Establishing arbitrary comparative relations and referential transformations of stimulus function in individuals with autism
Children with autism can learn abstract size comparisons through equivalence networks and then apply the rule to brand-new items.
01Research in Context
What this study did
Two children with autism learned to pick the "bigger" or "smaller" picture when given made-up names.
The trainer first taught direct matches (A=B, B=C) with cartoon animals.
Later the kids saw new pairs and had to choose without further teaching.
What they found
Both children soon picked the correct bigger or smaller picture in every new pair.
The untrained choices showed up right away and stayed solid when the team retested the whole set.
How this fits with other research
Murphy et al. (2009) got the same jump from trained to untrained with "more-less" mands, proving comparative relations can emerge in autism.
Murphy et al. (2010) also showed derived mands, but for requesting items, not size; together the three studies say "train the network, not each item."
Chastain et al. (2025) pushed the idea further, building perspective-taking (I-You) frames; Belisle’s work is the earlier stepping-stone that showed abstract comparatives are possible.
Why it matters
You can stop drilling every bigger-smaller example. Teach a tiny network of arbitrary relations and let the child derive the rest. Start with two or three anchored pairs, test for emergent choices, and move to new stimuli only after the pattern holds. This saves hours of direct trial time and builds flexible language that transfers to novel pictures, objects, or even numbers.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →Pick two known pictures, teach A=bigger and B=smaller with nonsense names, add a B=C link, then probe untrained A-C size choices.
02At a glance
03Original abstract
Relational Frame Theory posits that complex language develops through arbitrarily applicable relational networks, with potential implications for individuals with autism. Responding relationally based on comparison occurs when participants respond to any number of comparative properties, such as "bigger" or "faster." Experiment 1 established two 3-member comparative networks, in which a stimulus A was conditioned as "bigger" or "faster" than a stimulus B, and the stimulus B was conditioned as "bigger" or "faster" than a stimulus C in 2 children with autism. Both participants met the mastery criterion for the trained relations and demonstrated the emergence of the untrained combinatorially entailed A-C and C-A relations. The participants could also match the arbitrary A stimuli with larger or faster objects and the C stimuli with smaller or slower objects. The results were replicated in Experiment 2 with the same participants, where a 5-member relational network was established for the bigger/smaller relation.
Journal of Applied Behavior Analysis, 2020 · doi:10.1002/jaba.655