Using video modeling to teach reciprocal pretend play to children with autism.

The team filmed short clips of two preschool peers acting out a pretend scene. Each clip showed the same script—like "let’s play restaurant"—with clear roles and lines.

Children with autism watched the clip, then practiced the same scene with a peer. The study used a multiple-baseline design across three different play scripts.

After only a few viewings every child could act out the full script with a peer. The kids also started using new, unscripted words during play.

Parents and teachers saw the children keep the new play and language skills weeks later without extra training.

Petry et al. (2007) used similar video clips two years earlier and saw the same jump in reciprocal play. Their work is a direct forerunner—Rebecca et al. simply added written scripts.

Lancioni et al. (2000) taught cooperative play with live modeling plus a verbal step. They found the verbal step was required; without it no child learned. Rebecca et al. skipped that step yet still got fast gains. The difference is the video format—visual models alone were enough when the scene was fully scripted.

Bloh et al. (2025) recently asked whether cartoon avatars work as well as real peers. They found mixed results: some kids liked cartoons, others needed real people. Rebecca’s peer-only clips sit at one end of that new continuum; Bloh gives you permission to try animated peers if a child tunes out human videos.

You can teach pretend play in one short session. Film two peers acting out a clear script, let the learner watch, then jump into the same scene. No extra verbal instruction needed. The child gains new words and keeps the game later, freeing you to target other social deficits.