| Abstract: | An outcome-expectancy mediation theory of learning was tested in a series of transfer experiments with pigeons. The principal technique employed was a delayed two-choice conditional discrimination in which the subject's correct choice yielded a food reward on one kind of trial and a feedback tone on the other. According to the theory, this task engenders two-link mediational chains in that each conditional cue controls a particular outcome expectancy (S-E link) and the expectancy, in turn, cues a specific choice response (E-R link). Relative to the hypothetical mediators established by preceding training, transfer problems required either the same mediational chains, the alteration of one link in each chain, or the alteration of both links in both chains for optimal performance. The results of four separate transfer experiments confirmed detailed theoretical predictions, including that of an unusual nonmonotonic transfer effect. The relation of the present theory to earlier mediation theories is discussed. |
