Ivan Pavlov, a Nobel laureate renowned for his work on digestion, is paradoxically more celebrated for an entirely different domain: his experiments involving a dog, a bell, and saliva. The phrase “Pavlov’s dog” is widely recognized, yet the profound implications of this discovery often remain underexplored. Pavlov’s groundbreaking research provides a fundamental framework for understanding a spectrum of human behaviors, from the unease some feel when faced with a crowded bus to the aversion triggered by the sound of a morning alarm, and even the instant dislike for a food tasted only once. Classical conditioning, also known as Pavlovian conditioning, stands as a cornerstone of our understanding of how we learn and interact with the world. More than just a learning theory, it arguably shapes our very identities. By grasping the principles of Classical Learning, you begin to see how deeply ingrained preferences – for music, fashion, even political ideologies – might stem from the same learning mechanisms that cause a dog to salivate at the mere sound of a bell.
Image: The Pavlov in All of Us: Is your dog begging for food a learned behavior reinforced by table scraps? [Image: David Mease, https://goo.gl/R9cQV7, CC BY-NC 2.0, https://goo.gl/FIlc2e]
At the dawn of the 20th century, scientists investigating animal and human behavior began to recognize the significance of two fundamental forms of learning. The first, meticulously studied by Russian physiologist Ivan Pavlov, is termed classical learning or Pavlovian conditioning. In his seminal experiment, Pavlov consistently paired the ringing of a bell with the presentation of food to a dog. Through repeated pairings, the dog began to associate the bell with the imminent arrival of food, leading to salivation in anticipation of the meal, even before the food was presented. This phenomenon has been consistently replicated in laboratory settings, utilizing a diverse array of stimuli – from auditory tones and visual lights to tastes and environmental contexts – paired with various significant events beyond food, including drugs, electric shocks, and induced illness.
Modern understanding posits that this very learning process is at play in numerous human experiences. It explains how individuals form associations between drug use and the environment where it occurs, link stimuli like vacation symbols (e.g., a beach towel) with emotional responses (like joy), or connect food flavors with negative experiences like food poisoning. Despite its apparent simplicity and historical roots, classical learning remains a vibrant field of study for two key reasons. Firstly, it provides a clear, testable model for associative learning, serving as a foundation for exploring more intricate behaviors. Secondly, given its continuous and pervasive nature in our daily lives, classical learning profoundly influences both typical and atypical human behavior, making its study essential for understanding both normalcy and disorders.
In essence, classical learning emerges whenever neutral stimuli become linked with psychologically significant occurrences. Consider food poisoning: while eating fish might typically be a neutral event, if it leads to illness, the once-neutral stimulus (fish) becomes associated with the significant event of sickness. To describe these pairings universally, specific terms are employed.
In Pavlov’s experiment, the dog food is identified as the unconditioned stimulus (US) because it naturally and automatically triggers an unconditioned response (UR). This means no prior training is required for the stimulus to elicit this instinctive reaction. In Pavlov’s case, food (US) inherently causes the dog to drool (UR). Other examples include loud noises (US) causing a startle response (UR) or a warm shower (US) producing pleasure (UR).
Conversely, a conditioned stimulus elicits a conditioned response. A conditioned stimulus (CS) is initially neutral, holding no inherent meaning for an organism until it’s paired with something of significance. In Pavlov’s setup, the bell is the conditioned stimulus. Initially, the bell sound (CS) is meaningless to the dog until it is repeatedly paired with food (US). However, after these repeated pairings, the dog begins to drool at the sound of the bell alone. This drooling in response to the bell is the conditioned response (CR). While potentially confusing, the conditioned response is often very similar to the unconditioned response. It is termed “conditioned” because its occurrence is conditional upon or dependent on its association with the conditioned stimulus (e.g., the bell). Think about feeling hungry upon seeing a fast-food restaurant logo. The actual consumption of food (US) naturally causes salivation (UR), but merely seeing the logo (CS) can trigger the same physiological response (CR).
