The Avoidance Learning in the Shuttlebox Experiment – Solomon and Wynne

The study of aversive conditioning stands as one of the most intellectually challenging and clinically consequential chapters in the history of experimental psychology. During the middle decades of the twentieth century, as behaviorism sought to formalize the laws governing how organisms interact with, adapt to, and survive an indifferent or hostile environment, the phenomenon of avoidance learning emerged as an enduring theoretical enigma. How can an organism learn to perform a complex, effortful behavioral act in response to a harmless signal, solely to prevent an aversive event that never actually occurs? In classic Pavlovian and operant formulations, learning required reinforcement—the presentation of an unconditioned stimulus or the delivery of a primary reward. Active avoidance seemed to defy this fundamental axiom: once acquired, it appeared to be reinforced by nothing at all, sustained across hundreds or thousands of trials in the total absence of the original punishing stimulus.

To untangle this mechanistic puzzle, Richard L. Solomon and Lyman C. Wynne embarked on a series of landmark investigations at Harvard University in the early 1950s. Utilizing a specialized apparatus known as the shuttlebox, Solomon and Wynne placed canine subjects into an environment featuring discrete warning signals, electrifiable grid floors, and a hurdle over which the animals could leap. Their seminal 1953 monograph, along with subsequent papers, documented a behavioral trajectory that astonished the psychological community. Dogs subjected to intense, traumatic electric shocks rapidly shifted from frantic escape responses to swift, highly stereotyped avoidance maneuvers. More radically, once these avoidance responses were established, they proved virtually impossible to extinguish under ordinary laboratory conditions. The animals continued to jump within split seconds of warning signal onset, trial after trial, block after block, without ever experiencing another electric shock.

The implications of Solomon and Wynne’s shuttlebox findings reverberated across every branch of behavioral science. The empirical data provided the vital testing ground for O. Hobart Mowrer‘s emerging two-factor learning theory, challenged Hullian drive-reduction frameworks, prompted the development of cognitive expectancy models, and catalyzed evolutionary re-evaluations concerning species-specific defense reactions. Beyond basic theory, the paradigm laid the empirical bedrock for understanding the etiology and maintenance of human anxiety disorders, phobias, and obsessive-compulsive rituals. It directly inspired Victor Meyer’s revolutionary clinical protocol of Exposure and Response Prevention (ERP), which remains the gold standard of cognitive-behavioral intervention today. This treatise provides an exhaustive, multi-dimensional analysis of Solomon and Wynne’s shuttlebox experiments, tracing their historical foundations, technical architecture, behavioral dynamics, neurobiological substrates, theoretical controversies, and enduring clinical legacy.

1. Historical Context and Theoretical Foundations of Aversive Conditioning

1.1 The Rise of Neobehaviorism and Instrumental Learning Paradigms

The early twentieth century witnessed a profound shift in American psychology, transitioning away from the introspective paradigms of Wilhelm Wundt and Edward Titchener toward an objective, quantifiable science of behavior. At the vanguard of this revolution was Edward Thorndike, whose 1898 dissertation on animal intelligence introduced the foundational Law of Effect. Thorndike posited that behavioral responses followed by satisfying states of affairs are stamped in, becoming more likely to recur, whereas responses accompanied or closely followed by discomfort to the animal will have their connections to that situation weakened. While Thorndike initially conceived this law as symmetrical, later empirical revisions led him to recognize that punishment did not merely reverse the stamping-in process of reward, revealing early asymmetries between appetitive and aversive behavioral controls.

By the 1930s and 1940s, neobehaviorism had ascended to academic dominance, driven largely by the hypothetico-deductive formulation of Clark L. Hull. Hull’s drive-reduction theory attempted to unify all mammalian learning under a single, rigorous mathematical framework. In Hull’s system, learning could occur only when an organism’s biological need (drive, D) was reduced, which in turn strengthened the habitual connection (habit strength, sHr) between the prevailing stimulus environment and the emitted motor response. While Hullian theory operated smoothly in appetitive paradigms—where hungry rats negotiated mazes for caloric rewards that visibly reduced primary physiological drives—it encountered immense theoretical friction when applied to defense conditioning and defensive avoidance.

Early behavioral researchers had recognized that appetitive reward learning and aversive defense conditioning operated under markedly disparate ecological and chronological imperatives. In an appetitive paradigm, an animal can afford multiple errors; a missed food pellet results in hunger, but rarely immediate biological destruction. In sharp contrast, aversive conditioning simulates predatory encounters or environmental hazards where a single operational failure can result in mortal injury or death. This biological asymmetry demanded the development of highly controlled laboratory models of fear acquisition that could systematically isolate the mechanics of defense without killing the organism. Pioneering experiments by Neal E. Miller and O. Hobart Mowrer began to investigate whether acquired fear could function as an internal drive state possessing motivational and reinforcing properties fully analogous to hunger or thirst, setting the empirical stage for the Harvard investigations.

1.2 Theoretical Ambiguities in Early Avoidance Research

The central conundrum that haunted early avoidance research was the fundamental dilemma of negative reinforcement in the total absence of an immediate primary reward. In classical escape learning, the explanatory apparatus was straightforward: an animal was subjected to an unconditioned aversive stimulus (such as a thermal, mechanical, or electrical insult), initiated frantic motor output, accidentally made contact with a manipulandum or escaped into a safe zone, and experienced immediate termination of the physical insult. Here, the primary drive reduction was obvious, temporal contiguity was preserved, and standard Thorndikian or Hullian frameworks could readily explain the strengthening of the escape reflex.

Avoidance learning, however, introduced an intractable paradox for classical stimulus-response (S-R) theory. In a typical active avoidance sequence, an external warning cue is presented, the organism executes an adaptive motor response before the onset of the noxious event, and as a direct consequence, the noxious event is cancelled. The organism never experiences the physical pain. If reinforcement requires the reduction of a primary biological drive, or if learning is driven strictly by the contingent presentation of an unconditioned stimulus (US) as articulated in Ivan Pavlov‘s classical conditioning framework, what maintains the persistent execution of the avoidance response? The non-occurrence of an event—a non-event—appeared to serve as the functional reinforcer.

Classical conditioning could explain how an organism acquired an anticipatory conditioned response (CR)—such as salivation to a metronome or pupillary dilation to a tone—because the conditioned stimulus (CS) was reliably paired with the unconditioned stimulus (US). Yet Pavlovian models struggled fundamentally to explain active, self-initiated, motoric defense responses that actively altered the physical environment and decoupled the CS from the US. If the animal successfully avoids the US, the CS is presented in isolation, which, according to the inviolable laws of Pavlovian conditioning, must inevitably lead to experimental extinction. Therefore, classical theory predicted that avoidance behavior must be inherently cyclical: the animal learns to avoid, successfully avoids the shock, experiences the CS without the US, undergoes extinction, fails to avoid, receives the shock again, and reacquires the response. Early researchers observed, to their theoretical dismay, that avoidance behavior in the laboratory rarely exhibited such neat cyclic oscillations; instead, it often persisted with stubborn, immutable stability.

1.3 Solomon and Wynne’s Research Objectives at Harvard

Entering this theoretical quagmire at the Harvard University Department of Social Relations, Richard L. Solomon and Lyman C. Wynne designed a programmatic series of experiments explicitly engineered to subject avoidance learning to its most rigorous, extreme empirical test. Operating in an intellectual milieu characterized by vibrant cross-pollination among experimental psychology, anthropology, and clinical psychiatry, Solomon and Wynne sought to examine the behavioral parameters of what they termed “traumatic avoidance learning.”

Their primary experimental objective was to measure the boundary conditions of resistance to extinction when an organism was conditioned under states of exceptionally intense, biologically unambiguous primary punishment. Solomon and Wynne were deeply dissatisfied with the ambiguous, fragile avoidance habits generated by low-intensity aversive stimuli in contemporary rodent studies. They hypothesized that intense, traumatic unconditioned stimuli would recruit deeper, more indelible learning mechanisms that could reveal the fundamental architecture of defensive motivational systems. By observing how higher mammals responded to inescapable primary trauma and subsequent opportunities for behavioral control, they aimed to chart the precise chronological transition from reflexive pain-driven escape to pre-emptive signal-driven avoidance.

Beyond mapping basic acquisition curves, their ultimate conceptual goal was to identify the psychological and physiological engines driving chronic resistance to extinction. They intended to evaluate whether a learned behavioral habit could achieve total functional permanence—an outcome that would force a radical re-evaluation of Hullian, Pavlovian, and Skinnerian learning doctrines. In doing so, Solomon and Wynne aimed to construct an ecologically grounded, standardized, and strictly repeatable animal model for human psychopathology, providing an empirical bridge to illuminate the enigmatic persistence of severe anxiety disorders, phobias, and chronic compulsive rituals.

2. The Shuttlebox Experimental Apparatus and Technical Design

2.1 Physical Architecture of the Shuttlebox

To realize their demanding experimental objectives, Solomon and Wynne engineered an exceptionally robust, mechanically sophisticated apparatus known as the shuttlebox. Constructed within the animal testing facilities of the Harvard Psychological Laboratories, the shuttlebox was a symmetrical, two-compartment enclosure tailored to accommodate medium-to-large canine subjects. The overall dimensions of the chamber measured approximately eight feet in length, three feet in width, and four feet in height, providing ample space for vigorous motor movement, rapid acceleration, and high-clearance jumping.

The chamber was divided transversely into two identical compartments separated by an adjustable central barrier, or hurdle. This hurdle consisted of a solid wooden board whose vertical height could be calibrated to the specific shoulder height and jumping capacity of each experimental subject, typically set at roughly the level of the dog’s withers. Immediately above the hurdle hung a set of double guillotine gates. These gates were operated by silent pneumatic or mechanical counterweight systems that could drop rapidly to seal off one compartment from the other or lift simultaneously to expose the hurdle and permit free transit between the two halves of the shuttlebox.

