Ivan P. Pavlov · 1901–1936

Pavlov's conditioned reflexes

Can a signal that means nothing to an animal come to trigger a bodily response, just because it reliably comes before food?

  • Institute of Experimental Medicine, St Petersburg (Petrograd, later Leningrad), Russia
  • Replicated

The question

In the 1890s Ivan Pavlov was a physiologist studying digestion, not a psychologist. His lectures on the digestive glands, published in English in 1902, brought together much of the research behind the 1904 Nobel Prize in Physiology or Medicine, which was awarded for his research on the physiology of digestion.

Along the way his team kept noticing a nuisance. Dogs began to salivate before any food reached their mouths: at the sight of food, or of the person who usually fed them. This was called “psychic secretion”. Around 1901 Pavlov decided to stop treating it as noise and to study it. His question: can a signal that means nothing to a dog come to trigger salivation, just because it reliably comes before food? And if so, what rules govern that learning?

How it was done

The fistula. Each dog first had a small operation. The opening of one salivary duct was moved from inside the mouth to the outside of the cheek or under the chin. Once the wound healed, saliva dripped out where it could be collected. A small glass funnel was glued over the opening, and the saliva was counted in drops, measured by volume, or recorded automatically.

The room. Early on, the slightest disturbance (a shadow, a sound in the corridor, an unfamiliar person) could throw off a result. So a special building was put up at the Institute of Experimental Medicine, paid for by a Moscow businessman. It was surrounded by a trench, and each room was divided into a soundproofed compartment for the dog and a separate one for the experimenter. Stimuli were started and responses recorded from outside, electrically or by air pressure.

The signals. The lab used a wide range of stimuli: a metronome, an electric buzzer, tones and organ pipes, a bubbling-water sound, coloured lights and luminous shapes on a screen, a rotating object, touch on a particular spot of skin, warmth and cold. The method was always the same in outline. A neutral signal came first; food (or a weak acid in the mouth) followed a few seconds later; and after a number of pairings the experimenters tested the signal alone and counted the drops.

What came out

A signal starts to work. In one example from Pavlov’s 1927 lectures, a metronome produced saliva after 9 seconds, 11 drops in 45 seconds, with no food in sight. Order mattered a great deal. An electric buzzer sounded just before food worked after a single pairing. The loud ring of an electric bell sounded 5 to 10 seconds after food never became a signal, even after 374 pairings.

Extinction. When the metronome was sounded again and again without food, the flow of saliva shrank over the repetitions. Pavlov argued this was not erasure but an active holding back, which he called inhibition.

Spontaneous recovery. The evidence for that view: after a pause, the faded response came back on its own. In one demonstration, 23 minutes after partial extinction, the delay before salivation dropped from 13 to 5 seconds and the output doubled from 3 to 6 drops.

Generalisation and discrimination. A dog trained to one tone also salivated to nearby tones, less so the further they were from the original. Touching nearby patches of skin worked too, more weakly with distance. But if one stimulus was always followed by food and a similar one never was, dogs learned to tell them apart: a circle from a square, clockwise from anticlockwise rotation, even tones very close in pitch.

Experimental neurosis. The best-known example is an experiment by Dr. Shenger-Krestovnikova. A dog learned that a luminous circle meant food and an ellipse meant none. The ellipse was then made rounder step by step, from 2:1 to 9:8. At that point discrimination failed, and after three weeks it got worse instead of better. The dog, calm until then, began squealing and struggling in the stand, bit at and pulled off the equipment, and barked fiercely whenever it was led into the room. Even the easy 2:1 distinction was lost. Other co-workers produced similar breakdowns with very long delays before food, or by turning a strong electric shock to the skin into a food signal and then applying it to more and more places. After the 1924 Petrograd flood, some dogs that had to be rescued from flooded kennels lost their conditioned reflexes for weeks or months.

What really happened (replications and critiques)

The bell: rarer than the legend, but real. In popular retellings, Pavlov rang a bell and the dog drooled. His biographer Daniel Todes, who worked through the lab’s archives, says the image is largely wrong: over thousands of experiments the lab used bells only rarely, because a bell could not be controlled precisely. Todes links the myth to translation, since the Russian word zvonok can mean a buzzer as well as a bell. Michael Specter’s 2014 New Yorker essay quoted Todes’s stronger version: that Pavlov never trained a dog to salivate to a bell.

Not everyone accepts the strong version. In 1997 psychologist Roger K. Thomas pointed out that Pavlov himself mentioned a bell in his 1906 Huxley lecture (as a strong stimulus that conditioned slowly), wrote in 1923 that conditioned reflexes to electric bells could be formed, and that a 1920s film made in Pavlov’s lab was described in Time (1928) as showing dogs that “dripped saliva at the sound of a bell”. Pavlov’s own 1927 lectures, in Anrep’s translation, mention an “electric bell”, described as buzzing, and in at least one study it served as a working signal. The fair summary: the metronome and the buzzer were the workhorses, and bells were real but marginal.

“Conditioned” or “conditional”? Pavlov’s Russian term is closer to conditional reflex: a reflex that depends on conditions. Todes argues that “conditioned” is a mistranslation that changes the meaning of Pavlov’s idea. Early American reporters Yerkes and Morgulis wrote in 1909 that they used “conditioned” because they believed Pavlov had accepted it.