Another familiar example is your alarm clock. For most, waking up early is unpleasant. Here, waking up early (US) generates a natural feeling of grumpiness (UR). Instead of waking naturally, you likely use an alarm with a specific tone. Initially, this tone is neutral (no prior meaning). However, using it daily to wake up “pairs” the tone (CS) with morning grumpiness (UR). After enough pairings, the tone (CS) alone will trigger grumpiness (CR). The link between the unconditioned stimulus (US; waking early) and the conditioned stimulus (CS; the tone) becomes so robust that the unconditioned response (UR; grumpiness) transforms into a conditioned response (CR; hearing the tone anytime might induce grumpiness). Contemporary classical learning studies use a wide variety of CSs and USs and measure diverse conditioned responses.
Image: Rewards and Conditioning: Remember the childhood promise, “Be good in the supermarket, and you’ll get a treat on the way out”? This is a form of conditioning. [Image: Oliver Hammond, https://goo.gl/xFKiZL, CC BY-NC-SA 2.0, https://goo.gl/Toc0ZF]
While classical learning offers a powerful framework for understanding various forms of learning, a second type, operant conditioning, also plays a crucial role. First studied by Edward Thorndike and later expanded upon by B.F. Skinner, operant conditioning focuses on how a behavior (rather than a stimulus) becomes associated with a significant event. In a classic example, a rat in a “Skinner box” learns to press a lever to receive food. There is no innate connection between lever pressing and food for the rat; this association must be learned. Initially, the rat might explore the cage randomly. Eventually, in its exploration, it accidentally presses the lever, and a food pellet is dispensed. This voluntary action is termed an operant behavior because it “operates” on the environment, representing an action initiated by the animal.
Once the rat recognizes the connection between lever pressing and food, the lever-pressing behavior is reinforced. The food pellets act as reinforcers, strengthening the rat’s inclination to interact with its environment in this specific way. Consider a street-racing video game. As you repeatedly navigate a city course, you explore different routes to the finish line. Discovering a shortcut that significantly improves your time is learning through operant conditioning. Your actions within the environment (operant responses) – trying different streets – led to a positive reinforcement (finding the fastest route). Now, having learned this optimal route, you will repeat this sequence of driving behaviors to achieve the reward of a quicker finish, much like the rat consistently presses the lever.
Operant conditioning research investigates how the consequences of a behavior impact the likelihood of its recurrence. The effect of the rat’s lever pressing (receiving food) increases the probability of future lever pressing. Thorndike’s Law of Effect states that behaviors with positive or satisfying outcomes are more likely to be repeated, while behaviors with negative or unpleasant consequences are less likely to be repeated. Outcomes that increase behavior are reinforcers, while those that decrease behavior are punishers.
A common example of operant conditioning is striving for good grades. For students, good grades are often a positive reinforcer, producing a satisfying emotional response. To achieve this reward (like the rat getting food), students must modify their behavior. A student might learn that participating in class earns participation points (a reinforcer), leading to increased class participation. Conversely, they might learn that irrelevant comments cost points, thus shaping their classroom behavior through reinforcement and punishment.
A key distinction of operant conditioning is its focus on how consequences influence “voluntary” behavior. The rat’s lever pressing is voluntary; it can choose to press or not. Classical learning, in contrast, involves “involuntary” behaviors (e.g., a dog doesn’t consciously decide to drool). While the rat actively engages to gain a reward, Pavlov’s dog is a passive recipient. Operant conditioning highlights that voluntary behavior is strongly shaped by its consequences.
Image: Classical vs. Operant Conditioning: This illustration contrasts the core elements of classical and operant conditioning. [Image courtesy of Bernard W. Balleine]
The illustration above summarizes the fundamental differences between classical and instrumental conditioning. While distinct in several ways, modern perspectives emphasize that their primary difference lies in what is learned. In classical learning, an animal learns to associate a stimulus with a significant event. In operant conditioning, the association is between a behavior and a significant event. Another distinction is that in classical conditioning, the response (like salivation) is elicited by a preceding stimulus, whereas operant responses are not elicited by a specific stimulus but are emitted, further emphasizing their voluntary nature.
