← All study notes
Behavior-Change Procedures

Extinction Burst Is Not a Catch-All Term: Four Extinction Effects You Need to Stop Mixing Up

8 minute read

A behavior gets louder. A new topography appears. An old behavior suddenly returns. A response that disappeared last week shows up again Monday.

All four can happen in relation to extinction. They are not the same phenomenon — and the BCBA exam knows exactly where that confusion lives.

Extinction burst is one of the most recognizable terms in behavior analysis, and that familiarity becomes a liability. See extinction, see behavior change, pick extinction burst. It feels safe because you know extinction bursts are real. But extinction can be associated with several different changes in responding, and which name applies depends on what changed, when it changed, and the reinforcement history of the behavior you're looking at.

That distinction separates four concepts students routinely collapse into one: extinction burst, extinction-induced response variability, resurgence, and spontaneous recovery. The trick isn't memorizing four definitions. It's learning to reconstruct what happened to the behavior.

Exam Trap

An extinction burst is one possible effect of extinction, not another name for everything unusual that happens after extinction begins. In one influential analysis of 113 extinction data sets from applied research, an initial increase in the target response occurred in only 24% of cases — and bursting was even less common when extinction was combined with other procedures (Lerman & Iwata, 1995).

Extinction does not mean extinction burst. The fact that extinction is occurring tells you the contingency changed. It doesn't yet tell you which behavioral effect you're observing.

Extinction Burst: More of the Previously Reinforced Response

A child has historically obtained attention by yelling. Attention is now withheld following yelling. Shortly after extinction begins, the child yells more than before — that's the classic extinction-burst pattern. The important feature isn't merely that something happened after extinction. It's that the previously reinforced response itself temporarily increased.

For exam purposes: same previously reinforced response, temporarily increased responding. In Lerman and Iwata's (1995) analysis, extinction bursting was operationalized as an initial increase in the frequency of the target response following extinction. Don't let the word "extinction" do the answering for you — look for evidence that the established response itself has increased.

Extinction-Induced Response Variability: Different Ways of Responding

Now imagine the child doesn't simply yell more. Instead, once yelling stops producing attention, new forms of responding appear — vocalizations, gestures, movements that weren't previously observed. That's extinction-induced response variability, and it's not just textbook terminology: Lalli, Zanolli, and Wohn (1994) trained children on specific toy-play topographies, placed those responses on extinction, and documented the emergence of untrained toy-play forms. Later work by Grow and colleagues (2008) examined whether that variability could be clinically useful — evaluating whether variations in problem behavior under extinction could surface alternative mand topographies during functional communication training.

The useful discrimination: extinction burst is more of the previously reinforced response; extinction-induced variability is greater variation in response form. And notice what you don't need to claim — you don't need the learner to be "searching for" a new response. The observable fact is enough: when previously reinforced responding contacts extinction, variability may increase and previously unobserved forms may emerge.

The Door-Handle Analogy
Someone always opens a stuck door by turning the handle clockwise. It stops working. They turn it clockwise again. And again. Harder. Faster. Repeatedly.
That's a burst — increased occurrence of the established response.
Now imagine instead they turn it clockwise, pull upward, jiggle it, push the door, turn the handle backward, lift the door, try something else entirely.
That's variability — the repertoire itself has become more varied.

Resurgence: An Old Response Gets Another Chance

Resurgence is different again — here, the behavior that reappears isn't a novel variation. It has a reinforcement history. The classic sequence: Response A is reinforced. Then Response A stops being reinforced while Response B is reinforced instead. Then reinforcement for Response B is discontinued or worsens. Response A returns.

Epstein's (1983) foundational work with pigeons demonstrated exactly this: previously reinforced behavior reappeared once more recently reinforced responding was placed on extinction. That gives you one of the most useful discriminations in this whole topic — extinction-induced variability doesn't require the newly observed form to have been reinforced before. Resurgence does. When a question describes a behavior reappearing during extinction, ask about its history. New or varied form with no clear past? Think variability. A response that worked before, stopped, and returned as conditions worsened for its replacement? Think resurgence.

Spontaneous Recovery: Time Changes the Picture

Spontaneous recovery adds one more wrinkle — once again, an old behavior returns, but the event immediately preceding its return is different. A response is reinforced, then placed on extinction, and responding decreases substantially. Treatment ends for the day. Time passes. At a later session, the previously extinguished response temporarily reappears.