The flooring of the entire apparatus was composed of an electrifiable grid made of high-grade, heavy-gauge stainless steel rods spaced precisely to prevent the subjects’ paws from slipping through or avoiding contact. These rods were wired to an elaborate electrical distribution network designed to prevent “standing on one rod” or exploiting dead spots. The internal walls of the shuttlebox were constructed of smooth, non-conductive, shock-resistant materials painted an unreflective neutral gray to eliminate visual distractions. The entire enclosure was housed within an acoustically insulated, light-tight experimental room. Behavioral tracking was achieved via an array of mechanical switches and photobeam sensors embedded along the perimeter of the hurdle, enabling the automatic, objective registration of compartment entries, barrier approaches, and successful hurdle crossings with millisecond precision.

2.2 Conditioned and Unconditioned Stimuli Configurations

The environmental stimuli utilized in Solomon and Wynne’s 1953 experiments were calibrated to maximize sensory clarity and emotional salience. The conditioned stimulus (CS) was a compound, multi-modal sensory alteration designed to blanket the subject’s immediate perceptual field. It consisted of two distinct components: a drastic illumination shift and an auditory signal. In the baseline inter-trial state, the compartment occupied by the dog was illuminated by two 100-watt ceiling lamps. At the onset of a trial, these lights were abruptly extinguished, plunging the compartment into darkness, while simultaneously, the overhead lights in the opposite, unoccupied compartment were illuminated. Concurrently, an auditory stimulus—often a high-frequency buzzer or a 1000-Hz tone broadcast through overhead speakers at approximately 80 decibels—was activated, providing an unmistakable, urgent warning signal.

The unconditioned stimulus (US) was an exceptionally severe, traumatic electric shock delivered through the stainless steel grid floor. Unlike the low-current shocks standard in rodent exploratory chambers, Solomon and Wynne utilized a variable-ratio high-voltage, low-amperage alternating current circuit powered by a 1000-volt step-up transformer. The electrical circuit was routed through a current-limiting resistor bank designed to deliver between 10 to 15 milliamperes of scrambled, pulsating alternating current directly to the dog’s footpads. This current was sufficiently intense to induce profound physiological activation, violent involuntary motor spasms, and vocal distress, ensuring that the stimulus functioned as a truly traumatic unconditioned insult rather than a mild irritant.

The temporal architecture of the trial followed a rigid, fixed-interval delayed conditioning protocol. The conditioned stimulus preceded the unconditioned stimulus by an invariant temporal window of exactly 10.0 seconds. If the subject executed a hurdle-clearing response during this 10-second interval, the compound CS was instantly terminated, the drop gates descended to seal the animal in the newly entered compartment, and the scheduled electric shock was canceled entirely for that trial. If the dog failed to jump within the 10-second window, the grid floor of the occupied compartment was immediately electrified, with the shock and the compound CS persisting continuously until the animal successfully scrambled over the hurdle into the safe, non-electrified compartment.

2.3 Measurement Variables and Data Collection Protocols

Data collection in the Solomon and Wynne shuttlebox was rigorous, exhaustive, and quantitatively granular. The primary behavioral metric recorded on every single trial was response latency, defined with absolute operational precision as the elapsed time in seconds from the exact onset of the conditioned stimulus (light dimming and buzzer activation) to the moment the dog completely cleared the central hurdle and tripped the mechanical contact switches on the opposite side. Response latencies were measured using automated electric stop-clocks wired directly into the stimulus-delivery relays.

These latency measurements enabled an absolute, non-overlapping operational dichotomy between two distinct behavioral classes:

• Escape Responses: Defined as any hurdle-clearing action occurring with a latency greater than 10.0 seconds—meaning the behavior took place only after the unconditioned electric shock had already been activated through the grid floor. The reinforcement for an escape response was the primary physical cessation of pain.
• Avoidance Responses: Defined as any hurdle-clearing action occurring with a latency strictly less than 10.0 seconds—meaning the motor response was completed entirely during the warning interval. In this instance, the dog completely prevented the onset of the unconditioned electric shock.

Beyond digital latency measurements, Solomon and Wynne maintained comprehensive observational logs documenting the somatic, physiological, and emotional indices of their subjects. Observers viewed the animals through a one-way mirror, systematically recording gross autonomic discharge such as spontaneous defecation, urination, and continuous salivation. Somatic defense patterns—including high-pitched yelping, screeching, shivering, rigid crouching, frantic wall-climbing, and stereotypic searching movements—were categorized trial by trial. This dual-stream protocol, wedding millisecond chronometric data with nuanced ethological observation, permitted the longitudinal tracking of acquisition trajectories, plateau phases, and the subsequent, baffling failure of behavioral extinction.

3. Methodological Framework: Solomon and Wynne’s 1953 Experimental Protocol

3.1 Subject Selection, Handling, and Habituation

The subjects for Solomon and Wynne’s classic 1953 investigations were mongrel dogs, predominantly medium-sized canines weighing between 10 and 20 kilograms, chosen specifically for their robust physical constitution, high-level behavioral plasticity, and well-developed mammalian affective repertoires. Canines represented an ideal evolutionary model: unlike laboratory-bred rodents, which often exhibit narrow, rigid, stereotypic defensive topographies such as profound catatonic freezing, dogs possess a rich behavioral matrix of vocal, postural, and locomotive defense strategies that closely mirror human somatic expressions of panic and acute anxiety.

Prior to the initiation of experimental testing, the subjects underwent an extensive, standardized habituation protocol designed to neutralize extraneous environmental novelty and establish a stable behavioral baseline. Upon arrival at the Harvard facility, the dogs were housed in uniform individual kennels under controlled temperature and circadian lighting conditions. They were maintained on a strictly standardized nutritional regime, receiving balanced rations and free access to water. For several days preceding the experiments, human experimenters handled the dogs repeatedly, leading them on leashes through the laboratory corridors to habituate them to the human handlers, collars, and laboratory sounds.

Once preliminary handling was complete, each dog was introduced to the unpowered shuttlebox for extended free-exploration sessions. During these habituation trials, the drop gates were raised, the hurdle was set at a lowered height, and the dog was allowed to roam freely between the two compartments in the absence of any auditory warning signals, illumination changes, or electrical currents. The experimenters recorded baseline rates of spontaneous hurdle crossing, ensuring that the animals possessed the requisite physical agility to clear the barrier and that spontaneous transit rates dropped to near-zero once exploratory curiosity waned. Only after an animal demonstrated complete composure and non-reactivity within the unpowered apparatus was the formal conditioning protocol inaugurated.

3.2 Trial Sequence and Contingency Schedules

The formal experimental protocol followed a rigid, trial-by-trial contingency schedule executed with relentless precision. A dog was brought from its home kennel, placed carefully into one of the shuttlebox compartments, and allowed a three-to-five-minute resting period in the fully illuminated chamber with the guillotine gates lowered. Once the subject exhibited stable resting behavior, the trial sequence commenced.

A single trial unfolded across the following deterministic steps:

1. Trial Onset: The central guillotine gates lifted silently, exposing the hurdle, while simultaneously, the overhead lights in the dog’s compartment were extinguished, the lights in the opposite compartment were activated, and the auditory warning buzzer sounded.
2. The Critical Decision Window: A 10.0-second temporal interval elapsed. During this window, the dog was entirely free to jump the hurdle. If the dog jumped, the buzzer fell silent, the lights instantly reset to baseline conditions, the drop gates lowered, and the trial terminated immediately with zero shock delivery.
3. The Punishment Phase: If the 10.0-second clock expired without a hurdle crossing, the grid floor of the occupied compartment was instantly energized with the 10-15 mA high-voltage shock. The shock, darkness, and auditory buzzer remained active simultaneously until the dog cleared the barrier into the opposite compartment.
4. Inter-Trial Interval (ITI): Once the dog entered the opposite compartment—whether via escape or avoidance—the drop gates fell, sealing the subject in the new, safely illuminated compartment. A variable, randomized inter-trial interval ranging from two to five minutes was then enforced. This temporal jitter was vital: it prevented the dogs from acquiring a temporal conditioned reflex, ensuring that behavioral responses were tied strictly to the discrete sensory warning signals rather than an internal circadian or chronometric rhythm.

Testing sessions typically consisted of ten to twenty discrete trials per day, administered systematically across successive days until the subject reached the designated performance criteria. The physical height of the hurdle was incrementally adjusted upward during early training to ensure that the jump required a deliberate, effortful motor output rather than an incidental step or stroll.

3.3 Operational Definitions of Escape Versus Avoidance

To quantify the learning trajectories without theoretical ambiguity, Solomon and Wynne maintained rigorous operational boundaries separating escape performance from true avoidance execution. An escape response was defined as any barrier clearance where the recorded latency was greater than or equal to 10.01 seconds. In the initial trials, 100% of responses fell squarely into this category. The functional significance of the escape response lay in its direct relationship to unconditioned physical nociception: the animal moved because the external environment was actively inflicting severe physiological trauma. The learning mechanism here was classical negative reinforcement governed by the immediate termination of the primary unconditioned stimulus.

An avoidance response, conversely, was operationally defined as a hurdle crossing completed with a latency between 0.01 and 9.99 seconds post-CS onset. The functional significance of this response was profoundly different: the animal moved through a physically benign environment (the grid floor was completely inert) to terminate a neutral or conditioned cue and prevent an event that existed purely as a future probability. The transitional phase between escape and avoidance was characterized by an abrupt, non-linear collapse in response latencies, typically occurring within a narrow band of two to three consecutive trials.

