Association for Behavior Analysis International

The Association for Behavior Analysis International® (ABAI) is a nonprofit membership organization with the mission to contribute to the well-being of society by developing, enhancing, and supporting the growth and vitality of the science of behavior analysis through research, education, and practice.


33rd Annual Convention; San Diego, CA; 2007

Event Details

Previous Page


Symposium #173
International Symposium - Conditional Discriminations: Conceptual Issues and New Findings
Sunday, May 27, 2007
10:30 AM–11:50 AM
Randle A
Area: EAB/TPC; Domain: Basic Research
Chair: John A. Nevin (University of New Hampshire)
Abstract: Performance in conditional discriminations such as matching-to-sample and signal detection is affected by reinforcement and by the stimuli defining the task. Many recent results can be explained by a theory which assumes that probabilities of attending to the stimuli depend on overall reinforcement but are independent of the properties of those stimuli. Nevin will review the theory and its application to various paradigms including delayed matching-to-sample (DMTS). Three new studies with pigeons as subjects will be presented. Odum, Ward, Jimenez-Gomez, Shahan, and Nevin will report that DMTS accuracy is higher but less resistant to change in a condition with differential outcomes but lower reinforcer probabilities, challenging previous findings relating both accuracy and resistance to change to reinforcement. Ward and Odum will describe studies showing that resistance to change of accuracy depends on the stimulus modality as well as reinforcer probability, challenging their independence. Davison will report both local and molar analyses of the effects of differential reinforcement in relation to the difference between conditional stimuli that challenge the interpretation of some terms in the theory of attending. Conceptual and empirical questions raised by these studies will lead to changes in theoretical assumptions and prompt future research on conditional discriminations.
Overview of Theory and Conceptual Issues.
JOHN A. NEVIN (University of New Hampshire)
Abstract: In conditional discriminations such as matching to sample (MTS), reinforcement for a response to a comparison stimulus depends on the value of a preceding sample stimulus. A theory of conditional-discrimination performance suggests that when a reinforcer is presented, it strengthening effects depend on the similarity of the sample and comparison stimuli and that accuracy depends on the probabilities of attending to those stimuli, which are directly related to reinforcement. The theory accounts for the accuracy of discrimination, its resistance to change, and the allocation of choices in relation to reinforcer ratios, and with remembering construed as a form of attending, it may be extended to accuracy in delayed matching to sample (DMTS). Although it explains many archival findings, the theory does not account for the effects of differential outcomes, to be presented by Odum et al; for the effects of stimulus modality, to be presented by Ward and Odum; or for the interdependence of theoretical parameters when the difference between sample stimuli is varied, to be described by Davison. These new results may suggest ways to conceptualize the behavioral processes underlying conditional discriminations more effectively.
Persistence of Accuracy and Response Rate in Delayed Matching-to-Sample with Differential Outcomes.
AMY ODUM (Utah State University), Ryan D. Ward (Utah State University), Corina Jimenez-Gomez (Utah State University), Timothy A. Shahan (Utah State University), John A. Nevin (University of New Hampshire)
Abstract: The Differential Outcomes Effect is a robust phenomenon in which performance in conditional discriminations is superior when different outcomes follow correct choices of different comparisons than when non-differential outcomes follow choices. Performance with different outcomes is superior even when the overall rate of reinforcement for correct choices is lower than with the same outcomes. This experiment investigates the persistence of responding maintained by differential and non-differential outcomes. Four pigeons were trained in a multiple schedule where key pecking produced delayed matching-to-sample (DMTS) trials on variable-interval (VI) schedules. Reinforcer probability for correct matches was .1 following S1 and .9 following S2 in Differential Outcomes (DO) component, and .9 following both in the Same Outcomes (SO) component. Under baseline conditions, accuracy of the conditional discrimination was higher and declined less steeply with delay in the DO component. Response rates during the VI were similar or higher in the SO component. With the introduction of standard disruptors like extinction and pre-feeding, response rates during the VI and matching accuracy were both relatively more persistent in the SO component with the higher overall reinforcement rate, despite the higher level of accuracy in the DO component.
Stimulus Dimension and Resistance to Change of Conditional Discrimination.
RYAN D. WARD (Utah State University), Amy Odum (Utah State University)
Abstract: Recent data suggest that conditional-discrimination accuracy is more resistant to change in a high-reinforcement probability context than in a low reinforcement-probability context. The present experiments assessed whether this outcome depends on the nature of the sample stimuli. Five pigeons responded under a multiple schedule of conditional-discrimination procedures. In one component, correct choice responses were reinforced with a high probability (.8), while in the other component correct choice responses were reinforced with a lower probability (.2). The sample stimuli in Experiment 1 were temporal durations (2 and 8 s) while in Experiment 2 the sample stimuli were different key colors. In Experiment 1, there was no systematic difference in resistance to change between the high and low reinforcement probability components. In Experiment 2, resistance to change of color-matching accuracy was greater in the high reinforcement-probability component that in the low reinforcement-probability component, replicating previous results. These findings suggest the nature of the sample stimuli may affect resistance to change in conditional-discrimination procedures.
Conditional Discrimination at Three Levels of Stimulus Difference: Implications for Theory.
MICHAEL C. DAVISON (University of Auckland, New Zealand)
Abstract: Six pigeons were trained on a simple conditional discrimination. Reinforcer ratios for correct responses were varied across conditions at each of 3 levels of conditional-stimulus differential. Many sessions of data were collected for each condition, allowing both extended and local analyses of choice. Control by differential reinforcement increased as stimulus differential decreased both at the local and extended levels in a series of different analyses. The data pinpoint problems in current quantitative formulations, and define the direction of future developments.



Back to Top
Modifed by Eddie Soh