Within-subject analysis of a prevention strategy for problem behavior

Fahmie and her team ran quick sensitivity tests on one child. They wanted to find the exact situations that made problem behavior worse.

Each test lasted only a few minutes. They changed one thing at a time, like blocking a toy or adding loud noise. They watched which changes made the child hit or scream more.

After they found the hot spots, they used differential reinforcement right away. The child got fun stuff for any other response before problem behavior could blow up.

When the prevention plan was on, problem behavior stayed mild. During the no-plan phases, the same triggers made the behavior spike fast.

The data lines looked like a saw: low during DR, high during baseline. The team reversed the plan three times. The pattern held every time.

Allison et al. (2012) also used differential reinforcement to cut problem behavior, but they worked on food refusal in kids with autism. Both studies got good results, showing DR works across very different topographies.

Bensemann et al. (2015) warn that DRO can accidentally make other odd behavior pop up. Fahmie’s team avoided this by reinforcing ANY appropriate response, not just the absence of problem behavior.

Kettering et al. (2018) later showed how to unpair conditioned motivating operations. Their work extends Fahmie’s idea: once you spot the MO, you can weaken it instead of just riding it out.

You can copy the whole package in one afternoon. Run 5-minute probes with common triggers, graph the results, and pick the highest point. Then start DR for ANY appropriate behavior before the client has a chance to escalate. It’s an easy front-line tool for new intakes or when you feel a storm coming but don’t have full FA data yet.