By Matt Harrington, BCBA · Behaviorist Book Club · April 2026 · 12 min read

Using Cool vs. Not Cool to Teach Social Interaction Skills to Children with Autism

Overview & Clinical Significance

Social interaction deficits are among the defining features of autism spectrum disorder and among the most impactful in terms of quality of life. Children with autism frequently have access to social environments — classrooms, playgrounds, community settings — without the skills to navigate them successfully. The result is often social exclusion, reduced peer relationships, and diminished access to naturally occurring reinforcement that typically developing children rely on for skill development.

This study by Dr. Justin Leaf applies the Cool vs. Not Cool social discrimination procedure specifically to the domain of social interaction skills. Rather than teaching learners to execute social behaviors in isolation, the study evaluates whether the procedure can teach children to evaluate and produce socially appropriate interactions — a qualitatively different instructional objective that targets the contextual judgment layer of social competence.

The clinical significance of this application is substantial. Social interaction is not a single skill but a dynamic, context-dependent process. A learner who can greet a peer in a role-play but does not know when greeting is appropriate or what type of greeting fits the social context has acquired a limited version of a socially functional skill. The Cool vs. Not Cool framework provides a way to teach the discrimination that supports adaptive, context-sensitive social interaction — a goal that is not only educationally important but is also directly linked to inclusion outcomes, peer acceptance, and long-term social participation.

Background & Context

Social interaction skills occupy a complex position within ABA's evidence base. Behavioral interventions have demonstrated the ability to increase specific social behaviors such as eye contact, greetings, joint attention, and turn-taking, but the literature has consistently noted that these skills often do not generalize or maintain in naturalistic social contexts without explicit programming. One explanation is that these skills are taught as isolated behaviors without the contextual discrimination that makes their use flexible and appropriate.

The Cool vs. Not Cool procedure, as applied to social interaction, builds on a recognition that social behavior is rule-governed in a functional sense — there are contextual regularities that determine when a specific interaction behavior is appropriate, and learners can be taught to identify and apply those regularities. This is consistent with behavior analytic concepts of rule-governed behavior and stimulus generalization, and it connects to broader cognitive-behavioral frameworks that emphasize social cognition and social problem-solving.

Within ABA, video modeling has emerged as a well-supported procedure for social skills instruction, particularly for autistic individuals who may respond well to visual and self-modeling formats. The Cool vs. Not Cool procedure leverages video modeling in a specific way — not to demonstrate a correct behavior for the learner to imitate, but to present both correct and incorrect social interaction examples and develop the evaluative discrimination. This represents a meaningful extension of the video modeling literature into the domain of social judgment.

Clinical Implications

Teaching social interaction skills through the Cool vs. Not Cool lens changes the structure of social skills programming in several important ways. First, it requires that practitioners identify not just what a target social interaction looks like but what makes it appropriate versus inappropriate — which contextual features are the critical stimuli for the discrimination. This analysis often reveals that social skills goals are underspecified in existing programming, where the expected response is described but the contextual conditions for appropriate use are not.

Second, it implies that social skills instruction should include negative examples — demonstrations of what not to do — as a systematic part of teaching. Many social skills programs rely exclusively on positive models of the target behavior, which may inadvertently teach the behavior without the discrimination. Adding not cool examples to instruction, presented and labeled explicitly, provides the contrast necessary for the learner to develop a functional discrimination.

Third, the Cool vs. Not Cool approach supports generalization by training across multiple exemplars and contexts. Practitioners designing Cool vs. Not Cool programs should deliberately include variety in the video examples: different actors, different settings, different relationship contexts (peer, adult, stranger), and different emotional contexts. This variety increases the probability that the discrimination will generalize to the novel social situations the learner encounters in their daily life.

For schools and inclusion settings, the Cool vs. Not Cool approach can be adapted into a naturalistic coaching model where the practitioner — or a trained peer — provides in-the-moment cool or not cool feedback during social situations, supporting real-time application of the discrimination that has been built in structured instruction.

