Secondary Effects of Extinction and Punishment

Summary: Extinction and punishment both reduce behavior, and both come with side effects you’ll see in the room before anyone else does. Extinction can produce a burst, induced aggression and emotional responding, and spontaneous recovery. Punishment can produce emotional and aggressive reactions, escape and avoidance of you or the setting, and the learner copying the punitive behavior. None of these mean the plan failed. Your job is to recognize them, stay consistent, keep the data, and tell your supervisor.

Every procedure that reduces behavior does more than just reduce that one behavior. It ripples. The learner reacts to the change, and some of those reactions are predictable enough that the field has names for them. These are the secondary effects, and the third-edition task list pulls them out as something you specifically need to recognize.

Here’s why that matters for you and not just for your BCBA. You’re the person in the room when the side effect shows up. Your BCBA designed the plan from a desk, days ago. You’re the one watching a kid who’s suddenly screaming louder, or flinching when you walk over, or hitting a peer for the first time. If you don’t know these effects are normal and expected, you’ll read them as the plan backfiring, and you’ll either cave or improvise. Both make things worse. If you do know them, you stay steady, you write it down, and you flag it. That’s the whole skill here.

One thing up front. You won’t be designing or independently running punishment procedures, and good programs lean on reinforcement and extinction first. But you have to understand punishment’s side effects, partly for the exam and partly because the field treats them as a major reason to avoid punishment when something gentler will do the job.

The side effects of extinction

We cover extinction’s mechanics in depth on the extinction page. Here we’re focused on the fallout, the things that happen to the learner when you stop delivering the reinforcer.

Extinction burst

The burst is the big one. When you stop reinforcing a behavior that used to work, the behavior usually gets worse before it gets better. More often, harder, longer. You’ll frequently see new variations of the behavior too, the learner trying different versions to see if one still pays.

Think about it from the learner’s side. The behavior worked dozens of times. Now it stopped working. The natural human response is to do it harder, the way you jab an elevator button that didn’t come when you pressed it once. That escalation is the burst.

The trap is that the burst looks exactly like failure. The behavior is getting worse right after you started doing the right thing, so every instinct says stop. But the burst is actually a sign extinction is taking hold. The danger is caving during it, because if you give in when the behavior is at its most intense, you’ve just taught the learner that escalating is what works. You’ve reinforced the worst version of the behavior.

Extinction-induced aggression and emotional responding

Cutting off a reinforcer is frustrating, and frustration comes out. You’ll see emotional responses, crying, whining, looking visibly upset, and sometimes aggression that wasn’t there before. A kid who used to just scream for the tablet might start grabbing or hitting when the screaming stops working. That’s extinction-induced aggression, and it’s tied to the same frustration that drives the burst.

This one has a safety dimension the others don’t. New aggression can hurt someone. The response is the same as for any safety concern: there should be a plan in place before extinction starts, hazards cleared, trained staff, and heavy reinforcement available for appropriate behavior so the learner has a path that isn’t aggression. If it gets genuinely dangerous, you stop and bring it to your BCBA, because the plan may need to change.

Spontaneous recovery

This one shows up later, after the behavior already dropped to near zero and seemed gone. Spontaneous recovery is the behavior reappearing after a stretch of not happening. It’s usually brief and weaker than the original, and it fades again on its own as long as you keep extinction in place.

The reason it matters is the same as the burst: it looks like the plan failing, this time after you thought you’d won. Don’t read it that way. Hold the line and it goes away again.

On the exam: Don’t mix up the burst and spontaneous recovery. The burst happens at the start of extinction, right when reinforcement first stops. Spontaneous recovery happens later, after the behavior had already faded and seemed extinguished. Same “the behavior came back” surface, different timing. That timing is usually how the question tells you which one it wants.

The side effects of punishment

Punishment means delivering a consequence after a behavior that makes that behavior less likely in the future. It can work, sometimes fast, and that speed is exactly why people reach for it. But it carries a set of side effects serious enough that the field treats them as a strong argument for trying everything else first. You need to be able to name them.

Emotional and aggressive responses

Punishment tends to produce emotional fallout, fear, crying, anger, general distress. And like extinction, it can trigger aggression. A learner who’s punished may lash out, sometimes at the person delivering the punishment, sometimes at whoever’s nearby. The aversive event provokes it. This is a known, documented effect, not a sign you did something unusual.

Escape and avoidance of the punisher or the setting

This is the one that quietly corrodes everything you’re trying to build. When a person, place, or activity becomes associated with punishment, the learner starts working to escape and avoid it.

That doesn’t just mean avoiding the punished behavior. It means avoiding you, or the therapy room, or the table where the hard work happens. And here’s the painful part: as an RBT, your effectiveness rests entirely on the learner wanting to be around you and engage with you. The moment you become the source of something aversive, you’ve damaged the relationship that makes teaching possible. The learner may comply less, hide behavior instead of stopping it, or shut down when they see you coming.

Modeling of punitive behavior

Learners copy what they see. If a child watches adults use hitting, yelling, or other punitive responses to control behavior, the child learns that this is how you make people do what you want. So they do it, to peers, to siblings, to younger kids. You’ve reduced one behavior and accidentally taught a new and worse one. This is one of the strongest reasons to be careful about what behavior you model in front of the learners you work with, every minute, not just during programmed procedures.

