Conditioned suppression of an avoidance response by a stimulus paired with food.

Davis et al. (1972) worked with lab rats pressing a lever to avoid shocks. They added a tone that always came right before food pellets. The question: would this happy signal mess up the animals’ careful shock-avoidance rhythm?

Each rat already had a steady pattern: press within 10 s and no shock occurs. The researchers then turned on the tone-food pairings for some sessions and later removed them. They counted shocks, lever presses, and little pauses.

When the food tone sounded, rats stopped pressing for a moment and took more shocks. Shock rate almost doubled. After eating, they often pressed in quick bursts, so daily press totals stayed flat.

Once the tone stopped bringing food, the shock rate dropped back down. Traditional suppression ratios missed the change, but raw shock count told the story.

Schmidt et al. (1969) saw the opposite: rats kept pressing even after shocks no longer followed the lever. Both studies show shocks can rise when the schedule changes, but for different reasons—W’s shocks became non-contingent while H’s shocks increased because a food cue froze the avoidance response.

Malagodi et al. (1975) later proved long, variable warning intervals speed up avoidance learning. H’s food cue acted like a short, fixed warning and did the reverse: it slowed the rat down and raised shocks. Same lever-shock preparation, opposite outcomes, because the added stimulus pulled the rat toward food instead of safety.

Lambert et al. (2017) and Stevens et al. (2018) thin reinforcement in children and see response resurgence. H’s rats show a basic version: when the food cue disappears, the old efficient avoidance returns—an early lab model of relapse after schedule change.

If you run concurrent or embedded schedules, watch for cute stimuli that predict other reinforcers. A praise phrase that signals candy during a DRO session could freeze the target response and let problem behavior slip through, just like the food tone let shocks slip through. Track the real-life consequence count (hits, bites, shocks) not just response rate; the cost may hide in a flat line.