It has been suggested that cognitive style interacts with performance on tasks measuring formal thought. It is hypothesized that cognitive style interacts with task context rather than with the underlying formal thought processes. Experiments are reported to clarify the type of task that elicits an interaction and to determine whether such interactions can be altered by instruction in formal thought processes. About 250 students age 12 to 16 participated in one of three experiments. Two tasks involving the separation of variables schema but contrasting contexts were administered to all subjects. In one context, only information necessary for applying the separation of variables schema was presented. In another context, the relevant information was embedded in irrelevant additional information. Cognitive style was measured in two experiments. In one experiment, subjects taught the separation of variables schema were compared to uninstructed subjects. Results supported the hypothesis of the study: task context interacted with cognitive style and training or separation of variables did not appreciably affect performance on irrelevant context tasks. (Author/RD)