Four Ways to Prevent Resurgence, Tested Head to Head. The Popular One Backfired.

Four behavioral-momentum-informed tactics for reducing resurgence have been studied one at a time over five years. A new Journal of Applied Behavior Analysis paper puts them side by side, pooling 24 data sets from 24 children (18 boys, 6 girls) across three university-affiliated severe-behavior clinics.

They do not perform alike, and the one with the most intuitive appeal is the one that went backwards.

The results

• Lean reinforcement for the target response in baseline: reduced resurgence for 6 of 6 participants, median 67.14%.
• Lean reinforcement for the alternative response during treatment: helped 1 of 3 (59.30%) and increased resurgence for the other two, by −96.95% and −154.57%.
• Longer treatment duration: 4 of 6.
• All three combined: 6 of 6, median 63.88%.

The authors' summary of the combination is unusually direct: “it seems premature to suggest that clinicians ought to be packaging lengthy treatment duration and lean alternative reinforcement with lean target reinforcement given the limited efficacy of those first two variables in isolation.” The combined package works, but the two weak ingredients are along for the ride.

The price of the tactic that won

Lean target reinforcement means reinforcing destructive behavior less often in baseline. That has a cost, and the authors put a number on it: the lean schedule produced destructive behavior “ranging between 2.40 and 11.26 responses per minute and a grand mean of 6.30 responses per minute.”

Their own framing of the trade: “a clinician must consider whether a leaner schedule of reinforcement that engenders a fair amount of destructive behavior in baseline is more desirable than an increased possibility of resurgence later on in treatment.” And their decision rule: consider it “for challenging behavior that does not pose imminent harm, such as with disruptive behavior (e.g., ripping papers, vocal protests), or when providing consequences for precursor behavior.”

What the authors say it cannot tell you

Three limitations, in their terms. “First, the sample size was small and groups did not have equal numbers of participants. This limitation is particularly germane to the alt-Sr group that consisted of only three participants.” Second, the source experiments were translational, using VI schedules to control confounds, so “the outcomes of our analysis may not correspond to typical clinical practice” — real FCT usually involves schedule thinning or reinforcer-quality manipulations that were not in play here. Third, other analyses of the same data might find different relations.

Two more things to hold onto. Resurgence still happened: at least one session showed it in 83.33% of the lean-target cases and 83.33% of the combined cases. And when the same comparison was run on raw response rates rather than as a proportion of baseline, the effect shrank.

What you must know or do

• Stop treating “thin the FCR schedule” as resurgence insurance. On this evidence it is not, and for two of three children it ran the other way. If your team does it for that reason, that is the belief to retire.
• Pick the topography before you pick the schedule. The authors' own rule is that a lean baseline schedule is defensible for behavior that does not risk imminent harm, or for precursors. Apply it to paper-ripping and protests, not to head-banging.
• Write the extinction challenge into the plan. Every group here showed clinically significant resurgence in at least one session. Plan for it happening, rather than for preventing it.
• If you cite this to justify a change, cite the proportion caveat too. The effect is largest as a proportion of baseline and smaller on raw rates, and a reviewer reading the paper will find that.