Association for Behavior Analysis International

Behavior is controlled by antecedent stimuli and by contingent consequences. In the natural world, these antecedent stimuli and consequences often consist of multiple elements (e.g., color and line orientation) or dimensions (e.g., reinforcer magnitude and frequency). In such environments, behavior may come under the control of more than one element and/or dimension. Understanding what affects such divided control is critical for identifying the stimuli and consequences that maintain behavior, as well as for establishing control over behavior (as in behavior modification procedures). The research presented in this symposium explores divided control over behavior by stimuli and outcomes. Presentations will cover topics such as how informative stimuli control behavior, how stimuli that signal redundant or conflicting information compete for stimulus control, and the role of reinforcer frequency in divided control. The findings presented in this symposium are likely to be relevant to both experimental and applied researchers interested in how stimuli and outcomes can jointly control behavior in complex environments.

When behavior is controlled by multiple stimuli, stimuli associated with relatively higher reinforcer rates exert greater control over behavior. It is presently unclear whether this is also the case when stimuli signal conflicting contingencies. We investigated whether relative reinforcer rates determine divided stimulus control when stimuli signaled conflicting information about the location of the next reinforcer. Pigeons chose a left or right key conditional on the presentation of a sample stimulus. We varied the reinforcer rate associated with each stimulus in separate conditions. To investigate divided stimulus control, two stimuli were presented simultaneously in unreinforced test trials (Experiment 1), or the sample stimulus itself consisted of two simultaneously presented stimuli (Experiment 2). In Experiment 2, one stimulus (chosen randomly) signaled the next reinforcer location. When simultaneously presented stimuli signaled the same reinforcer location, subjects chose this location. When the stimuli signaled different locations, response ratios to the sample stimuli and to the choice keys approximately matched reinforcer ratios. That is, stimuli associated with higher reinforcer rates exerted greater control. Thus, divided control by conflicting stimuli also depends on relative reinforcer rates.

Cue conflict has a long history in the study of learning. Paradigms using cue-conflict procedures, such as overshadowing and blocking, have played an important role in the development and assessment of diverse theories of learning. We report a series of experiments investigating overshadowing of spatial cues to understand the role associative processes play in acquisition and expression of cued spatial behavior. Pigeons were reinforced with grain for pecking a visual target presented on a touchscreen. The location of the target could be signaled by one or more visual cues (i.e., landmarks) located elsewhere on the screen. Using this procedure, we investigated the theoretical basis of spatial learning in overshadowing. Our results indicate that, like conventional overshadowing, spatial overshadowing is governed by an associative process and likely reflects a learning deficit. Next, we report a series of experiments demonstrating similar overshadowing-like cue conflict effects at the time of test, which cannot be explained by a learning deficit but instead by competition between signals based on their relative salience. Both lines of research bear on the central issue of learning and performance processes that govern stimulus control of behavior.