Last updated: Fri, Feb 7, 2025
Pavlovian conditioning is a type of learning that associates an innate response with a learned stimulus. Pavlovian conditioning is also called respondent or sometimes classical conditioning. The name "pavlovian" comes from Ivan Pavlov, a Russian researcher who noticed the phenomenon while researching the digestive process using dogs early in the twentieth century.
The pavlovian model of learning begins with an innate or built-in response to a biologically-significant stimulus. As an example, if an object flies at your eye (the stimulus) you will blink (the response). The original or built-in stimulus and response are called "unconditioned." The flying object is an unconditioned stimulus for the protective eye blink, its unconditioned response. Another unconditioned stimulus/response pair is the withdrawal of a limb when it experiences pain (see Spinal Reflexes). A third unconditioned stimulus/response pair is salivating (the response) in the presence of appetizing food (the stimulus). The unconditioned stimulus is commonly abbreviated as "US" and the unconditioned responsed as "UR."
| US | UR |
|---|---|
| Object flying at the eye | Eyelid closes |
| Sudden pain in an extremity | Withdrawal of the extremity |
| Presence of food | Salivating |
When the unconditioned stimulus (US) is consistently preceded by some other event or condition in the environment, the organism learns to associate the other event or condition with the US, and eventually it will blink (or take some other appropriate action) when the other event or condition occurs. This new event or condition is then called a conditioned stimulus or CS. Let's imagine that an experimenter rings a bell immediately before launching some projectile toward the eye of an experimental rat, and the rat learns to close his eyelids when the bell rings. In this case, the conditioned stimulus (CS) is the ringing of a bell. The rat's response is now a conditioned response (CR), since the rat has learned to react to a conditioned stimulus.
| Before conditioning | |
|---|---|
| US: Projectile toward the eye | UR: Eyelid closes |
| After conditioning | |
| CS: Bell rings | CR: Eyelid closes |
In this scenario the conditioned response was the same as the unconditioned (innate) response, but the conditioned response can be some other response that is appropriate. For example, if the projectile followed the bell after some delay, and if the projectile always came from the same direction, the experimental rat might learn to turn his head instead of closing his eyes.
| Before conditioning | |
|---|---|
| US: Projectile toward the eye | UR: Eyelid closes |
| After conditioning | |
| CS: Bell rings | CR: Head turns away |
Pavlovian learning can occur quickly or slowly. The rate of learning is faster when the unconditioned stimulus and response are more important to the organism. Severe pain or poisoning can result in pavlovian learning after only one exposure. Learning is faster when the conditioned stimulus is easier to detect, when the unconditioned stimulus (the projectile) occurs soon after the conditioned stimulus (the bell), and when the unconditioned stimulus always folows the conditioned stimlus.
Pavlovian learning is not necessarily permanent. The association can be extinguished, for example, if the conditioned stimulus (the bell) occurs often enough without the unconditioned stimulus.
Pavlovian conditioning exists to increase the fitness of the organism by allowing it to predict biologically-significant events from environmental cues. It has been found in organisms as simple as certain sea slugs and even in garden peas. Under laboratory conditions the subject (often a rodent or a bird) can be placed in a controlled situation and exposed to a single controlled conditioned stimulus to determine the rate of learning, the rate of extinction, and so on. More-complex situations can also be contrived. For example, conditioned stimuli can be chained together: CS1 can be associated with the US, then CS2 can be associated with CS1. As a result, the subject may learn to perform the conditioned response when CS2 occurs.
As another example, the subject may first be taught to associate CS1 with the US. It can then be further trained by presenting both CS1 and a new stimulus, CS2, at the same time, before the US. Unlike the previous example, CS2 will not evoke the conditioned response as readily as CS1 does. This phenomenon is called "blocking."
Natural (non-laboratory) conditions can be much more complex than these simple phenomena, and hence more difficult to predict. Successful learning requires the organism to accurately learn precisely which combinations or sequences of cues in which situations predict the biologically-significant event, and to distinguish combinations which are more likely to predict from those that are less likely to. An organism that learns to perform the unconditioned response when it is not necessary will pay a cost in fitness, depending on the particular circumstance. For example, an organism that empties its stomach mistakenly will lose a certain amount of calories and time. An organism that fails to predict the unconditioned stimulus when it could have, based on environmental information, will pay another cost. For example, an organism that fails to detect a predator. Organisms can and do adopt general strategies to help with this problem. It may, for example, work out well to always assume the worst. Michael Schermer's idea of patternicity (The Importance of Models) may be an example of this. Notwithstanding, there is a great benefit to the organism in learning to make accurate predictions.