Solomon and Wynne established a stringent stabilization criterion to designate full mastery of the avoidance habit. An animal was deemed to have achieved complete operational mastery when it completed ten consecutive successful avoidance trials—executing ten consecutive jumps with latencies under 10.0 seconds, thereby enduring ten successive trials without receiving a single electrical shock. Once a subject reached this criterion, the experiment shifted from the acquisition phase to the critical extinction phase, where the apparatus was altered to measure how long this learned behavior could survive in the total absence of primary reinforcement.

4. Behavioral Acquisition: The Shift from Escape to Active Avoidance

4.1 Early Trial Dynamics and Panic Reactions

The behavioral phenotype displayed by canine subjects during the initial exposures to the shuttlebox conditioning schedule was characterized by utter behavioral disorganization, acute autonomic shock, and uncoordinated physical panic. On Trial 1, when the 10-second CS window elapsed and the high-voltage electrical current surged through the stainless steel grid floor, the dogs did not engage in directed, adaptive problem-solving. Instead, they exhibited a generalized, explosive unconditioned defense reaction.

The dogs shrieked, yelped at piercing volumes, defecated profusely, urinated, and engaged in frantic, undirected thrashing against the smooth walls of the enclosure. Many subjects attempted to climb the vertical wooden surfaces, bit violently at the metal drop gates, or crouched low to the floor in a state of partial motor tetany induced by the alternating current. The discovery of the central hurdle was entirely serendipitous. Amid the chaotic, violent scramble around the perimeter of the electrified chamber, the animal would accidentally careen into the barrier, scramble over the wooden hurdle, and tumble awkwardly into the dark, non-electrified opposing compartment.

Upon clearing the hurdle, the electrical shock instantly ceased, the buzzer stopped, and the drop gates fell. Despite the immediate alleviation of physical pain, the animals remained in a state of acute autonomic hyperarousal, trembling uncontrollably, panting heavily, and exhibiting pronounced tachycardia. Over the course of Trials 2 through 5, this disorganized panic began to undergo a progressive, rapid behavioral streamlining. The random thrashing and wall-biting extinguished as the animal rapidly formed an association between the spatial locus of the central hurdle and the cessation of agony. The dog’s escape movements became directional: upon shock onset, the animal immediately oriented toward the barrier and bounded across, dropping escape latencies from initial chaotic durations of 30 to 60 seconds down to a reliable two or three seconds post-shock onset (total latencies of 12 to 13 seconds from CS onset).

4.2 The Critical Transition Point and Latency Shifts

Between Trials 5 and 15, an extraordinary behavioral metamorphosis occurred across virtually all experimental subjects. This phase represented the critical transition point—the operational emergence of true active avoidance. The transition was rarely smooth or gradual; rather, it manifested as an abrupt, discontinuous cliff in the chronometric data, frequently referred to in behavioral literature as a sudden “insight-like” or catastrophic shift.

Typically, after several trials of efficient escape responding, a subject would exhibit a visible change in its demeanor during the 10-second warning interval. Instead of waiting passively for the shock to strike, the dog would become intensely vigilant upon the dimming of the lights and the onset of the buzzer. Its muscles would tense, its ears would pin back, and its gaze would lock onto the central hurdle. Then, suddenly, on a single trial—often Trial 7, 8, or 9—the dog would spring across the hurdle at the 7th or 8th second of the warning signal. The barrier was cleared before the grid floor was energized. The shock never occurred.

What made this transition point so empirically striking was its immediate, permanent consolidation. Once a dog executed its first successful avoidance jump, it almost never reverted to escape responding. On the subsequent trial, the response latency would not wander back to 12 or 13 seconds; instead, it plummeted radically. Latencies plunged from 8 seconds to 4 seconds, and then to 2 seconds. The dog had crossed a behavioral Rubicon. From that single trial onward, the subject was no longer reacting to physical trauma; it was proactively leaping into the dark to preempt an impending catastrophe.

4.3 Plateau Phase and Overlearned Motor Execution

Following the rapid consolidation phase, the behavior entered an exceptionally stable plateau phase characterized by highly stereotyped, overlearned motor execution. Within 20 to 30 trials, the subjects were executing hurdle jumps with breathtaking speed and mechanical consistency. Latencies hovered within an ultra-rapid window of 1.5 to 3.5 seconds following the presentation of the compound conditioned stimulus.

The physical topography of the response became intensely streamlined. The moment the drop gates began their upward mechanical ascent and the lights dipped, the dog did not hesitate, explore, or scan the environment. It launched itself across the hurdle in a single, fluid, athletic leap, clearing the barrier cleanly without its paws touching the wooden ridge, landing squarely in the opposite compartment, and spinning around to face the closed gate. All traces of the initial disorganized panic had vanished from the animal’s motor execution.

Crucially, as the avoidance habit solidified into an overlearned motor sequence, the gross autonomic expressions of terror that had characterized early acquisition underwent a marked behavioral dissociation. During the anticipatory two-second interval before the jump, the dogs no longer whimpered, shrieked, or evacuated their bowels. Their facial expressions and motor carriage settled into an uncanny, flat, mechanical efficiency. The avoidance response had been perfected: across hundreds of consecutive trials, the subjects successfully prevented 100% of scheduled shocks, sustaining this flawless defensive record across weeks of daily testing without suffering a single electrical contact.

5. Two-Factor Learning Theory: Pavlovian Conditioning and Instrumental Reinforcement

5.1 Mowrer’s Theoretical Architecture as Applied by Solomon and Wynne

To provide a rigorous theoretical account of the extraordinary behavioral stability observed in their shuttlebox, Solomon and Wynne turned to the dual-process framework originally formulated by O. Hobart Mowrer (1947), universally designated in contemporary psychology as Two-Factor Learning Theory. Mowrer had proposed that avoidance learning could not be adequately explained by either Pavlovian classical conditioning or Thorndikian operant conditioning operating in isolation; rather, it represented a dynamic, interdependent sequence combining both forms of associative learning.

The two factors of the theoretical architecture operated as follows:

• Factor 1: Pavlovian (Classical) Conditioning of Fear: In the early trials of the shuttlebox protocol, the neutral warning signals (dimming of lights, activation of the buzzer) were repeatedly and contiguously paired with the traumatic unconditioned stimulus (high-voltage electric shock). Through standard classical conditioning, the CS acquired secondary, aversive motivational properties. The CS was transformed into a conditioned fear signal capable of directly eliciting an involuntary, conditioned emotional response (CER)—a central state of autonomic fear and physiological distress.
• Factor 2: Instrumental (Operant) Negative Reinforcement: Once the CS became a powerful inducer of fear, the internal state of emotional distress itself functioned as a potent, aversive drive state (an internal stimulus drive, S_D). When the animal executed the instrumental motor response of jumping over the hurdle, this action instantly terminated the compound CS. The abrupt disappearance of the warning signal brought an immediate cessation or attenuation of the acute, visceral fear state. Therefore, the hurdle-clearing response was instrumentally reinforced not by the avoidance of the future shock, but by the immediate, negative reinforcement derived from escaping the current, internal agony of conditioned fear.

By framing the problem through this dual-process lens, Solomon and Wynne successfully rejected single-factor contiguity models. The animal was not acting for the sake of an abstract future non-event; it was acting in the immediate present to escape a terrifying conditioned sensory stimulus and its corresponding visceral autonomic turmoil.

5.2 The Internal Drive Reduction Hypothesis

At the very heart of the Two-Factor interpretation lay the Internal Drive Reduction Hypothesis. This theoretical construct represented an ingenious extension of Clark Hull’s drive-reduction mechanics into the realm of purely psychological and emotional states. In standard Hullian theory, an organism’s behavior is fueled by biological deficits (primary drives like hunger, thirst, or direct physical tissue damage), and reinforcement is defined as the homeostatic reduction of that biological tension.

Mowrer, Solomon, and Wynne expanded this formulation by demonstrating that acquired or conditioned drives possessed motivational and reinforcing capabilities fully equal to primary physiological drives. When the shuttlebox lights dimmed and the buzzer sounded, the external CS triggered a massive activation of the subject’s sympathetic nervous system. The dog experienced a rapid spike in epinephrine release, peripheral vasoconstriction, elevated heart rate, and an intense central emotional state of terror. This internal emotional state was intensely aversive; the organism possessed an urgent biological imperative to eradicate it.

Consequently, the instrumental hurdle jump functioned mechanically as an escape response directed against the warning signal itself. The animal was not “avoiding” an intellectualized shock that was scheduled to strike several seconds in the future; the animal was frantically escaping the present, intolerable sensory-emotional matrix created by the CS. The functional reinforcing event was the immediate offset of the light-dimming and the buzzer. Because this motor response led directly to the subjective relief of drive reduction, its associative habit strength (sHr) was powerfully reinforced, cementing the hurdle jump into the dog’s behavioral repertoire.

5.3 Empirical Verification of CS Termination as a Reinforcer

The internal validity of Two-Factor Theory hinged entirely on an empirical question: was the termination of the conditioned stimulus truly necessary and sufficient to reinforce the avoidance response, independently of the primary shock? If Two-Factor Theory was correct, modifying the contingency surrounding CS termination should radically alter the rate of learning and the stability of the avoidance habit.

Solomon, Wynne, and their contemporaries subjected this prediction to direct experimental manipulation through a series of elegant control designs. In one critical experimental variation, the apparatus was rewired so that when the dog jumped over the central hurdle during the 10-second warning window, the physical electric shock was successfully prevented, but the conditioned stimulus (darkness and buzzer) was programmed to persist for the full remaining duration of the 10 seconds. Under this contingency—where shock was avoided but the CS was not immediately terminated—the acquisition of the avoidance response was severely crippled. The animals learned at an agonizingly slow pace, and many failed to achieve the stable avoidance criterion entirely, remaining stuck in chronic escape patterns despite hundreds of trials.

