The Autoshaping Experiment (Sign Tracking) – Paul Brown and Herbert Jenkins

In the middle of the twentieth century, experimental psychology was dominated by a resolute theoretical partition. On one side stood the classical reflexology of Ivan Pavlov, conceptualized as an associative engine through which involuntary, autonomic responses became tethered to neutral predictive cues via stimulus-reinforcer contingencies. On the other side stood the formidable operant edifice constructed by B.F. Skinner and Edward Thorndike, which maintained that skeletal motor actions—the voluntary movements an organism executes within its spatial environment—were sculpted and maintained solely through response-reinforcer contingencies. Within the standard behaviorist canon, skeletal movements could not be unconditionally elicited by mere predictive signals; rather, they were emitted by the organism and subsequently selected by their post-hoc environmental consequences through the progressive mechanism of reinforcement. This foundational paradigm rested upon the belief that organisms were malleable, tabula rasa biological engines whose somatic topographies could be molded arbitrarily by carefully calibrated schedules of reinforcement.

This long-standing theoretical consensus was profoundly destabilized in 1968 by two researchers working at McMaster University: Paul L. Brown and Herbert M. Jenkins. Their classic paper, entitled “Auto-shaping of the Pigeon’s Key-Peck,” documented a phenomenon so deceptively simple yet theoretically recalcitrant that it fundamentally altered the trajectory of associative learning theory. By exposing naive, food-deprived domestic pigeons (Columba livia) to a completely non-contingent procedure—wherein a small visual key was illuminated for eight seconds immediately prior to the automatic presentation of grain, irrespective of what the bird was doing—Brown and Jenkins observed that the animals spontaneously began to approach, orient toward, and vigorously peck the illuminated key. There was no human experimenter patiently shaping the bird through successive approximations; there was no instrumental requirement stipulating that the bird must contact the key to receive the food. The physical signal itself appeared to draw the motor behavior out of the bird, transforming a neutral piece of illuminated ground glass into a powerful magnet for targeted consummatory behavior.

Initially labeled “autoshaping” because the procedure appeared to automate the traditionally laborious manual shaping process utilized by operant conditioners, the phenomenon quickly exposed a profound biological and associative reality: the animal was not learning an operant response under the control of an accidental consequence, but was exhibiting what later became widely known as “sign-tracking.” The illuminated key—a conditioned stimulus (CS)—had absorbed the motivational and consummatory qualities of the grain—the unconditioned stimulus (US)—compelling the pigeon to engage the predictive sign as if it were the consummatory goal itself. The following treatise provides an exhaustive, multi-dimensional analysis of the Brown and Jenkins experiment, tracking its historical antecedents, methodological architecture, physiological and morphological intricacies, theoretical fallout, neurobiological substrates, computational models, and profound translational relevance to contemporary neuropsychiatry, addiction, and human impulse control disorders.

1. Historical Context and Theoretical Foundations of Conditioning Prior to 1968

1.1 The Classical Skinnerian Paradigm of Operant Conditioning

To understand the profound disruption provoked by the Brown and Jenkins discovery, one must first inhabit the intellectual climate of American comparative psychology in the mid-1960s. For decades, the dominant theoretical engine driving experimental behavior analysis was operant conditioning, formulated in its most rigorous and dogmatic iteration by B.F. Skinner. Skinner’s theoretical model rested upon the foundational construct of the three-term contingency: the discriminative stimulus ($S^D$), the operant response ($R$), and the reinforcing consequence ($S^R$). Within this operational triad, the discriminative stimulus did not elicit the behavior in the manner of a physiological reflex. Instead, it merely set the occasion upon which a given behavioral emission would be met with an appetitive or aversive outcome. The causal nexus of learning was unequivocally located in the backward-acting influence of the reinforcer upon the emitted response.

Central to this methodological worldview was the process of manual successive approximation, commonly termed “shaping.” Because complex or localized skeletal motor actions—such as a pigeon pressing its beak against a small translucent plastic disc, or a rat depressing a stainless-steel lever—were vanishingly rare within the baseline operant repertoire of naive organisms, experimenters had to engineer the emergence of the target behavior. The experimenter sat beside the conditioning chamber, observing the subject’s spontaneous exploratory movements. When the pigeon casually oriented toward the wall containing the key, the researcher triggered the food magazine. When the bird stepped closer, another reinforcer was administered. Successive stages required the animal to face the key, raise its head, step within striking distance, peck in the general spatial vicinity, and ultimately strike the translucent disc with sufficient kinetic force to actuate a microswitch. This tedious, manually mediated protocol reinforced the prevailing theoretical doctrine: novel, targeted skeletal topographies were wholly dependent upon response-contingent reinforcement schedules.

Moreover, the Skinnerian paradigm enforced a strict, non-overlapping ontological division separating emitted voluntary behaviors from elicited involuntary reflexes. Skeletal musculature, mediated by the somatic nervous system, was considered the exclusive domain of operant learning, governed by the Law of Effect. In contrast, visceral, glandular, and autonomic effectors were relegated to Pavlovian classical conditioning. This theoretical partition maintained that while an animal could be classically conditioned to salivate or alter its cardiac rhythm in response to a neutral cue, it could never be made to execute localized, targeted skeletal adjustments toward a stimulus purely through stimulus-reinforcer contingencies. The organism had to emit the motor program voluntarily; the environment then selected it.

