When human subjects lose money in 50% of multipleschedule components, the most common finding is a decreased response rate in the other components (i.e., punishment induction). When money is lost in only 10% of components, however, the most common finding is an increased response rate in the other components (Le., punishment contrast). The purpose of the present study was to provide an experimental assessment of the relation between proportion of punishment components and contrast and induction. Four adult humans pressed a lever or pulled a plunger for money on a multiple VI VI schedule arranged with green and red rectangles. The proportion of red rectangles was either 10% or 50%. After responding was stable in both components, every response in the red component was followed by money loss. Across 10 punishment conditions, there was no systematic relation between proportion of punishment components and contrast and induction. With both the 10% and 50% proportions, induction occurred in four punishment conditions and contrast occurred in one. There were, however, systematic effects of condition sequence. For all subjects, the pattern across punishment conditions was induction, then contrast or less induction, then induction. Present results show that proportion of punishment components cannot account for discrepancies found in previous studies, and that effects of punishment on unpunished responses change over time, perhaps because of punisher novelty.

No abstract

First and foremost, I must thank my mentor John Crosbie for his gentle, patient, and unfailing guidance. So much of who I am I owe to John. Thanks also to my committee members Tracy Morris, Kent Parker, Mike Perone, and Dick Walls for their helpful comments and insights. Additionally, I thank Mike and Kent for their roles in shaping me into a behavior analyst. Thanks to Brandon Aragona, Holly Blanc, and Erin Kane for assistance with running subjects, analyzing data, and producing figures. And thanks to my parents for providing the support to make it all possible.