Representation and Transfer in Problem Solving

Abstract : The work reported here consists of two series of experiments that investigate the role of problem representation in transfer of skill from one to another. The first series showed that increasing the inter-problem similarity of important representational features increased transfer, and allowed us to localize the transfer effect. In accord with earlier work the solution process was found to consist of two distinct phases a fairly lengthy exploratory phase, that takes most of the solution time and results in no net progress toward the goal, and a final path phase in which the subjects rapidly close on the goal. The results confirmed our prediction that the locus of the differing amounts of transfer would be in the exploratory part of the solution process: the transferred skill effectively substituting for some of the learning that occurs during problem exploration. The second series of experiments compares the effects on transfer of the subjects internal representation of the problem and of the external task domain or stimulus situation presented to them. The problems were constructed so as to be isomorphic. The subjects were given different interpretations or representations of the problems in some cases and identical ones in other cases. The amount of transfer was shown to be dependent on the similarity of the internal representation rather than on external stimulus similarities.