Pavlov's Bell
In classical conditioning, a neutral signal that reliably comes before an automatic response starts to trigger that response on its own. · 10 min
Some responses need no learning. Food in the mouth makes saliva flow; a puff of air at the eye makes it blink. These are reflexes — wired in, automatic. Classical conditioning is the process by which a signal that merely predicts one of these reflexes comes, all by itself, to set it off. Nothing about the signal mattered at first. What changes it is timing: again and again, the signal arrives just before the thing that already triggers the reflex. The nervous system is a prediction machine, and this is a prediction being built.
Guess before you learn
Ivan Pavlov rang a bell just before feeding his dogs, over and over. Then one day he rang the bell but brought no food. What did the dogs do at the sound of the bell alone?
They salivated to the bell alone (Pavlov, early 1900s). The bell began as a neutral sound — no dog salivates to a bell by nature. But because it had reliably preceded food, it became a signal the body treated as if it were food itself. If you guessed the dogs would ignore it, you assumed the response needs the real thing present. The whole discovery is that it no longer does.
9–12
3–5
Some things make your body react without you trying — food makes you drool, a loud bang makes you jump. Now pair a signal with one of those, again and again: a bell right before food. After enough pairs, the signal alone sets off the reaction. The bell makes the dog drool, with no food in sight.
6–8
Classical conditioning has four parts worth naming. Food is an unconditioned stimulus — it triggers salivation with no training. That salivation is the unconditioned response. A bell starts as a neutral stimulus — it does nothing. Pair bell-then-food enough times and the bell becomes a conditioned stimulus: now it triggers salivation on its own, a conditioned response.
The word conditioned just means learned. Nothing about the bell changed physically. What changed is that it now predicts food, and the body responds to the prediction.
9–12
The four terms lock together. The unconditioned stimulus (US) and its unconditioned response (UR) are the pre-wired reflex — food, salivation. A neutral stimulus paired repeatedly with the US becomes a conditioned stimulus (CS), eliciting a conditioned response (CR) that resembles the UR. The engine is contingency: the CS must reliably predict the US.
Mere co-occurrence is not enough. Rescorla showed that if a stimulus is paired with food no more often than food appears without it, no conditioning develops — the stimulus carries no information. Learning tracks predictive value, not the raw count of pairings.
K–2
A dog smells food and drools — that just happens. Ring a bell, then feed it, again and again. Soon the bell alone makes the dog drool. The bell learned to mean food.
Undergrad
Pavlovian conditioning is not the stamping-in of an association by contiguity but the learning of a predictive relation. The Rescorla–Wagner model formalizes this: associative strength updates in proportion to prediction error, so a CS gains strength only to the extent that the US is still surprising. Once the US is fully predicted, learning halts.
This explains blocking (Kamin): if one cue already predicts the US, a second cue added alongside it gains little strength — there is no prediction error left to drive learning. Conditioning is the nervous system's error-correcting estimate of what predicts what.
Postgrad
Rescorla–Wagner captures much but not all. It is trial-level and silent about the CS–US interval and the timing of the CR; real-time models (temporal-difference learning) extend it and connect directly to dopaminergic reward-prediction-error signals in the midbrain (Schultz). Pavlov's dogs and modern reinforcement learning share one update equation.
Conditioning is also not one process. It dissociates into stimulus–stimulus (cue–outcome) and stimulus–response learning, revealed by post-conditioning US devaluation: devalue the US and an expectancy-based CR weakens while a directly evoked one need not. Whether the CR reflects an expectancy or a reflex depends on the paradigm — a distinction invisible in the bare bell-and-drool picture.
conditioned stimulus
A once-neutral signal that, after reliable pairing with an automatic trigger, comes to set off the response by itself. In Pavlov's study, the bell.
Follow the change in three phases. Before: food (the unconditioned stimulus) makes the dog salivate (the unconditioned response); the bell does nothing. During: the bell rings, then food arrives — paired, again and again, always in that order. After: the bell alone (now the conditioned stimulus) makes the dog salivate (the conditioned response). Order and reliability are everything. If the bell came after the food, or rang at random times unrelated to feeding, no learning would take hold. The signal has to earn its meaning by predicting what comes next.
Why is this true?
Why must the bell come before the food, not after?
Because conditioning builds a prediction, and only a signal that arrives first can predict what follows. A bell sounded after the food announces nothing the dog does not already know, so it never becomes a signal worth responding to.
Learning can also come undone. Ring the bell many times with no food, and the conditioned response fades — the dog stops salivating to a bell that no longer predicts a meal. This fading is extinction. But the learning is not erased, only overwritten. Let a day pass, ring the bell again, and a weaker burst of salivation often returns on its own — spontaneous recovery. The original association is still there, beneath the newer 'the bell means nothing now' learning. This is why a fear you have calmed can resurface after time: extinction adds a lesson, it does not delete the first one.
Classical conditioning is narrow but deep. It builds only automatic, reflex-like responses — salivation, blinking, fear, nausea, a rush of arousal — never deliberate actions. But those responses run under much of daily life: the tightening you feel at the whine of a dentist's drill, the ease that comes with a familiar smell. Wherever a signal reliably precedes something your body reacts to, this quiet learning is at work.
Note
The next folio turns from reflexes to chosen actions — how consequences, not signals, shape what you deliberately do.
Practice — new ink and old, interleaved
1.A word is on the tip of your tongue: you know you know it, but cannot produce it. Which of the three stages has failed?
2.A child gets sick hours after eating an unfamiliar fruit and afterward cannot stand even its smell. What has the smell become?
3.Put these three phases of conditioning in order, from earliest to latest.
- The bell rings just before food, again and again
- The bell alone makes the dog salivate
- The bell means nothing to the dog
4.Match each part of the neuron to its job.
5.In one sentence, explain what extinction is in classical conditioning.