Conversely, in experimental paradigms where the hurdle jump resulted in the immediate termination of the CS, but the overall probability of shock omission was systematically manipulated, the immediate offset of the warning signal proved to be the single most potent variable driving acquisition. Furthermore, researchers demonstrated that the visual and spatial cues of the safe, un-electrified compartment acquired powerful secondary reinforcing properties. The animal was simultaneously propelled by the aversive drive of the warning signal and drawn toward the conditioned safety signals (the fully illuminated, quiet chamber across the hurdle). These empirical demonstrations provided formidable confirmation for the Two-Factor architecture, establishing that negative reinforcement operating upon conditioned fear cues was the foundational engine driving shuttlebox avoidance acquisition.

6. The Avoidance Paradox and Extreme Resistance to Extinction

6.1 The Conceptual Dilemma of Self-Perpetuating Behavior

While Two-Factor Theory provided an elegant, coherent explanation for how avoidance behavior was initially acquired, it collided head-on with a massive theoretical crisis when confronted with the long-term performance of Solomon and Wynne’s dogs. This crisis became universally celebrated in learning literature as The Avoidance Paradox.

The logical paradox can be stated plainly: If Factor 1 (Pavlovian conditioning) is the absolute prerequisite for Factor 2 (instrumental reinforcement), then the entire theoretical system ought to contain the seeds of its own destruction. In classical Pavlovian conditioning, if a conditioned stimulus is presented repeatedly in the absence of the unconditioned stimulus, the conditioned response inexorably undergoes experimental extinction. This is one of the most fundamental, universally replicated laws of behavioral psychology. Once a dog mastered the shuttlebox avoidance response, it jumped within two seconds of CS onset. Because it jumped, the electric shock was never delivered. Therefore, trial after trial, the dog experienced the conditioned stimulus entirely alone, with no primary US reinforcement whatsoever.

Under strict Pavlovian principles, this continuous stream of CS-alone presentations should have caused rapid, progressive extinction of the conditioned fear response. As the fear elicited by the CS extinguished, the internal drive (S_D) motivating the hurdle jump should have evaporated. Without fear, there could be no fear reduction; without fear reduction, there could be no negative reinforcement; and without reinforcement, the instrumental hurdle jump should have ceased. The dog should have stopped jumping, stayed in the compartment, received a shock, reacquired the fear, and resumed jumping. Yet Solomon and Wynne observed nothing of the sort. Their dogs did not exhibit cyclic oscillations. Instead, the animals continued to leap the hurdle with unbroken, millisecond precision across 100, 200, 400, and in some subjects, more than 650 consecutive trials without receiving a single shock. The behavior appeared entirely self-perpetuating, immune to the universal law of extinction.

6.2 Conservation of Anxiety Hypothesis

Faced with this theoretical breakdown, Solomon and Wynne formulated an ingenious, highly influential hypothesis to reconcile Two-Factor Theory with their empirical observations: the Conservation of Anxiety Hypothesis (sometimes termed the anxiety preservation model). This hypothesis located the solution to the paradox within the precise chronometric latencies of the avoidance response itself.

Solomon and Wynne pointed out that experimental extinction of a Pavlovian conditioned response is not instantaneous; it requires cognitive and physiological exposure time. For an organism to extinguish its fear of a conditioned stimulus, the organism must remain in the presence of that CS long enough for the autonomic fear response to fully peak, mobilize, and subsequently experience the absence of the predicted catastrophe. Only when the animal is trapped in the presence of the CS and discovers that no trauma ensues can the underlying associative bond between the CS and the US undergo inhibitory restructuring.

In the overtrained shuttlebox dog, however, the response latency was extraordinarily brief—often measuring between 1.2 and 2.0 seconds. The dog leaped over the hurdle almost the instant the lights dipped and the buzzer sounded. Because the dog jumped so fast, the CS was terminated almost immediately after its onset. The subject effectively cut off the sensory signal long before the central nervous system could mobilize its full autonomic fear reaction, and long before the animal could ever register the reality that the grid floor was no longer capable of delivering a shock. The avoidance response acted as a temporal shield: by aborting the CS prematurely, the dog preserved or “conserved” its internal store of anxiety in a latent, pristine, unextinguished neurobiological state. The animal lived in an eternal subjective reality where the shock was always just a fraction of a second away from destroying it, ensuring that the behavior remained perpetually reinforced by the immediate relief of cutting the signal short.

6.3 Irreversibility and Functional Autonomy Hypotheses

Despite the elegance of the Conservation of Anxiety hypothesis, Solomon and Wynne were compelled by the sheer extremity of their empirical data to entertain an even more radical, controversial proposition: the Principle of Traumatic Irreversibility. In their 1953 monograph, they formally postulated that when an emotional reaction is conditioned under conditions of extreme, life-threatening primary trauma, the resulting central Pavlovian fear trace may become biologically irreversible. They suggested that high-intensity electric shock alters the central neural architecture in such a profound manner that the core S-R or S-S associative connection can never be truly eliminated through passive non-reinforcement.

This hypothesis bore a striking intellectual resemblance to the concept of “functional autonomy” introduced decades earlier by the personality theorist Gordon Allport. Allport had argued that behavioral habits originally acquired to serve basic biological drives could, over time, become independent motivational systems that sustain themselves entirely apart from their historical origins. Solomon posited that the overtrained avoidance jump might have transformed from an active fear-reduction mechanism into an autonomous, self-reinforcing motor habit. The motor act itself, hardwired through hundreds of rapid repetitions, might have become so deeply stamped into the subcortical motor pathways that it no longer required any mediational fear state whatsoever to trigger its execution.

The Irreversibility Hypothesis ignited ferocious debate across the psychological establishment. B.F. Skinner and other strict operant behaviorists vehemently resisted the notion of absolute behavioral irreversibility, arguing that claiming a behavior could never extinguish violated the foundational axioms of experimental behavior analysis. They asserted that if an avoidance habit appeared permanent, it was merely an artifact of an incomplete or improperly constructed extinction protocol. This methodological challenge pushed Solomon and Wynne to design a series of aggressive, intrusive experimental interventions designed to forcibly test whether this seemingly invincible avoidance habit could ever be shattered.

7. Extinction Procedures: Response Prevention, Flooding, and CS Termination

7.1 Failure of Conventional Extinction Paradigms

To systematically evaluate the Irreversibility Hypothesis, Solomon and Wynne instituted standard, conventional laboratory extinction procedures with their overtrained canine cohorts. In classical and operant conditioning, the universal protocol for extinction is the total, permanent termination of reinforcement: the food dispenser is emptied, or the shock generator is permanently disconnected from the experimental apparatus. Solomon and Wynne switched off the electrical power supply to the shuttlebox grid floors, ensuring that it was physically impossible for any animal to ever receive another shock.

The result of this conventional intervention was absolute empirical failure. The permanent disconnection of the shock generator produced virtually zero decline in the dogs’ avoidance performance. The animals continued to vault the hurdle the instant the lights dimmed and the buzzer sounded, clocking identical latencies of 1.6 to 2.5 seconds across day after day of testing. Because the dogs jumped with such hyper-efficient speed, they never stayed on the grid long enough to discover that the shock had been deactivated. They were executing an avoidance response against a ghost.

Even when experimenters introduced minor environmental variations or introduced brief scheduled breaks spanning days or weeks, the behavior remained rock-solid. Following testing interruptions, the dogs exhibited massive spontaneous recovery, bounding across the barrier with renewed vigor on the very first presentation of the compound CS. The standard rules of experimental extinction—derived primarily from appetitive rodent studies—were completely inapplicable to traumatic canine avoidance. The avoidance habit was impenetrable to passive non-reinforcement, demonstrating conclusively that so long as the animal possessed behavioral control over the termination of the conditioned stimulus, natural corrective learning could never take place.

7.2 Response Prevention (Flooding) as an Intervention

Realizing that passive extinction was totally impotent, Solomon, Wynne, and their research team engineered a radical, mechanically coercive experimental procedure: Response Prevention, an experimental paradigm that would later become famously known in clinical literature as Flooding. If the dog’s hyper-rapid jumping was the behavioral barrier preventing the extinction of fear, then the experimental solution was clear: the dog had to be physically prevented from jumping.

To accomplish this, the experimenters altered the central architecture of the shuttlebox. Immediately prior to a trial, a solid, transparent plate of heavy glass or an impenetrable wooden partition was dropped directly over the central hurdle, physically sealing the dog into a single compartment and completely eliminating any physical pathway to the safe side. The trial was then initiated according to standard protocol: the overhead lights were extinguished, the opposing lights flared, and the piercing auditory warning buzzer erupted through the speakers. The critical difference was that the 10-second temporal window was suspended. The dog was trapped in the presence of the active, terrifying conditioned stimulus for prolonged, uninterrupted exposure periods, ranging from several minutes to upwards of an hour, with the primary shock completely disconnected.

The behavioral phenotype observed during these early response-prevention trials was dramatic, visceral, and emotionally grueling to witness. During the initial minutes of forced exposure, the dogs exhibited a profound resurgence of the acute panic reactions seen on Trial 1. They launched themselves repeatedly and violently against the transparent glass barrier, clawed frantically at the wooden edges, shrieked, yelped, voided their bowels, and shivered violently in the darkened enclosure. They were trapped in the presence of a stimulus that their entire biological system screamed was an immediate harbinger of agonizing physical pain. However, because the shock generator was disconnected, the scheduled electrical disaster never arrived.

7.3 Mechanisms of Extinction Under Forced Exposure

As the minutes ticked by under forced response prevention, an unmistakable biphasic emotional curve began to unfold. After 10, 15, or 20 minutes of sustained, unremitting exposure to the active CS, the dog’s violent panic reactions began to wane. The somatic signs of autonomic hyperarousal slowly subsided: the vocalizations stopped, the frantic barrier-clawing ceased, the dog’s respiration shifted from rapid hyperventilation to heavy panting, and eventually, the animal collapsed onto the non-electrified grid floor in a state of behavioral exhaustion, resting its chin on its paws while the buzzer continued to sound and the room remained dark.

