Frequency of reinforcement as a determinant of extinction-induced aggression during errorless discrimination learning.

The team worked with pigeons learning a red-green discrimination. Birds got food for pecking the red key. Green key pecks never paid off.

Before the test, half the birds earned food every 30 seconds. The other half got food only every 3 minutes. Then the red key went dark and all reinforcement stopped.

During this sudden extinction, the birds could also peck a small plastic model. Hard pecks were counted as attack.

Birds on the rich 30-second schedule attacked the model far more. Their attack bursts peaked right after the red key turned off.

Birds on the lean 3-minute schedule showed little or no aggression. Less reinforcement history meant safer extinction.

Morse et al. (1966) first showed that extinction alone can spark aggression. Harrison et al. (1975) adds the rule: the richer the past pay, the harder the bite.

McIntyre et al. (2002) looked at the same rich-versus-lean setup but measured key-peck persistence. Rich histories made the pecking last longer. Together, the two papers show rich reinforcement strengthens both resistance and aggression during extinction.

Neisworth et al. (1985) flipped the finding to humans. They used a dense-then-extinction schedule to cut self-stimulation in adults with disabilities. Aggression was not reported, but the same CRF/EXT sequence was used. The difference: the humans had no toy to attack, highlighting that topography matters when predicting side effects.

If you must place a skill on extinction, first thin the reinforcement schedule. Move from continuous or dense pay to leaner ratios or longer intervals over several sessions. This simple step can prevent the burst of hitting, kicking, or property destruction that sometimes follows sudden non-reinforcement. Track the first minute after the SD ends; that is when attack is most likely to show.