AD ABAdegree.org
ABA Concepts
Last updated: November 14, 2026

Extinction in ABA: The Extinction Burst and What Research Shows

Extinction in ABA means a behavior stops producing its reinforcer. How often extinction bursts happen, why function sets the procedure, and how to soften them.

Taylor Rupe

Founder & Editor

Published

A toddler at a wooden table cries and reaches for a cup of applesauce held by an adult in a denim shirt

Key Takeaways

  • 01Extinction means a behavior no longer produces the reinforcer that maintained it. What that looks like depends on the function: no attention, no escape from the task, no item, or a blocked sensory result.
  • 02An extinction burst is a temporary increase in the behavior right after reinforcement stops. Lerman and Iwata found one in 24% of 113 published data sets.
  • 03Bursts were rarer when extinction was combined with other procedures: 12% of those cases, against 36% when extinction was used alone.
  • 04Iwata and colleagues showed in 1994 that extinction only reduced self-injury when it withheld the specific reinforcer that maintained it.
  • 05RBTs implement extinction under task D.4 of the RBT outline. BCBAs are tested on it in B.11, G.2 and H.4.

Extinction in ABA means a behavior that used to produce a reinforcer stops producing it, so over time the behavior happens less. The best-known side effect is the extinction burst: a short-lived jump in the behavior right after the reinforcer stops, when it can get more frequent, longer or more intense before it fades. Bursts are common but not universal, and the studies below show when they're more likely and what makes them smaller.

What Is Extinction in ABA?

Operant extinction has two parts. A behavior has a history of producing a reinforcer, and then that reinforcer stops following it. The procedure is withholding the reinforcer. The process is the decline in behavior that follows. The BCBA exam asks about both.

Extinction is not punishment. Nothing aversive is added after the behavior and nothing the person already has is taken away. The behavior simply stops working. That's also why extinction is usually gradual: the person has to contact the new contingency enough times to learn it.

It's also not the same as ignoring. Ignoring is extinction only when attention is the reinforcer. If a child screams to get out of brushing their teeth, ignoring the screaming while letting them leave the bathroom isn't extinction at all. The child still escapes. To find out which reinforcer to withhold, you need to know the function, which is the job of a functional behavior assessment.

The BCBA outline also asks you to tell operant extinction apart from respondent extinction. Respondent extinction is the Pavlovian version: a conditioned stimulus is presented without the unconditioned stimulus until it stops eliciting the response. Operant extinction works on behavior controlled by its consequences.

What Is an Extinction Burst?

An extinction burst is a temporary increase in the target behavior when extinction starts. The behavior that used to pay off doesn't, so the person does more of it, harder, or for longer. A child who whined for attention whines louder. A client whose hitting ended a task hits more often for the first few sessions. Here's how that looks on a graph.

What an extinction burst looks like on a graph A worked example. Baseline averages 8.2 responses per minute. In the first extinction session responding jumps to 15 per minute, then falls below every baseline session by session 10 and reaches zero in session 14. 0 4 8 12 16 Baseline Extinction Session 1 (baseline): 8 per minute Session 2 (baseline): 9 per minute Session 3 (baseline): 7 per minute Session 4 (baseline): 8 per minute Session 5 (baseline): 9 per minute Session 6 (extinction): 15 per minute Session 7 (extinction): 13 per minute Session 8 (extinction): 10 per minute Session 9 (extinction): 8 per minute Session 10 (extinction): 6 per minute Session 11 (extinction): 4 per minute Session 12 (extinction): 3 per minute Session 13 (extinction): 1 per minute Session 14 (extinction): 0 per minute burst: 15/min Session Responses per minute
A worked example, not study data. The dashed line marks the session where the reinforcer stopped.

In this example, responding averages 8.2 per minute across the five baseline sessions. The first extinction session, session 6, jumps to 15, which is the burst. Responding then drops below every baseline session by session 10 and reaches zero in session 14. If someone on the team had given in during session 6, the learner would have been reinforced for the highest rate yet. Giving in now and then also puts the behavior on intermittent reinforcement, which tends to make it more resistant to extinction, not less.

How often does this actually happen? Lerman and Iwata (1995) analyzed 113 sets of extinction data from applied research and found bursting in 24% of them. Bursts were less likely when extinction was combined with other procedures, such as reinforcement for another behavior: 12% of those cases, against 36% when extinction was used alone.

Higher numbers come from a different sample and a stricter count. Lerman, Iwata and Wallace (1999) looked at 41 people treated for self-injurious behavior and found extinction bursts or increases in aggression in nearly half of the cases, again with lower rates when extinction was part of a treatment package. According to the review by Fisher, Greer, Shahan and Norris (2023), 21 of those 41 received extinction alone, and 13 of the 21 (62%) showed a burst, counted as any rise above baseline in the first three treatment sessions. So the 24% figure covers published data sets of many kinds, while the 62% covers people treated for self-injury with extinction alone.