Understanding both classical and operant conditioning provides psychologists with vital tools for analyzing learning and behavior beyond the laboratory. These forms of learning are continuous in our lives. As noted by Spreat & Spreat (1982), “much like the laws of gravity, the laws of learning are always in effect.”
Key Insights into Classical Learning
The Far-Reaching Effects of Classical Conditioning on Behavior
A classical conditioned stimulus (CS), such as Pavlov’s bell, doesn’t just trigger a simple, singular reflex. While Pavlov focused on salivation, his bell likely initiated a complex system of responses preparing the organism for the anticipated unconditioned stimulus (US), food (Timberlake, 2001). Beyond salivation, food-related CSs also trigger the release of gastric acid, pancreatic enzymes, and insulin, all crucial for digestion. The CS also evokes approach behaviors and a state of heightened excitement. Remarkably, a food CS can even prompt animals with full stomachs to eat more if food is available. In modern society, food CSs are ubiquitous, influencing humans to eat or feel hunger in response to cues like the crinkling of a chip bag, familiar logos like Coca-Cola, or even the comfort of the couch in front of the TV.
Classical learning also influences other aspects of eating. Flavors associated with nutrients like sugar or fat can become preferred without conscious awareness. For instance, protein acts as a US, naturally increasing cravings upon consumption (UR). Since meat is protein-rich, its flavor becomes a CS, signaling incoming protein and perpetuating meat cravings (now a CR).
Conversely, flavors linked to illness or stomach pain become disliked and avoided. Taste aversion conditioning exemplifies this: someone sickened by tequila may develop a strong aversion to its taste and smell. This flavor-consequence association is vital for animals encountering new foods. It’s also clinically relevant. Chemotherapy drugs often cause nausea, leading patients to develop aversions to foods eaten before treatment or even to the chemotherapy clinic itself (Bernstein, 1991; Scalera & Bavieri, 2009).
Classical learning extends beyond food-related contexts. If a tone precedes a mild shock to a rat’s feet, the tone will, after a few pairings, elicit fear or anxiety. Fear conditioning of this type plays a role in human anxiety disorders, such as phobias and panic disorders, where cues (like enclosed spaces or shopping malls) become associated with panic or trauma (Mineka & Zinbarg, 2006). Here, the CS triggers an emotional response rather than a physical one like drooling.
Drug ingestion also involves classical learning. Drugs become associated with environmental cues present during use (e.g., rooms, smells, paraphernalia). A specific smell linked to drug sensation can later trigger responses (physical and emotional) related to drug intake. Intriguingly, drug cues can elicit conditioned compensatory responses (Siegel, 1989), which are physiological responses that counteract the drug’s expected effects. For example, morphine suppresses pain, but for regular users, cues signaling morphine’s arrival can paradoxically increase pain sensitivity. The body anticipates pain relief and becomes more sensitive in preparation. These compensatory responses can reduce the drug’s actual impact, as the body is already adjusting.
Conditioned compensatory responses have significant implications. Drug tolerance is often highest in familiar environments with associated cues because these cues elicit compensatory responses. Overdoses are often not due to increased dosage but to drug use in unfamiliar settings lacking these familiar cues, which would normally enable tolerance (Siegel, Hinson, Krank, & McCully, 1982). Discomfort from conditioned compensatory responses (like heightened pain sensitivity and decreased body temperature) can also contribute to drug dependence by motivating continued drug use to alleviate these withdrawal-like symptoms. This is a key way classical learning contributes to addiction.
Finally, classical cues motivate ongoing operant behavior (Balleine, 2005). A rat operantly conditioned to press a lever for drugs will press harder in the presence of cues signaling “drugs are coming” (like lever squeaking) compared to when those cues are absent. Similarly, food-associated cues increase effort for food, and negative cues (signaling fear) increase effort to avoid trauma. Classical CSs thus have diverse effects contributing to complex behaviors.
