Process generalization and the prediction of performance on mental imagery tasks.

Kosslyn (1980, 1983) theorized that performance measures on imagery tasks may vary as a function of the existence of independent processes in imaging ability. The present study determined whether improvement can be made in performance on such tasks with practice. It also considered whether performance on such tasks improves with practice and whether this improvement generalizes. Experiment 1 determined whether improvement in a mental rotation task generalizes to improvement in a geometric analogies task, with both tasks weighted in Kosslyn's find process, but not in a line drawing memory task weighted in Kosslyn's regenerate process. In Experiment 2, we examined generalization in improvement from a geometric analogies task to a mental rotation task. In Experiment 3, we tested whether improvement in an animal imagery task (Kosslyn, 1975) generalizes to improvement in a line drawing memory task, with both tasks weighted in Kosslyn's regenerate process, but not to improvement in a mental rotation task. Performance improved with practice on all tasks. Furthermore, performance improved from one task to another only if both tasks loaded on the same process.