Theories of skill acquisition assume that the effects of practice on t
ask performance are due to either qualitative changes in the task stru
cture, an increased efficiency of performing individual task component
s, an increased efficiency of performing sequences of task components,
or some combination of these mechanisms. We propose an extension to t
he existing theories by arguing that for many tasks, practice affects
which information is processed. More specifically, we argue that peopl
e learn, over the course of practice, to separate task-relevant from t
ask-redundant information, and to limit their processing to relevant a
spects of the task. In three experiments, subjects verified alphabetic
strings, such as M [4] R S T. Strings were correct if they followed t
he alphabet when the number of letters, given by the digit in parenthe
ses, was skipped. Strings were constructed such that errors occurred o
nly within the initial ''letter-digit-letter'' triplet. Analyses of su
bjects' RTs for strings of varying lengths demonstrated that: (a) subj
ects were able to distinguish relevant from redundant task information
, and to limit their processing to the relevant information, (b) the a
bility to reduce the amount of information that is processed takes tim
e and develops gradually over the course of practice, and (c) the mech
anism underlying this ability appears to be largely stimulus-independe
nt in the sense that structural components of a task are ignored, rath
er than specific task information. The findings and their implications
for general theories of skill acquisition are discussed. (C) 1996 Aca
demic Press, Inc.