Proceedings of the Annual Meeting of the Cognitive Science Society

Contrary to a widely held belief, experts recall randommaterial better than non-experts. This phenomenon, predictedby the CHREST computational model, was first establishedwith chess players. Recently, it has been shown through ameta-analysis that it generalises to nearly all domains wherethe effect has been tested. In this paper, we carry outcomputer simulations to test whether the mechanismpostulated with chess experts – the acquisition and use of alarge number of chunks – also applies to computerprogramming experts. The results show that a simplifiedversion of CHREST (without the learning and use of high-level schemata known as templates) broadly captures the skilleffect with scrambled programs. However, it fails to accountfor the differences found in humans between different types ofrandomisation. To account for these differences, additionalmechanisms are necessary that use semantic processing.