Effects of Training Structure and the Passage of Time on Trained and Derived Performance

Abstract

Conditional discriminations established using a linear series training structure were shown in a previous study to be intact when unreinforced trials assessing such performance were presented interspersed among stimulus equivalence test trials. Experimenter-defined correct performance on these trials was seen both for participants responding in accordance with stimulus equivalence and for some participants who did not. The goal of the current study was to investigate the status of previously trained relations at the time of stimulus equivalence tests in further contexts. Experiment 1 investigated this following many-to-one conditional discrimination training with 20 adult participants. Experimenter-defined correct performance on probes for trained relations was observed for 16 out of 17 of participants who formed stimulus equivalence classes, as well as for all participants not responding in accordance with stimulus equivalence. In Experiment 2, 18 adult participants received linear series, many-to-one, or one-tomany conditional discrimination training and tests for stimulus equivalence with interspersed training trials. Subsequently, participants experienced three additional tests for derived and trained performance within a month of the original training. Continued testing and the passage of time had little effect on the stability of both learned and derived performance for participants initially responding in accordance with stimulus equivalence. Participants who failed to form stimulus equivalence classes in the initial test, but responded in line with the trained relations, displayed more varied performance over time. Results from both experiments indicate that the lack of stimulus equivalence performance is not commonly caused by a breakdown of trained relations at the time of the stimulus equivalence test.Establishing the prerequisite behavior for stimulus equivalence commonly involves training multiple conditional discriminations consisting of arbitrarily related stimuli using matching-to-sample (MTS) procedures. Conditional discriminations are defined by a discriminative stimulus, a comparison, being a behaviorally effective part of a three-term contingency only when a specific conditional stimulus, a sample, has been presented. In commonly used stimulus equivalence procedures, stable experimenter-defined correct performance in an MTS procedure establishing conditional discriminations is followed by a test for the emergence of derived performance having the properties of reflexivity, symmetry, and transitivity (Sidman and Tailby 1982). If performance according to these properties is observed, a stimulus equivalence class is considered established. Three common variations of MTS found in the stimulus equivalence literature are the linear series (LS), many-to-one (MTO), and one-to-many (OTM) training structures. In the LS structure, sets of conditional discriminations are linked by having the same stimuli appear as samples and comparisons on different trials in the MTS procedure; for example, by training s...

An examination of transformation of evaluative and consequential functions through derived relations with participant-generated values-relevant stimuli

“…In addition, the MTS procedure can vary in structure of initial training among linear protocols (consequating relating each stimulus as both sample to one comparison and comparison to another sample), oneto-many (consequating relating a single sample to different comparisons, as in the current study), or many-to-one (consequating relating many samples to single comparison). Data have been mixed with regard to the relative effectiveness of these structures at establishing mutual and combinatorial entailment (e.g., Arntzen & Holth, 1997Arntzen & Nikolaisen, 2011;Arntzen & Vaidya, 2008;Eilifsen & Arntzen, 2015;Fields et al, 1999;Hove, 2003;Saunders et al, 2005;Smeets & Barnes-Holmes, 2005). The present study data provide some support for extant literature on the appropriateness of one-to-many matching-to-sample procedures for establishing mutual and combinatorial entailment (e.g., Bordieri et al, 2016;Keenan et al, 2015;Stewart et al, 2015).…”

Time, Sleep, and Stimulus Equivalence-Based Relational Memory

“…Moreover, a future study employing the stimulus equivalence paradigm could increase the size of the relational classes from 4 to 5 or 6, increase the number of relations from 2 to 3, and use a training structure other than simultaneous one-to-many (e.g., linear series). These variables have all been shown to influence emergence of stimulus equivalence (e.g., Arntzen, 2012;Ellifsen & Arntzen, 2015;Fields et al, 1997) and should therefore reduce the likelihood of a ceiling effect, allowing for the predicted effects of sleep to be identified.…”