A preliminary investigation of fixed and repetitive models during object imitation training

Deshais and team asked a simple question. When you teach a child with autism to copy an action with toys, should you show the exact same model every time or change it a little each time?

They ran an early test with a few kids. Each child got two kinds of teaching: fixed models that never changed and repetitive models that varied slightly across trials. The team tracked how fast each child learned to imitate.

Repetitive models won. Kids reached the mastery goal faster when the model changed a bit from trial to trial. Adding a fun end result, like a tall stack of blocks, did not help. The small changes in the model itself seemed to drive the speed-up.

Halbur et al. (2023) ran almost the same study three years later. They found the opposite: fixed models worked better in most kids. The two papers look like they clash, but the 2023 study used preschoolers and tighter trial timing. Those small design shifts may flip which style wins.

Earlier work by Staats et al. (2000) and MacDonald et al. (2009) showed that video models, whether fixed or looped, also teach imitation fast. Their success fits the idea that clear, repeatable models matter; the live-versus-video choice may be less important than how stable the model looks within a lesson.

McAuliffe et al. (2020) reminds us that autistic children often show slower, bumpy learning curves. Deshais’s speed-up with repetitive models may help smooth those bumps by keeping attention high.

If a child stalls during imitation lessons, try switching the model slightly each time instead of keeping it identical. Track the data for that learner; some kids will do better with fixed models, others with repetitive. Let the child’s progress, not the protocol name, pick the winner.