Does Guessing Incorrectly Impair Fact Learning?

In learning new information through testing with feedback, is it best for the learner to guess when unsure, or does that cause errors to become engrained? On day 1, subjects learned 80 facts, and were then tested on each. When a subject reported being unable to give an informed response, on a randomly chosen half of all trials the computer insisted upon a guess and provided corrective feedback (either immediately, in Experiment 1, or at a delay, in Experiment 2). Subjects returned for a test on day 2. Forced guessing neither impaired nor facilitated day 2 recall. However, items that spontaneously elicited erroneous guesses on day 1 were more likely to be correctly recalled on day 2 than were items on which the subject refused to guess. Why would spontaneous guessing bode well, while forced guessing had no effect? An explanation in terms of error-correction learning algorithms is suggested, and simulations are presented to show the potential viability of this account. Practical implications are also discussed. 2 More than a century of research on learning and memory has provided students