Using RFT to Promote Generative Language: Understanding the HDML
As you know, I’ve been working on some new writing for our updated and expanded text on using RFT to promote generative language (which is still very much a work in progress). One area that has emerged in RFT since we published our first handbook is the Hyper-Dimensional Multi-Level (HDML) framework (Barnes-Holmes and Harte, 2022; Barnes-Holmes et al., 2020; Barnes-Holmes et al., 2017). While the HDML framework was originally conceived as a way of capturing and organizing the variables that conceptual and experimental work in RFT has considered, and for orienting basic researchers to avenues for future work, we also find the framework helpful to organize our thinking with respect to planning intervention. Over the next few posts, I’ll describe how this conceptualization can be viewed in the context of language development and intervention in early childhood.
The HDML conceptualizes arbitrarily applicable relational responding (AARR) in terms of levels and dimensions, and considers the unit of analysis in RFT to be the ROE-M—referring to relating, orienting, evoking, and motivating. In today’s post, I’ll explain the levels that we can take into consideration when planning early intervention. Stay tuned for the other aspects of the HDML in my next posts.
Levels in the HDML
In the HDML framework, the levels of AARR refer to the relational developmental level of the response: mutual entailment, combinatorial entailment, relational framing, relational networking, relating relations, and relating relational networks. As you may suspect, we find this particularly important in terms of how to sequence instruction for establishing new relational responding—whether in very early language development, or as new patterns of relating are targeted.
While the HDML “officially” starts at mutual entailment, we think it is consistent with the HDML framework to include the earliest developmental foundations for AARR, beginning with joint attention (conceptualized as mutually entailed orienting and evoking). Similarly, non-arbitrary relational responding is foundational for arbitrarily applicable relational responding, and should be considered as an earlier level of relational responding, emerging before AARR.
Mutual entailment is then the first form of generalized contextually controlled arbitrary relational responding to be acquired. Relational framing, the next level, simply refers to deriving based on the combination of two relations—the classic “triangle” of equivalence or any other pattern of responding.
Relational networking refers to deriving based on the combination of relational frames, whether involving the same type or many different patterns of framing, with increasing complexity as more patterns and frames are combined. Early relational networking can begin as soon as two relational frames are combined—for example, a cat (A) says meow (A→B) and has paws (A→C) (so an animal that says meow has paws, B→C), and my cat’s name is Emi (A→D), and Emi likes fish (D→E), and so an animal that says meow eats fish (B→E). Relational networking is involved in simple problem solving games like “Clue” (e.g., “it was the maid in the parlor with a knife!”) or “Twenty Questions” (e.g., relating a dog as “bigger than a breadbox” and “not edible” and “an animal” and “a pet” and “something that you take on walks”). Relational networking is also involved in following rules and instructions (e.g., “when the bell rings take the cake from the oven”).
Relating relations refers to the derivation of relations between relations themselves, as typified by analogy and metaphor. For instance, in the analogy ‘Apple’ is to ‘Orange’ as ‘Dog’ is to ‘Sheep’, apple and orange can be related as the same in the context of fruit while dog and sheep can be related as same in the context of animals, and as these are both sameness relations then they themselves can also be related as the same.
Finally, relating relational networks refers to deriving relations between not just two or more relational frames, but between complex relational networks. “False belief” and other perspective taking tasks provide examples of these, as one is required to derive relations between one’s own perspective and that of another person; relating relational networks would also be seen in more complex problem solving and academic and narrative-based tasks requiring comparing and contrasting of complex concepts or theories.
This sequence helps us to think developmentally in quite precise ways about our assessment and teaching plans—starting at the earliest developmental level before proceeding to more advanced developmental levels. By considering the earlier levels of mutually entailed orienting and evoking, and non-arbitrary relational responding, along with the HDML, we can thoroughly ground our work in RFT, while still meeting children where they are at—whether that is prior to any language skills, or at a highly complex level of problem solving.
I hope you’ve found this explanation helpful—I’d love to hear your thoughts and questions on this subject.