1.2 Pavlovian Conditioning and the Stimulus-Substitution Doctrine

Parallel to the American operant tradition stood the classical conditioning paradigm developed by Ivan Petrovich Pavlov. Pavlov’s conceptual universe was premised upon the reflex arc. By pairing an antecedent neutral stimulus—a Conditioned Stimulus (CS), such as an acoustic metronome or a visual geometric placard—with a biologically significant Unconditioned Stimulus (US), such as meat powder infused into the buccal cavity, the neutral stimulus gradually acquired the capacity to evoke a Conditioned Response (CR) resembling the native Unconditioned Response (UR). Pavlov interpreted this empirical finding through the physiological doctrine of “stimulus-substitution.” According to this hypothesis, repeated, contiguous temporal pairings between the CS and US established functional, excitatory neural pathways across the cerebral cortex, linking the sensory projection area of the conditional cue directly to the subcortical and cortical centers regulating the unconditional consummatory reflex.

Under the strict interpretation of stimulus-substitution, the CS becomes an internal functional surrogate for the US. The organism responds to the conditional signal not merely as an informational beacon announcing an impending event, but as if the physical properties of the US were incarnated within the sensory signature of the CS itself. Pavlov observed this phenomenon predominantly within glandular and secretory outputs, such as the flow of saliva from the parotid duct or gastric acid secretion, reinforcing the perspective that classical associative networks operated primarily below the threshold of voluntary motor coordination.

Despite the conceptual power of the stimulus-substitution model, classical learning theory historically conceded the sovereignty of skeletal motor navigation to instrumental mechanics. Although Pavlov documented generalized restlessness, postural orienting reflexes, and prancing movements directed toward the food source in his experimental dogs, these skeletal dynamics were largely classified as preparatory orienting responses or dismissed as incidental epiphenomena. The prevailing consensus among both American functionalists and European reflexologists asserted that while an animal’s visceral architecture could be seized by a classical CS, precise, directed skeletal engagement with an inanimate cue required the selective pressure of response-contingent reinforcement. Skeletal actions directed systematically at predictive artifacts remained firmly classified as operants.

1.3 Pre-1968 Anomaly Accumulation in Behavioral Psychology

As the mid-twentieth century progressed, the rigid walls demarcating operant and classical conditioning began to exhibit profound microfractures. An escalating body of anomalous experimental findings emerged, suggesting that biological organisms did not conform to the clean, mechanistic divisions mandated by behavioral dogma. Ethologists and heterodox comparative psychologists repeatedly observed non-reinforced, idiosyncratic skeletal motor actions manifesting in conditioning chambers, behaviors that resisted extinction despite running directly counter to programmatic reinforcement contingencies.

The most famous and devastating early indictment of this theoretical orthodoxy came in 1961 with the publication of “The Misbehavior of Organisms” by Keller Breland and Marian Breland. Former students and collaborators of B.F. Skinner, the Brelands had founded Animal Behavior Enterprises, an commercial venture focused on training thousands of animals across diverse species for commercial displays and zoo exhibits. Applying orthodox operant shaping techniques, they repeatedly encountered what they termed “instinctive drift.” When pigs and raccoons were reinforced with food for depositing coins into a metal piggy bank, the animals initially acquired the instrumental sequence. However, as training progressed and the coins became strongly associated with the appetitive primary reinforcer, the instrumental performance completely disintegrated.

Instead of rapidly dropping the coins into the bank to collect their food, pigs would drop the coins onto the substrate, root them with their snouts, toss them into the air, and repeatedly trample them; raccoons would endlessly rub the coins together, dip them into the receptacle, pull them back out, and perform prolonged, compulsive washing rituals. These behaviors were not shaped; worse, they actively delayed or prevented the delivery of the food reward. The Brelands concluded that innate, evolutionarily preformed behavioral repertoires specific to foraging and consummatory sequences were being elicited by the cues paired with food, systematically overriding the artificially imposed operant contingencies. The animal’s evolutionary history was invading the chamber, disrupting the pristine mechanics of the Law of Effect.

Concurrently, theoretical dissatisfaction was mounting regarding the arbitrary distinction between Pavlovian and Thorndikian/Skinnerian paradigms. Researchers such as Robert Rescorla and Richard Solomon were investigating two-process theories, exploring how Pavlovian emotional conditioning (such as conditioned fear or conditioned anticipation) exerted continuous, modulating control over baseline instrumental avoidance and appetitive responding. The conceptual landscape was becoming fraught with tensions: was an animal’s interaction with a conditioned cue truly a purely emitted voluntary act selected by its consequence, or could an unadulterated classical association bypass the instrumental architecture entirely and compel skeletal motor performance? The empirical stage was set for an experiment that would definitively strip away the response-reinforcer requirement to isolate the pure, generative power of the stimulus-reinforcer relationship.

