Effects of reward magnitude, response requirement, and food-cup locus on response speeds within homogeneous behavior sequences
Colsher, Patricia Louise
MetadataShow full item record
An experiment was performed to test the validity of the assumptions made by Platt and Senkowski (1970) concerning the role of terminal responses (e.g., goal-entry responses) in determining the shape of the response-speed gradients of homogeneous behavior sequences. Three variables (locus of the food cup relative to the lever, response requirements for delivery of reinforcement, and reward magnitude) with two levels of each (near vs. far from the lever, completion of a fixed ratio 5 vs. completion of a fixed ratio 5 followed by a goal-entry response, and 2 vs. 6 ,045-g pellets) were factorially combined to yield eight experimental groups. The pattern of results was consistent with theoretical expectations. Response-speed gradients were in the shape of an inverted-U, although the peak response speed tended to occur earlier in the response sequence than had been expected. The pattern of goal-entry responses among subjects required to complete the ratio and perform a goal-entry response to receive reinforcement confirmed the predictions concerning the shape of the goal-entry gradient. Subjects for which reinforcement was contingent upon simple completion of the ratio requirement rarely made premature goal entries. Additionally, it was found that large reward resulted in more premature goal entries than did small reward. Further support for the proposed role of the terminal response as a determinant of lever-press performance was gained from the analysis of overall acquisition measures such as start speeds (reciprocal latency of the first response of the ratio) and two measures available only for the goal- entry terminated groups: the number of lever presses exceeding the fixed ratio requirement and the amount of time from the completion of the ratio to the performance of the goal-entry response that resulted in delivery of reinforcement. The pattern of results, while consistent with theoretical predictions, was not always statistically reliable. It was suggested that more robust effects might have been observed if the size of the fixed ratio were greater, the magnitude of the large reward were increased, and the effort required to perform the index response were increased.