Common mistake: Treating punishment side effects as the learner being “difficult” or “manipulative.” A kid who suddenly avoids the work table or flinches when corrected isn’t being defiant. Those are textbook side effects of an aversive contingency. Reading them as the kid’s character problem instead of a predictable response to the procedure leads you to the wrong fix, and it’s unfair to the learner.

A scenario that ties it together

You’re working with Marcus, an 8-year-old, on an extinction plan for screaming. The FBA says the screaming is tangible-maintained: historically, screaming got him his tablet back. The plan is extinction plus teaching him to hand you a picture card to request the tablet.

Day one, you put the tablet away and the screaming kicks off as usual. You hold the line, no tablet. Within a few minutes the screaming gets louder and longer than you’ve ever heard it, and then he throws his shoe across the room, which he’s never done before. Your gut says the plan is making him worse and you should give the tablet back to calm him down.

Walk through what’s actually happening. The louder, longer screaming is the extinction burst, expected, and a sign the procedure is working. The thrown shoe is extinction-induced aggression, the frustration from the burst spilling into a new behavior. If you hand over the tablet right now, you don’t fix anything. You teach Marcus that screaming this loud, plus throwing things, is what gets the tablet. You’d be reinforcing the worst version of the behavior at the worst possible moment.

So you don’t cave. You keep everyone safe, you make sure the picture card is right there and you prompt him toward it, and the second he uses it appropriately you hand over the tablet immediately. You log the burst, you log the throwing as a new topography, and you tell your BCBA at the next check-in that you saw induced aggression, because that’s a safety detail that might change the plan.

Now flip the example. Imagine the plan had instead used a punisher for the screaming, and a week in you notice Marcus has started refusing to come to the work area at all and gets teary when he sees you set up the materials. That’s escape and avoidance of the setting and the punisher, and it’s exactly the side effect that makes punishment a last resort. Same kid, very different fallout, and a clear reason the field reaches for extinction and reinforcement first.

What to do when you see these effects

Across both extinction and punishment, your response when a side effect appears comes down to the same three moves. Memorize these, because they’re what the exam wants and what the job actually requires.

Stay consistent. Don’t change the procedure on your own in the middle of a side effect. The burst, the induced aggression, the spontaneous recovery, these are the exact moments inconsistency does the most damage. If you cave during a burst, you reinforce escalation. If you quietly stop running the plan when spontaneous recovery shows up, you undo the progress. Run the plan as written, every time, the same way every other person on the team runs it.

Keep the data. The side effect itself is information. The shape of the burst, how long the aggression lasts, when spontaneous recovery shows up, all of it tells the team whether the procedure is working and whether it’s safe to keep going. “It felt awful today” isn’t data. Frequency, duration, and a note on any new behavior you saw, that’s data, and it’s what your BCBA uses to decide what happens next.

Tell your supervisor. You implement; your BCBA decides. New aggression, a side effect that’s becoming unsafe, a learner starting to avoid you or the setting, these are all things you report rather than solve on your own. You’re not authorized to modify the plan, and you shouldn’t want to be. Your job is to flag what you’re seeing clearly and early, so the person who designed the plan can adjust it. If something is dangerous in the moment, you follow the safety protocol and then you report.

On the exam: When a question describes a side effect appearing and asks what the RBT should do, the answer is almost never “change the procedure” or “stop the procedure.” It’s stay consistent, keep collecting data, and report to the supervisor. Watch specifically for distractors that have the RBT independently modifying or abandoning the plan. Those are wrong because that’s a BCBA decision, not yours.

Why this pushes you toward reinforcement

Step back and the pattern is obvious. Extinction’s side effects are rough but mostly temporary, and the procedure adds nothing aversive. Punishment’s side effects, the aggression, the avoidance, the modeling, the damage to your relationship with the learner, are heavier and more lasting. That’s the core ethical argument for least restrictive practice: when reinforcement-based and extinction-based approaches can do the job, you use those first, because the collateral damage is smaller.

You’re not the one making that call. But understanding why the field is cautious about punishment, and why even extinction gets paired with reinforcement instead of run alone, makes you better at the part you do own: running the plan with fidelity and flagging the side effects honestly when they show up.

Key points to remember

• Every behavior reduction procedure produces side effects. The third-edition task list expects you to recognize them, not just the main procedure.
• Extinction’s side effects: the extinction burst (a temporary increase in frequency, intensity, or duration at the start), extinction-induced aggression and emotional responding, and spontaneous recovery (a later, weaker reappearance after the behavior seemed gone).
• The burst happens at the start of extinction; spontaneous recovery happens later, after the behavior already faded. Don’t confuse them.
• Caving during a burst is the worst move, because you reinforce the most intense version of the behavior.
• Punishment’s side effects: emotional and aggressive responses, escape and avoidance of the punisher or the setting, and modeling, the learner copying the punitive behavior they see.
• Escape and avoidance can quietly damage your relationship with the learner, which is the foundation of being able to teach at all.
• When any side effect appears, the response is always the same: stay consistent, keep the data, tell your supervisor. Don’t modify or abandon the plan on your own.
• The weight of punishment’s side effects is a central reason the field favors reinforcement and extinction first, under least restrictive practice.