The neurobiological and psychological mechanisms operating during this forced exposure were profound:

1. Exhaustion of Sympathetic Arousal: The visceral sympathetic nervous system cannot sustain peak adrenergic activation indefinitely. Through natural neurochemical depletion and peripheral receptor desensitization, autonomic arousal inevitably decayed.
2. Reality Testing and Expectancy Disconfirmation: By being physically forced to endure the conditioned stimulus for prolonged durations without receiving an electrical shock, the animal was finally granted the necessary exposure time to register the complete absence of the unconditioned stimulus. The cognitive expectancy of impending agony was fundamentally disconfirmed by raw, incontrovertible reality.
3. Full Pavlovian Extinction: Because the CS was presented for sustained periods completely uncoupled from the US, true classical extinction occurred. The conditioned stimulus was systematically stripped of its acquired secondary aversive valence.

The proof of this transformation came when the experimenters finally removed the glass partition, restored the apparatus to its original configuration, and presented the compound CS once again. The effect was astonishing: the seemingly invincible, permanent avoidance habit had collapsed. The dogs did not jump. They stood calmly in the darkened compartment, listening to the buzzer without fleeing. When latencies were recorded, they stretched past the 10-second mark into infinite delay. Response prevention had achieved what hundreds of shock-free avoidance trials could never accomplish: the definitive, enduring eradication of the traumatic avoidance response.

8. Alternative Formulations: One-Factor Theory and Cognitive-Expectancy Models

8.1 Herrnstein and Hineline’s One-Factor Operant Model

Although Solomon and Wynne championed Mowrer’s Two-Factor framework, the inherent complexities and theoretical vulnerabilities of the model soon sparked the development of powerful alternative formulations. Foremost among these was the One-Factor Operant Theory, pioneered by Richard Herrnstein and Philip Hineline in the late 1960s. Herrnstein was deeply skeptical of the necessity of invoking an unobservable, subjective internal state—namely, “fear”—as a required intervening variable to explain avoidance learning.

Herrnstein and Hineline challenged the core foundation of Two-Factor Theory by constructing an ingenious experimental paradigm known as the nondiscriminated or free-operant avoidance schedule (often utilizing variations of the Sidman avoidance schedule). In these experiments, animals were placed into chambers where no discrete warning signals (no lights dimming, no buzzers sounding) were ever presented. Instead, shocks were programmed to occur randomly or at fixed temporal intervals (the Shock-Shock or S-S interval). If the animal pressed a lever or cleared a hurdle, it reset the clock, postponing the next shock by a specified duration (the Response-Shock or R-S interval). Under these conditions, where there was no external CS to terminate, animals nonetheless acquired and maintained stable, highly effective avoidance responding.

Based on these findings, Herrnstein argued for a parsimonious, single-factor operant model. The primary reinforcer in avoidance conditioning was not the immediate reduction of an internal fear drive elicited by an external warning cue; rather, it was the molar, mathematical reduction in the overall frequency of aversive events over time. Organisms were sensitive to environmental probability vectors: if emitting a specific motor response significantly lowered the statistical likelihood of experiencing pain per unit of time, that behavioral response was strengthened according to standard operant principles. One-Factor theory entirely eliminated the Pavlovian factor, asserting that avoidance was governed solely by instrumental contingencies operating directly on rate reduction.

8.2 Seligman and Johnston’s Cognitive-Expectancy Theory

In 1973, Martin Seligman and J. Bruce Johnston launched a radical cognitive critique of both One-Factor and Two-Factor formulations by publishing their Cognitive-Expectancy Theory of Avoidance. Seligman and Johnston argued that purely mechanical S-R associations and molecular drive-reduction models were fundamentally inadequate to explain the behavioral realities of the overtrained shuttlebox subject. They pointed out that well-trained avoidance animals do not look afraid; their heart rates do not spike, their endocrine profiles remain flat, and their motor output is casual and precise. If fear is no longer present, how can fear reduction be reinforcing the behavior?

Seligman and Johnston replaced mechanical drives with cognitive constructs: mental representations, subjective probabilities, and expectancies. They posited that during the acquisition phase in the shuttlebox, an animal constructs two distinct, competing cognitive expectancies regarding environmental outcomes:

• Expectancy A: If I execute the hurdle jump within 10 seconds of CS onset, no electric shock will occur.
• Expectancy B: If I do not execute the hurdle jump within 10 seconds of CS onset, an agonizing electric shock will occur.

In addition to these expectancies, the animal possesses an affective preference: the biological state of “no shock” is profoundly preferred over the state of “shock.” Once these cognitive expectancies are formed, the animal’s behavior is guided by rational utility maximization: it jumps because it expects that jumping leads to safety, while staying leads to pain.

This cognitive architecture resolved the Avoidance Paradox with stunning elegance. Why does the behavior resist extinction when the shock is turned off? Because every time the animal jumps, it validates Expectancy A (jumping leads to no shock). However, to discover that the contingency has changed—that the shock generator has been deactivated—the animal must remain in the compartment and allow Expectancy B to be put to the test. So long as the animal jumps, it never challenges Expectancy B. The cognitive belief that non-jumping equals catastrophe remains eternally untested and intact. Extinction can occur only when the animal is physically prevented from jumping (via response prevention), forcing it to experience the non-occurrence of shock following a non-jump, thereby shattering Expectancy B through direct cognitive disconfirmation.

8.3 Safety-Signal Hypotheses (D’Amato and Rescorla)

A third major theoretical counterweight emerged in the form of Safety-Signal Hypotheses, developed independently by researchers such as M. R. D’Amato and further enriched by Robert Rescorla‘s work on conditioned inhibition. These theorists suggested that psychologists had spent decades focusing on the wrong side of the shuttlebox: they had fixated exclusively on the fear and terror driving the animal away from the warning compartment, while ignoring the powerful appetitive and inhibitory forces pulling the animal into the safe compartment.

When a dog clears the hurdle and lands in the opposite chamber, it experiences a dramatic sensory transformation. The dark compartment is left behind; the piercing buzzer ceases; the physical barrier is now behind the dog; and the animal lands on an inert grid floor in a brightly illuminated space. These distinct environmental, proprioceptive, and spatial stimuli are uniquely correlated with one absolute biological reality: the total absence of shock. According to Rescorla’s framework, these cues become conditioned inhibitors of fear—formally designated as “safety signals.”

Safety signals do not merely inhibit fear; they acquire positive conditioned reinforcing properties. Organisms actively seek out and find comfort in safety cues. D’Amato demonstrated experimentally that if researchers introduced an explicit, salient feedback stimulus (such as a distinct light flash or chime) the instant the avoidance jump was completed, the acquisition of avoidance responding accelerated dramatically, and its resistance to extinction deepened profoundly. Viewed through this theoretical lens, shuttlebox avoidance was reconceptualized not merely as a desperate, terrified flight from dread, but as an active, motivated approach toward safety and security. The animal was leaping toward a conditioned haven of peace.

9. Evolutionary and Biological Constraints: Species-Specific Defense Reactions

9.1 Robert Bolles’ Theory of Species-Specific Defense Reactions (SSDRs)

By the early 1970s, the entire behaviorist paradigm was rocked by an ethological revolution that challenged the foundational assumption of the “equipotentiality premise”—the long-held dogma that any stimulus could be conditioned to any response with equal ease across any animal species. At the forefront of this conceptual overhaul in avoidance literature was Robert C. Bolles, whose 1970 paper introduced the theory of Species-Specific Defense Reactions (SSDRs).

Bolles argued that traditional laboratory learning theories were biologically naive. In the savage theater of evolutionary natural selection, an animal cannot afford to spend tens of trials engaging in trial-and-error instrumental learning to figure out how to escape a predator. If a wild rodent or canine had to rely on Thorndikian reinforcement or gradual drive reduction to survive its first encounter with a hawk or a wolf, it would be consumed before the first learning trial was ever completed. Evolution could not have left defense to the slow, clumsy mechanics of general operant conditioning.

Instead, natural selection pre-wires every vertebrate species with a tightly organized, hierarchical repertoire of innate defense behaviors—the SSDRs. When an organism detects sudden, imminent, life-threatening danger, its general appetitive and operant learning systems are immediately overridden and suppressed by ancient subcortical survival circuits. The primary mammalian SSDR repertoire consists of three fundamental, hardwired behavioral outputs:

• Flight: Explosive, rapid locomotion directed away from the threat source.
• Freezing: Complete, catatonic motor immobility designed to evade detection by motion-sensitive visual predators.
• Fighting: Explosive, aggressive defensive attack when cornered or physically restrained.

Bolles pointed out that the speed and success of avoidance learning in any laboratory apparatus is strictly determined by how closely the required experimental manipulandum aligns with the animal’s natural, innate SSDR hierarchy. The reason Solomon and Wynne’s dogs learned the shuttlebox avoidance jump with such breathtaking speed was not because their general-purpose instrumental learning machinery was exceptionally brilliant; it was because running and leaping over a physical barrier is the direct laboratory analogue of evolutionary Flight. The apparatus demanded an action that sat at the very pinnacle of the canine defense hierarchy.

9.2 Biological Preparedness and Contrapreparedness (Seligman)

Complementing Bolles’ ethological insights, Martin Seligman formulated the concept of Biological Preparedness. Seligman posited that evolutionary pressures have genetically prepared organisms to form specific associative connections with immense rapidity and minimal reinforcement, while leaving them “unprepared” or actively “contraprepared” to learn others.

