Analysis of response class hierarchies with attention-maintained problem behaviors.

The team worked with two children who hit, screamed, and flopped for adult attention. They first checked that attention was the real pay-off. Then they stopped giving attention only for the mildest behavior. They timed how long each action lasted during this extinction phase.

The goal was to see if shutting down the small stuff also made the big stuff drop. If it did, the behaviors were in the same response class.

When mild behavior no longer got attention, it stopped fast. The severe behaviors fell at the same time. Latency data lined up like steps: mild went first, then the next, then the worst. This order showed a clear response-class hierarchy.

Lalli et al. (1995) did the same thing earlier with escape-maintained behavior. They also saw a tidy latency ladder. Lejuez et al. (2001) now proves the ladder idea works for attention-maintained topographies too.

Schmidt et al. (2020) take it further. They show you can spot precursors without evoking the worst behavior. Combine a quick correlation scan with a short test and you stay safe while still mapping the class.

Sullivan et al. (2020) add the relapse angle. After you stop extinction, the whole hierarchy can pop back up together. This supports the idea that the behaviors are truly linked, not just coincidentally dropping.

If your functional analysis is muddy, test mild topographies first. Put only the safest one on extinction and watch the clock. When severe behavior falls in step, you have proof of shared function. You can then treat the whole class through the easiest member, saving time and risk.