Centrifugal selection of signal-directed pecking.

Michael (1974) worked with three pigeons that pecked a lit key for grain. First the birds got food every peck. Then the rules flipped: any peck when the key was lit cancelled the next piece of food.

The switch happened two ways. For one bird the change was sudden. For the others the authors first gave short 'tastes' of the new no-food rule before making it permanent.

All birds stopped pecking the lit key. The bird that got the sudden switch quit fastest. The birds with the gradual taste took a little longer.

Once the key peck died out, the birds rarely tried again. The food-omission contingency worked like a brake on the old reinforced behavior.

Bacon et al. (1998) later used the same extinction logic with three speech-delayed preschoolers. Instead of losing food, the children got brief breaks from demands whenever they stayed calm. Disruptive behavior dropped fast, showing the pigeon rule applies to kids in a clinic.

The monkey study by Terrace (1969) is an earlier cousin. Shock was removed when the animals pulled a lever, so the lever pull was strengthened. Michael (1974) flips the coin: food was removed when the pigeons pecked, so the peck weakened. Same process, opposite direction.

Oliver et al. (2002) review warns that extinction and omission tricks still need more lab work before we trust them for severe problem behavior. The 1974 data are clean, but the reviewers remind us to keep measuring in real-world settings.

If you want to cut a reinforced behavior quickly, skip the slow fade. Move straight to the omission contingency—no reinforcement after the target response. Start baseline measurement first, then flip the rule in one shot. Watch the rate drop; if it stalls, check for accidental reinforcement.