2. The Landmark 1968 Brown and Jenkins Experiment: Design and Methodology

2.1 Experimental Apparatus and Subject Preparation

In their seminal 1968 study, Paul L. Brown and Herbert M. Jenkins set out to systematically examine whether an operant skeletal response—specifically, the standard avian key-peck—could be generated entirely through Pavlovian forward-pairing procedures, completely devoid of human intervention or instrumental dependencies. To execute this investigation with rigorous precision, they utilized a standardized operant conditioning chamber, historically termed a Skinner box, engineered specifically for the housing and automated testing of Columba livia (the domestic pigeon). The chamber was sound-attenuated, ventilated with a continuous white-noise background to mask external laboratory disturbances, and equipped with a small, circular, translucent Plexiglas response key mounted on the front intelligence panel, positioned at approximately the height of the pigeon’s head.

Behind the response key sat an electrical microswitch capable of registering physical deflections requiring a minimal actuation force, alongside a miniature projection system that could illuminate the clear key from behind with a distinct, homogeneous visual field (a bright white light). Mounted directly below this key, near the floor of the chamber, was a rectangular aperture providing access to a grain hopper. Under resting conditions, the hopper was dropped and plunged in darkness, rendering its contents completely inaccessible and invisible to the subject. When actuated by an automated electromechanical relay, the hopper was driven upward by a solenoid, presenting a plentiful mixture of nutritious grain (such as vetch and hemp seeds), simultaneously illuminated by an internal hopper lamp to provide clear visual feedback of reward availability.

Subject preparation followed standardized behavioral deprivation protocols. Thirty-six naive adult pigeons were housed in individual cages and subjected to a regulated dietary restriction protocol until their somatic mass stabilized at precisely 70 to 80 percent of their ad libitum (free-feeding) body weight. This degree of chronic nutritional deprivation was essential to establish high appetitive drive, activating the physiological seeking systems associated with feeding. Prior to experimental trials, the birds underwent hopper-training sessions to habituate them to the mechanical operation and acoustic signature of the ascending grain dispenser. During this pre-training, the hopper was raised intermittently at unpredictable intervals until the pigeons reliably moved toward the aperture and consumed grain immediately upon its presentation, establishing the sound and sight of the elevated hopper as a potent primary reinforcer and discriminative environment.

2.2 The Core Autoshaping Protocol

The experimental architecture instituted by Brown and Jenkins was remarkable for its mathematical purity and absolute refusal to implement any response-reinforcer contingency. The paradigm unfolded as a classical forward-delay conditioning arrangement. The naive, hopper-trained pigeon was placed into the dark, silent chamber. The experimental procedure was run automatically via solid-state programming logic and relay racks, entirely removing human bias or manual observational shaping from the equation.

The sequence of temporal intervals within each conditioning trial was structured according to a rigid Pavlovian forward-pairing timeline:

• Inter-Trial Interval (ITI): An unlit baseline period lasting an average of 60 seconds (with variable durations across test cohorts to prevent temporal conditioning), during which the experimental chamber was kept in dim illumination or blackout, and the response key remained completely dark. No food was ever delivered during this interval.
• Conditioned Stimulus (CS) Presentation: The translucent key was suddenly illuminated from behind with a bright light for a discrete duration of exactly 8 seconds. This visual cue functioned as the predictive conditioned stimulus.
• Unconditioned Stimulus (US) Delivery: At the precise moment the 8-second key-light terminated, the grain hopper was actuated, rising for 4 seconds, during which the pigeon was permitted unhindered consummatory access to the grain. Upon the elapsing of the 4 seconds, the hopper dropped back into its recessed position, the chamber returned to its baseline state, and the next inter-trial interval commenced.

The critical theoretical parameter of this design was its complete lack of an instrumental requirement. The pigeon was under no procedural obligation to observe the key, approach the key, or touch the key. If the bird stood completely stationary at the back of the chamber, groomed its feathers, or stared at the floor, the 8-second key-light would reliably illuminate, terminate, and the grain hopper would rise with unyielding fidelity. Conversely, pecking the key did not accelerate the arrival of the food, nor did it increase the magnitude or duration of the grain presentation. The key illumination was an absolute, non-contingent predictor of food presentation—a textbook Pavlovian CS-US relationship implemented upon an apparatus designed specifically for measuring operant behavior.

2.3 Empirical Findings and Behavioral Emergence

The empirical outcomes documented by Brown and Jenkins defied traditional operant predictions. Naive pigeons exposed to this non-contingent forward-pairing sequence did not remain passive recipients of food; rather, they exhibited the rapid, spontaneous emergence of directed key-pecking behavior. Typically, within a median acquisition window of only 40 to 50 trials—amounting to less than an hour of automated conditioning—the pigeons began to visually orient toward the illuminated key, approach the key panel across the chamber floor, and execute sharp, ballistic, forceful pecks directly against the illuminated surface of the plastic key.

The rate of acquisition was strikingly rapid. Once the first initial key contact occurred, response rates accelerated across successive experimental sessions without any external shaping interventions or differential reinforcement schedules. The pigeons did not merely peck the key occasionally; they pecked it with remarkable frequency, often striking the surface dozens of times during the 8-second window, continuing right up until the exact millisecond the key-light extinguished and the hopper ascended. The birds transformed an entirely non-contingent predictive signal into an intense focus of direct physical, consummatory assault.

