Operant extinction is what happens when a behavior that was previously reinforced stops producing the expected reward, and as a result, the behavior gradually decreases and eventually stops. If a rat has learned to press a lever for food pellets and the pellets stop coming, the rat will eventually stop pressing the lever. The same principle applies to human behavior in countless everyday situations.
What makes extinction interesting, and sometimes counterintuitive, is that the behavior doesn’t simply vanish the moment reinforcement ends. It follows a predictable pattern that includes temporary escalation, gradual decline, and occasional reappearance.
How Extinction Actually Works
Extinction isn’t the same as forgetting. The original learning remains intact. Instead, extinction is a form of new learning that competes with the old. When a behavior no longer produces its expected outcome, your brain doesn’t erase the original association. It builds a new, competing association: “this behavior no longer works here.” Researchers describe this as a retroactive interference process, where new information conflicts with previously learned information.
This distinction matters because it explains why extinguished behaviors can come back under certain circumstances. The original learning is still stored. Extinction essentially teaches the brain to inhibit a specific behavior in a specific context, rather than deleting the behavior from memory entirely.
Context plays a major role. A behavior that has been extinguished in one setting can reappear in a different setting where extinction never took place. This is called “renewal,” and it happens because the inhibition learned during extinction is tied to the environment where it occurred. A child who has stopped throwing tantrums at home might still try it at a grandparent’s house, where the old rules might still apply.
The Extinction Burst
Before a behavior fades, it almost always gets worse first. This temporary spike is called an extinction burst, and it’s one of the most important things to understand about the process. When reinforcement suddenly stops, the immediate response isn’t to give up. It’s to try harder.
An extinction burst can show up in three ways. First, the target behavior itself may increase in frequency, duration, or intensity. A child who cries for candy in the store may cry louder or longer when they first realize the behavior no longer works. Second, variations of the same type of behavior may appear. If head-banging was the original behavior, face-slapping might partially replace it. Third, entirely new problem behaviors can emerge, including aggression that wasn’t part of the original picture.
Research on the timing of extinction bursts shows that the target behavior typically peaks during the first minute after reinforcement is removed. Aggression, if it appears, tends to peak during the second and third minutes. The burst is temporary by nature, but it can be intense enough that people give in, which resets the entire process and actually teaches the person or animal that escalation works.
Spontaneous Recovery
Even after a behavior has been successfully extinguished, it can reappear after time has passed. This is called spontaneous recovery, and it happens when enough time elapses between the extinction experience and the present moment that the behavior resurfaces, sometimes at near-original strength.
In laboratory settings, spontaneous recovery has been observed after intervals as short as 24 hours and as long as several days. In one study, researchers tested for spontaneous recovery eight days after extinction training was complete and still found the behavior returning. This doesn’t mean extinction failed. It means the original learning is still stored and can temporarily override the newer extinction learning, especially when the context shifts (including a shift in time).
Each time spontaneous recovery occurs and the behavior again goes unreinforced, the recovery episodes tend to be weaker and shorter. Over repeated cycles of extinction, the behavior loses more and more of its strength.
Why Some Behaviors Are Harder to Extinguish
Not all behaviors extinguish at the same rate. One of the strongest predictors of how long extinction will take is whether the behavior was reinforced every single time or only some of the time. This is known as the partial reinforcement extinction effect, and it’s one of the most reliable findings in behavioral psychology.
A behavior that was rewarded every time it occurred (continuous reinforcement) extinguishes relatively quickly. The change from “always rewarded” to “never rewarded” is dramatic and easy to detect. But a behavior that was rewarded unpredictably, only some of the time (partial reinforcement), is far more resistant to extinction. The shift from “sometimes rewarded” to “never rewarded” is harder to distinguish, so the organism keeps responding longer before giving up.
This explains why slot machines are so effective at maintaining gambling behavior, and why intermittent attention from a parent can make a child’s demanding behavior remarkably persistent. If whining for a treat works one out of every five times, that behavior will be much harder to eliminate than if it worked every single time and then suddenly stopped.
The rate of prior reinforcement also matters, but in a more nuanced way than you might expect. Research shows that when different reinforcement rates are experienced in alternating contexts, the behavior associated with more frequent reinforcement is actually more resistant to extinction in its respective context, even when response rates were lower during training. The richness of the reinforcement history strengthens the association between the setting and the reward, making the behavior stickier.
What Happens in the Brain
Extinction learning involves several brain regions working together. Three structures play central roles: the amygdala (which processes emotional responses and threat signals), the prefrontal cortex (which exerts top-down control over behavior), and the hippocampus (which encodes contextual information).
During extinction, the prefrontal cortex essentially acts as a brake. It activates inhibitory cells in the amygdala that function like an “off switch,” suppressing the previously learned response. This is why extinction is context-dependent: the hippocampus tracks where and when extinction occurred, and the prefrontal cortex uses that contextual information to determine whether the inhibition should be applied.
At the cellular level, extinction requires calcium signaling in the amygdala, modulated by the brain’s own cannabinoid and opioid systems. The brain’s natural opioid receptors in a region called the periaqueductal gray are necessary for extinction to take hold. Blocking those receptors prevents extinction from occurring. Meanwhile, the inhibitory neurotransmitter GABA helps suppress the old response once extinction has been consolidated.
Like other forms of learning, extinction has three phases: acquisition (the initial decrease in behavior during the extinction session), consolidation (the stabilization of that new learning over hours), and retrieval (successfully recalling and applying the extinction learning later). Each phase depends on different molecular processes, which is why extinction can fail at different points.
Practical Applications
Extinction is one of the core tools in applied behavior analysis, particularly for addressing problem behaviors in children. The basic strategy is to identify what reinforcement is maintaining a behavior and then remove that reinforcement consistently. In practice, this looks different depending on what’s driving the behavior.
For attention-seeking behavior, extinction means not providing the attention that the behavior is designed to produce. If a child whines to get a parent’s focus, the parent withholds that response. For escape-motivated behavior, like a child acting out to avoid doing homework, extinction means not allowing the behavior to result in escape from the task. For tangible-driven behavior, like throwing a tantrum to get a toy, it means holding the boundary and not handing over the toy.
In each case, effective use of extinction pairs the removal of reinforcement with teaching a replacement behavior. If a child yells to get attention, extinction works best when the child is simultaneously taught an appropriate way to request attention. If a child tantrums for a toy, you can hold the boundary while offering comfort and redirecting to an alternative activity. Extinction without a replacement strategy is incomplete, because the person still has the underlying need that the problem behavior was serving.
The extinction burst is the most challenging phase for anyone implementing this approach. Knowing it’s coming helps. The temporary escalation is a sign that the process is working, not that it’s failing. The critical rule is consistency: if you give in during the burst, you’ve taught the person that escalation is the strategy that pays off, creating a more intense version of the original problem.