Influence of Training Procedure on Multiple Choice Learning

Abstract

Three training techniques, a guidance only (correct answer only), a guidance with knowledge of alternative wrong answers (all the alternatives with the correct answer underlined) and a knowledge of results technique, were compared for their efficiency as training procedures in a multiple choice learning task. Subjects learned under one of five different conditions: (1) one trial of guidance only; (2) one trial of guidance with knowledge of alternatives; both followed by three trials with knowledge of results; (3) two trials of guidance only; (4) two trials of guidance with knowledge of alternatives, followed by two trials with knowledge of results; (5) all four trials with knowledge of results only. Knowledge of results was supplied by a modified Prewey-type punchboard.

Subjects from the four guidance conditions committed fewer errors and had more correct answers when transferred to the knowledge of results condition, than did subjects who learned with knowledge of results only, on corresponding trials. No significant differences occurred between the two guidance conditions (i.e. guidance only or guidance with knowledge of alternatives), although subjects from the latter condition achieved slightly better results. Two trials of guidance produced a very small improvement over one trial of guidance, but the differences did not approach significance. An incorrectly placed hole due to the faulty manufacture of the punchboard provided positive information for an incorrect alternative for one question, on the second trial only. The subsequent performance of subjects provided data that was interpreted as lending support to a ‘ level of confidence ’ hypothesis as a variable affecting the difficulty or otherwise of unlearning incorrect responses. The more confidence an individual places in a response, the more difficult it will be to unlearn if subsequently shown to be incorrect.