Effects of testing on learning of functions.

Is learning of a complex functional relationship enhanced by trying to predict what output will go with a given input, as compared to studying an input-output pair? We examined learning of a bilinear function and transfer to new items outside the trained range. Subjects either saw the input-output pairs (study-only condition) or attempted to guess the output and then saw the pair (test/study condition). The total study times were equated, and motivation was enhanced with a monetary bonus. Performance was markedly better for the test/study condition, both within the trained range and in the transfer test. This benefit of testing during training was observed on a criterial test administered shortly after training. Testing has long been shown to enhance the explicit learning and retention of verbal material; our present findings reveal a novel domain for which testing can also be advantageous-that is, function learning.