Response-independent food delivery and behavioral resistance to change.

Harper (1996) gave pigeons free food while they pecked a key for other food. The birds worked on two schedules at once. One schedule gave tiny food pellets. The other gave bigger pellets.

The team then made the free food last longer or come faster. They watched which schedule broke down first.

When free food got longer or more frequent, birds pecked less on both schedules. The tiny-pellet schedule crashed the hardest.

The result fits behavioral momentum theory: lean reinforcement = weak behavior = easy to disrupt.

Edwards et al. (1970) saw the same drop, but only after many sessions or doubling the free food. Their effect was small and slow. Harper (1996) shows the drop can be large and fast when baseline reinforcers are small.

McIntyre et al. (2002) later flipped the test. Instead of adding food, they removed all food (extinction). Rich histories kept responding alive longer. Together the studies book-end momentum: lean histories break easier under extra food; rich histories survive longer under no food.

Costa et al. (2024) repeated the logic with college students and point-loss. High reinforcement rates again shielded responding. The pattern now spans species, reinforcers, and disruptors.

You now have a quick probe for lean vs rich components in a client’s program. If noncontingent snacks or iPad time crash one skill but not another, the crashed skill probably runs on thinner reinforcement. Thicken that component first, then retest.