Automaintenance in guinea pigs: effects of feeding regimen and omission training.

Neef et al. (1978) worked with guinea pigs in a small lab cage. A lever sat near a food cup.

Every few minutes a light came on. While the light glowed, food dropped into the cup no matter what the pig did. The pigs soon began to press the lever whenever the light appeared. This is called automaintenance.

Next the rule changed. Now any lever press during the light blocked the food pellet. This is omission training. The team counted how fast the lever pressing stopped.

Lever presses almost vanished once the press blocked food. Omission training wiped out the automaintenance effect in every animal.

The result shows that a response strengthened only by stimulus-food pairings can be undone by one simple contingency: the response costs you the reinforcer.

King et al. (2025) extends the same logic to children with autism. They used non-contingent food deliveries to prevent resurgence, a cousin of automaintenance. Both studies prove that controlling when food arrives can strengthen or weaken behavior without extra rewards or punishers.

Austin et al. (2015) found that returning to the old reinforcement context brings the response back. Their renewal effect looks like a contradiction: if context can revive the lever press, did omission training really work? The answer is yes, but you must stay in the new no-food-for-pressing context. Move the child back to the old room or teacher and the response may return.

O'Reilly et al. (2008) used the opposite move. They gave free bites of the reinforcer before play time. This also cut problem behavior, showing that either front-loading or blocking the reinforcer can serve as an abolishing operation.

You can copy the omission principle today. If a child's stereotypy is maintained only by sensory feedback, arrange the sensory outcome to disappear when the behavior occurs. For example, if tapping a tablet brings a preferred sound, program the app to mute on any tap. The 1978 data say the movement should drop fast, and the 2025 follow-up shows the tactic works with kids. Just keep the context steady or renewal may sneak in.