Psychology of Learning

Module 04: Classical Conditioning 1

Part 2: Acquisition & Extinction

Looking Back

In Part 1, we explored the basics of classical conditioning, learning how Ivan Pavlov accidentally discovered that neutral stimuli could acquire the power to elicit responses through association with unconditioned stimuli. We defined key terms (US, UR, CS, CR), examined how conditioning proceeds through CS-US pairing during acquisition, and explored real-world examples demonstrating that classical conditioning is a fundamental learning process occurring constantly in everyday life. Now we’ll examine the factors that make conditioning stronger or weaker, what happens when conditioning is reversed through extinction, and how different types of unconditioned stimuli produce different types of conditioning.

Factors Affecting Acquisition

Once researchers established the basic classical conditioning procedure, they quickly began investigating what factors influence the strength of conditioning. What makes a CR develop more quickly or become stronger? Four major factors emerged from this research.

Number of CS-US Pairings

Generally, greater numbers of CS-US pairings produce stronger CRs. Early conditioning trials produce rapid learning, but as conditioning progresses, each additional pairing adds less and less. The learning curve shows rapid initial acquisition that gradually plateaus as the CR approaches maximum strength (Pavlov, 1927).

However, this generalization has important exceptions. Some types of conditioning—especially taste aversion learning—can produce strong CRs after just one pairing. If you eat novel food and become violently ill, you may develop a lasting aversion after that single experience. This rapid, one-trial learning makes adaptive sense for food aversions, where a second “learning opportunity” might be fatal (Garcia & Koelling, 1966).

Strength of the US

Stronger USs produce faster and more robust conditioning. A loud noise produces stronger fear conditioning than a soft noise. A larger food reward produces stronger appetitive conditioning than a small reward. The more intense the US, the more quickly and strongly the organism learns to respond to the CS (Rescorla, 1968).

This makes intuitive sense from an adaptive perspective. Biologically significant events—those involving strong stimulation—are precisely the events organisms should learn about quickly. A barely audible rustling might not warrant strong learning, but a loud roar definitely does.

Does the CS Precede the US?

Timing matters crucially. If the US precedes the CS (backward conditioning), very little learning occurs. The most effective arrangement is forward conditioning, where the CS precedes and predicts the US. Think about it: a stimulus can only signal an upcoming event if it occurs before that event. A metronome ticking after food arrives can’t predict food delivery (Pavlov, 1927).

Several timing arrangements exist. Delay conditioning (CS begins before US, continues during US) produces the strongest learning. Trace conditioning (CS ends before US begins, leaving a gap) produces weaker learning. Simultaneous conditioning (CS and US occur together) produces even weaker learning. Backward conditioning (US precedes CS) produces minimal learning (Smith, Coleman, & Gormezano, 1969).

Contiguity & CS-US Timing

An interval of approximately 0.5 seconds between CS onset and US onset is generally optimal for many types of conditioning. Shorter or longer intervals typically result in weaker conditioning. This half-second window represents the “sweet spot” for many reflexive responses like eyeblink conditioning or salivation (Smith et al., 1969).

But here’s a puzzle: Does that statement seem strange? Have you already encountered something that contradicts it? Yes! In taste aversion learning, the delay between CS (taste) and US (illness) can be as long as 24 hours, yet strong conditioning still occurs. This dramatic exception suggests that different types of conditioning follow different rules—a theme we’ll explore more when discussing biological constraints on learning (Garcia, Ervin, & Koelling, 1966).

Extinction: When the CS No Longer Predicts the US

Extinction is the presentation of the CS without the US, leading to a reduction in the strength and frequency of the CR. Because the US is not presented when researchers test for the CR, the CR gradually decreases in strength (Pavlov, 1927).

Why does extinction make adaptive sense? Right after conditioning, the CS is an excellent predictor that the US is going to occur. The CS reliably signals an important upcoming event. But after several CS-only (extinction) trials, the CS is no longer a good predictor of US occurrence. The environment has changed—the CS-US relationship no longer holds. The organism’s ability to detect this change and reduce the CR demonstrates adaptive flexibility (Rescorla, 1968).

Extinction doesn’t simply erase learning. Instead, the organism learns something new: that the CS no longer predicts the US. The original learning remains but is suppressed by new learning. We know this because of a phenomenon called spontaneous recovery.

Resistance to Extinction

Resistance to extinction is a measure of the strength of the CR and indicates how much learning has occurred. The more resistant a CR is to extinction, the stronger the original conditioning was (Pavlov, 1927).

Researchers measure resistance to extinction in several ways. Number of CS-only trials until extinction asks how many times the CS must be presented alone before the CR disappears; more trials indicate stronger conditioning. Amplitude of the CR measures how strong the response is; more salivation, larger startle responses, or more vigorous reactions indicate stronger conditioning. Latency of the CR measures how quickly the CR appears after CS presentation; with strong conditioning, latency is short—the CR appears almost immediately, while weak conditioning produces longer latencies. These measures allow researchers to quantify learning and compare conditioning across different procedures, species, or stimuli.

Spontaneous Recovery: The Return of the Extinguished

Spontaneous recovery is the reappearance of a CR that was thought to be extinguished. After extinction, if time passes before the next test session, the CR often reappears—though typically weaker than before extinction (Pavlov, 1927).

The greater the passage of time between extinction sessions, the greater the spontaneous recovery. This phenomenon can occur multiple times—each extinction session weakens the CR, but time allows partial recovery. Spontaneous recovery is more prevalent with stronger initial conditioning.

How can we explain this mysterious reappearance? A case can be made that the CS predicted the US for one session, then the CS did not predict the US for one session. Given a new session, the organism doesn’t know which type of session it will be—one in which the CS predicts the US or not. The organism essentially hedges its bets, showing some responding until it determines what kind of session is occurring (Bouton, 1993).

