Induction, contrast, and resistance to extinction.

The team used a two-part schedule with pigeons. In one part, pecks paid off every few seconds. In the other part, pecks never paid off.

They then stopped all payoff. They asked: does the still-paying part change how long the birds keep pecking when nothing pays?

Birds kept pecking longer in the extinction part if the other part had kept paying. Positive contrast before extinction made the behavior stickier.

Even when the alternate part quickly dropped to zero pecks, the first part still resisted extinction.

de Rose (1986) later showed the same pattern with fixed-interval schedules. Longer no-pay segments made the next reinforced segment busier, extending the 1968 finding to new timing rules.

Hineline et al. (1969) added that order matters. The component that comes right before extinction drives the size of the contrast bump, refining the original picture.

Cox et al. (2015) seemed to clash. They said brief exposure to steady payoff does not build resistance; only many stable sessions do. The difference is time: 1968 looked at momentary contrast, 2015 looked at long-term momentum. Both can be true.

When you fade reinforcement, think about the whole schedule, not just the target response. Keeping payoff alive in one area can accidentally make another behavior harder to extinguish. Check your program for hidden contrast before you start thinning.