Abstract
Two experiments were conducted to examine the generality of automatic/controlled processing training principles to rich, complex tasks. In both experiments, subjects' tasks were modeled after a job function performed in the telecommunications industry. These tasks required subjects to process conjunctions of information. Large quantitative and qualitative differences were found between the consistently and variably mapped training conditions. The need for determining trainable consistent components of complex tasks is discussed. The present experiments indicate that theories of automatic and controlled processing can be expanded to include many domains of rich, complex industrial tasks.