Abstract
Chunking was previously defined (Rosenbloom & Newell, 1983) as a process that acquires chunks that generate the results of a goal, given the goal and its parameters. The parameters of a goal were defined to be those aspects of the system existing prior to the goal’s creation that were examined during the processing of the goal. Each chunk was represented as a set of three productions, one that encoded the parameters of a goal, one that connected this encoding in the presence of the goal to (chunked) results, and a third production that decoded the results. These chunks were learned bottom-up in the goal hierarchy; only terminal goals — goals for which there were no subgoals that had not already been chunked — were chunked. These chunks improved task performance by substituting efficient productions for complex goal processing. This mechanism was shown to work for a set of simple perceptual-motor skills based on fixed goal hierarchies (Rosenbloom, 1983).
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Author information
Authors and Affiliations
Rights and permissions
Copyright information
© 1986 Kluwer Academic Publishers
About this chapter
Cite this chapter
Laird, J., Rosenbloom, P., Newell, A. (1986). Chunking in Soar. In: Universal Subgoaling and Chunking. The Kluwer International Series in Engineering and Computer Science, vol 11. Springer, Boston, MA. https://doi.org/10.1007/978-1-4613-2277-1_16
Download citation
DOI: https://doi.org/10.1007/978-1-4613-2277-1_16
Publisher Name: Springer, Boston, MA
Print ISBN: 978-1-4612-9405-4
Online ISBN: 978-1-4613-2277-1
eBook Packages: Springer Book Archive