Varieties of contrast: a review of incentive relativity by charles f. Flaherty.

Williams (1997) wrote a narrative review that maps the incentive-contrast and behavioral-contrast worlds. He pulled together decades of pigeon and rat data to see where the two literatures overlap and where they clash.

The paper pays special attention to anticipatory contrast — the bump or dip that shows up before the next reinforcer. Flaherty argues this timing effect breaks the old story that contrast is just about reward value going up or down.

The review shows that simple ‘rich-to-lean’ or ‘lean-to-rich’ reward shifts cannot explain every contrast curve. Anticipatory effects pop up too early and in the wrong direction for a pure value-shift account.

In plain words: the animal is not only reacting to how good the payoff is; it is reacting to when it expects that payoff.

REYNOLDS et al. (1961) first showed contrast with FI and DRL schedules, proving you do not need high-rate reinforcement to see the effect. Williams (1997) folds that finding in, but notes that even those early data look odd if you only track reward size.

Parsons et al. (1981) found contrast only after they scrubbed out accidental reinforcement — a warning that procedure details can hide or create the effect. The 1997 review uses this to caution: if your contrast disappears, check your stimuli first.

Dixon (2014) later offered a new model that pins contrast on response competition and schedule-induced side behaviors, not on reward value at all. This theory directly answers the puzzle Williams (1997) raised, giving practitioners a fresher lens than the 1960s value-shift view.

When you see a sudden response burst or slump during a mult schedule, do not jump to ‘must be the reinforcer size.’ Look at timing, stimulus change, and possible adjunctive behaviors. Try holding reward size constant while you shift only the SD or the interval before reinforcement. If contrast still shows up, you have evidence that value-shift is not the driver — and you can design leaner, cleaner interventions that rely on stimulus control instead of bigger candy.