Image: Blocking in Classical Learning: This diagram illustrates how prior learning about one stimulus (the bell) can block learning about a new stimulus (the light) when both are paired with a reward. [Image courtesy of Bernard W. Balleine]
Understanding the Classical Learning Process
Counterintuitively, simply pairing a CS and US is not enough for learning. Blocking demonstrates this (Kamin, 1969). An animal first learns to associate one CS (stimulus A, say a bell) with a US (food). Once learned, a second stimulus (stimulus B, a light) is presented along with stimulus A, both paired with the US. However, because the animal already associates stimulus A (bell) with food, it doesn’t learn to associate stimulus B (light) with food. The conditioned response occurs only to stimulus A, as prior learning about A “blocks” learning about B. Stimulus A already predicts the US, making the US unsurprising when paired with stimulus B.
Learning depends on surprise or prediction error – a mismatch between what’s predicted and what actually occurs. To learn via classical learning, there must be an initial prediction error, a chance that the CS won’t lead to the expected outcome. In the bell-light example, the bell always predicts food, so adding light doesn’t reduce prediction error. But if food only appears when both bell and light occur, the bell alone now creates prediction error, driving new learning.
Blocking and related effects show that learning prioritizes the most reliable predictors of significant events and ignores less useful ones. Imagine a supermarket using star stickers for sales. You learn stars mean discounts. Then, a similar store uses stars and orange price tags for discounts. Due to blocking (stars already signal discounts), you ignore the color system. Stars are sufficient predictors, eliminating prediction error for discounts, making color irrelevant.
Classical learning is stronger when CS and US are intense or salient, and when they are novel. It’s also enhanced when an organism is biologically prepared to associate specific CS-US pairs. Rats and humans readily associate illness with flavors, not lights or tones. Taste is the primary sense for food experience, so linking flavor (not appearance) to illness is more adaptive for future food avoidance. This evolutionary predisposition is preparedness.
Numerous factors influence classical learning strength, a focus of extensive research (Rescorla & Wagner, 1972; Pearce & Bouton, 2001). Behavioral neuroscientists also utilize classical learning to study fundamental brain processes involved in learning (Fanselow & Poulos, 2005; Thompson & Steinmetz, 2009).
Extinguishing Classical Learning
After conditioning, the CR can be eliminated if the CS is repeatedly presented without the US. This is extinction. For example, if Pavlov kept ringing the bell without food, the dog’s drooling (CR) would eventually cease because the bell no longer predicted food. Extinction is crucial, forming the basis of therapies for maladaptive behaviors. For a spider phobia, exposure therapy might involve showing spider pictures in neutral settings. Initially, spiders (CS) elicit fear (CR). Repeated exposure without negative consequences leads to extinction; spiders no longer predict fear.
However, extinction doesn’t erase original learning (Bouton, 2004). Imagine associating chalk smell with school detention. Years later, the smell no longer evokes detention (extinction). But suddenly, in a new building, chalk smell triggers detention agony – spontaneous recovery. After extinction, a lapse in CS exposure can lead to the CR reappearing upon CS re-exposure.
The renewal effect is related: after extinction, testing the CS in a new context (different room) can also revive the CR. In the chalkboard example, entering a new building (unexpected chalk smell) renews detention sensations. These effects suggest extinction inhibits rather than erases learning, primarily in the extinction context.
Extinction remains a valuable therapy. Clinicians can enhance effectiveness by using learning research to counter relapse effects (Craske et al., 2008). Conducting extinction therapy in relapse-prone contexts (e.g., work) can improve therapy success.
Key Insights into Instrumental Conditioning
Many factors influencing classical learning also affect instrumental learning – associating actions with outcomes. Larger reinforcers/punishers lead to stronger learning. Unreinforced instrumental behaviors also undergo extinction. Most associative learning principles apply to both, but instrumental learning has unique aspects.
