become components of the current goal (because they need 
to be maintained as the reader constructs the meaning of the sentence 
across words). Each of these "star" goal components would likely activate a 
star-concept chunk in declarative memory (among other associated chunks) by 
virtue of the goal's source activation propagating. However, this does not 
mean that there could only be one star chunk in ACT-R. Indeed, after the 
sentence is read, there will probably be additional "star" chunks (perhaps 
episodic in nature) after the current goal is popped.

I should also mention that Lynne and Christian and I (along with past and 
current post-docs Larry Daily and Melanie Cary) have written about how we 
have successfully expanded ACT-R's theory of working memory to account for 
individual differences in working memory capacity (and the impact on 
performance). Using the W parameter in ACT-R (goal activation) as our 
measure of capacity, we have estimated individuals' W based on their 
performance of one task and then apply this value to predict their 
performance on another task.

Lovett, M. C., Reder, L. M., & Lebiere, C. (1999).  Modeling working memory 
in a unified architecture: An ACT-R perspective.  In A. Miyake & P. Shah 
(Eds.)  Models of Working Memory. pp. 135-182. Cambridge, MA: Cambridge.

Daily, L. Z., Lovett, M. C., & Reder, L. M. (2001).  Modeling individual 
differences in working memory performance: A source activation account in 
ACT-R. Cognitive Science, 25, 315-353.

Lovett, M. C., Daily, L. Z., & Reder, L. M. (2000). A source activation 
theory of working memory: Cross-task prediction of performance in ACT-R. 
Cognitive Systems Research, 1, 99-118.






More information about the ACT-R-users mailing list