Key Takeaways
- Behavior analysts have long described play deficits as a core feature of autism, but the field has often treated creativity and imagination as if they were unobservable internal states. Dr. Robert Ross argues these terms fail the basic behavior analytic test of being observable and measurable.
- What looks creative in typically developing children is, on closer inspection, a large repertoire of object, action, and language responses combined in novel ways. Children perceived as imaginative simply have more in their repertoires than children perceived as not.
- The teachable unit is the object-action-language routine, and behavior analysts can build creativity systematically through multiple modeling, point-of-view video modeling, activity schedules, and matrix training across peers, settings, and materials.
- Reducing challenging behavior without adding functional skills leaves the learner worse off. The field’s most important challenge may be to shift from being known as the people who stop bad behavior to being known as the people who are best at teaching new skills.
Defining Play, Creativity, and Imagination in Behavior Analytic Terms
Play, creativity, and imagination are three critical targets in early autism intervention that are often not taught, and this may be due to the fact that they are typically poorly defined. At the MassABA Annual Conference in Marlborough, Massachusetts, Robert K. Ross, Ed.D., BCBA-D, LABA, opened his keynote by asking the audience what those three words had in common. The answer, he said, is that they describe no observable behavior at all.
“Where is your imagination located? What does it look like? How much does it weigh?” Ross asked the room. The terms, he said, fail the standard that a behavior analyst applies to any other target. “I can’t observe your imagination. I can observe you engage in behavior, and it is this behavior that is the evidence of your imagination.”
Ross, former senior vice president of curriculum and research at Beacon ABA Services and a founding past president of MassABA, and current vice president of clinical integration and development at Biostream Technologies, has spent more than three decades training clinicians to translate behavior analytic principles into classroom and clinic practice. The argument he laid out in Marlborough is one he has refined for years: when a young child with autism is described as lacking creativity or imagination, the practitioner’s first job is to throw the labels out and look at what is actually happening. It is also an argument with implications for how the next generation of behavior analysts is trained, at a moment when the federal RISE rule has capped graduate student loan access for master’s-level BCBA programs and the pipeline of new clinicians faces its own structural pressures.
To make the point concrete, Ross described picking up a stick and pretending it was a gun, or driving a toy car around while singing “Happy Birthday” to it. Each behavior is observable. Each combines an object, an action, and language. Each is reinforced or not by the social environment. Teaching it effectively does not require positing an internal faculty called imagination.
Then he showed a video of two children playing. One moved through an elaborate kitchen routine with running commentary; the other sat with a single object and repeated himself. Ross said the audience’s intuition was the point. “Every kid who you know is creative because they have more language, they have more actions, they have more behavior in the repertoire. The kids with less aren’t creative.” When observers do not know a child’s learning history, he added, they are more likely to perceive that child as creative. The more they know about how each skill was taught, the more the perception shifts. Creativity, in this account, is a judgment about a repertoire, not a property of a mind.
Object-Action-Language Routines: The Teachable Unit of ABA Play Skills
The instructional mantra Ross returned to throughout the talk was “object-action-language routines.” A child picks up a car (object), pushes it across a mat while making a vroom sound (action), and labels what is happening, what is coming next, or who else is involved (language). String enough of those routines together across enough materials and peers, and what emerges in the child’s behavior is what observers call play.
Ross walked through four examples of procedures that can be used to build the unit. One procedure, multiple modeling, he said, is among the most underused. The structure is simple: line up four or five children, hand the first the materials, prompt the target action and language, then move down the line. By the time the fifth child takes a turn, they have observed four models with slight variations: louder, softer, faster, slower, higher, lower. The peer-to-peer variability is the point. “If you want to get observational learning, you have got to make it something that is habit, that is routine, and that gets reinforcement across a range of topographies and conditions,” Ross said.
Ross then explained how to use video modeling procedures. He discussed how point-of-view video modeling, in which the camera films from the actor’s perspective rather than facing the actor, produced stronger imitation than scene-based video modeling in a master’s thesis Ross supervised. Eleven actions and eleven language responses were targeted; point-of-view modeling moved the needle more on both, with the language gains the larger of the two. The reason, he suggested, is repertoire economy: it is easier to imitate a hand and a voice you experience as your own than a third-person view of someone else’s body.
The third workhorse procedure is the picture activity schedule, used to chain object-action-language routines into longer sequences. Ross described teaching a child to play with cars using a binder of pictures, each picture cueing a discrete action such as “drive,” “back up,” “broom-broom,” or “zoom.” Rearranging the order of the pictures forces the child to attend to the stimulus rather than to the sequence itself, which is the difference between teaching a routine and teaching generative play. Activity schedules can be faded from a binder to a checklist to nothing at all, and the same materials can be transferred from clinician to parent to grandparent without losing stimulus control.
Finally, Ross also pointed to matrix training as a way to multiply skills without multiplying instructional time. A 4×4 matrix of objects and actions yields sixteen combinations from a handful of directly taught exemplars. The point, again, is that what looks like spontaneous variation is actually the predictable output of well-designed teaching.
Where ABA Practitioners Are Misallocating Effort in Autism Play Instruction
The final third of the keynote turned to what Ross sees as misallocated effort across the field. The first target was reliance on verbal prompts to do the work that the environment and routine should be doing. Picture cues, peer models, and visible materials, he said, support production language more reliably than verbal redirection and correction from an adult. “Stop reminding kids what not to do, after they do it wrong or not at all,” he told the audience. “Raising your voice doesn’t teach new skills and new behavior.”
The second target was the field’s emphasis on decreasing challenging behavior at the expense of teaching functional replacements. The example was deliberately uncomfortable. “If I’ve got ten skills in my repertoire and nine of them are crappy, and you take one of them away, I have eight skills in my repertoire,” Ross said. “I’m actually worse off for having met you.” Reduction without replacement, in his framing, is not treatment. The behavior analyst’s job is to add to the repertoire, and adding new skills is something the science is particularly good at.
The third target was the gap between clinic-based skill demonstrations and outcomes that change a child’s life. Ross described decades of programs that produced impressive within-session data on non-functional tasks: rote labeling, simple discrimination, academic skills that never transferred home. The core deficits of autism, he said, are in social and interactional language, and those are exactly the skills that have to be built through object-action-language routines embedded in peer interactions. The argument also has implications for the ongoing debate over RBT retention and the ABA workforce shortage: the skills Ross says practitioners ought to be teaching are the same skills that require sustained, peer-rich therapeutic environments, which are exactly the environments that suffer when frontline staff turnover outpaces hiring. “We’ve got to make sure that our focus is on the core deficits of autism: social language, interactional skills. We want to help that kid develop the skills to be happy and to develop and maintain friendships. We want that kid to have a life.”
The structural argument running underneath the talk is one Ross has made for years across multiple ABA forums. Play is behavior. Behavior is selected by its consequences. Consequences are arranged by the people in the child’s environment. If the field accepts that, the implication is that creative play is buildable, that it is no harder to teach than any other complex repertoire, and that the responsibility for whether a child reaches it sits with the practitioner, not the child. Ross has been making a parallel case in his role as vice president of clinical development and integration at BioStream Technologies, a paid advisory position with a company whose Lookware platform uses eye-tracking and gaze-contingent gaming to teach the social orienting behaviors that conventional discrete trial instruction has consistently failed to deliver.
Ross closed where he began. Play is behavior. It has to be directly taught at the basic levels. The fundamental unit is the object-action-language routine. The more practitioners teach it, the more they get it. “It’s not hard stuff,” he said. “We overthink it because we use terms like creativity and imagination to pretend it’s in the imagination area.”
