Memory and Hypothesis in Solving Alternation Problems with Random Competition

Abstract

Three experiments are described dealing with solution attempts for single, double, and quadruple alternations in one or two of three possible symbol-sequence classification dimensions. Results showed that alternation length did not alter solution difficulty. The conclusion is formulated that the way in which Ss tackled alternation problems could not lead to solutions on the basis of “reading off” from memory or learning on an incremental basis. Therefore the concept of an hypothesis must be brought in to account for Ss with incomplete evidence often obtaining successful solutions and yet not differing in problem-solving approach from nonsolvers. There is also some indication that single alternation may be easier than double alternation when coding of symbol sequences is most straightforward, but there is no indication of any circumstances that favor double over quadruple alternation.