Spontaneous recovery demonstrates that extinction doesn’t erase original learning. The CS-US association remains; extinction simply overlays new learning on top of it. Time allows the older learning to resurface, though repeated extinction sessions eventually produce more permanent suppression.

Contemporary Understanding of Extinction and Relapse

Recent comprehensive reviews have integrated decades of research on extinction learning and its neural mechanisms. Bouton, Maren, and McNally (2021) reviewed extensive behavioral and neurobiological evidence demonstrating that extinction depends on new inhibitory learning that is primarily expressed in the context where it is learned. This explains why extinguished behaviors can “relapse” through several mechanisms: spontaneous recovery (passage of time changes the temporal context), renewal (returning to the original conditioning context), reinstatement (re-exposure to the US), and rapid reacquisition (when CS-US pairings resume). The prefrontal cortex plays a crucial role in extinction retrieval, while hippocampal-prefrontal circuits mediate context-dependent relapse phenomena. Understanding these mechanisms has important implications for clinical interventions like exposure therapy, where the goal is to help extinction learning generalize beyond the treatment context.

Appetitive & Aversive Unconditioned Stimuli

While Pavlov’s research used food (an appetitive US), many other USs have been employed in classical conditioning research. These fall into two broad categories.

An appetitive US is a US, such as food, that is sought out and elicits approach behaviors. Appetitive USs are desirable, pleasant stimuli that organisms work to obtain (Konorski, 1967).

An aversive US is a US, such as electric shock, that is avoided and elicits avoidance or withdrawal behaviors. Aversive USs are unpleasant stimuli that organisms work to escape or prevent (Konorski, 1967).

This distinction is important because appetitive and aversive conditioning differ in crucial ways. Appetitive conditioning produces approach to the CS; aversive conditioning produces withdrawal from the CS.

Examples of Appetitive USs & Their URs

In the voluntary muscles, an object touching lips elicits sucking—this innate reflex in infants ensures feeding. In the digestive system, good food elicits salivation and other digestive responses; Pavlov’s original research exploited this reflex. In the emotional system, sexual stimulation elicits erotic feelings, and these pleasant emotional responses can be conditioned to associated stimuli. In the reproductive system, genital stimulation elicits physiological arousal responses including vaginal lubrication, penile erection, and orgasm; these reflexive responses can become conditioned to contextual cues.

Examples of Aversive USs & Their URs

In the voluntary muscles, sharp or hot stimuli elicit jerking away and crying; blows, shocks, and burns elicit withdrawal. These protective reflexes remove the body from harm. In the circulatory system, high temperature elicits sweating and flushing to dissipate heat, while sudden loud noises elicit blanching (blood leaving the skin) and heart pounding—components of the startle response. In the digestive system, bad food elicits sickness, nausea, and vomiting; these responses expel potentially toxic substances. In the respiratory system, irritation in the nose elicits sneezing, throat clogging elicits coughing, and allergens elicit asthma attacks; these reflexes clear airways of irritants. In the emotional system, painful blows elicit fear; this emotional response prepares the organism for defensive action.

Common Aversive Conditioning Procedures

Another favorite US with classical conditioning researchers is a brief puff of air administered to the open eye of a test subject—frequently a rabbit or a human infant. The puff of air reflexively elicits an eye blink that can be conditioned easily. Researchers pair a tone or light (CS) with the air puff (US), and soon the CS alone elicits blinking (Gormezano, 1966).

Likewise, some researchers use mild electric shock as the US. Often they apply shock to one foot of a test dog; the dog reflexively raises its foot. This foot-raising behavior is easily conditioned to a CS, such as a tone or blinking light. The conditioned foot withdrawal demonstrates aversive conditioning (Smith et al., 1969).

Experimental Neurosis: When Conditioning Creates Conflict

Experimental neurosis is the neurotic behavior that Pavlov created in some laboratory animals by bringing excitatory and inhibitory tendencies into conflict. When organisms face impossible discrimination tasks, their behavior can break down dramatically (Pavlov, 1927).

Pavlov would present food (reinforcement) following presentation of a red circle, but not following presentation of a black circle. The red circle produces excitation of salivation (a CS+), and the black circle produces inhibition of salivation (a CS-). Dogs learned this discrimination easily.

But then Pavlov made the task progressively harder. He made the red circle continually darker until it became indistinguishable from the black circle. Now excitation and inhibition occur simultaneously—the organism can’t tell which stimulus is present. The animals’ behavior broke down: they became agitated, whined, bit at equipment, refused to eat, and showed signs of extreme distress—experimental neurosis.

Different animals responded to this impossible situation differently. Pavlov identified four types based on their responses, corresponding roughly to temperament types: those with very strong excitatory tendencies became aggressive; those with moderately strong excitatory tendencies became anxious; those with moderately strong inhibitory tendencies became passive; those with very strong inhibitory tendencies became withdrawn.

Pavlov believed how animals respond to conflict depends on their nervous system characteristics. He extended this to humans, proposing that abnormal human behavior results from breakdown in the brain’s inhibitory processes. While his specific neurological theory hasn’t held up, experimental neurosis remains an important phenomenon showing how learning procedures can create psychological distress when organisms face unsolvable conflicts (Pavlov, 1927).

Looking Forward

We’ve examined the factors that strengthen conditioning (number of pairings, US strength, CS-US timing), how conditioning is reversed through extinction (with the mysterious spontaneous recovery), and how appetitive versus aversive USs produce different types of learning. In Part 3, we’ll explore generalization and discrimination—how organisms respond to stimuli similar to the CS and how they learn to differentiate among stimuli—as well as more complex forms of classical conditioning including higher-order conditioning, sensory preconditioning, sign tracking, and conditioned emotional responses.