The Burst That Wasn't Only a Burst
You are supervising a functional communication training (FCT) protocol for 4-year-old Noor, whose aggression is maintained by access to a tablet. The plan specifies that a prompted mand ('tablet please') produces 30 seconds of access, and that aggression produces extinction with the mand prompted again. Week 1 aggression averaged 3 per hour; Week 2 it averaged 9 per hour, and independent mands fell from 12 to 4 per session. Integrity checks across 10 observations show staff delivered the tablet after aggression on two occasions and used a 2-minute time-out, which is not in the plan, on three occasions. The lead technician wants to add a response-cost component and 'ride out the burst.'
1. Is this an extinction burst? What evidence would you need before concluding that?
Read the worked answer
It may be, but the data as they stand cannot support that conclusion, and one finding actively argues against it.
What fits a burst: aggression rising sharply in the first two weeks of extinction is the textbook pattern, and bursts are usually temporary and followed by decline.
What does not fit, or fits something else:
- Integrity failures are the wrong kind. Delivering the tablet after aggression on 2 of 10 checks means aggression was intermittently reinforced. Intermittent reinforcement during extinction is precisely the schedule that produces the greatest resistance to extinction and resurgence - so the rise may be a product of the plan being partially implemented, not of extinction itself.
- The time-out is a confound. An unplanned 2-minute time-out is a punishment contingency layered onto an escape-maintained repertoire. It could suppress the mand, inflate aggression, or both.
- The mand drop is the most informative number. Independent mands falling from 12 to 4 is not characteristic of a clean burst in which the alternative response is strengthened. It looks like the alternative response stopped working - check whether access is immediate, whether it is actually 30 seconds, whether only prompted mands are being honored, and whether the response effort has crept up.
Before calling it a burst, require: integrity at 90% or better across several consecutive sessions, then look at the aggression curve and the mand curve together. A burst with a rising mand curve is survivable; a rising aggression curve with a collapsing mand curve is a plan problem.
2. What do you do before making any change to the plan, and what specifically do you refuse to do right now?
Read the worked answer
Refuse: do not add response cost yet, and do not tell staff to 'ride it out.' Adding a punisher to a protocol with 60–70% integrity will bury the signal: if behavior changes, you will not know which component did it, and if it worsens you will not know whether the cause was the punisher, the integrity failure, or the extinction. Response cost on top of an unplanned time-out also moves the plan toward restriction without a review.
Do, in order:
- Retrain to the protocol using BST - explain the contingency, model it, rehearse with staff, then observe live and give feedback. Target the two specific failure modes observed, not 'the plan' in general.
- Raise integrity checks to every session until integrity is at or above 90% on three consecutive sessions, and graph integrity on the same chart as aggression and mands.
- Repair the alternative response. Honor independent mands immediately and consistently, keep access at the full 30 seconds, accept approximations, and confirm nothing in the environment has made manding harder.
- Collect ABC data on the high-rate sessions to see what aggression is contacting now, including whether brief tablet access is still being delivered.
- Bring the BCBA in on whether the time-out and any plan modification move forward.
If integrity is high for two to three weeks and aggression stays elevated or the mand curve does not recover, then the plan is the problem and a modification is warranted - but that decision must be made on clean data.
3. What safety and consent obligations arise from a rise of this size, and how do you handle them?
Read the worked answer
A threefold increase in aggression is a risk event, not just a data point.
- Quantify the risk. Are there injuries, to Noor or to staff? Where and when do the 9-per-hour sessions occur? Is anyone working alone with her?
- Notify and update consent. Tell the parents what the data show, why a rise can occur early in FCT, what you are doing about integrity, and what the alternatives are. Do not wait for them to notice a bruise. If the plan's risk profile has changed, the consent conversation needs to be refreshed and documented.
- Protect staff and client. Review safe response and blocking procedures, ensure no unauthorized holds or time-outs occur again, arrange a second adult during high-risk blocks, and set a threshold at which the session is paused and the BCBA is called.
- Check for medical involvement. Any injury or change in Noor's physical status gets referred, not managed behaviorally.
- Keep the plan least restrictive. The fact that a procedure is failing is not a justification for escalating to punishment; it is a justification for implementing the current plan correctly first.
Document the risk review, the parent communication, the retraining, and the decision about what happens if integrity does not improve.