Parametric reinforcement effects in a programmed activities environment for the severely retarded.

The team watched adults with severe intellectual disability during room activities.

A room manager gave tokens for staying on task. They tested four schedules: a token every 65 s, every 100 s, every 200 s, or none at all.

Each adult cycled through the schedules while staff tracked on-task minutes and disruptive acts.

Tokens every 60-100 s kept engagement near 80 percent.

Stretching to 200 s or stopping tokens cut work time and sparked more yelling, hitting, and wandering.

Art and workshop tasks held up slightly better than music when reinforcement thinned.

Schwarz et al. (1970) saw the same drop-off when they stopped reinforcing simple ball games in withdrawn children. Both studies warn: if you fade reinforcement too fast, behavior crashes.

Jessel et al. (2017) later echoed the finding with momentary checks instead of fixed intervals. Their teen with autism also stayed on-task only when reinforcement stayed tight.

Cividini-Motta et al. (2013) tested size instead of timing. They showed that giving the big reinforcer only for independent responses beats giving it every time. Together the three papers map two levers you can turn: how often and how big.

You now have a clear rule for adults with severe ID: deliver a reinforcer at least every 60-100 s during group activities. If you must thin, do it gradually and watch for disruption. Pair this with Catia’s size trick to avoid prompt dependency and you get steady engagement without escalation.