To confirm that this behavioral emergence was genuinely associative and not an artifact of generalized arousal, exploratory motor excitation, or spontaneous motor recovery, Brown and Jenkins conducted rigorous control procedures:

• Unpaired/Explicitly Unpaired Controls: When the 8-second key-light and the 4-second grain presentations were separated by substantial, non-predictive temporal intervals, the birds never developed key-pecking behavior.
• Random Control Procedures: When key-light illuminations and hopper deliveries were presented according to completely independent, random schedules (the Rescorla random control design), key-pecking failed to emerge.
• Simultaneous Conditioning Controls: Delivering the key-light and the grain concurrently produced dramatically attenuated or nonexistent acquisition compared to the forward-delay pairing.

The empirical verdict was unequivocal: the emergence of the key-peck was definitively tethered to the predictive, forward contingency between the conditioned stimulus (the key light) and the unconditioned stimulus (the grain). Autoshaping was not an illusion of baseline exploratory activity; it was an authentic associative phenomenon wherein a skeletal motor act, traditionally deemed an operant, was elicited by an antecedent Pavlovian contingency.

3. Topography and Morphological Signatures of Autoshaped Responses

3.1 Consummatory Topography: Food vs. Water Conditioning

The theoretical shockwaves unleashed by the 1968 Brown and Jenkins study intensified dramatically when researchers began scrutinizing the physical form—the precise behavioral topography—of the autoshaped response. If the pigeon was merely acquiring an instrumental operant through some unobserved, accidental adventitious reinforcement mechanism, the physical mechanics of the peck should have been arbitrary, conforming only to whatever physical motion was sufficient to actuate the electrical microswitch. If, however, autoshaping was governed by Pavlovian stimulus-substitution, the autoshaped response should morphologically mirror the specific consummatory reflex elicited by the unconditioned stimulus itself.

This empirical test was definitively conducted in a landmark 1973 study by Herbert M. Jenkins and Bruce R. Moore. Jenkins and Moore exposed separate cohorts of naive, deprived pigeons to autoshaping protocols that differed exclusively in the qualitative nature of the unconditioned stimulus: one group received access to dry grain, while the other group, placed under water deprivation schedules, received access to a liquid water reservoir. Using high-speed cinematography and force-transducing key mechanisms, the researchers captured the precise spatial, temporal, and physical dynamics of the conditioned pecks directed at the illuminated key.

The photographic and kinematic analyses revealed a profound divergence in peck morphology that corresponded with absolute fidelity to the unconditioned consummatory responses directed toward the respective reinforcers:

• Food-Conditioned Pecks: When the key-light signaled food, the pigeon’s conditioned peck was characterized by a rapid, forceful, ballistic strike. The bird’s beak was opened wide prior to contacting the key, its eyes were partially or fully closed at the moment of impact (a protective reflex common in avian feeding to avoid flying dust or grain fragments), and the contact duration with the key was exceedingly brief, terminating in a sharp, grasping retraction. This motor program was an exact replica of the physical action used to seize, crush, and ingest dry grain seeds from the earth.
• Water-Conditioned Pecks: When the identical key-light signaled the delivery of water, the conditioned response transformed entirely. The bird approached the key with a markedly lower approach velocity. Its beak remained completely closed or only slightly agape; it pressed its bill gently against the illuminated plastic surface with sustained contact duration, often executing rhythmic pumping movements of the throat, tongue protrusion, and overt swallowing motions. The animal was literally attempting to drink the illuminated disc.

The Jenkins and Moore findings provided overwhelming, visually unassailable evidence in support of Pavlov’s stimulus-substitution hypothesis within the domain of skeletal motor performance. The conditioned response was not an arbitrary operant that had been accidentally reinforced; it was a reinforcer-specific consummatory motor program elicited by the conditional cue. The illuminated key had effectively transformed into a sensory stand-in for the primary reward, driving the pigeon to project its evolutionary ingestive motor repertoires directly onto the inanimate predictive signal.

3.2 Spatial and Physical Manifestations of Sign-Tracking

As empirical investigations into the autoshaping phenomenon expanded throughout the 1970s, researchers moved beyond simple key-mounted microswitches to trace the wider spatial and bodily trajectory of the subject within the experimental space. These studies exposed a broader behavioral constellation that Hearst and Jenkins (1974) formally termed “sign-tracking.” Sign-tracking referred to the pervasive, overarching tendency of an organism to physically orient toward, approach, and engage in direct tactile and consummatory interaction with a localized conditioned stimulus that accurately signals the delivery of an appetitive reinforcer, even when that stimulus is physically isolated or spatially separated from the ultimate site of reinforcement delivery.

