Manipulating parameters of reinforcement to reduce problem behavior without extinction

The team ran DRA without extinction on three kids with developmental delay. First they tested which reinforcement detail each child cared about most: better snack, bigger portion, or faster delivery.

Then they used that detail to make the new behavior pay off more than the problem behavior. No ignoring, no blocking—just sweeter reinforcement for the right response.

Problem behavior dropped for every child once the chosen parameter was tilted. The kids kept getting the same reinforcers; only the timing, size, or quality changed in favor of the alternative.

Stephens et al. (2018) reviewed ten similar studies and saw the same pattern: tweak magnitude, immediacy, or quality and you can skip extinction. Kunnavatana’s work is one of the single-case examples inside that review.

Rey et al. (2020) tested DRO without extinction in a lab and also cut problem behavior. Their data say the drop happens because other responses get accidentally reinforced, backing up the idea that contingency, not extinction, drives the change.

Attwood et al. (1988) saw the same thing with pigeons: differential reinforcement beat plain free food. Across species and procedures, the story matches—make the right response pay better and the wrong one fades.

You can drop extinction from your DRA plan when safety or consent makes ignoring risky. Run a quick preference test to see if the client wants the treat faster, bigger, or fancier, then load that feature onto the replacement behavior. You keep reinforcing, they keep learning, and problem behavior loses its edge without a battle.