Assessing and training analogical reasoning in young children using relational frame theory

Abstract

Analogical responding is pervasive in everyday language and cognition and is a key component in learning. However, despite this, there is as yet relatively little behavioural research on 1) the age of emergence and 2) assessment and training of this repertoire. The aim of the present thesis was provide additional insight into these issues by extending previous research that used relational frame theory (RFT) to investigate analogical relations in young children. Study 1 aimed to assess the age of acquisition during the typical development of analogical relations in young children between ages three and seven. Studies 2–4 aimed to investigate an RFT-based procedure to test and train analogy in young children; in Experiment 2 of Study 4, the procedure was extended to test and train analogy in children with autism spectrum disorders. Given the apparent importance of analogy for intellectual development, cognitive-developmental psychologists have examined the emergence of this skill in young children. Early researchers believed that analogical reasoning developed at the age of 12 or later (Levinson & Carpenter, 1974; Lunzer, 1965; Piaget et al., 1977; Sternberg & Nigro, 1980). More recently, however, it has been argued that children as young as four can show analogical reasoning (Goswami & Brown, 1990), with prior knowledge playing a critical role. Research on analogy has been mostly the province of cognitive psychologists, but behaviour analysts have also begun to research analogy during the last two decades. The impetus for this has primarily come from researchers who take an RFT perspective. Carpentier et al. (2002) found that 5-year-olds initially failed to show analogical responding and required additional training before doing so. Study 1 investigated a multi-stage training and testing protocol that allowed for assessing analogical responding in the context of the assessment of participants’ relational responding more broadly. Participant analogical relational performance was correlated with their age and intellectual performance as assessed on a standardised test of intellectual functioning. A second aim of Study 1, a cross-sectional study, was to measure relational responding of various types and at various levels of complexity in order to provide more comprehensive data on the emergence of basic framing patterns in 3- to 7-year-old children. The primary focus was on analogy, or the relating of relations, as one particularly important pattern of relational responding. Relational frame theory (RFT) views the operant acquisition of various patterns of relational framing (frames) as key to linguistic and cognitive development, and it has explored the emergence of a range of psychological phenomena (e.g., analogy, perspective-taking) in these terms. Despite the growing evidence that relational framing 1) underlies human cognition and language and 2) is operant behaviour, there is little research on the normative development of relations in young children. One potentially important advance for RFT research is to obtain more detailed information on the normative development of relational framing in childhood. Study 1 examined a range of frames, including coordination, comparison, opposition, temporality, and hierarchy at four different levels of complexity, of which two levels looked specifically at analogical responding (nonarbitrary relating, nonarbitrary relating of relations, arbitrarily applicable relating, and arbitrarily applicable relating of relations). The relational evaluation procedure (REP)-based training and testing format utilised in the current study was employed in the context of the multi-stage protocol relational assessment that allowed testing for a range of different types of relations. Study 2 assessed and trained analogical responding in young, typically developing children. Three 5-year-old children were assessed and trained in relating relations using an RFT-based REP protocol in a combination multiple baseline design across participants and a multiple probe design across behaviours. The study included a relational pre-assessment to screen potential participants; a baseline condition in which analogy was tested; and a training condition in which analogical responding was trained and generalisation probe trials (including three different probes; CE, DMC, and D-Cue Probes) were presented. After training in relating combinatorially entailed relations, all three participants showed analogical responding according to RFT’s conception of analogy as the derived relating of relations. Study 3 was a replication of Study 2; however, the correction procedure was modified to provide more training opportunities for incorrect responding. In Study 2, the probe trials were presented whether the participant emitted a correct or incorrect response during the training correction procedure. In Study 3, the training procedure required participants to respond correctly before the probe was re-presented. Three 5-year-old children were assessed and trained in relating relations using the RFT-based REP protocol in a combination multiple baseline design across participants and a multiple probe design across behaviours. As in Study 2, following multiple exemplar training, correct responding increased to criterion levels for all three children, and both generalisation and maintenance were observed. In Study 4, the testing and training procedure was modified to include a larger array of stimuli, and directly trained and mutually entailed relations within relations were assessed in addition to combinatorially entailed relations as in Studies 2 and 3. In Experiment 1 of Study 4, two 5-year-old typically developing children were assessed and trained in relating relations in a multiple baseline design. Following training, both participants successfully showed analogical responding during CE Probe sets, including the original CE Probe Set 1 used during baseline testing, a novel CE Probe Set 2, and the generalisation probe, CE+D Probe. Experiment 2 was a replication of Experiment 1; however, Experiment 2 sought to investigate analogical responding in children diagnosed with autism spectrum disorders. Two children with ASD were assessed and trained in relating relations. Following training, both participants successfully showed analogical responding during CE Probe sets, including the original CE Probe Set 1 used during baseline testing, a novel CE Probe Set 2, and the generalisation probe, CE+D Probe. These results suggest that this format can be used to successfully train children with ASD to respond to analogical relations as defined by RFT. The present thesis offers further insight into the testing and training of anlogical relations in young children. The analogy assessment data provide further evidence that analogical relations develop around age five, and the training data show that analogy can be successfully trained when the repertoire is weak or missing in five-year-old children. The data from the present studies suggest the experimental and applied potential of the REP format. The REP format permitted multiply controlled studies in which we could target analogy testing and training directly while maintaining experimental control. Furthermore, it afforded us with quick and effective stimulus control, allowing us to implement multiple baseline designs to examine the efficacy of multiple exemplar training to establish the core repertoire. Future work could extend the REP format to test and train analogical relations beyond coordination and distinction. A closely related possibility for further research could be to examine the effects of training sameness relations on the emergence of other relations. The present thesis contributes to the extant behavioural research on relational language assessment and training. The resulting data suggest that regardless of the level of complexity of the derivation required (i.e., whether directly presented, mutually or combinatorially entailed), relations between relations should be considered analogy. Thus, derived relations between nonderived relations are also analogies, albeit simpler than derived relations between derived relations. The present data also constitute an important addition to the literature on analogical relations beyond behaviour analysis. Although cognitive researchers have an extensive literature on analogy, they have not yet presented a functional analytic model analogy that might lend itself to training this repertoire. Considering the relevance of analogy to intellectual potential, future researchers could investigate the generalised effects of training analogical responding on socially valid measures such as mainstream analogy tests, academic achievement tests, or standardised tests of cognitive performance. Future work could examine if training children with ASD or other developmental delays using procedures such as the present one might result in generalisation to the understanding and creation of novel figurative language in a more naturalistic context. Considering the potential for improving language and cognition, further research into relational assessment and training, in general, is undoubtedly warranted.