|
| Choice and Levels of Analysis |
| Monday, May 30, 2016 |
| 8:00 AM–9:50 AM |
| Zurich FG, Swissotel |
| Area: EAB; Domain: Translational |
| Chair: Stephanie Gomes-Ng (University of Auckland) |
| Discussant: Sarah Cowie (University of Auckland, New Zealand) |
| Abstract: Traditionally, choice has been analyzed by aggregating responses to each alternative across several experimental sessions. Such extended-level analyses have shown that response ratios tend to match reinforcer ratios (the generalized matching law; Baum, 1974). However, extended-level analyses may not reveal the processes that may control choice. Thus, more recently, choice has been analyzed on a less-temporally-extended (‘local’) time-scale. These local analyses have shown that there are local-level regularities in choice, and these local choice processes may underlie extended-level choice. Some researchers have argued that these local-level regularities reflect the local effects of reinforcers on behavior, while others have argued that local-level regularities, such as the preference pulse, are merely artifacts of analysis type that arise due to more extended-level variables.
This symposium will present research and theory examining choice on different levels of analysis. Presentations will discuss whether local-level regularities in choice reflect the local effects of reinforcers, importance of contingency discriminability, the local- and extended-level effects of changeover delays on choice, local-level processes underlying suboptimal choice, and whether phylogenetic and ontogenetic explanations of behavior can be united to explain choice. |
| Keyword(s): local choice, multi-scale selection, preference pulse, suboptimal choice |
| |
Evolution as a General Theoretical Framework for an Explanation of Behavior |
| CARSTA SIMON (Oslo and Akershus University College) |
| Abstract: Since biologists no longer limit themselves to studying the evolution of physical bodies but have developed theories of complex human behavior such as altruistic and cooperative actions (cf. group selection theory,) their area of study overlaps considerably with that of behavioral scientists. Behavior analysists study how environmental events during an organism's ontogeny correlate with changes in that organism's allocation of time to different activities. Those events cause changes in behavior because of their effect on the individual's relative fitness. Despite the overlap in topic areas, both biologists and behavioral scientists are largely uninformed about each other's work. How do biology and behavior analysis relate to each other? How can biology benefit from behavior analysis by paralleling the behaviorist's approach to developing explanations of behavior omitting human agency as a causal factor? How can behavior analysis benefit from Baum's (1994) introduction of the Multi-Scale View, which paves the road between the two disciplines by arguing for a selection of nested activities through their correlation with phylogenetically important events? By breaking ground for uniting phylogenetic and ontogenetic explanations of behavior, the answers to those questions not only benefit basic knowledge but can also inform effective public policy making. |
| |
| The Effects of Changeover Delays on Choice |
| STEPHANIE GOMES-NG (University of Auckland) |
| Abstract: In concurrent schedules with a changeover delay (COD), the cause of preference pulses (transient, extreme preference towards the just-reinforced alternative) is unknown. When a COD is arranged, reinforcers can only be obtained from the just-reinforced alternative in the seconds after a reinforcer; a switch to the not-just-reinforced alternative instigates the COD. This change in the local reinforcer differential may produce preference pulses. Alternatively, preference pulses may arise because the COD increases mean visit length, hence decreasing the probability of a switch after a reinforcer. We investigated which of these explanations best accounts for the COD’s effects on choice. Pigeons participated in four conditions, in which the COD either did not operate, only operated after switches not preceded by a reinforcer, only operated after the first switch since a reinforcer, or operated after all switches. Preference pulses were obtained in conditions with the longest visits; changes in the local reinforcer ratio were not always accompanied by changes in local choice. Thus, preference pulses may be attributed to the COD’s effects on mean visit length. However, subjects appeared unable to discriminate the local reinforcer ratio. The present results therefore highlight the importance of contingency discriminability in control by time-based contingency changes. |
| |
| Local Choice Processes Underlying Melioration: Extending the Findings of Vaughan (1981) |
| VIKKI J. BLAND (The University of Auckland) |
| Abstract: Studies show that even when negative or harmful outcomes of choice patterns are signaled, humans and animals may continue to distribute choice sub-optimally. Sub-optimal choice may be defined as choice that results in less response-contingent reinforcement across time, relative to the overall availability of reinforcement. Sub-optimal choice may be the outcome of both global and local choice processes. One theory that attempts to explain sub-optimal choice is melioration theory (Herrnstein & Vaughan, 1980; Vaughan, 1981). Melioration theory suggests that organisms allocate their time between competing options in an attempt to equilibrate local rates of reinforcement to obtain a higher per unit return, rather than the greatest overall return. Whilst several studies have investigated and challenged melioration theory, there has been no direct replication of Vaughan’s seminal 1981 study. The present study used pigeons to replicate that study. Global and local choice analyses of results suggest that pigeons use different choice strategies to maximize overall rates of reinforcement, resulting in matching. However, when maximization and matching contingencies are placed in opposition to melioration contingencies, pigeons meliorate. These findings provide a platform for investigating operant procedures that may subvert the process of melioration, and potentially reduce the negative impact of sub-optimal choice. |
| |
| Local Effects of Reinforcement in Corrected Preference Pulses |
| ANTHONY P. MCLEAN (Canterbury University), Randolph C. Grace (University of Canterbury), Rebecca Bodeker (University of Canterbury) |
| Abstract: Preference pulses are intended to reveal what reinforcers do, locally, to responding in a choice situation. The characteristic shape of the preference pulse is determined largely by local probability of switching away from the currently-engaged choice alternative. Because this probability is generally low, preference pulses appear even in responding that occurs after an unreinforced response. Because switching probability may be further reduced immediately after reinforcement, preference pulses may be enhanced when constructed from post-reinforcer responding, but only the enhancement can be attributed to recent reinforcement. Thus, the difference between post-reinforcement and post non-reinforcement preference pulses might give a more accurate assessment of local reinforcement effects. In the present experiment, four pigeons responded on concurrent schedules in which some of the arranged reinforcers were delivered, and some were withheld. Separate preference pulses were constructed from responding that followed delivered and withheld reinforcers. In all four subjects, the difference between these preference pulses was initially very small, then increased and later, decreased. Thus, the local effects of reinforcement on do not appear immediately, as suggested by uncorrected preference pulses. This pattern is consistent with a visit-lengthening effect of reinforcement described by Buckner, Green and Myerson, and more recently by Baum. |
|
| |