Association for Behavior Analysis International

The first paper by Vaidya will review the nonhuman literature on concept formation, functional class organization, and other instances of complex learning and ask whether Sidmans theory provides a framework within which these data can be interpreted. The second paper by Fields et al. describe an experiment in which 3 3-node 5-member classes were trained in 30 adult participants. All participants were given a sorting task with 15 cards, each of which was one of the stimuli in 1 of the 3 classes. These results suggest the sorting test may be more sensitive to partial class formation than the emergent relations test. In the third paper by Tomanari et al., 3 adults were taught conditional relations and tested for equivalence-class formation in a 2-choice matching-to-sample task. Equivalence tests suggest that select control was established for all participants; reject control was established for two of them. In the fourth paper by Arntzen et al., they asked how equivalence class formation was influenced when training 3 7-member classes in different training structures. The results show small differences in number of responses to criterion and responding in accordance with stimulus equivalence.

Sidmans (1994, 2000) conceptualization of the origins of equivalence relations makes a number of counterintuitive predictions about outcomes involving equivalence relations with human subjects. The paper has been responsible for much experimentation and many, though not all, of its predictions have been tested. Some have been verified, others have been disconfirmed, and yet others await further research. Though not often formally appreciated, Sidmans theory also has the potential to organize some findings from the nonhuman literature including the fact the differential observing responses facilitate the development of conditional relations and the differential outcomes effect. This paper will review the nonhuman literature on concept formation, functional class organization, and other instances of complex learning and ask whether Sidmans theory provides a framework within which these data can be interpreted. It will be suggested that Sidmans theory is able to handle some findings from the nonhuman literature. It is hoped that an extended discussion along these lines might lead to a reconsideration of the human and nonhuman data by researchers working with one or the other population

Thirty participants attempted to form three 3-node 5-member classes under the simultaneous protocol. Class formation was assessed with the performances occasioned by the emergent relations test performances. Regardless of class formation, all participants were given 15 cards, each of which was one of the stimuli in one of the three classes, and was asked to sort them into clusters of related stimuli. Two noteworthy findings were obtained by comparing the outcomes of both tests. 1) The 13 participants who formed the three equivalence classes sorted the cards into clusters that corresponded to the three experimenter defined classes. The performances evoked by the two tests were consistent with each other. The sorting test, however, took less than 2 minutes while the emergent relation test took about 30 minutes. Thus, the sorting test measured the emergence of all of the trained equivalence classes an order of magnitude faster than the traditional test of class formation. 2) Three of the participants who did not form classes sorted the cards into three clusters, one of which contained the five stimuli from experimenter defined class-1 while the other two contained unsystematic mixtures of stimuli from classes 2 and 3. A reanalysis of the emergent relations tests for these participants did not show the presence of equivalence class 1; rather, the performances showed not class indicative pattern. These results suggest that the participants do not form all of the equivalence classes, the sorting test may be more sensitive to partial class formation than the emergent relations test. Factors that might influence both of these finding will be considered in the presentation.

During the training of conditional relations in a matching-to-sample task, select (sample/S+ relations) and reject control (sample/S-) may yield different outcomes in equivalence tests. Would eye movements predict these controls? Three adults were taught conditional relations and tested for equivalence-class formation in a two-choice matching-to-sample task. In two different conditions, either select or reject control was biased during training. Different proportions of S+/S- and a delayed cue procedure were combined as biasing procedures. During the whole procedure, eye movements were recorded. Equivalence tests suggest that select control was established for all participants; reject control was established for two of them. Different patterns of observing characterized select and reject control. Under select control, whenever S+ was the first gazed stimulus, participants tended to respond without gazing S-; whenever S- was the first gazed stimulus, participants tended to gaze S+ before responding. Under reject control, whenever the S- was the first gazed stimulus, participants tended to gaze both comparisons before responding; one participant tended to observe comparisons more than once, and two of them made their choices having gazed only S-. By identifying select and reject control during training, factors that affect equivalence classes may be manipulated in future studies.

Saunders and Green (1999) put forward a discrimination analysis in which they stated that number of discriminations established in conditional discrimination training differed across different training structures, i.e, many-to-one, one-to-many, and linear series. The difference in number of discriminations among training structures will increase with number members in a class and should, therefore, affect the outcome more. Relatively, few studies have investigated this notion. Hence, the question we asked was how equivalence class formation was influenced when training three 7-member classes in different training structures. The design in the present study was to expose participants to different sequences of training structures: (1) MTO OTM, (2) OTM MTO, and (3) LS MTO. In addition to record number of trials to criterion, responding during tests for stimulus equivalence and reaction time, we also recorded eye fixations. The results show small differences in number of responses to criterion and responding in accordance with stimulus equivalence. However, responding in accordance with stimulus equivalence in LS is lower than for MTO and OTM.