This biological continuum explained decades of baffling laboratory anomalies in aversive conditioning:

• Prepared Responses: In the shuttlebox, jumping a hurdle, running through an open doorway, or bolting down an alleyway are biologically prepared responses for quadrupeds encountering pain or predatory warning signals. The neural circuits linking the perception of threat to the motor programs of locomotion and leaping are evolutionarily pre-assembled. Consequently, avoidance acquisition is instantaneous, robust, and virtually permanent.
• Contraprepared Responses: Conversely, if an experimenter attempts to train an animal to perform an arbitrary, unnatural operant response to avoid an electric shock—such as requiring a rat to press a lever, spin a running wheel backward, or requiring a dog to scratch its nose or yawn—learning is exceptionally difficult, and frequently impossible. When terrified, an animal’s innate SSDRs actively suppress fine-motor manipulation. A rat in the throes of fear will freeze or frantically bite the lever, but it cannot calmly execute an arbitrary operant press.

Solomon and Wynne’s canine data had to be reinterpreted through this evolutionary matrix. The apparent invincibility and hyper-fast consolidation of their shuttlebox jumping response was an emergent property of testing an animal on an evolutionary prepared task. The hurdle jump was not an arbitrary, neutral operant stamped in by raw reinforcement; it was an ancient, highly canalized mammalian survival program unlocked by the experimental parameters of the Harvard shuttlebox.

9.3 Comparative Avoidance Across Avian and Mammalian Taxa

The ethological critique gained further weight as comparative psychologists extended the shuttlebox paradigm across diverse taxonomic classes, revealing dramatic interspecies differences that completely shattered the notion of a universal, monolithic learning curve. The behavioral strategies deployed in the shuttlebox varied wildly depending on the specific ecological niche and predatory imminence strategies of the subject species.

When rodents (such as laboratory rats or mice) were placed into a two-way shuttlebox—where the animal must repeatedly run back and forth between two compartments that both deliver shock on alternating trials—acquisition proved notoriously difficult. Rodents rely heavily on spatial territoriality and scent marking; being forced to run back into a compartment where they were violently shocked only two minutes prior directly violates their natural defensive instincts. Furthermore, when exposed to unescapable footshock, the rat’s primary SSDR is profound, immotile freezing. In a standard shuttlebox, a freezing rat simply sits on the grid, absorbing continuous electrical shocks because the immobility reflex prevents the execution of the running response. To get a rat to learn avoidance, researchers had to switch to “one-way” avoidance boxes (where the animal always flees to a dedicated, permanently safe compartment) or design elevated runways that trigger flight.

Pigeons, operating under an avian sensory-motor framework, exhibit entirely different defense dynamics. When threatened with shock, a pigeon’s primary SSDR is wing-flapping and flight; attempting to train a pigeon to peck a key to avoid a shock is notoriously ineffective, whereas training it to flap its wings to terminate a warning signal occurs almost immediately. Non-human primates and canines, possessing advanced spatial mapping, high locomotive agility, and sophisticated behavioral flexibility, adapt to the two-way shuttlebox with far greater ease, rapidly overcoming the paradox of returning to a previously electrified chamber. Solomon and Wynne’s data, therefore, captured a specific convergence: a large, predatory carnivore endowed with powerful athletic leaping capacity, placed into an apparatus that perfectly accommodated its evolutionary flight response.

10. Neurobiological Mechanisms Underlying Shuttlebox Avoidance Learning

10.1 The Amygdaloid Circuitry in Fear Conditioning and Flight

While Solomon and Wynne conducted their behavioral experiments decades before the dawn of modern neuroanatomy and optogenetics, the past thirty years of translational neuroscience have mapped the precise neural architecture executing the shuttlebox avoidance program. At the absolute center of this neurobiological network lies the amygdala, a complex nuclear group embedded deep within the temporal lobe, functioning as the central hub for the acquisition, consolidation, and expression of aversive associations.

During the early acquisition phase in the shuttlebox, sensory inputs processing the conditioned stimulus (auditory projections from the medial geniculate nucleus of the thalamus and auditory cortex; visual projections from the visual cortex) converge simultaneously with the noxious somatosensory pathways carrying the unconditioned shock (spinothalamic tract inputs via the posterior thalamus) onto individual pyramidal neurons within the basolateral amygdala (BLA). Through NMDA-receptor-dependent long-term potentiation (LTP), the synaptic connections conveying the harmless CS are dramatically strengthened, allowing the warning cue alone to trigger robust neuronal firing throughout the BLA complex.

The downstream flow of information within the amygdaloid architecture governs the critical behavioral transition from passive terror to active instrumental avoidance:

• The Central Amygdala (CeA) and Passive Freezing: In the early trials, the BLA strongly projects to the central nucleus of the amygdala (CeA). The CeA orchestrates the primitive, unconditioned autonomic and somatic fear cascade via projections to the midbrain periaqueductal gray (PAG; driving freezing and vocalization), the lateral hypothalamus (activating the sympathetic nervous system, inducing tachycardia), and the paraventricular nucleus of the hypothalamus (triggering the HPA axis and massive cortisol/corticosterone release).
• The Amygdalo-Striatal Shift to Active Avoidance: As the animal shifts from passive escape to active avoidance, modern optogenetic studies demonstrate that synaptic drive is rerouted. While the CeA must be transiently suppressed to permit movement (disinhibiting the ventrolateral PAG to abolish freezing), the BLA establishes powerful, direct glutamatergic projections to the ventral striatum (nucleus accumbens) and the dorsal striatum. This BLA-to-accumbens projection is the critical neurobiological bridge converting a Pavlovian fear signal into an active, motivated instrumental motor command.

10.2 Striatal and Cortical Interactions in Active Instrumental Escape

Once the amygdalar output engages the striatal machinery, the execution of the avoidance jump transitions into the domain of the basal ganglia and the fronto-striatal loops. The nucleus accumbens functions as the primary limbic-motor interface, integrating the emotional valence routed from the basolateral amygdala with motivational drive, before signaling the dorsal striatum to organize and release the complex motor sequences required to jump the central barrier.

This process is heavily regulated by discrete regions of the prefrontal cortex (PFC):

• Prelimbic Cortex (PL) and Action Initiation: In rodents and carnivores, the prelimbic cortex (homologous to human dorsolateral and anterior cingulate networks) is essential for the early expression of active avoidance. Neurons within the prelimbic cortex sustain elevated firing rates during the 10-second warning interval, maintaining working memory of the threat contingency and driving downstream striatal targets to initiate the motor response before the timer expires.
• The Dorsomedial to Dorsolateral Striatal Transition (Habit Formation): In early avoidance training, the behavior is goal-directed, highly conscious, and encoded within the dorsomedial striatum (caudate nucleus), which processes explicit action-outcome (A-O) contingencies. However, as Solomon and Wynne documented during the hundreds of plateau-phase trials, the behavior undergoes profound overlearning. Neurobiologically, this corresponds to an anatomical handoff of control from the dorsomedial striatum to the dorsolateral striatum (putamen). The avoidance jump ceases to be a flexible goal-directed action and transforms into a rigid, automatic, stimulus-response (S-R) motor habit that is executed with lightning speed and requires negligible prefrontal cortical oversight.
• Dopaminergic Omission Signaling: Elegant modern in vivo fiber photometry studies have revealed that when an animal successfully clears the hurdle during the CS interval, a massive, transient burst of dopamine is released within the nucleus accumbens. This dopamine burst is not triggered by a physical reward; it is an intrinsic neurochemical reward prediction error signaling the successful omission of expected trauma. This dopaminergic surge directly reinforces the corticostriatal synapses encoding the hurdle jump, providing a precise biological substrate for the reinforcement of non-events.

10.3 Neuroplasticity and the Neural Traces of Extinction

When Solomon and Wynne implemented their forced response prevention (flooding) procedure, they were unknowingly engaging the complex neurocircuitry of fear extinction—a process that modern neuroscience has proven is not the literal erasure of the original fear trace, but the de novo synthesis of an active, inhibitory memory trace that overrules the original fear association.

The neuroanatomical epicenter of extinction learning is the ventromedial prefrontal cortex (vmPFC), specifically the infralimbic cortex (IL) in mammals. When the dog is trapped behind the glass partition during response prevention, forced to endure the sustained CS without shock, the infralimbic cortex begins to exhibit sustained, plastic activation. The IL projects dense, powerful glutamatergic axonal fibers directly into the amygdaloid complex, but crucially, it does not synapse onto the principal fear-generating cells of the BLA. Instead, it terminates on a specialized meshwork of inhibitory interneurons known as the intercalated cell masses (ITC).

The intercalated cells are GABAergic neurons that act as a high-security gatekeeper sitting between the basolateral amygdala and the central amygdala. When the infralimbic cortex fires, it excites the ITC cells, which in turn release massive waves of the inhibitory neurotransmitter GABA directly onto the central nucleus, effectively clamping down on CeA output and extinguishing autonomic fear expression. Concurrently, the vmPFC sends inhibitory projections to the striatal regions that execute the habitual motor jump.

At the molecular level, this forced extinction process requires robust neuroplasticity, driven by the activation of NMDA receptors, brain-derived neurotrophic factor (BDNF) synthesis, and structural remodeling of dendritic spines within the vmPFC-ITC pathway. This modern neurobiological understanding has opened exciting translational frontiers: pharmacological agents such as D-cycloserine (a partial NMDA receptor agonist) have been demonstrated to dramatically accelerate the extinction of avoidance responses in both laboratory shuttleboxes and human clinical trials, proving that molecular facilitation of prefrontal neuroplasticity can permanently dissolve the most stubborn traumatic avoidance habits.

11. Clinical Implications: Phobias, Anxiety Disorders, and Exposure Therapy

11.1 The Shuttlebox as an Etiological Model for Human Neuroses

The true genius of Solomon and Wynne’s 1953 experiments lay not merely in their contributions to pure animal behaviorism, but in their immediate, profound explanatory power when applied to human clinical psychiatry. For decades, psychoanalytic doctrine had dominated the conceptualization of neuroses, attributing phobias, chronic panics, and compulsive behaviors to complex, unprovable unconscious conflicts, repressed psychosexual dynamics, and symbolic ego defenses.

