Lever-contact responses in rats: automaintenance with and without a negative response-reinforcer dependency.

The team worked with rats in a small lab cage. A lever sat on one wall. A light and food cup sat above it.

Some rats got positive automaintenance: light came on, then food, no matter what. If the rat touched the lever during the light, it still got food.

Other rats got negative automaintenance: light came on, then food, but only if the rat did NOT touch the lever. Lever touches canceled the meal.

Rats in both groups learned to touch the lever. The positive group pressed more, but even the negative group kept pressing even though it cost them food.

The result shows that simple pairing of light and food is enough to make lever contact stick, even when the response hurts.

Hodos et al. (1976) repeated the setup with pigeons and crows. They also saw positive automaintenance, but the birds quickly stopped pressing when it canceled food. The clash looks real: rats stay high, birds drop low. The difference is likely species; rats may be more stubborn or less sensitive to lost pellets.

Rutland et al. (1996) zoomed in on pigeons and tracked minute-by-minute changes. They found that slower food rates flattened the within-session response curve. Their work extends the 1974 finding by showing that timing details, not just yes/no learning, follow similar rules across species.

Harrison et al. (1975) kept the lever-press-for-stimulus idea but swapped automaintenance for duration control. Rats worked hard to turn on a positive stimulus and start a schedule, while escape from a bad stimulus barely held. Together the papers say: stimulus production can drive pressing, but the way you link the stimulus to food sets the strength.

You now know that automaintenance is not just a pigeon oddity; it also hits rats and probably humans. If you pair a cue with reinforcement, clients may keep touching, approaching, or repeating even when the response cuts off the good thing. Watch for this in token boards, sensory bins, or response-cost systems. Before you call a behavior "maladaptive," test if it’s an automaintenance artifact. If it is, change the cue-reinforcer pairing first instead of adding more punishment.