An evaluation of a punisher assessment for decreasing automatically reinforced problem behavior

Four children with autism kept hitting or biting themselves. The behavior happened for no outside payoff—no attention, no toys, no escape.

The team first tried non-contingent reinforcement plus DRA. They gave the kids fun items every 30 seconds and praised any hand-down behavior. When that failed, they ran a five-minute punisher test.

Each session they delivered one possible punisher—brief restraint, a loud ‘no,’ or a spray bottle—right after every self-hit. They watched which punisher cut the hits the most.

NCR plus DRA alone did nothing. Self-injury stayed high for every child.

Adding the best punisher dropped the behavior by 80-a large share within one or two sessions. One kid needed restraint, another just needed a quick ‘no.’

Fahmie et al. (2013) showed that multielement tests can pick the right control condition. Verriden copied that same rapid test idea, but swapped in punishers instead of ignore or DRO conditions.

Robertson et al. (2013) also blended reinforcement with a second procedure—microswitch clusters—to cut self-injury. Both studies prove that reinforcement alone often fails for automatic behavior; you must stack on a consequence.

Pascale et al. (2025) used DRO plus self-monitoring in prison and saw big drops in aggression. Their DRO is cousin to Verriden’s DRA, yet Verriden had to add a punisher while Pascale did not. The difference: prisoners could wait for rewards; kids with autism and automatic reinforcement could not.

If your NCR or DRA plan is flat-lining, run a quick multielement punisher screen. Five minutes per consequence can show you the one that actually works. Start with mild, ethical options first—verbal ‘no,’ brief hands-down—then fold the winner into the behavior plan. You save weeks of stalled treatment and avoid heavier restraints later.