Psychology ETDs

Publication Date



The averaging hypothesis states that intermediate rates in generalization tests represent a composite of two interresponse time (IRT) topographies established during prior acquisition. A short-IRT topography, characteristic of SD responding, and a long-IRT topography, characteristic of SΔ responding, occur in different proportions as a function of generalization test stimulus value. Demonstrations of averaging have had limited generality due to departures from the typical discrimination training and test procedures. The present experiment investigated averaging in a context distinguished from typical discrimination training and test procedures only by the use of separate manipulanda for the collection of SD and SΔ responsesacross a range of acquisition stimuli conditions. Four groups of three pigeons each were trained to respond either left or right, depending upon the stimulus which appeared simultaneously on both keys. Left and right responses, respectively, were required to vertical lines and diagonal lines in Groups I and IV; to vertical lines and blank keys in Group II; and to blank keys and diagonal lines in Group III. During the initial 14 sessions, correct responses to either key were reinforced according to identical random interval (RI) 30 sec. schedules. For the next 7 sessions, an RI 120 sec. schedule prevailed for correct responses to the right key for Groups I-III, while left key responses were reinforced as before. Group IV continued to receive the RI 30 sec. schedule on the right key. In terms of combined response rate to both keys, behavioral contrast was observed following the transition from RI 30 to RI 120 for Groups I, II, and III. Mean IRTs decreased in SD and increased in SΔ, reflecting overall changes in the IRT distributions. That is, a short-IRT topography appeared in SD and a long-IRT topography appeared in SΔ. The post-discrimination gradients observed during three extinction test sessions were similar to those obtained in comparable single-key studies. No major changes in the form of the IRT distributions occurred at any test stimulus value. However, each test stimulus controlled a particular proportionality of SD and SΔ responding. Predictions made from the response rate averaging hypothesis were verified and implications for post-discrimination peak shift and rate gradients in general were discussed.

Degree Name


Level of Degree


Department Name


First Committee Member (Chair)

Douglas Peter Ferraro

Second Committee Member

Frank Anderson Logan

Third Committee Member

G. Robert Grice



Document Type


Included in

Psychology Commons