Stimulus Control of Instrumental Responses
In operant labs, lever pressing for food is classic. But, lever pressing can be reinforced only when a stimulus is present. For example, food only appears when a Skinner box light is on. Rats learn to discriminate, pressing only when the light is on (light-off responses extinguish). In real life, think of traffic lights. Green means go, but only a green arrow allows turning. Operant behavior is now under stimulus control. Stimulus control is likely the norm in real-world scenarios.
The controlling stimulus is a discriminative stimulus, setting the “occasion” for the operant response, not eliciting it like a classical CS. A canvas before an artist doesn’t compel painting but allows/enables it.
Stimulus control is used to study animal perception and cognition. Rats discriminating light vs. dark show they can see light. Such methods test animal color vision, ultrasound hearing, and magnetic field detection by pairing discriminative stimuli with known behaviors (lever pressing). Researchers test if animals learn to lever press only when ultrasound plays.
“Higher” cognition can also be studied. Pigeons can learn to peck different buttons for flower, car, chair, or people pictures on a screen (Wasserman, 1995). Button 1 (only) is reinforced with flowers, button 2 with chairs, etc. Pigeons learn these discriminations and even categorize new images, demonstrating categorization learning. Stimulus-control methods study categorization learning.
Choice in Operant Conditioning
Image: Operant Choice: Pigeons in Skinner boxes are often used to study choice behavior in operant conditioning experiments.
Operant conditioning always involves choosing one behavior over alternatives. A student going to a bar chooses drinking over studying. A rat chooses lever pressing over sleeping. Alternative behaviors are linked to different reinforcers. The tendency to perform an action depends on its reinforcers and reinforcers for alternatives.
Choice is studied in Skinner boxes with multiple levers/buttons, each with its own reinforcement rate. The Quantitative Law of Effect (Herrnstein, 1970) emerged from such studies. It states that reinforcement effectiveness for one behavior depends on reinforcement available for alternatives. If one light yields two food pellets and another only one, pigeons favor the first. But if the first is harder to reach, will the energy cost outweigh the extra food? Generally, a reinforcer is less potent if many alternatives exist. Alcohol, sex, or drugs may be weaker reinforcers if other fulfilling options like work achievement or family love are present.
Cognition in Instrumental Learning
Modern research shows reinforcers don’t just “stamp in” behaviors, as Thorndike thought. Animals learn about specific consequences of actions and act based on the current “value” of those consequences.
Image: Reinforcer Devaluation: This image represents the reinforcer devaluation effect, where a learned preference for a reward can be altered by changing the value of that reward. [Image courtesy of Bernard W. Balleine]
The reinforcer devaluation effect (Colwill & Rescorla, 1986) illustrates this. Rats learn two lever presses (left/right), each for a different reinforcer (sucrose/food pellet). Initially, they press both, alternating between rewards. Then, one reinforcer (sucrose) is paired with illness, creating taste aversion. In a test, rats can press either lever, but no reinforcers are given. Rats avoid the lever for the now-aversive sucrose. They learned and remembered each lever’s reinforcer and combined this with the reinforcer’s current “bad” value. Reinforcers aren’t just stamps; animals learn more. Behavior is goal-directed (Dickinson & Balleine, 1994), influenced by the current value of its goal (how much the rat wants/doesn’t want the reinforcer).
However, frequent, repeated instrumental actions can become automatic habits. After months of lever pressing, the action becomes routine. This goal-directed action (pressing for sucrose/food) can become habitual. Even if sucrose is again paired with illness, habituated rats will still press the sucrose lever (Holland, 2004). Habitual responses become insensitive to reinforcer devaluation. They respond automatically, ignoring sucrose-induced sickness.
Habits are common and useful. We don’t relearn coffee making or teeth brushing daily. Instrumental behaviors can become habitual, freeing cognitive resources for other tasks.
