Behavioral contrast and the automaintained key peck.

The team worked with eight pigeons in a standard chamber. A key light came on for several seconds, then grain followed no matter what the bird did. This is called positive automaintenance — the bird learns to peck even though pecking is not required.

Two colors alternated. One color was always followed by grain (food CS). The other color was never followed by grain (no-food CS). The birds never had to peck to get grain.

Six out of eight birds pecked more during the food color and less during the no-food color. This is positive behavioral contrast: responding goes up where food is signaled and down where it is not.

When both color and grain were removed, the contrast vanished. The effect returned when the colors and grain came back.

Sadowsky (1973) already showed contrast when birds simply sat through timeout or blackout. Wesp et al. (1981) extends that idea — contrast still happens even when the bird does not have to work at all.

Van Houten et al. (1980) looked at the same key-peck response but with real reinforcement for pecking. They found contrast did not change how long each peck lasted. Wesp et al. (1981) now shows contrast can emerge without any peck-reinforcer link, so topography and contingency are separate issues.

Hassin-Herman et al. (1992) later explained that Pavlovian contrast (driven by signaled food versus no food) can override other effects. Wesp et al. (1981) is an early lab proof of that Pavlovian path.

If you run Pavlovian pairings — CS followed by reinforcer no matter what the client does — watch for contrast. A stimulus that signals “no reinforcer coming” can suppress responding, while a stimulus that signals “reinforcer coming” can raise it, even when the client did nothing to earn it. This can help or hurt your program. For example, a brief “no-token” card during a token stretch might drop work rate more than you expect, while a “token soon” card could boost it. Plan your stimulus pairs and check data component by component.