“Practice makes perfect” is no mere idiom. Practice has been shown to significantly improve performance in almost all domains tested (For review see Macnamara, Hambrick & Oswald, 2014; Heathcote, Brown & Mewhort, 2000). While this improvement is well-documented, the mechanisms by which this improvement occurs are still poorly understood.

This project seeks to test the idea that the primary driver of improvement via practice is simple memorisation of stimulus-specific features, and does not reflect a more general improvement in the task. I present the results of an experiment showing that the performance improvement following extensive training on a novel task was highly specific to stimuli with direct featural overlap to those practiced. Moreover, only exactly practiced stimuli reached maximal performance, even when compared with stimuli where all the components recieved subtaintial practice. These results suggest that mastery in a task stems from specific practice of a finite number of stimuli, not general task improvement.