Is a consistent goal structure sufficient to solve difficult unfamiliar problems in a familiar domain? Constraints on teaching for transfer

In today's rapidly changing world, educators put a premium on developing instructional methods that improve novices’ performance on unfamiliar problems and foster development of adaptive expertise. In this experiment, 44 undergraduate participants completed three hours of training and then had 15 minutes per problem to solve two test problems. Three training sequences taught subskills for a complex problem-solving domain encompassing many problem categories, and a control group did equivalently challenging tasks.; All three domain-specific training groups were taught a top-level goal structure applicable to solving any problem in the domain and did 25 practice problems progressing from easy to difficult. The three training sequences were equivalent in most respects, but there were important between-group differences. The first group solved type A practice problems and were taught type A methods for accomplishing two of the seven top-level goals. The second group's training was similar, substituting type B practice problems and type B methods, which were less explicit than type A. The general-within-domain training taught methods flexible enough to apply to all types and substituted practice problems drawn from many types.; All three domain-specific training sequences proved equally effective, but only the type A group solved the type A test problem, and only the type B group solved the type B test problem. Detailed action protocols of test problem performance demonstrated that variation in mean solution times was caused by differences in the specific subskills acquired during training. The results provide strong support for three theoretical propositions: (1) explicit, use-specific procedures are essential to narrow the search space of difficult problems, (2) learning a consistent domain-specific goal structure is insufficient for narrowing the search space of difficult problems in the domain, and (3) the difficulty of learning and implementing solution procedures increases as the explicitness of the procedures decreases. People typically solve unfamiliar problems by analogy. To solve difficult unfamiliar problems they need access to well-chosen worked-out examples representing the unfamiliar type of problem.