Extinction and Spontaneous Recovery

Pavlov's dogs learned that a tone meant food was coming, so the tone alone made them salivate. If you keep ringing the tone but stop giving food, the salivation response gradually shrinks and eventually seems to disappear. That's extinction: the conditioned stimulus (tone) is no longer followed by the unconditioned stimulus (food), so the prediction it carried stops paying off, and the brain lowers its bet. The response curve drops because the association's signal value weakens, not because the wiring is ripped out.

Put it on a timeline. During acquisition, repeated tone-food pairings build up a strong conditioned response, salivation climbs. During extinction, the tone keeps firing alone and salivation declines toward zero. Then comes a rest period with no trials at all. Test the tone again after that break, and the dog salivates again, weaker than the original response but clearly present. That reappearance is spontaneous recovery. It happens without any new pairings, which is the key clue that the original learning was stored, not destroyed.

This matters because it corrects a tempting misconception: extinction looks like erasing, but it behaves like suppression. The learned association stays on file somewhere; extinction just builds a competing, newer memory that says 'tone now means nothing,' and that newer memory can weaken with rest, letting the old one peek through. Each time you run another round of extinction, the recovered response comes back smaller and fades faster, showing the suppression is accumulating. This sets up stimulus generalization and discrimination next, since how broadly or narrowly that original association still applies depends on what exactly got learned underneath the extinguished surface.