Participants in 2 experiments practiced solving complex arithmetic problems
with or without external memory aid (paper and pencil). Participants with
the memory aid more often developed routines that corresponded to the conce
ptual structure of the task. The availability of a memory aid also slowed t
he rate of settling on a stable routine. In Experiment 2, the availability
both of an external memory aid and of a worked example varied between parti
cipants. Examples had the greatest influence on initial problem-solving str
ategies but did not override the effects of the memory aid. The results pro
vide evidence about the roles of situational and cognitive constraints in s
haping problem-solving routines.