Ethical Considerations

Teaching social interaction skills raises a set of ethical considerations that practitioners must navigate carefully. At the most fundamental level, BCBAs must ensure that the social skills being targeted reflect goals that are valued by the client and their family rather than goals imposed by the practitioner based on normative assumptions. Code 2.09 and related standards establish that intervention goals should be determined collaboratively and should address the client's welfare and stated priorities.

For children, whose preferences and values may be difficult to directly assess, caregiver involvement in goal selection is essential. However, BCBAs should also consider the learner's own perspective to the extent possible, including information from direct observation about which social situations the learner engages with most readily and what types of interactions appear functionally reinforcing. Social skills instruction that targets behaviors the learner finds aversive or meaningless is unlikely to produce genuine generalization and may undermine motivation.

Code 1.05 (Non-Discrimination) requires cultural humility in the definition of appropriate social interaction. What is considered a cool social interaction in one cultural context may be cool, not cool, or irrelevant in another. Practitioners should work with families to ensure that the social norms encoded in Cool vs. Not Cool examples reflect the learner's actual community and relational environment.

There is also a philosophical ethical consideration that the field is increasingly attending to: the right of autistic individuals to have their neurodivergent social styles respected rather than extinguished. BCBAs should be thoughtful about the distinction between teaching skills that enhance the learner's ability to navigate social environments on their own terms versus teaching conformity to neurotypical social norms. The former is a legitimate clinical goal; the latter may not be.

Assessment & Decision-Making

Designing a Cool vs. Not Cool social interaction program requires a multilevel assessment process. The first level is a social interaction needs assessment that identifies which types of social interactions are most relevant and challenging for the specific learner. This should draw on direct observation, caregiver interviews, teacher reports (when relevant), and review of previous social skills data. The output is a prioritized list of social interaction targets for which the Cool vs. Not Cool discrimination is clinically appropriate.

The second level is a stimulus development process for each target. Practitioners or their teams must create or identify video examples of each target interaction in both cool and not cool versions. For social interactions, this often requires scripted video shoots with actors who portray realistic peer or adult interactions. The quality and realism of these examples affect the procedure's clinical utility significantly. Stimuli should be reviewed by the clinical team and by caregivers before use.

The third level is a prerequisite assessment to confirm that the learner can engage with the video format, make verbal or symbolic judgments about video content, and attend sufficiently to process the relevant features of the examples. Learners who cannot yet engage with video at the level required may need preparatory instruction.

Once instruction begins, data collection should track the accuracy of cool versus not cool judgments, error patterns, generalization probe outcomes, and any changes in naturalistic social interaction behavior tracked through direct observation. This multi-method data system provides a comprehensive picture of whether the program is producing its intended effects.

What This Means for Your Practice

This study extends the Cool vs. Not Cool procedure into one of the most clinically important and practically relevant domains in autism intervention: social interaction. For BCBAs, the key takeaway is that social interaction programs benefit from an explicit discrimination training component, and the Cool vs. Not Cool procedure provides a structured way to deliver it.

Reviewing your current social skills curriculum through the lens of this procedure may reveal opportunities to strengthen existing targets. If a learner has mastered the performance of a greeting behavior but does not reliably use it appropriately in natural contexts, adding a Cool vs. Not Cool discrimination phase targeting the contextual features of appropriate greeting may be the missing piece.

For practitioners working in inclusive school settings, the Cool vs. Not Cool framework can be adapted for brief naturalistic teaching opportunities. A practitioner observing a social interaction can pause to ask the learner whether what just happened was cool or not cool, provide brief feedback, and use the interaction as a teaching trial without removing the learner from the natural environment. This embedded instruction approach supports generalization and is feasible within school inclusion models.

For caregiver training, Cool vs. Not Cool gives families a simple, intuitive framework and language for discussing social situations at home and in the community. When parents can say — in the same terms the learner has been trained with — whether an interaction they just observed was cool or not cool, they are extending the instructional effect of clinic-based training into the natural environment. Training caregivers to use this language consistently and correctly is one of the highest-leverage generalization strategies available for social skills programs.