Types of Extinction, Matched to Function

Because extinction withholds a specific reinforcer, there's no single extinction procedure. There's one for each function. Iwata, Pace, Cowdery and Miltenberger (1994) showed why that matters. They studied three children whose head banging looked the same but served different functions: attention for one, escape from tasks for the second and automatic reinforcement for the third. Each child was exposed to at least two forms of extinction. Head banging went down only when the procedure withheld the reinforcer that had been shown to maintain it.

The table below lays out what extinction means for each of the four functions of behavior.

Function What maintains it What extinction means here Example
Attention Someone looks, talks, scolds or comforts The behavior no longer gets attention. Attention comes for other behavior instead. A child who whines for a parent's attention gets no reaction to whining, and a quick, warm reply to a calm request.
Escape The task, demand or situation goes away The behavior no longer ends the task. The demand stays, with help if needed. Also called escape extinction. A student who sweeps materials off the desk still finishes the worksheet, with prompting, instead of being sent out of the room.
Tangible The person gets an item or activity The behavior no longer produces the item. Screaming at the checkout doesn't get the candy bar. Asking politely might, on days it's on offer.
Automatic The behavior produces its own sensory result The sensory result is blocked or masked. Also called sensory extinction. Padding a table edge so that banging a hand on it no longer makes the sound or feeling that kept it going.

Escape extinction means continuing a demand while the learner protests, so it's often paired with prompting: the learner finishes the task with help and the protest doesn't end it. Sensory extinction requires first working out what sensory result the behavior produces, then blocking it without making the setting unsafe.

Other Side Effects of Extinction

The burst isn't the only thing to plan for. Extinction can bring out new problem behavior, and an extinguished behavior can come back.

Effect What you see What the research found
Extinction burst The behavior gets more frequent, longer or more intense right after reinforcement stops 24% of 113 published data sets (Lerman & Iwata, 1995)
Extinction-induced aggression New problem behavior, such as hitting, appears alongside the old one Bursting or more aggression, counted together, in nearly half of 41 self-injury cases (Lerman, Iwata & Wallace, 1999)
Spontaneous recovery The behavior comes back for a while after it seemed to be gone Examined during treatment of disruption, along with its attenuation when large amounts of reinforcement were delivered (Lerman et al., 1999)
Resurgence The old behavior returns when the replacement behavior stops paying off Returned for 4 of 5 children in FCT treatment (Volkert et al., 2009)

Extinction-induced aggression can make a procedure unsafe for the learner or staff, which makes it a reason to revisit the plan. Spontaneous recovery can happen even when a plan is working. Lerman, Kelley, Van Camp and Roane (1999) examined it during treatment of disruption, along with its attenuation when large amounts of reinforcement were delivered. Resurgence matters most later in treatment. In Volkert, Lerman, Call and Trosclair-Lasserre (2009), problem behavior came back for 4 of 5 participants when their new communication response was put on extinction or a thin schedule.

How to Reduce the Extinction Burst

Across these studies, bursts were less common when extinction was part of a package than when it was used alone (12% against 36% in Lerman and Iwata's data sets). That's why it's usually paired with differential reinforcement, so the learner has another way to get the same outcome. When that other way is a request, the package is functional communication training.

Fisher and colleagues (2023) go further. Their account, built on a refinement of the matching law, predicts that bursts follow a drop in reinforcement when treatment starts, and that the bigger the drop, the bigger the burst. In their simulations, extinction alone produced the largest burst, a lean schedule for the replacement behavior a smaller one, and a dense schedule that kept reinforcement at baseline levels produced no burst at all. The authors note these predictions still need to be tested. The first step below comes from their recommendation; the rest are standard practice that fits the same logic:

  1. Match the reinforcement the problem behavior was getting. Fisher and colleagues advise keeping the rate, size, immediacy and quality of reinforcement for the replacement behavior at least as high as baseline when treatment starts.
  2. Make the replacement easy. If asking for a break takes more effort than hitting and pays off less, expect hitting to win.
  3. Thin later, and slowly. Thinning is when resurgence shows up, so it belongs after the problem behavior is low and stable.
  4. Get everyone on the same plan. One adult who gives in during a burst reinforces the escalation. Parents, teachers and every RBT on the case need the same instructions.
  5. Plan for safety first. If the behavior can injure someone, the plan needs blocking or protective procedures, and the supervising BCBA decides whether extinction is appropriate at all.

All of this belongs in the written behavior intervention plan, including what staff should expect in the first sessions so a burst isn't mistaken for failure.

Can You Predict How Long Behavior Will Persist?

Not reliably, at least not yet. A study published in the Journal of Applied Behavior Analysis in November 2026 tested whether a measure of how hard someone will work for a reinforcer could predict how long their responding would persist during extinction. Lambert, Hayward, Price and Logan worked with 90 adults with disabilities and found statistically significant correlations, but even their fullest model never explained more than 30% of the variation. They concluded the measure isn't ready for clinical use. Details are in our news story on the study.

The same paper notes that the efficacy of reinforcement-based treatments often depends on contact with extinction, which is why avoiding extinction on principle has costs too. Weighing those risks for a given learner is the supervising BCBA's call.