Although pavlovian learning has been studied intensively for nearly a hundred years, no single model has yet been found that successfully explains all of the studied laboratory scenarios. This may be in part because there is more than one mechanism of pavlovian conditioning. As an example of this, compare the flexor reflex (Flexor and Withdrawal Reflexes) with the eye blink response. In the first case, the flexor and withdrawal reflexes, we've seen that one or two synapses in the spinal cord are sufficient to cause the unconditioned response. In the eye blink response, an array of synapses is required to detect motion of an object toward the eyeball. Because the pavlovian learning model is a black box model, it incorporates no knowledge of the actual physiological mechanism of learning.
The Rescorla-Wagner equation, formulated in 1972, is the classic model used to organize the analysis of pavlovian conditioning experiments. It has been successful in a number of ways, but has been superceded by newer, more-complex models that explain additional phenomena. Nevertheless, it reflects the perspective of pavlovian research. It predicts how much learning will occur in one exposure to the conditioned stimulus, and it calculates that amount based on how strongly the CS is already associated with the US, the "salience" of the CS, the strength of the US, and the association between the US and other stimuli that may be present in the experimental tableau.
Although pavlovian learning has been demonstrated in humans, relatively less is known about when and how our possession of a 100-billion neuron central nervous system affects this process.
When a conditioned stimulus is associated with an unconditioned response, the conditioned response is "biologically appropriate" but not necessarily the same as the unconditioned response. I offered the example of a rat that might learn to turn his head rather than blink when a projectile headed for the eye is associated with a bell. The pavlovian conditioning model provides no guidance that allows predicting precisely what the conditioned response will be to a particular combination of cues.
The name "cognitive-behavioral therapy" reminds us that CBT is at least partiallly based on behavioral theory. As discussed earlier, behaviorism is an approach to psychological understanding founded on the observable behaviors of actual subjects, human or otherwise. It represented a rejection of the "mentalism" or introspection that was used in other approaches, notably by Sigmund Freud and his school. Introspection relied heavily on the researchers' perceptions of what went on internally, and so couldn't be tested and verified.
Behavioralists in the twentieth century became interested in learning processes, which could be studied in controlled environments. Learning theory developed from this effort, and attempted to discover how animals adapt to their environments. The behavioral research upon which learning theory is built is all "black box" research. It studies the organism from the outside and develops generalizations based on observable behaviors. This was well-suited to the research technologies that were available to researchers around the middle of that century. Although there were electroencephalograms (EEGs), there were no MRIs. The neurotransmitters were only beginning to be identified in the 1950s. The black-box methods of learning theory matched the technology available.
Learning, as the word is understood in behavioral psychology, is rather different from what laypeople usually think of as learning, and so it's necessary to review the psychological concepts in order to understand the CBT theory around learning and pain.1
To begin, short-term memory and long-term memory are two different capabilities based on two different processes. Short-term memory is our ability to hold a small number of facts in consciousness for a short length of time as we work out a problem. This is how, for example, we can remember a telephone number while we dial it. Short-term memory is effortful and limited to "seven plus or minus two items."
Learning theory is concerned with and based upon long-term memory, which can store an unknown but practically unlimited amount of information for as long, potentially, as the life of the subject.
Long-term memory has two forms, explicit and implicit. Explicit memory is also called declarative memory. Explicit memory shows itself in the retrieval of prior experiences and factual knowledge about people, places, and things. Explicit memory is flexible. It allows experiences and facts to be associated with others in complex ways, to be consciously recalled and manipulated, and allows disparate pieces of information to be recalled together. Much of what we're taught from textbooks and lectures is explicit knowledge.
Explicit memory can be divided into two forms. Episodic memory is the memory of personal experiences. Semantic memory is the memory of facts and concepts, the sort of things we learn in school. The learning and use of explicit memories involves four neural processes: encoding, storage, consolidation, and retrieval. Each of these processes influences the operation of explicit memory.
Explicit memory is "declarative, explicit, cognitive, and conscious."2
Implicit memory is also called nondeclarative or procedural learning. Implicit memory shows itself in the performance of tasks, and is activated automatically. Skill acquisition, habits, conditioning, and priming are all forms of impliciy memory. The behaviors acquired through Implicit learning are tightly associated with the conditions under which learning happened.
Implicit memory can be nonassociative or associative. Nonassociative learning is learning about a single stimulus or condition. Associative learning is about the relation between two stimuli, or between a behavior and a stimulus/state. Implicit learning occurs gradually through many repetitions. It uses different neural processes than those used in explicit learning.