In spatial tracking studies where the illuminated key was mounted on one side of an elongated runway or oversized chamber and the food hopper was permanently fixed at the opposite end (several meters away), sign-tracking manifested with astonishing vigor. Upon key illumination, the pigeon would turn its back on the food magazine, sprint across the length of the chamber toward the illuminated key, vigorously peck the illuminated glass during the predictive CS period, and then, only upon key termination, turn and race frantically back toward the food hopper to collect the grain before the access cycle lapsed. This dynamic occurred despite the reality that approaching the key imposed a massive energetic tax on the animal and placed it at a severe spatial disadvantage for obtaining the primary reinforcer.

Physical quantification of these responses using force-displacement transducers, linear accelerometers, and tracking video cameras revealed remarkable properties regarding response duration, kinetic force, and latency:

• Response Latencies: Following initial acquisition, response latencies dropped precipitously; animals would initiate approach movements toward the CS within milliseconds of its sensory onset, demonstrating complete attentional and motor capture.
• Kinetic Excursion and Force: The force exerted against the key often exceeded by multiple orders of magnitude the minimal threshold required to register an electrical contact. Pigeons struck the key with sufficient kinetic energy to cause audible mechanical rattling and wear on the apparatus, indicating that the motor output was not calibrated for mechanical efficiency, but was driven by an autonomic, intense appetitive urgency.
• Tactile and Visual Fixation: Eye-tracking and photographic measurements documented sustained visual fixation. Pigeons aligned their binocular field with obsessive precision upon the center of the key, maintaining continuous foveal tracking while pacing, bobbing the head, and executing preparatory intention movements prior to launching the strike. The inanimate sign exercised total hegemony over the animal’s spatial orientation and physical engagement.

4. Sign-Tracking Versus Goal-Tracking: Individual Differences and Topographical Divergence

4.1 Phenotypic Segregation in Sign-Tracking and Goal-Tracking

For several decades following the initial Brown and Jenkins publication, sign-tracking was largely evaluated as a general, species-typical characteristic of vertebrate conditioning. However, when the paradigm was systematically translated into outbred mammalian models—specifically outbred Sprague-Dawley, Wistar, or Long-Evans laboratory rats (Rattus norvegicus)—an entirely new theoretical dimension materialized. When an animal was presented with an insertion of a retractable lever or an illuminated visual cue (CS) that reliably predicted the subsequent delivery of a food pellet into an adjacent food receptacle (US), the animal population did not respond uniformly. Instead, as pioneered by researchers such as Boakes (1977) and later extensively characterized by Terry E. Robinson and Kent C. Berridge, the subjects exhibited pronounced, stable individual differences, segregating into distinct behavioral phenotypes.

These phenotypes fall into two primary divergent behavioral trajectories, alongside a distributed intermediate cohort:

• Sign-Trackers (ST): Upon presentation of the predictive conditioned stimulus (e.g., the extension of a mechanical lever), sign-trackers immediately orient toward, approach, and intensely interact with the cue itself. They grasp the lever with their forepaws, gnaw on it, bite it, lick it, and vigorously deflect it throughout the CS exposure period, completely ignoring the food magazine where the reinforcer will actually be delivered. Only when the lever retracts do they divert their attention to the magazine to ingest the already-delivered pellet.
• Goal-Trackers (GT): Goal-trackers perceive the identical predictive contingency with equal associative accuracy, yet their behavioral topography is completely reversed. Upon the extension of the lever, goal-trackers pay minimal, fleeting attention to the cue; instead, they immediately pivot and charge directly to the food hopper (the goal). They insert their heads into the magazine receptacle, waiting with anticipatory sniffing and mouth movements for the food pellet to drop while the lever continues to cycle harmlessly behind them.
• Intermediate Phenotypes: A subset of any outbred animal population displays an intermediate or shifting profile, vacillating between cue engagement and goal engagement depending on environmental stress, deprivation depth, or fluctuating motivational variables.

Longitudinal testing has demonstrated that these phenotypes represent remarkably stable, trait-like biological distributions within outbred populations. Once established, an individual’s classification as a sign-tracker or goal-tracker remains invariant across months of testing, persists through progressive extinction and re-acquisition cycles, and generalizes across varied sensory modalities, establishing the sign-tracker/goal-tracker dichotomy as an authentic behavioral endophenotype of profound significance to personality theory, neurobiology, and clinical psychiatry.

4.2 Incentive Salience Attribution Hypothesis

To explain why two organisms exposed to the exact same associative contingency, with identical learning rates and identical predictive knowledge regarding the CS-US pairing, exhibit completely different motor outputs, Kent Berridge and Terry Robinson developed the Incentive Salience Attribution Hypothesis. This theoretical model bifurcates appetitive reward into distinct psychological components: cognitive predictive information (“learning”), the hedonic sensory impact of consumption (“liking”), and the attribution of dynamic motivational magnetism to mental representations and physical stimuli (“wanting”).

According to this framework, both sign-trackers and goal-trackers acquire the cognitive, informational association: both animals learn that CS termination predicts US arrival. This predictive knowledge is mediated by standard associative networks and cognitive mapping systems. However, in sign-tracking phenotypes, a distinct neurochemical event occurs: the conditioned stimulus undergoes an intense process of incentive salience attribution. The cue does not merely act as an informational street sign pointing down the road toward food; it is transformed into an incentive stimulus—a “motivational magnet.”

