RBT Assent Withdrawal Protocol: The 3-Strike Rule

If you are an RBT (Registered Behavior Technician, the person running the session at the table), the rule is simple: after three assent withdrawals on the same task, stop and call the BCBA (Board Certified Behavior Analyst, the clinician who wrote the plan). Do not run the same trial a fourth time. Do not wait it out. Do not push through. Three strikes, then phone. This is the RBT-facing protocol that keeps you from becoming the part of the session your client wants to escape from.

The 3-withdrawal rule (and where it came from)#

A client pulls away. Maybe they push the cards off the table. Maybe they get up. Maybe they say no. That is an assent withdrawal. The client is telling you the current task is something they want to move away from. You back off, you wait for calm, and you try again. That is trial one.

The rule is this: you get three of those. If the client withdraws assent three times on the same task, you stop. You log it, you note what you saw, and you message your BCBA before the next trial. You do not run trial four with the same setup.

The reason this rule exists is honest and a little embarrassing. Matt set the standard for himself first, then for his team, after watching what happened when the rule was vague. When he ran session, he changed things after two. When his team ran the same kid, they ran it six, eight, ten times before flagging it. Same client, two different sessions, two completely different experiences. The fix was a number everyone could count to.

"I ended up setting a standard both for myself of the RVTs of three ascent withdrawals, and then call me three ascent withdrawals and then call the BCBA." From the talk — Matt Harrington

Why no protocol = the RBT becomes aversive#

Here is what happens when there is no number. The BCBA pops in for an hour, reads the room, changes the task, and the kid has a good time. The RBT runs the same kid for fifteen hours a week and follows the old plan because nobody told them when to stop. The kid figures out fast that one of those two people changes things and one does not. Guess which one they start avoiding.

This is pairing in reverse. Pairing is the work you do to become a person the client wants to be around. When you run the same dead-end trial ten times in a row, you pair yourself with the bad feeling of that task. The client is not mad at you. They are mad at the loop you are running.

"When I was working with my RVTs, I told them at the beginning to do it through a session. If you get ascent withdrawal, wait until we're calm, try again. And so the reality was they were waiting for six to 10 times." From the talk — Matt Harrington

"I was becoming super preferred. My clients loved me, everything was going great. And these RVTs were had a ton of aversion associated with them." From the talk — Matt Harrington

The three-strike rule fixes this. It puts you and the BCBA on the same page about when to change the plan, so the client does not learn that you are the one who never updates.

When to break the rule (severe behavior risk)#

Three is the default. It is not the law. The number gets smaller when the risk gets bigger.

If the assent withdrawal could turn into headbanging, self-biting, hitting, or anything that hurts the client or someone else, you do not wait for strike three. You stop after one. You call. You wait. The point of the rule is to keep the session safe and consistent. Pushing into a fourth trial when the second one had teeth in it is not consistency. It is gambling.

"A loose guideline, very loose, is that the more severe the behavior was, the faster I would modify my intervention." From the talk — Matt Harrington

A useful way to think about it: if the worst case at trial four is the client saying no louder, finish out your three. If the worst case at trial four is a trip to the nurse, the rule is already done.

What to say when you call the BCBA#

Calling the BCBA is not a confession. It is a data point. You do not need to apologize for it. You need to give them four things, fast:

1. What you were running. The exact task. Name of the program, target, materials. Not "tabletop." Say "two-step receptive instructions, sitting at the kitchen table, with the picture cards."
2. What withdrawal looked like, each time. Did they push the cards away? Get up? Cry? Say no? Did it escalate across the three trials, or stay the same?
3. What you tried between trials. Did you wait for calm? Move locations? Offer a break? Change reinforcement?
4. What you think might be driving it. You do not have to be right. You see this kid more than anyone. Your guess matters.

Keep it short. A two-sentence Slack message is enough. The BCBA does not need a treatment plan from you. They need enough information to decide what to change.

What the BCBA should change next#

You are not making this call. But you should know what they are weighing, so you can answer the questions they ask.

The session ran the same trial three times and the client said no three times. Something has to change before trial four, or the next three trials will go the same way. The BCBA is looking at two levers: shrink what is bad about the task, or grow what is good about it.

Shrinking the bad means making the task itself easier to be in. Shorter trial. Smaller step. Different chair. Quieter room. Pencil instead of marker. Tablet instead of flashcards. Anything that takes a thing the client is trying to escape from and makes it smaller.

Growing the good means making the task more worth doing. Better reinforcer. Faster reinforcement. A reinforcer the client actually picked today, not the one on the plan from three months ago. For older clients, this can mean tying the task to something they care about (friends, a job, getting out of the house) instead of just tokens.

Most of the time the answer is both. A little less bad, a little more good. The BCBA may also add a skill (toleration, communication, asking for a break) so the task itself gets easier to handle next week. Your job is to run whatever comes back with the same care you gave the original plan. If the new version hits three withdrawals, you call again. Same rule.

Writing the protocol into the treatment plan#

A rule that lives in a Slack message gets lost. The 3-strike rule needs to be in the written plan, in the section the RBT actually reads, so a new tech on day one runs the same protocol as the tech who has been there a year.

Three things to put in writing:

• The default. "After 3 consecutive assent withdrawals on the same target within a session, RBT will pause the program and notify BCBA before re-presenting."
• The exception. List the behaviors that drop the count. "If withdrawal includes [self-injury, aggression, elopement, or named precursor], RBT will pause after 1 withdrawal and notify BCBA."
• The reset. State whether the count resets between sessions or carries over. (Usually it resets. Say so anyway.)

This is the part most plans skip. Write it once and every tech on the case runs the same play.

Frequently asked questions#

Is 3 withdrawals an ethics requirement or just a guideline? It is a guideline, not a rule from the BACB (Behavior Analyst Certification Board, the group that writes the ethics code). The code requires you to respond to assent withdrawal and not just push through. It does not name a number. The number is a team agreement that makes the ethics requirement easy to follow in the moment. Your BCBA can set it at two, three, or one for high-risk cases. The point is that there is a number, and everyone on the case uses the same one.

What if the BCBA isn't reachable during the session? You still stop. You do not run a fourth trial without a change. Move to a different program the client has already mastered, take a break, or do a preferred activity until the BCBA gets back to you. Log what happened, log when you tried to reach them, log what you did instead. A short pause is always safer than a fourth trial of something that already failed three times.

Does the count reset between sessions? Usually yes, but only if the plan says so in writing. Default to reset. A kid who pulled away from a task on Tuesday morning might be fine with the same task on Thursday afternoon. The exception is when the BCBA has flagged the task as a known hot spot. In that case, the plan should say "count carries to next session" and the BCBA should already be working on a change before you run it again. If you are not sure, ask. Do not guess.

Want the full version?#

This protocol comes from the back half of the talk. The full session walks through what the BCBA does with the information you give them: the shrink-the-bad / grow-the-good framework, the school bus example, the personal hygiene teen, and how skill acquisition shrinks the aversive part of a task over time. Watch the full CEU above.