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.


First International Conference; Italy, 2001

Event Details

Previous Page


Symposium #105
The Next Generation of Equivalence Research
Friday, November 30, 2001
3:00 PM–3:50 PM
Palladian Refectory Hall
Area: EAB; Domain: Applied Behavior Analysis
Chair: Lanny Fields (Queens College)
Abstract: In the last five years, research spawned from the study of equivalence classes has begun to extend its scope in two directions. Some studies have begun to explore how equivalence classes can be formed when training is conducted with procedures other than matching to sample. These studies provide explicit linkages between the study of equivalence classes and the processes that influence classical conditioning. Other studies have begun to study the formation of complex perceptual categories and have linked them to the study of equivalence classes, and the traditional domains of concept formation and categorization. These experimental results illustrate how emergent performances are accounted for from three different explanatory perspectives: a relational frames, stimulus control topographies, and environmental parameter spaces. Finally, the papers in this symposium illustrate how the study complex human behavior from a behavioral analytic perspective can make meaningful contact with other traditional areas of psychological inquiry such as associative learning and cognition.
Three Determinants of a Generalized Ability to Categorize New Information
LANNY FIELDS (Queens College), James Belanich (Queens College), Priya Wadhwa (Queens College), Antonios Varelas (Queens College)
Abstract: Five groups of college students were exposed to different experimental conditions, each of which influenced the emergence of a generalized repertoire of categorizing stimuli in novel stimulus domains with no direct training. 102 stimuli were divided into 6 pairs. Each stimulus in a pair was used as an endpoint of one of 6 synthetic stimulus continua: A, B, C, D, E, and F. Intermediate stimuli in each continuum were produced using morphing software. In Group 1, subjects received identity conditional discrimination training with the endpoint stimuli in Domain A. In group 2, training involved the presentation of many stimuli in domain as samples and the endpoint stimuli as comparisons. In group 3, training involved the presentation of many stimuli in domain as samples and the endpoint stimuli as comparisons. Training was conducted with stimuli in Domains A and B. In group 4, training involved the presentation of many stimuli in domain as samples and many stimuli from the same domain as comparisons. As with Group 3, training was conducted with stimuli in Domains A and B. After training, generalization tests were conducted with all of the stimuli in all six domains. These generalization tests were conducted in the absence of any reinforcement. The generalization tests involved the presentation of many variants of any reinforcement. The generalization tests involved the presentation of many variants of stimuli in a given domain as the sample with the endpoint stimuli in the same domain as comparisons. Spontaneous categorization was demonstrated by the selection of one endpoint stimulus in the presence of many contiguous stimuli from the same end of the domain would constitute categorization, for the domains that were not used in training. Thus, two categories could be formed for each domain. Performances indicative of the formation of 10 categories (2 each from domains B-F) could emerge after training in Groups 1 and 2. Performances indicative of the formation 8 categories (2 each from domains C-F) could emerge after training in Groups 3 and 4. 17, 37, 72 and 95 of the possible untrained categories emerged after training in groups 1-4, respectively. A between group analysis then can be used to identify three independent variables responsible for the inducing a generalized ability to categorize information arrayed in novel stimulus domains: (a) multiple exemplar training with exemplars as samples, (b) multiple exemplar training with variants as samples and compar
Stimulus Equivalence Methodology in Modeling Categories: The Case of Hierarchically-Organized Categories
KRISTA M. WILKINSON (Eunice Kennedy Shriver Center for Mental Retardation), William J. McIlvane (Eunice Kennedy Shriver Center for Mental Retardation)
Abstract: Although Sidman's stimulus equivalence paradigm has been widely discussed as a potentially useful model of categorization, no one thus far has developed procedures that model the complexity of true, hierarchically based categories. To this end, we will describe a novel research methodology that marries elements of equivalence research with the widely known psycholinguistic methodology of sentence verification. Subjects are trained in a matching-to-sample format in which they match at two levels: (1) lexigram labels (B) with novel exemplars © and, (2) higher-order category labels (A) with the lower-order lexigram labels (B). In conventional stimulus equivalence research, teaching relations AB and BC would be expected to yield emergent relations BA, CB, AC, and CA. To model hierarchically organized categories, however, it is necessary to constrain the emergence of certain relations. Specifically, it is necessary to demonstrate an asymmetrical relation between A and B and between A and C, while maintaining the symmetrical relation between B and C. Sentence verification tasks allow us to asses directly the symmetricality of such relations. Results with normally developing adults indicate successful hierarchical category learning with the combination of equivalence training and sentence verification tests. The approach provides an alternative stimulus control analysis of performances typically attributed to "relational frames".
The Respondent-Type Training Procedure: Some Recent Developments
GERALDINE LEADER (University of Ulster at Jordanstown, Ireland)
Abstract: Over the past number of years we have developed a training procedure that does not involve matching-to-sample, but reliably generates emergent or derived matching performances in both adults and children. The basic procedure simply involved presenting subjects with a multiple random series of stimulus pairs with a within-pair-delay of 0.5 seconds and a between-pair delay of 0.3 seconds. This procedure has been seen to be more effective than the traditional matching-to- sample training and has been used in the educational setting to teach fraction- decimal equivalence to five-year-old children. The most recent experimental manipulations using this procedure will be outlined in this paper. The results of this experiment will be related to contemporary relative-time analyses of Pavlovian conditioning.



Back to Top
Modifed by Eddie Soh