When incentive salience is attributed to a conditional cue, the CS acquires three fundamental functional properties:

1. It becomes a magnet for physical approach, causing the animal to orient toward and navigate directly into physical contact with the cue itself.
2. It acquires the capacity to elicit direct consummatory reactions; the organism treats the cue as if it were the object of its appetitive desire, gnawing, licking, or pecking it.
3. It functions as a powerful conditioned reinforcer in its own right; the animal will emit novel operant responses (such as pressing a distinct, secondary lever) purely for the privilege of viewing or contacting the CS, even when no primary reward is ever delivered.

In goal-trackers, the CS retains its cold, predictive informational value (an epistemic signal), but fails to be imbued with this hot, visceral incentive salience. The incentive salience is instead directed exclusively toward the mental representation of the primary goal and its physical delivery site. For the sign-tracker, the sign and the substance become motivationally fused; the predictive proxy usurps the primary biological reward.

4.3 Psychometric and Behavioral Correlates of the Phenotypes

The divergence between sign-tracking and goal-tracking phenotypes extends far beyond mere spatial preference within a behavioral chamber; it is intimately correlated with a broad spectrum of psychometric, cognitive, and physiological traits. Systematic testing has revealed that the sign-tracking phenotype is inextricably linked to heightened trait impulsivity, deficits in cognitive inhibitory control, and distinct vulnerability profiles across various paradigms of environmental challenge.

When evaluated on standardized operant measures of impulsivity—such as the 5-Choice Serial Reaction Time Task (5-CSRTT) or delay-discounting protocols—sign-trackers exhibit profound impairments in motor inhibition. They produce significantly elevated rates of premature responding (inability to withhold an action prior to an environmental trigger) and show steep, hyperbolic discounting curves, selecting immediate, small rewards over larger, delayed alternatives. Their behavioral architecture is fundamentally biased toward immediate action execution triggered by environmental cues, reflecting an impaired capacity for top-down prefrontal suppression over cue-elicited motor cascades.

Furthermore, sign-tracking cohorts display marked elevations in novelty-seeking and sensation-seeking behaviors. When placed into novel, open-field environments, sign-trackers show hyper-exploratory locomotion and rapid entry into high-risk central zones. Most critically, this phenotype demonstrates an immense vulnerability to reinstatement, relapse, and compulsive engagement across a host of maladaptive paradigms. When exposed to stress, non-contingent reward primes, or environmental contexts previously associated with reward delivery, sign-trackers succumb to reinstatement at dramatically higher rates than their goal-tracking counterparts, marking the sign-tracking endophenotype as a prime preclinical model for understanding pathological vulnerability to psychiatric and addictive disorders.

5. Theoretical Implications: The Pavlovian-Operant Dichotomy Reassessed

5.1 Deconstruction of Traditional Reinforcement Learning Theories

The empirical verification of the autoshaping phenomenon dealt an irreversible blow to the foundational tenets of radical behaviorism and classical reinforcement learning models. Prior to 1968, the Law of Effect—codified by Edward Thorndike and canonized by B.F. Skinner—was widely promoted as an exhaustive, universally applicable law of nature governing all voluntary motor actions. The Law of Effect asserted that the frequency of an emitted behavior was determined solely by its post-hoc consequences: if a behavior was followed by a reinforcer, its associative connection to the surrounding stimulus situation was strengthened; if it was followed by an indifferent or aversive state, it withered.

Autoshaping dismantled this paradigm by demonstrating that complex, highly focused, skeletal motor actions could be synthesized de novo in the complete absence of a response-reinforcer contingency. Pigeons in an autoshaping chamber pecked the illuminated key not because previous pecks had yielded grain, but because the illumination of the key was reliably followed by grain. The motor pattern was driven by an antecedent stimulus-stimulus (S-S) relationship, fundamentally undermining the claim that skeletal responding requires the scaffolding of response-contingent selection.

This realization forced an epistemological revolution within comparative learning theory. It demonstrated that organisms do not operate merely as passive physical systems awaiting consequence-driven reinforcement to shape their somatic outputs. Instead, internal associative structures, triggered strictly by the predictive veridicality of the surrounding sensory environment, possess direct, unmediated access to skeletal motor generators. The clean, pedagogical dichotomy that had divided psychology for half a century—relegating autonomic reflexes to Pavlov and skeletal actions to Skinner—was revealed to be an artificial, ecologically invalid construct.

5.2 The Superstition Hypothesis and Its Systematic Refutation

Faced with this direct challenge to orthodox operant theory, orthodox behaviorists initially sought to assimilate autoshaping into Skinner’s famous “superstition” hypothesis (1948). Skinner had previously claimed that when food was delivered to pigeons at fixed temporal intervals regardless of their behavior, the animals developed idiosyncratic, repetitive rituals—such as turning counter-clockwise, tossing their heads, or swaying from side to side. Skinner theorized that whatever random skeletal movement the bird happened to be executing at the exact moment the automated feeder clicked was adventitiously reinforced. Through progressive, accidental contiguous reinforcement, these accidental movements were allegedly cemented into the bird’s behavioral repertoire as “superstitions.”

