An integrated model for guiding the selection of treatment components for problem behavior maintained by automatic reinforcement

The team worked with five people who hurt themselves for no clear social payoff.

First they ran two super-short tests.

Test one: give free toys while alone. Test two: check if the person wants people or things more.

Those two answers picked the next treatment: either attention given for free or extra play time.

They tracked the problem behavior for almost a year to see if it stayed low.

Every person’s self-injury dropped after the model picked the treatment.

Gains held up 10–12 months later without extra tweaks.

The quick pair of probes saved hours of full functional analysis sessions.

Webb et al. (1999) showed brief FAs can match long ones; Berg’s model goes further by linking those same short probes straight to the right intervention.

Saini et al. (2020) later pooled many studies and agreed that short, synthesized formats are the fastest way to find function—Berg’s two-test flow fits that advice.

Bottini et al. (2020) warn that treatment errors slow learning; Berg’s built-in probe keeps the plan simple and easy to run correctly, heading off those errors.

You can copy the two probes in under 15 minutes.

Pick NCR-social if attention wins the preference test, or run enriched play if alone-with-toys beats alone-without.

No extra hardware, no long FA grid—just data-driven treatment choice that lasts.