Key Highlights

• Cooperative play is the most advanced stage of play, and most difficulties come from targeting it before the earlier stages are solid.
• Prerequisite skills like joint attention, turn-taking, and tolerating a wait predict success more than the play activity itself.
• The most effective strategy is engineering activities where cooperation is structurally required rather than merely encouraged.
• Peer-mediated intervention, where the peer is taught too, has one of the strongest evidence bases in this area.
• Video modeling and script training with systematic fading help children learn social sequences without depending on adult prompts.
• Adults should exit the interaction early and deliberately, because prompt dependence is the most common reason play skills fail to generalize.
• The goal is genuine connection and access to peers, not making a child’s play look neurotypical.

Play is where a great deal of childhood social learning happens. Negotiation, perspective taking, flexibility, repair after conflict, and the simple experience of enjoying another person all develop inside games that adults often mistake for downtime. For many autistic children, this is also one of the areas where support makes the biggest practical difference, because peer access in school and community settings frequently depends on it.

Cooperative play is specifically the stage where two or more children work toward a shared goal with assigned or negotiated roles. Building a fort together. Playing a board game. Running a pretend restaurant where one child cooks and another takes orders. It requires holding a joint objective in mind while coordinating with someone whose actions you do not control.

That is a genuinely complex skill, and it is taught most successfully by breaking it into components and building deliberately. Here is how skilled behavior analysts approach it.

Start By Identifying the Real Entry Point

The most common error in this area is targeting cooperative play when a child has not yet consolidated the stages that come before it. The result is frustration on all sides and a conclusion that the child “is not interested in other kids,” which is usually inaccurate.

Play develops through a recognized progression, and each stage supports the next.

Assessing honestly matters here. A child who plays comfortably in parallel with a peer for fifteen minutes is doing well, and the next target is associative play, not a four-player board game. Skipping stages tends to produce a child who can perform a play routine when an adult directs it and cannot do anything with a peer independently.

Build the Prerequisite Skills First

Cooperative play sits on top of a set of component skills. When play instruction stalls, the cause is almost always one of these rather than the play activity itself.

• Joint attention. Sharing focus on the same object or event, and shifting gaze between the object and the person.
• Imitation. Copying a peer’s action with materials, which is how most play learning happens naturally.
• Requesting and commenting. Asking for a turn, a material, or help, in whatever communication mode the child uses.
• Turn-taking with waiting. Tolerating a delay between turns without leaving or escalating.
• Flexible use of materials. Accepting that a peer may use a toy differently than the child prefers.
• Tolerating losing and not getting the preferred role. Often the single hardest component, and worth teaching directly.
• Responding to a peer’s bid. Noticing and answering when another child initiates.

Behavior analysts typically assess these individually and target the weakest ones before layering on complexity. It is far more efficient to spend three weeks building tolerance for waiting than to spend three months failing at board games because waiting is the actual barrier.

Engineer Activities That Require Cooperation

This is the highest-leverage strategy in the entire toolkit, and it is underused.

Most play activities allow cooperation without requiring it. Two children can sit at the same building set and each build separately. When cooperation is optional, children do the easier thing, which is fair.

The alternative is arranging materials so that the activity cannot proceed without interaction. Behavior analysts sometimes call this contrived interdependence, and it does the teaching work that verbal prompting cannot.

Examples that work well:

• Split materials. One child holds the marble run pieces, the other holds the marbles. Neither can proceed alone.
• Two-person equipment. A seesaw, a tandem swing, a parachute, a wagon that needs a puller and a rider.
• Divided roles in pretend play. One child is the cashier with the register, the other is the customer with the money.
• Single-set puzzles or crafts where pieces are distributed between two children rather than pooled.
• Physical tasks requiring two people, such as carrying a large box, holding a sheet to make a tent, or a ball activity that needs a passer.

We used exactly this approach with a seven-year-old boy who could take turns beautifully with adults and almost never with peers. Every attempt at a shared activity dissolved within a minute. Rather than continuing to prompt him toward interaction, we changed the setup. We gave him a marble run track and gave a peer the marbles, then stepped back and said nothing.

The first session produced about ninety seconds of engagement. He looked at the peer, waited, and eventually held out his hand. That gesture was the whole target, and it happened without a single adult prompt because the activity itself created the need. Within six weeks he was initiating with that peer at recess, and the initiations looked like his own rather than a rehearsed script. Details have been changed for privacy, but the principle is one we return to constantly. Design the activity well and it teaches for you.

Teach the Peer, Not Only the Child

Peer-mediated intervention has one of the strongest research bases of any social skills approach, and it addresses a structural problem: typically developing peers often do not know how to include an autistic classmate, and their attempts fail in ways that discourage both children.

Peer training is straightforward and usually brief. Peers learn to get attention before speaking, wait longer than feels natural for a response, accept communication in whatever form it arrives, including a device or gesture, offer a choice of two activities, and keep trying after a first attempt does not work.

The effect is often immediate. Interactions that previously ended after one failed bid begin to continue, and the autistic child gets far more successful practice. It also shifts the social load off a single child, which is both more effective and more respectful.