Orthodox theorists argued that autoshaping was nothing more than an instance of accidental, adventitious reinforcement. They posited that during the 8-second key-light, the pigeon might randomly step toward or glance at the key; if the grain hopper raised at that moment, the forward step was reinforced. Across subsequent trials, this accidental chain was gradually pulled forward until the bird’s beak struck the key, at which point the key-peck was cemented via ordinary operant mechanics.

Brown and Jenkins, along with subsequent researchers, systematically dismantled this adventitious reinforcement defense through a series of elegant experimental controls:

• Terminal Response Stereotypy: Superstitious behavior, by Skinner’s own empirical observations, was highly variable, idiosyncratic, and differed wildly from bird to bird (one bird turned, another bobbed, another stretched its neck). In autoshaping, the behavior was rigorously uniform, invariant, and universally focused directly on the localized visual key across all subjects.
• Spatial Selectivity: If accidental reinforcement were driving the behavior, any movement occurring prior to hopper delivery (such as standing over the hopper, grooming, or turning away) should have been equally reinforced. Yet birds selectively bypassed movements near the food hopper during the CS period to direct their behavior exclusively toward the distal, non-reinforced visual cue.
• Kinematic Inflexibility: Adventitious reinforcement could not explain why pigeons pecked with an open beak for food signals and a closed, pumping beak for water signals. If any physical strike was sufficient to bridge the temporal gap, the emergence of reinforcer-specific consummatory topographies pointed unequivocally to Pavlovian elicitation, completely refuting the superstition account.

5.3 Two-Factor Theory Revisions and Unified Frameworks

The undeniable failure of both pure operant mechanics and the superstition hypothesis to account for autoshaping necessitated sweeping theoretical revisions. Scholars re-examined the classical Two-Factor Learning Theory originally formulated by O. Hobart Mowrer. Mowrer had proposed that avoidance conditioning required two distinct phases: first, the classical conditioning of fear to an antecedent warning cue; second, the instrumental reinforcement of an escape response that terminated the fear-inducing cue. In the wake of autoshaping, theorists like Robert Rescorla and Vincent Lolordo recognized that the interplay between classical and instrumental processes was far more integrated, ubiquitous, and bidirectionally permeable than previously conceived.

Autoshaping demonstrated that Pavlovian conditioning was not a secondary, peripheral system concerned merely with visceral support systems, but was an overarching, dominant cognitive architecture that dictated attention, incentive motivation, and motor execution. Mathematically formalized models of associative learning—most notably the revolutionary Rescorla-Wagner Model (1972)—were seamlessly applied to sign-tracking dynamics. The Rescorla-Wagner model formalized learning as the progressive reduction of prediction error:

$$\Delta V = \alpha \beta (\lambda – \sum V)$$

Where $\Delta V$ represents the change in associative strength between the conditioned cue and the reinforcer, $\alpha$ and $\beta$ represent the salience of the CS and US, $lambda$ represents the asymptotic limit of associative support supported by the reinforcer, and $\sum V$ represents the total associative strength already accumulated by all predictive cues present in the environment.

In autoshaping, the illuminated key absorbs associative strength precisely as predicted by this error-correction equation. When associative strength ($V$) reaches critical levels, it does not merely manifest as an internal expectancy; it drives somatic motor output directly. Autoshaping emerged as the theoretical bridge linking Pavlovian conditioning, cognitive expectancy theories, and instrumental action selection, forcing modern behavioral science to formulate unified frameworks where operant and classical systems operate as deeply coordinated, co-dependent components of a singular, adaptive neural engine.

6. The Negative Automaintenance Paradigm and the Omission Procedure

6.1 Methodology of the Sheffield Omission Training Procedure

Although the 1968 Brown and Jenkins experiment demonstrated that operant reinforcement was unnecessary for the emergence of key-pecking, a crucial, radical question remained: What would happen if an explicit, punishing instrumental contingency were pitted directly against the Pavlovian stimulus-reinforcer relationship? Could Pavlovian conditioning force an animal to execute a skeletal motor response even when executing that response caused the absolute loss of the primary biological reinforcer?

To answer this question, researchers implemented the elegant and unforgiving experimental design known as the Sheffield Omission Training Procedure (or negative automaintenance), originally conceptualized by Virginia Sheffield and rigorously applied to autoshaping by Williams and Williams (1969). The omission training protocol modifies the standard autoshaping schedule by introducing an explicit response-prevention or omission contingency:

• The translucent key illuminates for an 8-second interval (CS), signaling the impending presentation of the grain hopper (US).
• If the pigeon withholds from pecking the illuminated key for the entire 8-second duration, the key turns off, and the grain hopper rises, providing full access to food.
• If the pigeon strikes the illuminated key even a single time at any point during the 8-second window, the key immediately extinguishes, and the scheduled food presentation is completely canceled for that trial. The chamber immediately plunges into the inter-trial interval blackout, forcing the bird to wait in hunger for the next cycle.

