Psychology ETDs

Publication Date



Recent transfer of response differentiation studies have found that training Ss on easy differentiations and then transferring them to harder differentiations along the same continuum facilitates performance as compared to giving all training on the harder differentiations. These results are analogous to those found in transfer of stimulus discrimination and have been interpreted as evidence for the proposition that response differentiation is a special instance of stimulus discrimination. The only difference between the two is hypothesized to be in the nature of the cues used by the organism, interoceptive versus exteroceptive. More recent evidence, however, has suggested that the generalization of training on easy differentiations facilitating transfer to harder differentiations is oversimplified in that training on easy differentiations has led to negative transfer on harder differentiations and training on harder differentiations has led to positive transfer on easier differentiations. However, in these latter cases transfer was from easy differentiations requiring short duration responses to harder differentiations of longer response duration and from harder differentiations of long duration to easier differentiations of shorter duration. The present experiment was performed to eliminate the above confounding of length of duration requirement with difficulty of differentiation. Furthermore, various parameters derived from the S’s response distributions obtained in pretraining were correlated with transfer performance so that more specific statements could be made about what is transferred in transfer of response differentiation procedures. Response duration bandwidth requirements were first constructed to yield "easy" differentiations, one of which required short duration responses and one of which required longer duration responses, and "hard" differentiations, two of which required short duration responses and two of which required longer duration responses. Seven groups of six rats were pretrained on the above bandwidth requirements and then transferred to a duration requirement of "moderate" difficulty which required responses of an intermediate length. Subsequent to this part of the experiment, two groups of four rats were given similar pretraining on duration bandwidths, which were designed to further evaluate the predictive effectiveness of several measures developed with the first seven groups' data, and were transferred to the same transfer bandwidth as described above. The results indicated that (a) the groups trained on the longer bandwidth requirements, if not exceedingly long, transferred as well as or better than the group given all its training on the transfer requirement and (b) the groups trained on the shorter bandwidth requirements all transferred worse than the group given all its training on the transfer bandwidth requirement. It was concluded that in order to achieve optimal performance on differentiation bandwidths along quantitative response continua, bandwidth criteria should be implemented so as to generate: 1) a low probability of making responses short of the transfer bandwidth criteria; 2) a high probability of making responses within the transfer bandwidth; and 3) a response distribution which is displaced somewhat to the long side of the ultimate bandwidth requirement.

Degree Name


Level of Degree


Department Name


First Committee Member (Chair)

Douglas P. Ferraro

Second Committee Member

Frank Anderson Logan

Third Committee Member

David Wilmot Bessemer



Document Type


Included in

Psychology Commons