RBT Behavior Reduction Guide

Start with function, always

Interventions that ignore function are guesses. A behavior maintained by escape will not budge if you shower it with attention-based consequences, and an attention-maintained behavior laughs at your escape-proofing. Every question in this domain secretly asks: do you know the function first? See our functions of behavior guide for the full breakdown. The short version: attention, escape, tangible access, sensory. Find it with ABC data, then pick the intervention that fits.

Extinction: stop feeding the behavior

Extinction means withholding the reinforcement that has been maintaining the behavior. If tantrums have been earning escape from demands, extinction means the demand stays. If whining has been earning the iPad, extinction means the iPad stays away. The behavior was working, now it does not, so eventually it fades.

Two things the exam will absolutely ask about. First, the extinction burst: behavior often gets worse before it gets better, more intense, more frequent, sometimes with new topographies. This is normal and temporary, not a sign the plan is failing. Second, spontaneous recovery: the behavior can briefly pop back up after it seemed gone. Stay the course. The most dangerous mistake is giving in during the burst, because that teaches the learner that bigger behavior works.

Differential reinforcement: pay for better

Instead of only stopping the bad, you reinforce something better. The DR family:

• DRO (of other behavior): reinforce the absence of the problem behavior for a set time. "If you go 5 minutes without hitting, you earn the token." Simple, but it does not teach a replacement skill.
• DRI (of incompatible behavior): reinforce a behavior that cannot happen at the same time as the problem. Hands in lap is incompatible with hitting. Elegant when you can find a true incompatible.
• DRA (of alternative behavior): reinforce a specific alternative, usually a communication replacement like asking for a break instead of screaming. The exam favorite, because it teaches.
• DRL (of low rates): reinforce when the behavior happens less often. Used for behaviors you want to reduce but not eliminate, like asking questions in class.

Antecedent interventions: prevention beats reaction

The cheapest intervention is the one that prevents the behavior from starting. Antecedent strategies change what happens before the behavior: offering choices, breaking tasks into smaller steps, enriching the environment, adjusting demands to the learner's level, or changing a motivating operation, like giving a snack before a session so hunger is not driving behavior. Noncontingent reinforcement (scheduled freebies of the functional reinforcer) is another antecedent move the exam mentions.

When behavior escalates in session

Plans work until the day they do not, and your response in that moment matters more than any procedure name. First, protect people: follow the crisis and safety procedures in the plan, create space, remove others if the plan says so. Second, stay neutral: calm voice, minimal attention to the problem behavior, no lectures mid-crisis. Third, follow the plan exactly, even if your instincts suggest improvising; the plan was written with the full clinical picture and your improvisation was not. Fourth, document and report: objective notes on what happened, what you did, and the outcome, plus a prompt message to your supervisor. Escalation data often triggers a plan review, which is the system working as designed.

Quick-fire review

• Function first, intervention second, always
• Extinction burst = worse before better; never give in during it
• DRA teaches a replacement; DRO just pays for quiet
• Antecedent changes prevent; consequences react

Last reviewed: October 7, 2026 against the BACB RBT 3rd edition task list.