The impact of high versus low levels of derivation for mutually and combinatorially entailed relations on persistent rule-following

Abstract

The effects of rules on human behaviour have long been identified as important in the psychological literature. The increasing importance of the dynamics of arbitrarily applicable relational responding (AARR), with regards to rules, has come to be of particular interest within Relational Frame Theory (RFT). One feature of AARR that previous research has suggested may differentially impact persistent rule-following is level of derivation. However, no published research to date has systematically explored this suggestion. Across two experiments, the impact of levels of derivation was examined on persistent rule-following at two stages of relational development: mutual entailment (Exp. 1) and combinatorial entailment (Exp. 2). A Training IRAP was used to establish a mutually entailed relational network in Experiment 1 and a combinatorially entailed network in Experiment 2, and to train these networks to different levels of derivation. This was followed by a contingency switching Match-to-Sample (MTS) task to assess rule persistence. Results from both experiments were generally consistent with the suggestion that lower levels of derivation produce more persistent rule-following. Unexpectedly, however, the findings from Experiment 1 also indicated that persistence was moderated by the type of novel word employed. Variations in results across both experiments and their implications for future research are discussed.

Exploring a Training IRAP as a single participant context for analyzing reversed derived relations and persistent rule‐following

“…The results of Experiment 2 suggested that the derived rule had minimal impact on an MTS performance when the ‘meaning’ of the derived rule was reversed, but the original MTS contingencies were maintained. One important difference between the current study and the previous research reported by Harte et al (2017, 2018), and Harte, Barnes‐Holmes, Barnes‐Holmes, McEnteggart et al (2020) is that in these earlier studies, the MTS contingencies were reversed during exposure to the task but in the current study the contingencies remained unchanged (i.e., it was the derived rule that was reversed). Experiment 3 was a partial replication of Experiment 2, but the MTS procedure involved a reversal in the task contingencies.…”

“…Following the rule was initially reinforced but the contingencies subsequently reversed and thus the rule now specified a direct contingency that no longer applied. This study and others that followed provided evidence that rules involving derived relations could override the impact of contact with direct contingencies (Bern et al, 2020; Harte et al, 2018; Harte, Barnes‐Holmes, Barnes‐Holmes, McEnteggart et al, 2020).…”

“…Recent attempts to bridge the gap between these two research areas have involved training and testing a simple relational network and inserting a derived element of that network into a rule for responding on a contingency‐switching MTS task. The critical measures involved determining the extent to which participants would persist with the derived rule when the contingencies switched and were thus inconsistent with that rule (e.g., Harte et al, 2017, 2018; Harte, Barnes‐Holmes, Barnes‐Holmes, McEnteggart et al, 2020). Experiments 2 and 3 reported here employed the derived (persistent) rule‐following paradigm developed by (Harte et al, 2017, 2018; Harte, Barnes‐Holmes, Barnes‐Holmes, McEnteggart et al, 2020) to determine if flexibility in reversing the derived A‐C relations in Experiment 1 would generalize in a persistent rule‐following context.…”

“…Later in the current article, we will argue that persistent rule‐following in the face of competing reinforcement contingencies may be usefully interpreted as evaluating the various properties (in this case coherence) of two relational networks; specifically, the derived rule provided within the experiment versus the rule generated by direct exposure to the task contingencies. Finally, it is again worth noting that unlike previous research in this area (i.e., Bern et al, 2020; Harte et al, 2017, 2018; Harte, Barnes‐Holmes, Barnes‐Holmes, McEnteggart et al, 2020), the current study employed single‐case experimental designs in order to demonstrate high levels of behavioral prediction and influence.…”

“…The experiments reported in the current study aimed to address this question. In conducting this work, we adopted a methodology (a training version of the implicit relational assessment procedure [IRAP] 1 ), similar to that employed by Harte and colleagues (e.g., 2018; Harte, Barnes‐Holmes, Barnes‐Holmes, McEngeggart et al, 2020) for training and testing, and then reversing and retesting the relevant derived relations. We opted to use this methodology because this recent work had successfully demonstrated that derived rules could indeed override direct reinforcement contingencies, at least for a limited number of trials.…”