Integrating Classical and Instrumental Conditioning
Classical and operant conditioning are typically studied separately, but in reality, they often occur simultaneously. Reinforcement for drinking or overeating happens in specific contexts – pubs, restaurants, home couches. These stimuli also become associated with reinforcement. Classical and operant conditioning are thus intertwined.
The figure below illustrates this integration. Any reinforced/punished operant response (R) is paired with an outcome (O) in the presence of stimuli (S).
The figure shows learnable associations in this scenario. Organisms learn the response-outcome link (R-O) – instrumental conditioning, likely involving similar surprise and prediction error mechanisms as classical learning. R-O learning means the organism will perform the response if the outcome is desired. Reinforcer value is also influenced by alternatives. These are core aspects of instrumental learning.
Second, organisms learn stimulus-outcome links (S-O) – classical conditioning, with diverse behavioral consequences. Stimuli evoke preparatory responses (body temperature changes, salivation, insulin release) and approach/retreat behavior. Stimuli also prompt the instrumental response.
Image: Integrated Learning: Classical and operant conditioning often work together in real-world learning situations. This diagram illustrates the interplay between stimuli, responses, and outcomes.
Third, stimulus-response links (S-R) can form. With practice, stimuli directly elicit responses – habit learning, where responses are automatic, without much R-O or outcome value processing.
Finally, stimulus-response-outcome association [S-(R-O)] can be learned. Stimuli signal the R-O relationship, “setting the occasion” for operant response. A canvas signals painting behavior will be reinforced.
This framework helps analyze almost any learned behavior. Further learning in a course on learning will deepen understanding of classical, instrumental, habit, and occasion-setting mechanisms and their interactions.
Observational Learning
Not all learning is solely explained by classical and operant conditioning. Imagine a child watching other children play a new game. Instead of joining immediately, they observe, learning rules and strategies by watching. This is observational learning.
Image: Learning by Watching: Children learn through observation, especially from social models. [Image: David R. Tribble, https://goo.gl/nWsgxI, CC BY-SA 3.0, https://goo.gl/uhHola]
Observational learning is part of Albert Bandura’s Social Learning Theory (Bandura, 1977), stating people learn novel responses by observing others. It doesn’t require reinforcement, but relies on social models – often higher status figures like parents, teachers, or older peers. In the game example, experienced children are social models. Observing social models teaches how to act in situations. Examples include a child learning napkin placement by watching parents or a customer finding condiments by watching others.
Bandura’s observational learning involves four parts: attention (paying attention to the model), retention (remembering observed behavior), initiation (executing learned behavior), and motivation (wanting to learn). The child watching the game must want to learn to play.
Bandura’s famous Bobo doll experiment explored observational learning.
Image: Bobo Doll: The Bobo doll was used in Albert Bandura’s famous experiments on observational learning of aggression. [Image: © Sémhur / Wikimedia Commons / CC-BY-SA-3.0 (or Free Art License), https://goo.gl/uhHola]
In this experiment (Bandura, Ross & Ross 1961), children watched an adult interact with a Bobo doll. Some saw aggression (punching, kicking, hitting with a mallet), others saw non-aggression. Later, children could play with Bobo. Aggression-exposed children were more aggressive to Bobo than non-aggression-exposed children. Researchers concluded children learned aggressive behavior was acceptable by observation.
While reinforcement wasn’t needed in the first Bobo doll study, consequences are relevant. A later study (Bandura, Ross, & Ross, 1963) showed children seeing the adult punished for Bobo aggression displayed less aggression themselves. Bandura called this vicarious reinforcement – learning influenced by observing others’ reinforcement/punishment experiences, not direct experience.
Conclusion
We’ve explored classical learning, operant conditioning, and observational learning as key explanations for behavior. Reflecting on your experiences, how do these apply to you? Fashion choices influenced by compliments (operant)? Restaurant choices based on happy commercial music (classical)? Punctuality learned by observing punishment for lateness (observational)? Whatever the behavior, these theories likely offer insights into its origins.
