Behavior Modification

Extinction Bursts, Ratio Strain, and Resurgence

When reinforcement changes, behavior gets worse before it gets better — sometimes. The extinction burst is real but less universal than trainers think. Ratio strain and resurgence are more common and less discussed. All three are lawful, predictable, and manageable if you know what you're looking at.

Extinction Bursts: Not as Universal as You've Heard

When a previously reinforced behavior no longer produces reinforcement, the organism temporarily increases the frequency, duration, or intensity of that behavior. The logic: "This worked before — maybe I'm not doing it hard enough."

But the prevalence varies dramatically. Lerman and Iwata (1995) found bursts in 36% of cases with extinction alone, dropping to 12% when combined with DRA or DRO. Muething et al. (2024) reported 24% across 108 clinical cases. Katz and Lattal (2020) found near-zero prevalence in basic research with stringent criteria, concluding that "the extinction burst does not appear to be a ubiquitous effect of extinction."

Fisher et al. (2023) reconciled the discrepancies: burst likelihood depends on the magnitude of reinforcement-rate decrease. Dense DRA prevents bursts by maintaining overall reinforcement rates.

Resurgence Is the Bigger Threat

Resurgence is the return of a previously reinforced problem behavior when conditions for a replacement behavior worsen. Briggs et al. (2018) documented resurgence in 19 of 25 FCT schedule-thinning applications (76%). Muething et al. (2020) found post-thinning problem behavior was seven times higher than the preceding five-session average.

For dog training: if reinforcement for sitting-to-greet is abruptly reduced, jumping comes back. Gradual thinning with discriminative stimuli — signals that tell the dog when reinforcement is and isn't available — mitigates resurgence.

Ratio Strain: When the Deal Gets Too Lean

Ratio strain is the progressive degradation of behavior as schedule requirements increase. Vicars et al. (2014) observed this directly in dogs: as the number of nose-touches required per reinforcer increased, dogs showed pawing, barking, spinning, mouthing, and eventually ceased responding entirely.

The distinction matters because the interventions are different. For an extinction burst, hold steady and wait. For ratio strain, return to a richer schedule immediately.

Recovery and Prevention

From ratio strain: Return to the last schedule where the dog was responding fluently. Rebuild for 2-3 sessions. Thin again in a smaller step.

From extinction burst: Hold steady. Don't reinforce the escalation.

Prevention: Thin slowly. Use variable schedules. Monitor the 80% accuracy rule. Increase non-food reinforcement as food delivery decreases, so the total reinforcement value stays high even as the food schedule thins.

Spontaneous recovery — the reappearance of extinguished behavior after a rest period — is normal. Temporary relapses are features of extinction, not evidence of treatment failure.

The Welfare Question

Extinction-induced aggression is well-documented: Lerman et al. (1999) found aggression increases in 22% of extinction applications. The LIMA principle (CCPDT, IAABC) and Friedman's Humane Hierarchy place extinction below positive reinforcement and differential reinforcement in the intervention hierarchy. No published study has directly measured extinction bursts in dogs — all prevalence data come from human clinical populations or laboratory animals.

Sources: Lerman & Iwata (1995, 1996), Fisher et al. (2023), Briggs et al. (2018), Muething et al. (2020, 2024), Vicars et al. (2014), Katz & Lattal (2020). From the Data Dogs research briefs: ABA Methodology and Reinforcement in Practice.