Implicit memory is "procedural, implicit, and unconscious."3
Nonassociative learning is about a single stimulus type or a single state of the organism. Habituation is a reduction in response as a stimulus is encountered repeatedly. The organism's response to a pleasant or non-harmful stimulus tends to decrease when the stimulus is repeatedly encountered. This is habituation. Habituation reduces the degree to which the organism "orients toward" the stimulus, and so allows the organism to attend to other stimuli. Dishabituation is the reversal of habituation by a sensitizing stimulus.
Sensitization, on the other hand, is an enhanced response. As we saw earlier, for example, the response to a painful stimulus will generally increase if the organism is again subjected to that stimulus.
Associative learning, in its basics, is the learning of associations between stimuli and biologically significant outcomes (classical conditioning) or between behaviors and biologically significant outcomes (operant conditioning). Associative learning is well-suited to learning about probabilistic or variable relationships between events in the world. When event A often, but not always, precedes event B, it is more useful to learn the relationship ("event A often precedes event B") than to explicitly remember examples of event A. These relationships take numerous repetitions to learn. Since the relation between A and B is actually probabilistic, it can easily be learned in error. Additional experiences will in general lead to more accurate estimates of the probabilities. In associative learning, the timing (both sequence and time lag) of two stimuli or events is critical. It is this relationship that is associative.
Classical conditioning was first documented by Ivan Pavlov in the early 20th century. He trained laboratory dogs to salivate in response to a sound by repeatedly following the sound with feeding. In this example the food, which has innate biologic importance, is called the "unconditioned stimulus," and in shorthand is designated as the "US." The stimulus that the subject is trained to respond to, in this case a bell or buzzer, is called the "conditioned stimulus," and is designated as "CS." In Pavlov’s experiments, the US was food or a shock, which are believed to be naturally rewarding or punishing. Pavlov’s CS was an audio signal, a bell or such. The conditioned response, CR, was salivating or guarding the site of shock, in response to the audio signal. Classical conditioning (also referred to as respondent conditioning) then is a way that the subject learns to “predict” events, although no conscious cognition is implied. Animals that are very simple neurally, such as sea slugs, show classical conditioning in the lab and in their normal environments.
While it has been believed that classical conditioning is the simple result of US following CS after a brief interval, more study has disproven this simple view. The relationship between CS and US must also be consistent. The more often the subject experiences the CS without the US following, the less conditioning occurs.
Operant conditioning involves learning associations between behavior and results, rewarding or punishing. A behavior that results in reward will happen more often, behavior that results in punishment will happen less often. B. F. Skinner was the major explorer of operant conditioning in the mid-20th century. A paradigmatic example of operant conditioning might occur within a "Skinner box," a cage equipped with a lever or button and a device for dispensing food, and under the control of the trainer/experimenter. The pigeon inside the box might press the lever, either because of previous learning, or through play and random activity. If the pigeon promptly receives a positive reinforcer (food) after pressing the lever, it will begin to press the lever more often than before. The pigeon is said to have learned to press the lever for food, based entirely on behavioral observation. Logically, operant conditioning (also called instrumental conditioning) is the learning of a predictive relation between a behavior and an outcome significant to the pigeon or to a human.
Priming is a phenomenon of implicit memory. It is known in two forms. Conceptual priming makes relevant semantic knowledge more readily available. It is believed to work because the primed information has been utilized in similar situations before. Thus an implicit process can assist in the use of explicit learning. Perceptual priming works within a sensory modality (vision, hearing, etc.) and operates through modules of the cortex that resolve form and structure of objects as known in that modality.
In the discussion on this page we've talked about sensitization as a learning process. In that context, sensitization is an increased response to a repeated stimulus, and is nonassociative learning. There's another usage of "sensitization" in pain science which doesn't necessarily involve repetition. This other "sensitization" can be a temporary response to a single event. After an organism receives an unpleasant or an intense stimulus, its response to a wide variety of stimuli will become stronger. If you're startled by a loud noise, for example, your response to a wide variety of events may be increased for a while. This can happen whether or not the sensitizing stimulus is regularly associated with some other stimulus. This other type of sensitization is related to learning, even if it occurs independently of the stimulus of interest, because this type of sensitization is an increase in physiologic and neural arousal, and the state of arousal affects learning.
"The background arousal of the organism determines the speed of habituation, as well as other factors such as stimulus intensity, stimulus duration, and the frequency of presentation."4 Low levels of arousal promote desensitization (that is, the unlearning of learned sensitization).
You may recall that earlier (see The Psychological Perspective On Pain) we described pain sensitization as a physiological process that follows injury and/or repeated irritation. In fact there are several known forms of pain sensitization, including inflammatory sensitization, central sensitization, and others. It is important, as well as tricky, to distinguish these various uses of the word while attempting to evaluate the state of pain science, especially in discussions of pain psychology. What is the relation between physiological sensitization and learned sensitization?