Solomon and Wynne provided the empirical hammer that shattered these esoteric formulations. The shuttlebox demonstrated that severe, chronic, seemingly irrational “neurotic” symptoms could be generated, maintained, and permanently fixed purely through the objective, predictable laws of aversive conditioning and negative reinforcement. The canine bounding over a hurdle within two seconds of a light dimming was an exact mechanical analogue of a human patient fleeing an elevator, crossing the street to avoid an approaching dog, or bolting from a crowded subway station the instant an internal flutter of anxiety was detected.

In human anxiety pathology—whether in specific phobias, agoraphobia, or panic disorder—the cardinal maintaining factor is identical to the shuttlebox dynamic: behavioral avoidance. An individual experiences a traumatic panic attack in an enclosed space (the equivalent of the shuttlebox shock). The environmental cues of the enclosed space become conditioned stimuli evoking acute fear. The individual immediately flees the space (the escape response). Soon, the person begins to anticipate the danger, avoiding enclosed spaces altogether (the active avoidance response). Because the person flees at the first sign of discomfort, they never remain in the environment long enough to discover that the panic attack will peak, plateau, and naturally subside without causing death, insanity, or cardiac arrest. The avoidance response systematically protects the irrational cognitive expectancy from ever being disconfirmed, freezing the patient in an eternal cycle of anxiety conservation.

11.2 Mechanistic Parallels in Obsessive-Compulsive Disorder (OCD)

Perhaps the most direct, elegant clinical translation of Solomon and Wynne’s two-factor shuttlebox framework is found in the modern conceptualization and treatment of Obsessive-Compulsive Disorder (OCD). Historically viewed as an intractable, bizarre psychiatric illness, OCD maps with uncanny, one-to-one fidelity onto the two-factor avoidance paradigm.

The clinical architecture of OCD operates along the exact same two-factor axis:

• Factor 1 (The Obsession as Conditioned Fear Cue): An intrusive thought, image, or urge—such as the sudden thought “my hands are contaminated with a deadly pathogen” or “the stove has been left on to burn the house down”—erupts into consciousness. Through associative conditioning and cognitive biases, this intrusive thought functions precisely like the shuttlebox buzzer. It triggers an immediate, massive spike in internal emotional distress, catastrophic dread, and visceral autonomic panic.
• Factor 2 (The Compulsion as Instrumental Avoidance): Faced with this intolerable internal distress, the patient executes a rigid, stereotyped motor or mental ritual—such as scrubbing their hands with boiling water, checking the stove locks ten times, or chanting a mental phrase. This compulsive ritual functions as the exact clinical counterpart of the shuttlebox hurdle jump. The moment the compulsion is completed, the intrusive anxiety drops precipitously. The patient experiences immediate, profound emotional relief.

This drop in anxiety provides powerful, immediate negative reinforcement for the compulsive ritual. However, like the dog’s two-second jump, the compulsion carries a catastrophic long-term clinical cost. Because the patient washes their hands the instant the intrusive doubt strikes, they never give their nervous system the opportunity to discover that the pathogen would not have killed them, or that the anxiety would have naturally dissipated on its own. The compulsion conserves the obsession. Over months and years, these compulsive rituals undergo the exact same overlearning observed in the Harvard shuttlebox, transitioning from flexible, goal-directed relief-seeking into rigid, automatic, motoric habits that consume hours of the patient’s day, inflicting catastrophic functional impairment on their occupational, social, and personal lives.

11.3 Development of Exposure and Response Prevention (ERP) Therapy

The profound clinical value of Solomon and Wynne’s work materialized when behavioral clinicians realized that if the shuttlebox model accurately explained how human anxiety and compulsions were maintained, then the shuttlebox extinction protocols must contain the blueprint for their cure. The breakthrough came in 1966, when the British-South African psychologist Victor Meyer published his seminal clinical trials adapting Solomon and Wynne’s response-prevention techniques directly to human psychiatric inpatients suffering from severe, intractable OCD.

Meyer recognized that standard talk therapy, psychoanalysis, and passive reassurance were utterly useless—they were the clinical equivalents of turning off the shock generator in the animal shuttlebox, which had completely failed to stop the dogs from jumping. To cure the disorder, the clinician had to implement the human analogue of the glass partition: the patient had to be subjected to Exposure and Response Prevention (ERP).

The operational protocol of ERP represents a direct, pure translation of the Harvard laboratory procedures:

1. Exposure: The patient is deliberately, systematically brought into prolonged contact with the exact conditioned stimuli that trigger their catastrophic dread (e.g., intentionally touching a public bathroom doorknob, intentionally leaving the stove unchecked, or imagining the feared catastrophic scenario). This is the functional equivalent of plunging the shuttlebox into darkness and sounding the buzzer.
2. Response Prevention: The patient is strictly, absolute blocked from executing the avoidant motor or mental compulsion (e.g., the patient is barred from washing their hands, forbidden from checking the locks, and prevented from executing neutralizing mental rituals). The patient is forced to remain in the contaminated or uncertain situation with their avoidance pathways completely closed off.

Just as Solomon and Wynne documented with their dogs, the human patient undergoing ERP exhibits a profound biphasic emotional response. In the initial minutes, the patient experiences a massive surge of terror, panic, and an overwhelming physical urge to execute the ritual. Yet, if response prevention is rigorously maintained across 45, 60, or 90 minutes, the patient’s sympathetic nervous system inevitably exhausts itself. The physiological panic peaks, plateaus, and decays. More importantly, the patient undergoes profound cognitive restructuring: raw reality proves that the catastrophic consequence did not materialize, and the brain realizes that the obsession possesses zero intrinsic causal power. Today, ERP stands as the indisputable first-line psychological intervention for OCD worldwide, boasting recovery rates that revolutionized psychiatry—a therapeutic triumph whose direct lineage traces straight back to the Harvard shuttlebox.

11.4 Post-Traumatic Stress Disorder (PTSD) and Avoidance Topographies

The clinical reach of the shuttlebox paradigm extends with equal power into the understanding and treatment of Post-Traumatic Stress Disorder (PTSD). In PTSD, an individual has experienced an authentic, life-threatening trauma—a real-world unconditioned stimulus of overwhelming horror, such as combat violence, sexual assault, severe vehicular collisions, or natural disasters.

Following this trauma, a vast constellation of environmental, sensory, and internal stimuli become conditioned cues linked to the original traumatic memory. A veteran hearing a car backfire, smelling diesel exhaust, or walking through a crowded marketplace experiences an immediate, explosive activation of the amygdaloid fear circuits, reliving the original horror through intrusive flashbacks and acute somatic panic. In response to this torment, the trauma survivor develops pervasive, multi-layered avoidance topographies:

• Behavioral Avoidance: The individual strictly avoids physical locations, people, conversations, sounds, or dates that serve as sensory reminders of the traumatic event, severely constricting their geographic and social life.
• Cognitive and Emotional Avoidance: The individual resorts to substance abuse, emotional numbing, dissociation, and aggressive mental suppression to avoid experiencing the internal sensations of fear, grief, and hyperarousal.

Just as in the shuttlebox, this pervasive avoidance functions as the ultimate maintaining engine of the pathology. By constantly fleeing trauma-related cues, the survivor ensures that the traumatic memory network remains completely isolated from natural corrective information. The brain never learns that the war is over, that the assault is in the past, or that a car backfire is not an improvised explosive device. Drawing directly upon the principles of forced exposure and response prevention, contemporary evidence-based PTSD treatments—most notably Prolonged Exposure (PE) therapy developed by Edna Foa—utilize systematic imaginal exposure (repeatedly revisiting the traumatic memory in vivid detail) and in vivo exposure (gradually returning to safe, avoided real-world situations) without allowing cognitive or physical avoidance. By blocking avoidance and enforcing prolonged exposure, PE drives the complete Pavlovian extinction of traumatic fear, demonstrating once again the universal clinical power of Solomon and Wynne’s fundamental findings.

12. Contemporary Legacy and Modern Re-Evaluations of Solomon and Wynne’s Paradigm

12.1 Methodological and Theoretical Re-Evaluations

Looking back across more than seven decades since the publication of Solomon and Wynne’s 1953 monograph, the shuttlebox paradigm continues to command immense intellectual respect, even as contemporary behavioral science has critically re-evaluated its methodologies and theoretical constructs. From a methodological standpoint, modern researchers must acknowledge the radical historical shifts in animal welfare ethics. The use of companion animals (canines) subjected to 1000-volt alternating current shocks and pushed to states of acute, screaming panic would be completely impossible under modern Institutional Animal Care and Use Committee (IACUC) regulations. While these intense parameters were essential to generate the phenomenon of “traumatic” avoidance that Solomon sought, modern neuroscience has successfully transitioned these protocols to specialized rodent paradigms, utilizing carefully calibrated micro-currents that minimize suffering while preserving the essential behavioral dynamics.

Theoretically, the historical debate between One-Factor, Two-Factor, and Cognitive models has reached an elegant, contemporary synthesis within the framework of computational psychiatry and hierarchical Bayesian reinforcement learning. Rather than viewing these theories as mutually exclusive, modern computational models recognize that avoidance learning operates across multiple, nested parallel systems in the mammalian brain:

• The Model-Free System: Maps directly onto the classic S-R habits and two-factor autonomic drive reduction, mediated by the basolateral amygdala and the dorsolateral striatum. It operates rapidly, reflexively, and with minimal computational cost, driving the fast, two-second plateau jumps observed in the shuttlebox.
• The Model-Based System: Maps onto Seligman’s cognitive-expectancy theory, mediated by the prefrontal cortex and the dorsomedial striatum. It computes forward-looking probability trees, outcome expectancies, and counterfactual evaluations (e.g., “what would happen if I did not jump?”), governing the flexible acquisition and conscious restructuring of defense behaviors.

