Iwan Pawlow

The starting question was physiological, not psychological: why does a dog begin to salivate before it gets food? For Pavlov this was no curiosity but a disturbance in his digestion measurements — until, around the turn of the century, he decided to make the disturbance itself the subject. The decision was riskier than it looks in hindsight: a Nobel candidate of digestive physiology turned to a phenomenon his colleagues considered psychology — and psychology counted as a term of abuse in his laboratory. Pavlov forbade his co-workers mentalistic expressions; whoever said the dog “expected” the food paid, according to the anecdote, a small fine into the laboratory kitty. Whether the kitty existed is transmission; the language ban is not.

The scene that carried everything was a building. In the “Tower of Silence,” a laboratory erected for the purpose in St. Petersburg, the dogs stood in sound-damped chambers, harnessed on a stand, with a surgically created cheek fistula leading the saliva of one gland outward — into graduated tubes whose drops a co-worker counted in the next room. The experimenter was separated from the animal by walls, windows and pneumatic mechanics, so that no smell, no footstep, no glance entered as an uncontrolled stimulus. That was the real innovation: not the idea that animals learn — every dog owner had that — but an apparatus that translated learning into drops per minute.

The design is so simple it fits into every textbook. A stimulus that triggers no salivation on its own — the neutral stimulus — is presented repeatedly just before the food. After several trials it triggers the secretion alone. The neutral stimulus has become a conditioned one, the salivation a conditioned response.

What Pavlov described beyond this is the part textbooks cut: extinction, when the conditioned stimulus comes several times without food; spontaneous recovery, when the extinguished response returns by itself after a pause; generalisation to similar stimuli; and differentiation, when the animal learns to tell two similar stimuli apart. Extinction unlearns nothing — it layers something over. That is the most clinically consequential of his observations: whoever has “unlearned” a fear still carries it, overwritten, not erased — which is why it can return in a new context, under stress, or with time. Modern exposure research continues spelling at exactly this point.

From differentiation comes the darkest chapter of the series. When one of Pavlov’s co-workers taught a dog to distinguish a circle from an oval and then brought the oval closer to the circle step by step, the animal broke down: it whined, bit at the apparatus, refused the trial — the famous “experimental neurosis.” Pavlov, by then past eighty, took this as a bridge to psychiatry and spent his last years in clinics, trying to understand patients with his concepts. This extension has weathered time worse than the basic series; what remains is the insight that an organism can break on an unsolvable discrimination task.

What was measured, incidentally, was never “the dog” but always a particular one: the records carry the animals by name — Baikal, Zolotisty, Krasavietz — and some series hang recognisably on the peculiarities of a single experienced subject. The individuality the theory did not provide for stands in the footnotes of its own data tables.

And then there is the bell. In the textbook version Pavlov rings a bell and the dog salivates. In the laboratory records it barely occurs. Pavlov worked with metronomes, buzzers, tones, light signals and a rotating object — stimuli that can be dosed and repeated exactly. Daniel Todes, who worked through the Russian primary sources, records: Pavlov practically never used a bell. It is an invention of the reception, born on the way through translations and summaries.

A second inaccuracy sits in the English technical term itself. The Russian uslovny refleks means “conditional” — provisional, tied to a condition. The early English translation made it conditioned — completed rather than conditional. The difference is small and shifted the reception nonetheless: a provisional, always revocable connection became, linguistically, a finished “conditioned” animal — and the provisionality Pavlov stressed became the image of an imprint for life, which Watson and the psychology of advertising then took over.

The reception ran faster than the translations anyway. Watson made the conditioned reflex the foundation of behaviourism in 1913 without reading Russian; American learning psychology built on secondary sources for half a century. In the Soviet Union, Pavlov was elevated to state doctrine after his death — the “Pavlovian Session” of the academies in 1950 formally committed the sciences to his teaching and thereby damaged it more thoroughly than any critic could have, for a doctrine can no longer be tested, only cited.

The second half of the twentieth century then reinterpreted the reflex without dethroning it. Robert Rescorla showed in the 1960s that pairing alone is not enough: the conditioned stimulus must predict the food, not merely accompany it — an animal learns contingency, not contiguity. The Rescorla-Wagner model of 1972 turned this into an equation describing learning as the correction of prediction errors; and when neurophysiology in the 1990s found dopamine neurons signalling exactly such prediction errors, Pavlov’s preparation had arrived at the centre of modern brain science. “The dog salivates at the tone” has become "the brain is a prediction machine" — a reinterpretation Pavlov would presumably have liked, because it replaces mentalistic words with a rule of calculation.

For the human being Pavlov reserved, in his last years, an extension of his own: the second signal system. Where the animal reacts to signals of the world, the human reacts additionally to signals of signals — to words standing for stimuli. With him the idea remained a sketch; in Soviet psychology after 1950 it became a compulsory formula with which nearly everything human could be retroactively Pavlovised. That too belongs in the file: a concept can perish of its own decree.

How far from arbitrary the linkage is was shown in the 1960s by a Pavlovian in spirit: John Garcia let rats drink flavoured water and made them nauseous hours later by irradiation — the animals henceforth avoided the taste, but not the light or the sound presented at the same time; with electric shocks it was exactly the other way round. Taste links to nausea across hours, light to pain only across seconds — evolution has pre-sorted what can belong together. The journals at first rejected the findings; one reviewer, by tradition, compared their probability to bird droppings in a cuckoo clock. Today taste aversion stands in every textbook, as the place where biology reached into the reflex scheme’s workings.

What stands today: the thing itself, to an extent few findings in the field reach. Classical conditioning is reproduced across species, laboratories and decades — from the sea slug Aplysia, in which Eric Kandel found its cellular basis, to human fear in the MRI scanner. Its neural foundations are described down to the level of individual nuclei, and it carries the most effective treatment psychotherapy knows for anxiety disorders: exposure therapy. What does not stand is Pavlov’s own interpretation that the brain works by waves of excitation and inhibition running across the cortex — the cortex is not even necessary for the salivary reflex. It is the rare fate of a classic that its finding has outlived the ages and its theory not a single one.