The key clue isn't alternative reinforcement becoming unavailable, the way it is for resurgence — it's the passage of time following extinction. Spontaneous recovery has been demonstrated in applied research as the temporary reappearance of responding after behavior appeared to have been extinguished (Lerman et al., 1999). This is exactly why "resurgence equals old behavior comes back" is a dangerous shortcut: spontaneous recovery can also involve the return of previously extinguished responding. What separates them is the history immediately preceding the recurrence — a worsening alternative for resurgence, elapsed time following extinction for spontaneous recovery.

Stop Naming the Behavior. Reconstruct the Contingency.

Suppose a question simply says: "Problem behavior returned." Don't name the phenomenon yet — that's only an observation. Reconstruct what happened immediately before it instead.

PhenomenonWhat you observeWhat to look for
Extinction burstPreviously reinforced responding increasesExtinction has just been introduced
Extinction-induced variabilityNew or varied response topographies emergePreviously reinforced responding is contacting extinction
ResurgencePreviously reinforced behavior returnsReinforcement for a more recently reinforced alternative worsened or stopped
Spontaneous recoveryPreviously extinguished behavior returnsTime has passed following extinction

The surface descriptions overlap. The contingency histories don't. That's why the safest question to ask isn't "what does this behavior look like?" — it's "what happened immediately before this behavior changed?"

Try it yourself

Real exam items give you a full history and expect you to trace it, not match a keyword. Read this one slowly.

Quick check — try it before you keep reading

A client historically requested breaks by tapping the table, which staff reliably honored. The team then introduced a more efficient AAC button for break requests and stopped honoring the table-tap. Within days, table-tapping dropped out entirely and AAC button presses became the client's consistent way of requesting breaks. Two weeks later, the AAC device develops a technical fault and intermittently fails to register presses. During that same window, the client resumes tapping the table to request breaks — a response not observed in the two weeks since the AAC button took over. Which phenomenon does this describe?

Keep reading — we'll trace exactly why the other three don't fit, one detail at a time.

Why the other three don't fit

Option A — Extinction burst

Table-tapping was placed on extinction, so it's tempting to reach for the most familiar term. But the burst pattern is a previously reinforced response temporarily increasing right after its own extinction begins. Table-tapping didn't increase when it stopped being reinforced — it dropped out entirely and stayed gone for two weeks. Whatever brought it back happened much later, and for a different reason.

Option B — Extinction-induced variability

Variability describes increased variation in responding under extinction, including the emergence of response forms not previously observed in the relevant context. But table-tapping isn't simply a new variation appearing when reinforcement is withheld. The stem gives it a specific reinforcement history: table-tapping previously produced breaks, was replaced by a reinforced alternative, and then returned when reinforcement for that alternative worsened. That history makes resurgence the more precise description.

Option D — Spontaneous recovery

This is the closest call, because spontaneous recovery also involves previously extinguished responding returning after a period of reduced responding. But the critical clue for spontaneous recovery is recurrence following the passage of time after extinction. Here, the stem supplies a different contingency change immediately before table-tapping returns: the AAC device malfunctions and reinforcement for the more recently reinforced alternative worsens. Given that history, resurgence is the more precise description.

Correct — Resurgence

The full sequence matches the classic three-phase arrangement exactly: table-tapping (Response A) was reinforced, then extinguished once the AAC button (Response B) took over and was reinforced instead, then reinforcement for Response B worsened when the device malfunctioned. Response A returned. That's resurgence — and the AAC malfunction is the detail that makes it identifiable, not incidental scene-setting.

"Behavior changed during extinction" is an observation, not an explanation. Ask what changed, whether the response is new or previously reinforced, what happened to reinforcement immediately beforehand, and whether time passed. Then name the phenomenon.

How to catch these on exam day

Extinction doesn't erase history

There's a larger lesson underneath all four concepts. Extinction shouldn't be understood as deleting what an organism previously learned. The response may decline when the contingency changes, but prior learning can continue to matter — under the right conditions, responding can recur. A behavior reaching low levels during extinction doesn't mean its reinforcement history has been erased.

That's also why these questions are genuinely hard: the current behavior only makes sense once you account for what was reinforced before, what stopped being reinforced, what alternatives were reinforced, and what changed afterward. Behavior has a history. The exam expects you to use it.

BA Prep's explanations are built the same way this article is — tracing the contingency history behind a question, not just naming the right answer. If four terms that all mean "behavior changed during extinction" gave you pause, that's exactly the kind of discrimination our practice questions are built to train.

Start Free Trial

References