By leveraging contemporary optogenetic, chemogenetic, and two-photon calcium imaging tools in rodents navigating modified shuttleboxes, twenty-first-century neuroscientists are dissecting these computational circuits at the level of individual, genetically identified cell types, confirming that the dual-process dynamics intuited by Solomon, Wynne, and Mowrer possess profound physical reality within the mammalian connectome.

12.2 Impact on Translational Neuroscience and Pharmacology

The enduring vitality of the shuttlebox paradigm is nowhere more evident than in the fields of translational neuropsychiatry and behavioral psychopharmacology. For more than half a century, the automated active avoidance shuttlebox has served as an indispensable, high-throughput screening battery for the discovery and validation of novel psychiatric medications.

The shuttlebox provides a unique, highly sensitive behavioral assay capable of dissociating non-specific motor sedation from true clinical anxiolytic and antidepressant efficacy:

• Screening Anxiolytics: Compounds that modulate GABAergic neurotransmission (such as benzodiazepines) or serotonergic tone (such as SSRIs and buspirone) demonstrate characteristic profiles in the shuttlebox. A true anxiolytic selectively increases the rate of avoidance extinction during non-reinforced trials or accelerates habituation during response prevention, without impairing the animal’s physical capacity to leap the hurdle during baseline escape trials.
• Screening Antipsychotics: Historical neuroleptics (such as chlorpromazine and haloperidol) were originally characterized by their unique ability to cleanly block the conditioned avoidance response (the animal fails to jump during the 10-second warning signal) while leaving the primary unconditioned escape response completely intact (the animal jumps immediately upon shock delivery). This precise behavioral dissociation proved predictive of dopamine D2 receptor antagonism in human mesolimbic circuits.
• Biomarkers of Stress Resiliency vs. Vulnerability: In modern preclinical stress neurobiology, the shuttlebox is utilized to identify genetic, epigenetic, and neurochemical biomarkers that differentiate resilient organisms from vulnerable ones. Animals that fail to acquire avoidance, collapsing into passive freezing or learned helplessness, exhibit distinct neuro-inflammatory profiles, altered microglial activation, and downregulated neurotrophin levels in the hippocampus compared to subjects that successfully acquire proactive active avoidance, providing vital clues into the biological etiology of treatment-resistant major depression.

12.3 Enduring Contributions to Behavioral Science

The conceptual footprint left by Richard Solomon and Lyman Wynne extends across the entire architecture of modern psychological science. By constructing an unyielding, rigorous laboratory model of traumatic avoidance, they forced the psychological establishment to confront the deepest paradoxes of motivation, learning, and behavioral persistence. Their empirical work established the definitive gold standard for how to execute rigorous, chronometrically precise aversive conditioning research, providing the empirical foundation that validated dual-system learning theories over simplistic single-factor dogmas.

Their findings delivered an unforgettable demonstration of the terrifying power of behavioral history. They proved that an organism can become permanently trapped by its own adaptive successes: by mastering the ability to preemptively avoid pain, the organism inadvertently cuts itself off from the very environmental feedback required to discover that the world has become safe. The shuttlebox illuminated the tragic irony at the heart of all clinical anxiety—that the very behaviors we invent to protect ourselves from suffering end up becoming the prison that preserves our dread.

Ultimately, the legacy of Solomon and Wynne’s shuttlebox experiment is celebrated not merely because it solved a technical puzzle in mid-century learning theory, but because it bridged the vast chasm between basic animal behaviorism and human psychiatric healing. The drop gates, the dimming lights, and the wooden hurdle of their Harvard apparatus provided the direct intellectual lineage that led to Exposure and Response Prevention therapy, liberating millions of human beings worldwide from the crushing torments of phobias, panic, and obsessive-compulsive rituals. In the grand tapestry of scientific inquiry, few experimental paradigms have ever achieved such a magnificent, enduring synthesis of empirical precision, theoretical depth, and compassionate clinical utility.

Conclusion

The shuttlebox avoidance experiments conducted by Richard L. Solomon and Lyman C. Wynne in the early 1950s stand as a monumental achievement in experimental psychology and translational neuroscience. What began as an empirical inquiry into the boundaries of Hullian drive reduction and Pavlovian extinction revealed an unexpected, fundamental truth about mammalian defensive systems: once an organism learns to actively prevent primary trauma through the rapid execution of a signal-driven motor response, that behavior can achieve near-infinite operational persistence. By documenting the seemingly self-perpetuating nature of avoidance, Solomon and Wynne uncovered the Conservation of Anxiety phenomenon, proved that passive non-reinforcement is entirely impotent against overlearned defensive habits, and demonstrated that true extinction requires the rigorous, physical blocking of avoidance through response prevention.

Their work decisively elevated Two-Factor Learning Theory into one of the most influential frameworks of the twentieth century, while simultaneously stimulating the birth of One-Factor operant models, cognitive-expectancy paradigms, safety-signal hypotheses, and ethological theories of species-specific defense reactions. Decades later, modern molecular and systems neuroscience has validated the physiological reality of these behavioral insights, mapping the transition from amygdalar panic to striatal habit, and confirming the indispensable role of the ventromedial prefrontal cortex in orchestrating the inhibitory learning that underlies extinction.

Most profoundly, Solomon and Wynne provided the empirical and conceptual bedrock for modern clinical psychology. By demonstrating that avoidance acts as an impenetrable shield preventing the natural disconfirmation of catastrophe, they laid bare the core engine driving human phobias, panic disorder, obsessive-compulsive rituals, and post-traumatic stress disorder. The response-prevention techniques forged within their laboratory became the direct foundational architecture of Exposure and Response Prevention therapy, transforming clinical practice and providing an evidence-based pathway to psychological recovery for millions of patients across the globe. The Harvard shuttlebox remains an enduring testament to the power of basic behavioral science—a timeless reminder that by rigorously illuminating the darkest mechanics of fear and avoidance in the animal laboratory, we uncover the ultimate mechanisms to liberate the human mind from the prisons of its own design.

References

• Bolles, R. C. (1970). Species-specific defense reactions and avoidance learning. Psychological Review, 77(1), 32–48. https://doi.org/10.1037/h0028589
• D’Amato, M. R., Fazzaro, J., & Etkin, M. (1968). Anticipatory responding and avoidance discrimination as factors in avoidance conditioning. Journal of Experimental Psychology, 77(1), 41–47. https://doi.org/10.1037/h0025765
• Foa, E. B., & Kozak, M. J. (1986). Emotional processing of fear: Exposure to corrective information. Psychological Bulletin, 99(1), 20–35. https://doi.org/10.1037/0033-2909.99.1.20
• Herrnstein, R. J., & Hineline, P. N. (1966). Negative reinforcement as shock-frequency reduction. Journal of the Experimental Analysis of Behavior, 9(4), 421–430. https://doi.org/10.1901/jeab.1966.9-421
• Hull, C. L. (1943). Principles of behavior: An introduction to behavior theory. Appleton-Century-Crofts.
• LeDoux, J. E. (2000). Emotion circuits in the brain. Annual Review of Neuroscience, 23(1), 155–184. https://doi.org/10.1146/annurev.neuro.23.1.155
• Meyer, V. (1966). Modification of expectations in cases with obsessional rituals. Behaviour Research and Therapy, 4(4), 273–280. https://doi.org/10.1016/0005-7967(66)90023-4
• Miller, N. E. (1948). Studies of fear as an acquirable drive: I. Fear as motivation and fear-reduction as reinforcement in the learning of new responses. Journal of Experimental Psychology, 38(1), 89–101. https://doi.org/10.1037/h0058455
• Mowrer, O. H. (1947). On the dual nature of learning—a re-interpretation of “conditioning” and “problem-solving”. Harvard Educational Review, 17(2), 102–148.
• Mowrer, O. H. (1960). Learning theory and behavior. John Wiley & Sons. https://doi.org/10.1037/10802-000
• Rescorla, R. A. (1969). Pavlovian conditioned inhibition. Psychological Bulletin, 72(2), 77–94. https://doi.org/10.1037/h0027760
• Seligman, M. E. P. (1971). Phobias and preparedness. Behavior Therapy, 2(3), 307–320. https://doi.org/10.1016/S0005-7894(71)80064-3
• Seligman, M. E. P., & Johnston, J. C. (1973). A cognitive theory of avoidance learning. In F. J. McGuigan & D. B. Lumsden (Eds.), Contemporary approaches to conditioning and learning (pp. 69–110). V. H. Winston & Sons.
• Sidman, M. (1953). Avoidance conditioning with brief shock and no exteroceptive warning signal. Science, 118(3058), 157–158. https://doi.org/10.1126/science.118.3058.157
• Solomon, R. L., Kamin, L. J., & Wynne, L. C. (1953). Traumatic avoidance learning: The outcomes of several extinction procedures with dogs. The Journal of Abnormal and Social Psychology, 48(2), 291–302. https://doi.org/10.1037/h0058807
• Solomon, R. L., & Wynne, L. C. (1953). Traumatic avoidance learning: Acquisition in normal dogs. Psychological Monographs: General and Applied, 67(4), 1–19. https://doi.org/10.1037/h0093649
• Solomon, R. L., & Wynne, L. C. (1954). Traumatic avoidance learning: The principles of anxiety conservation and partial irreversibility. Psychological Review, 61(6), 353–385. https://doi.org/10.1037/h0054540
• Thorndike, E. L. (1898). Animal intelligence: An experimental study of the associative processes in animals. The Psychological Review: Monograph Supplements, 2(4), i–109. https://doi.org/10.1037/h0092987