The effect is robust; the explanation changed. Classical conditioning has been reproduced so often, in so many species, that it is a basic tool of research. Two studies from the 1960s changed how it is understood:

In 1988 Rescorla summed this up: textbook conditioning, a neutral stimulus simply borrowing the response of a meaningful one, is outdated. Modern theory treats it as learning the relationships between events, often in just a few trials.

Ethics

The conditioning work rested on surgery. The salivary fistula was a fairly minor operation, and dogs that had it lived longer than those used in the earlier operations. Pavlov’s earlier digestive research went much further: openings in the oesophagus and along the digestive tract so secretions could be collected from living dogs. According to Todes, many dogs died while these techniques were being developed. Sessions could last eight to ten hours, and some, as described above, were designed to make dogs break down.

The picture is not one-sided. Co-workers named their dogs and recorded their personalities in detail, and in 1935 Pavlov had a monument to the laboratory dog erected in the institute’s grounds. Its inscription expresses the view that dogs could be sacrificed to science, but without unnecessary cruelty.

Today, animal research in many countries is governed by the “3Rs” proposed by Russell and Burch in 1959: replace animals where possible, reduce their number, and refine procedures to minimise suffering. Under such rules, basic conditioning with a minor operation might still be approved. Procedures meant to induce lasting distress would face very strict scrutiny and are unlikely to be approved in their original form.

What it means today

Conditioning is quiet and everywhere: the jolt of a notification sound, a mood triggered by a smell, the nausea some people feel at the sight of a food that once made them ill. Rescorla’s 1988 review noted that the same mechanisms are studied in drug tolerance, stress-induced pain relief and the immune system, and that behaviour therapy grew out of this tradition.

A few practical thoughts follow, though they go beyond what Pavlov tested:

Would you have passed?

Coming soon. A series of sounds and shapes appears on screen, some followed by a reward and some not. Learn which ones predict it, then try telling apart signals that grow more and more alike, as the circle and the ellipse did. One honest caveat: you know the rules are a game, and nothing your body does is being measured, so this shows the logic of conditioning, not the reflex itself.

Questions people ask

Did Pavlov really use a bell?

Rarely, and this is still debated. His lab mostly used metronomes, buzzers, tones, lights and touch. His biographer Daniel Todes says the famous bell comes largely from translation and retelling. Psychologist Roger K. Thomas found that Pavlov himself reported bell stimuli in a 1906 lecture and a 1923 paper, and that a 1928 Time report described a film from his lab showing dogs responding to a bell.

What did Pavlov win the Nobel Prize for?

For digestion. The 1904 Nobel Prize in Physiology or Medicine recognised his work on the physiology of digestion. He had already begun his work on conditioned reflexes and mentioned it in his Nobel address, but it was not part of the prize.

What are extinction and spontaneous recovery?

If the signal keeps coming without food, the response fades: that is extinction. Pavlov showed it was not forgetting, because after a rest the response returned on its own, partly or fully. That return is spontaneous recovery.

Has Pavlov's conditioning been replicated?

Yes, countless times, in many species and with many kinds of signals. What changed is the explanation. Work from the 1960s showed that animals learn how well one event predicts another, and that some pairings, such as taste and illness, are learned far more easily than others.

Sources

  1. Pavlov, I. P. (1927). Conditioned reflexes: An investigation of the physiological activity of the cerebral cortex (G. V. Anrep, Trans.). psychclassics.yorku.ca/Pavlov/
  2. Yerkes, R. M., & Morgulis, S. (1909). The method of Pawlow in animal psychology. Psychological Bulletin, 6, 257–273. psychclassics.yorku.ca/Yerkes/pavlov.htm
  3. Nobel Foundation. The Nobel Prize in Physiology or Medicine 1904: Ivan Pavlov. www.nobelprize.org/prizes/medicine/1904/summary/
  4. Pavlov, I. P. (1902). The work of the digestive glands (W. H. Thompson, Trans.). Charles Griffin & Company.
  5. Thomas, R. K. (1997). Correcting some Pavloviana regarding "Pavlov's bell" and Pavlov's "mugging." American Journal of Psychology, 110(1), 115–125.
  6. Todes, D. P. (2014). Ivan Pavlov: A Russian life in science. Oxford University Press.
  7. Specter, M. (2014, November 24). Drool: Ivan Pavlov's real quest. The New Yorker. www.newyorker.com/magazine/2014/11/24/drool
  8. Adams, M. (2020, April 27). The kingdom of dogs. The Psychologist (British Psychological Society). www.bps.org.uk/psychologist/kingdom-dogs
  9. Golikov, Yu. P., & Sysuev, V. M. (2015). Something unknown about well-known monument 'Pavlov's Dog'. Medical Academic Journal, 15(3), 85–100. doi:10.17816/MAJ15385-100
  10. Rescorla, R. A. (1968). Probability of shock in the presence and absence of CS in fear conditioning. Journal of Comparative and Physiological Psychology, 66(1), 1–5.
  11. Garcia, J., & Koelling, R. A. (1966). Relation of cue to consequence in avoidance learning. Psychonomic Science, 4, 123–124. doi:10.3758/BF03342209
  12. Rescorla, R. A. (1988). Pavlovian conditioning: It's not what you think it is. American Psychologist, 43(3), 151–160.
  13. Russell, W. M. S., & Burch, R. L. (1959). The principles of humane experimental technique.