Signaled alternative reinforcement and the persistence of operant behavior

Bland et al. (2016) worked with neurotypical animals in a lab. They compared two kinds of differential reinforcement of alternative behavior (DRA).

In one condition a brief light or sound told the animal, 'other rewards are now available.' In the other condition no cue was given. The team then stopped everything to see how long the old target response would keep going.

Target responding lasted longer after the signaled DRA. The cue that said 'free treats here' accidentally glued the old behavior in place.

When the signal was absent, the same alternative rewards caused less persistence. The warning itself was the problem.

Craig et al. (2018) extends the story. They showed that putting DRA in its own DRO context (a special corner, different color) cuts later relapse in half. Their context cue helped; Bland's cue hurt. The difference is who controls the message. In Bland the cue said 'extra food any time.' In Craig it said 'food only here, then session ends.'

Cullinan et al. (2001) is a predecessor that foreshadowed the risk. Unsignaled free milk also made lever pressing stubborn during extinction. Bland proves that adding a signal makes the problem worse, not better.

Laugeson et al. (2014) looks like an apparent contradiction. They paired brief stimuli with response-independent food and saw no change in extinction resistance. The key difference is their food did not depend on any behavior and the cue was short. Bland's signal arrived during true DRA, when the animal could switch behaviors. Method tweaks turn a neutral result into a harmful one.

If you use DRA for problem behavior, think twice before adding a signal that marks when alternative rewards are available. A simple 'break' card, bell, or verbal 'good, you can have an iPad now' may make the problem return later. Instead, deliver the alternative reinforcer quietly or move to a separate space as Craig suggests. Test both ways in your next session and track which produces faster extinction when the reinforcer stops.