Is information reduction an item-specific or an item-general process?

An important aspect of cognitive skill acquisition is the ability, developed through task practice, to ignore task-irrelevant information (i.e., information reduction, Haider & Frensch, 1996 1996). By this account, practice not only affects how information is processed, but also what information is processed. Performance increases because, with practice, cognitive resources are used more and more efficiently as irrelevant aspects of the task are discarded from processing. Human–machine interaction, educational and sports psychology are just some contexts where behaviour consistent with the notion of information reduction has been reported. Information reduction applies in situations in which tasks contain both relevant and irrelevant information, and denotes a change from a processing-all-elements-of-a-task strategy to a processing-relevant-elements-only strategy. In two experiments, we tested empirically if information reduction is an item-specific or an item-general process by manipulating the frequency with which items were repeated during task practice. Participants verified alphanumeric strings containing task-relevant and task-irrelevant information. As the latter was not mentioned in the instructions, information reduction was studied in an incidental learning situation. We found that string-specific repetition frequency was related to the amount of time participants spent processing relevant portions of the stimuli, but was not related to the time spent processing the irrelevant portions. Thus, repetition frequency did not affect the degree of information reduction. The conclusion that information reduction is an item-general phenomenon was further supported by frequency-independent high rates of transfer errors when, late in practice, the former regularity of an irrelevant and relevant string-part no longer applied. The findings are incompatible with most data-driven theories of skill acquisition that attribute strategy changes during task practice exclusively to automatically occurring learning mechanisms. The results are compatible with theories that assume a top-down influence on skill acquisition.