Association for Behavior Analysis International

This symposium will consider the effects of meaning and training structure on equivalence class formation, the effects of relational type on the strength of relations among the stimuli in equivalence classes, and the induction of joint stimulus control by college level stimulus material. All four presentations address complex levels of stimulus control from different perspectives.

When a multi-nodal equivalence class is established by training AB, BC, CD, DE, EF, and FG, the class has a structure represented by A?B?C?D?E?F?G. Some emergent relations such as D?F and D?G are called transitive relations, while others such D?A and D?B are called equivalence relations. Two nodal stimuli separate the stimuli in the D?G and D?A relations while one nodal stimulus separates the stimuli in the D?F and D?B relations. According to a definitional view, all of the stimuli in an equivalence class are equally related to each other. Therefore, no preference should be shown for transitive over equivalence relations of the same nodal spreads. When evaluated using within class preference tests that pitted the 1-node transitive and equivalence relations against each other (D?B vs D?G) and the 2-node transitive and equivalence relations against each other, all participants showed virtually complete preferences for the transitive relations over the equivalence relations. Thus, relational type influences the relational strength of stimuli in transitive and equivalence relations. In contrast, when traditional cross class tests are conducted, all of the stimuli in an equivalence class function in an interchangeable manner, which reflects equal relatedness. The presentation will consider how both forms of relatedness can coexist. These data, along with preference data for nodal distance, support the view that2 structural parameters of equivalence classes control the differential relatedness of stimuli in equivalence classes.

When presented with complex stimulus arrays that have many features and descriptions of the features in those arrays, the selection of a correct description requires attending to all of the features, all of the components in the descriptions, and the correspondences of each feature and their corresponding descriptive components. When that occurs, the behavior is said to be under joint stimulus control. The present experiment describes how a joint stimulus control repertoire can be induced by use of matrix training, a minimal form of multiple exemplar training. This experiment was conducted using stimulus arrays that were graphs that depicted the interactive effects of two variables on behavior, and descriptions that were paragraphs that described the information contained in the graphs. Performances in a pretest showed a substantial amount of variation. For most participants, performances were much more indicative of joint stimulus control after matrix training, In addition, the post-training performances were similar to each other across participants, regardless of pre-test performances. The participants in the control groupshowed small changes in performances with a test retest procedure. These results then documented the efficacy of matrix training in the adduction of a generalized joint stimulus control repertoire, which resulted in participants identifying accurate and complete descriptions of the interactive effects of 2 variables on behavior for many novel graphs and corresponding descriptions.

When a2 3-node 5-member equivalence class is established by training AB, BC, CD, and DE on a concurrent basis, and all stimuli are meaningless, the classes are formed by 10% of the participants in Group 1. When the same classes are established using the same meaningless A, B, D, and E stimuli, along with a meaningful picture as the C stimulus, the classes are formed by 100% of the participants in Group 2. At the level of behavioral function, all meaningful cues serve as discriminative stimuli for some response. Thus, one of the factors that could account for the class-enhancing function of a meaningful stimulus might be its implicit discriminative function. This was assessed by having a third group attempt to form equivalence classes by training AB, BC, CD, and DE on a concurrent basis, where all stimuli were meaningless. Prior to class formation, however, the C stimuli acquired explicit discriminative functions by successive and simultaneous discrimination training. Thereafter, the classes were formed by 40% of the participants in Group 3. Thus, the implicit discriminative function served by a meaningful stimulus accounts for part of its class-enhancing effect. By implication, either overtraining of the discriminative function and/or other functions served by meaningful stimuli would have to account for the remaining class-enhancing effect shown by their inclusion as class members. Group 4 replicated the procedures used in Group 3 but with overtraining of the discriminative function, and 60% of the participants the formed classes. Therefore, overtraining of the discriminative function provided a modest increase in class enhancement. The remaining class enhancement effects of meaningful stimuli would have to be due to the other functions served by meaningful stimuli.

Lack of responding that is indicative of equivalence class formation may be related to a failure to respond according to the directly trained baseline relations at the time of equivalence tests. When establishing classes that has an LS structure, Eilifsen and Arntzen (2009) found that consistent responding according to the directly trained relations does indeed occur but is often not accompanied by class indicative performances by the derived relations probes. In this study unreinforced presentations of the baseline relations were interspersed among the derived relations test trials. Only one test block was presented. The current experiments determined whether similar findings would occur when the classes had other training structures. In Experiment 1, 62 college students were trained using either LS, OTM, or MTO structures. The experiment replicated the previous finding. Specifically, many participants responded in accordance with equivalence for both the directly trained relations and the derived relations. In contrast, other participants showed class indicative responding by the directly trained relations but the derived relations probes did not evoke responding that was in accordance with equivalence. These results are shown in Table 1. Therefore, the findings of Eilifsen and Arntzen (2009) were not contextually limited to classes with LS training structures. Rather, they occur regardless of the training structure of the equivalence classes. Experiment 1 did not involve continued testing for class formation. Thus, the results could not track the possible gradual emergence of responding in accordance with equivalence or the maintenance of the baseline relations during continued testing in classes that had different training structures or for classes of different sizes. To further illuminate the status of the directly trained relations in derived relations test blocks, both when equivalence performance is detected and when it is not, Experiment 2 will be conducted with an increased number of class members and with continued testing that will be conducted over several days. The results of this experiment extend the findings of Experiment 1, by permit us to track the maintenance of the baseline relations and the possible gradual emergence of derived relations over time and with co