Extinction on the RBT and BCBA Exams

On the RBT exam, the RBT Test Content Outline (3rd ed.) lists task D.4, "Implement extinction procedures," in the Behavior Reduction domain, right after D.3 on differential reinforcement. RBTs carry out extinction as written in a plan; designing it is BCBA work.

The BCBA Test Content Outline (6th ed.) covers it three times:

  • B.11: Identify and distinguish between operant and respondent extinction as operations and processes.
  • G.2: Design and evaluate differential reinforcement procedures with and without extinction.
  • H.4: Plan for and attempt to mitigate possible unwanted effects when using reinforcement, extinction, and punishment procedures.

The full outline is in our guide to the BCBA task list 6th edition.

Extinction Practice Questions

These four are original questions written for this page. They aren't taken from any BACB exam. For a full-length set, try our RBT practice exam.

1. A functional analysis shows a client's head hitting is maintained by escape from tasks. Which procedure is extinction for this behavior?

  1. A.Ignoring the head hitting
  2. B.Continuing the task when head hitting occurs
  3. C.Giving a short break after each instance
  4. D.Placing a helmet on the client
Show the answer

B. Continuing the task when head hitting occurs. Extinction means the behavior stops producing its maintaining reinforcer. For escape-maintained behavior that reinforcer is the end of the task, so the task continues. Ignoring would be attention extinction, and Iwata and colleagues found extinction reduced self-injury only when it withheld the reinforcer that maintained it.

2. In the first session after an RBT stops giving attention for yelling, the client yells twice as often as in baseline. What is this most likely to be?

  1. A.Spontaneous recovery
  2. B.Resurgence
  3. C.An extinction burst
  4. D.Stimulus generalization
Show the answer

C. An extinction burst. A temporary increase right after reinforcement is withheld is an extinction burst. Spontaneous recovery is a return after the behavior had already dropped, and resurgence is the return of an old behavior when a newer one goes unreinforced.

3. Six weeks into a successful extinction plan, a behavior that had dropped to zero shows up again at the start of a session after a weekend break. What is this called?

  1. A.Extinction burst
  2. B.Spontaneous recovery
  3. C.Response generalization
  4. D.Shaping
Show the answer

B. Spontaneous recovery. Spontaneous recovery is the temporary reappearance of a behavior after it looked fully extinguished. Responding to it the same way, with no reinforcement, is what keeps it temporary.

4. A BCBA wants to lower the chance of an extinction burst when treatment starts. Based on Fisher and colleagues' 2023 recommendations, what should the plan do?

  1. A.Start extinction alone, then add reinforcement once the burst passes
  2. B.Thin reinforcement for the replacement behavior starting in session one
  3. C.Keep reinforcement for a replacement at least as rich as baseline at the start
  4. D.Deliver all reinforcement in one block at the end of each session
Show the answer

C. Keep reinforcement for a replacement at least as rich as baseline at the start. Fisher and colleagues predict bursts follow a drop in reinforcement when treatment begins, so they advise keeping reinforcement at or above baseline levels, delivered for the alternative behavior, when treatment starts.

Frequently Asked Questions

What is an extinction burst in ABA? +
How long does an extinction burst last? +
Is extinction the same as ignoring? +
What is the difference between extinction and punishment? +
What is spontaneous recovery in ABA? +

Sources

Last checked November 14, 2026.

  1. Lerman and Iwata, Prevalence of the extinction burst and its attenuation during treatment (Journal of Applied Behavior Analysis 28(1):93-94, 1995)
  2. Lerman, Iwata and Wallace, Side effects of extinction: prevalence of bursting and aggression during the treatment of self-injurious behavior (Journal of Applied Behavior Analysis 32(1):1-8, 1999)
  3. Iwata, Pace, Cowdery and Miltenberger, What makes extinction work: an analysis of procedural form and function (Journal of Applied Behavior Analysis 27(1):131-144, 1994)
  4. Fisher, Greer, Shahan and Norris, Basic and applied research on extinction bursts (Journal of Applied Behavior Analysis 56(1):4-28, 2023)
  5. Lerman, Kelley, Van Camp and Roane, Effects of reinforcement magnitude on spontaneous recovery (Journal of Applied Behavior Analysis 32(2):197-200, 1999)
  6. Volkert, Lerman, Call and Trosclair-Lasserre, An evaluation of resurgence during treatment with functional communication training (Journal of Applied Behavior Analysis 42(1):145-160, 2009)
  7. Lambert, Hayward, Price and Logan, Relative reinforcer efficacy and persistence in extinction are significantly correlated but not clinically actionable (Journal of Applied Behavior Analysis 60(1):e70097, 2027; published online November 2026)
  8. BACB, RBT Test Content Outline (3rd ed.), tasks D.3 and D.4
  9. BACB, BCBA Test Content Outline (6th ed.), tasks B.11, G.2 and H.4

Keep going

Explore ABA programs

Ready to take the next step? Browse our expert-ranked program listings.

Sponsored

ABA Degree Programs

Schools pay to appear in these listings. They are separate from our rankings, which no school can pay to change.