Video Modeling and Script Training

Video modeling works particularly well for play sequences because play is visual and repeatable. A short clip of two children playing a specific game, watched a few times before the activity, gives a child a clear model without an adult narrating over the top of the moment.

Script training serves a similar purpose for the language component. A child learns a small set of phrases for entering and sustaining play, such as asking to join, offering a turn, or suggesting what to do next. The essential part is the fading. Scripts are systematically shortened and removed so that the child ends up with flexible language rather than a fixed routine. A program that teaches scripts and never fades them has produced a performance, not a skill.

Prompt, Then Get Out of the Way

Prompt dependence is the most common reason play skills look good in session and disappear at recess. When an adult is standing beside two children directing the interaction, the children are not playing together. They are both responding to the adult.

Effective practice uses the least intrusive prompt that works, delivers it quickly, and fades it aggressively. Beyond that, it uses proximity fading, which means the adult physically moves away on a plan: from beside the children, to a few feet back, to across the room, to out of sight. Each step is deliberate and data is collected at each distance.

Ask a provider how they fade themselves out of play activities. The specificity of the answer is a good indicator of whether generalization is being taken seriously.

Reinforce the Interaction, Not the Compliance

What gets reinforced determines what gets learned. If reinforcement follows adult-directed compliance, a child learns to follow adults during play. If it follows peer interaction, a child learns to interact with peers.

Practically, this means reinforcing initiations toward the peer, responses to the peer’s bids, and sustained engagement, while letting the activity itself carry as much reinforcement as possible. The strongest outcome is when the play becomes reinforcing on its own, which is why activity selection based on the child’s genuine interests matters so much. A child who loves dinosaurs will work harder inside a dinosaur game than inside any activity chosen for its social convenience.

Group contingencies, where two children earn something together, can also be useful because they align the children’s interests rather than putting them in competition.

Teach the Hard Parts Directly

Several components of cooperative play are difficult enough that they need their own instruction rather than being expected to emerge.

• Losing a game. Practice deliberately with brief games, high reinforcement for calm responses, and a graduated increase in how much the loss matters.
• Joining a game already in progress. A specific and teachable sequence of observing, waiting for a natural pause, and asking.
• Handling a peer who says no. Teaching an alternative response rather than treating refusal as a failure.
• Repair after conflict. Simple, genuine scripts for reconnecting, taught without shame attached.
• Sharing a highly preferred item, which is a much harder skill than sharing a neutral one and should be taught separately.

A Note on What Success Means

The purpose of teaching cooperative play is connection and access to peers, not making a child’s play resemble a neurotypical template. Some autistic children genuinely prefer parallel play alongside a friend, and that is a legitimate and satisfying way to be with someone. Some prefer one close friend to a group. Some love structured games and dislike open-ended pretend play, which is a preference many adults share.

Good programming expands options and lets the child choose among them. It does not treat solitary play as a deficit to be eliminated, and it does not require eye contact or scripted enthusiasm as proof of engagement.

Helping Your Child Connect With Peers

Cooperative play is built, not waited for. It comes from identifying the right starting stage, strengthening prerequisite skills, designing activities where cooperation is necessary, preparing peers, and fading adult support on purpose so that children are genuinely playing with each other.

At Kennedy ABA, our behavior analysts build play and peer skills into everyday routines and real settings, working with siblings, classmates, and community opportunities so that the skills hold up where they matter. We focus on giving the autistic children we serve real access to friendship on their own terms. Our teams provide services across North Carolina, Virginia, Georgia, and Alaska.

If you want support helping your child build genuine connections with peers, contact us today to talk with our team about where to start.

---

Frequently Asked Questions

1. At what age should we start working on cooperative play?

Follow developmental readiness rather than age. If a child is showing interest in other children, tolerating proximity, and playing in parallel, the foundation is there. Prerequisite work can begin much earlier and is valuable regardless of when full cooperative play emerges.

2. My child plays well with adults but not with other children. Why?

Adults adapt constantly. We wait longer, interpret unclear communication, follow the child’s lead, and repair breakdowns automatically. Peers do none of that. This gap is very common and is exactly why peer training and engineered activities matter more than additional adult-led practice.

3. Are siblings good practice partners?

Often the best available, because they are present daily and motivated. They also need brief coaching, and the arrangement works better when the sibling has a defined role rather than being asked to be a therapist. Sibling relationships should stay relationships.

4. How long does progress usually take?

Component skills such as turn-taking often improve within weeks. Sustained cooperative play with peers in natural settings typically develops across months and depends heavily on how much real peer practice a child gets. Frequency of genuine opportunities matters more than session intensity.

5. Should we push our child to play if they seem to prefer being alone?

Teach the skills so the option exists, and then respect the choice. A child who can join peers and sometimes chooses not to is in a very different position than a child who wants to join and cannot.

---

Sources:

• https://pubmed.ncbi.nlm.nih.gov/10817421/
• https://www.autismparentingmagazine.com/social-skills-cooperative-play/
• https://www.sciencedirect.com/science/article/pii/S1875952124000211
• https://pmc.ncbi.nlm.nih.gov/articles/PMC5087797/
• https://pubs.asha.org/doi/10.1044/2024_AJSLP-23-00479
• https://pmc.ncbi.nlm.nih.gov/articles/PMC11428823/