Noncontingent reinforcement competes with response performance

Kelley et al. (2017) tested how much noncontingent reinforcement hurts learning. One child with developmental delay and several pigeons served as participants.

The team varied the share of free reinforcers delivered each minute. They then tracked how quickly the child learned to ask for toys and how fast pigeons learned to peck a key.

More free goodies meant fewer correct responses. When 90 % of reinforcers were noncontingent, the child's new manding and the birds' keypecking both dropped.

The drop happened during teaching and later during maintenance. Free reinforcers were competing with the very behaviors the experimenters wanted to strengthen.

Wildemann et al. (1973) showed pigeons can tell when food is response-independent. Kelley extends that idea: once animals notice the free food, they stop working for more.

Griesi-Oliveira et al. (2013) found that delaying a conditioned reinforcer slows acquisition. Kelley adds a second way to slow learning: flood the session with reinforcers that require no response at all.

Nevin (1968) arranged reinforcement for not responding and also saw flatter response curves. Both studies warn that reinforcing 'doing nothing' can accidentally weaken desirable behavior.

If you run high-density NCR to kill problem behavior, watch the learner's new skills. Ask: are mands, toileting, or social bids fading? If yes, thin the free reinforcers or tie them to a simple contingency so the alternative behavior keeps strength.