Under this arrangement, the instrumental contingency was crystalline: pecking the key was purely maladaptive, counterproductive, and directly penalized by the immediate forfeiture of the primary appetitive reward. To maximize food consumption, the pigeon needed only to stand motionless and watch the light illuminate, wait for eight seconds, and then feast. If the pigeon’s key-pecking was an operant response governed by the Law of Effect, the response should have extinguished with extreme velocity, as every key-peck was directly punished by reinforcement loss.

6.2 Persistence of Counter-Productive Responding

The results of the Williams and Williams (1969) negative automaintenance experiments were astounding. Rather than ceasing their responding to preserve their access to food, the pigeons continued to relentlessly approach and peck the illuminated key. By striking the key, they repeatedly triggered the omission circuit, canceling their own food deliveries trial after trial, session after session. In many cases, pigeons sacrificed 30, 50, or even up to 80 percent of the total available food rewards because they could not refrain from pecking the conditioned visual sign.

The behavioral patterns observed during omission training were poignant manifestations of biological conflict. Pigeons would display clear signs of physiological agitation as the trial began. When the key illuminated, a bird would frequently turn its body away from the key, actively pacing toward the back of the chamber or pressing its beak against its breast feathers in an apparent motor effort to resist the cue. Yet, as the 8-second interval ticked away, the bird’s visual gaze would inexorably snap back to the illuminated disc. Driven by an overwhelming spatial attraction, the pigeon would orient, advance across the floor, and, as if magnetically pulled into the cue, execute a rapid peck against the glass, instantly plunging the chamber into darkness and forfeiting its sustenance.

Response rates under omission training did not remain permanently at maximum baseline levels; instead, they stabilized into an oscillatory equilibrium. An animal might peck the key and lose food for several consecutive trials, pushing its bodily deprivation state higher. Driven by mounting hunger, the predictive value of the CS would become even more intensely salient. Eventually, the animal might successfully withhold a peck for a trial or two, receiving the desperately needed grain. However, the receipt of this reinforcement instantly recharged the associative excitatory strength of the CS-US pathway, causing the pigeon to immediately resume pecking on subsequent trials, triggering another wave of self-inflicted food cancellation. The classical Pavlovian excitation generated by the forward CS-US pairing completely overwhelmed and overrode the explicit instrumental punishment contingency.

6.3 Theoretical Insights into Behavioral Inflexibility

Negative automaintenance stands as one of the most powerful empirical demonstrations of behavioral inflexibility in the annals of comparative psychology. It established that sign-tracking responses are fundamentally non-teleological; they are not executed with a conscious, instrumental eye toward future outcomes or utility maximization. The pigeon pecks not to produce an effect in the world, but because the internal architecture of its nervous system has identified a high-fidelity predictor of survival-relevant biological resources, unleashing an evolutionarily hardwired motor sub-routine that operates autonomously from higher-order cognitive override mechanisms.

This operational dissociation between the instrumental utility of an action and its Pavlovian motor elicitation provided deep theoretical insights into the autonomous control exerted by conditioned incentives. The animal’s somatic execution machinery is captured by the sign. The conditioned stimulus acts as a supernormal trigger, activating a localized behavioral program that refuses to yield to standard instrumental cost-benefit feedback loops.

Subsequent comparative investigations demonstrated the broad conservation of this Pavlovian dominance across diverse vertebrate lineages. When omission contingencies were applied to rodents utilizing retractable levers, or to quails utilizing localized visual cues, similar phenomena of counter-productive, persistent engagement were routinely observed. Organisms across broad evolutionary clades were shown to possess a profound, hardwired cognitive-behavioral vulnerability: whenever a localized cue reliably predicts a primary biological reinforcer, the elicited Pavlovian urge to physically approach and consummatorily engage that sign possesses the biological horsepower to overpower instrumental restraint, laying bare the deep evolutionary roots of compulsion.

7. Behavior Systems Theory and Ethological Perspectives

7.1 Timberlake’s Behavior Systems Approach

In the late 1970s and 1980s, the conceptual limitations of both radical Skinnerian operant mechanics and simplistic Pavlovian stimulus-substitution led to the development of a radically more nuanced, biologically grounded paradigm: Behavior Systems Theory, formulated by William Timberlake. Timberlake argued that laboratory conditioning paradigms do not write novel associations onto blank biological slates, nor do they merely glue arbitrary reflexes together via temporal contiguity. Instead, conditioning operates by tapping into, activating, and reorganizing pre-existing, evolutionarily conserved behavioral systems that evolved to solve specific ecological challenges within an animal’s ancestral niche.

According to Timberlake, an organism’s behavioral repertoire is structured into specialized functional modules—such as feeding, mating, predator defense, thermoregulation, and maternal care. Within each system, behaviors are hierarchically organized across distinct spatiotemporal modes of environmental engagement:

• General Search Mode: Manifested when an organism is in a state of biological deprivation (e.g., hunger) but has not yet identified a specific source of food. Behavior consists of broad, non-directed exploratory locomotion, heightened sensory scanning, and spatial dispersal across home ranges.
• Focal Search Mode: Triggered when the organism encounters a localized cue indicating that food is spatially proximal or temporally impending. Behavioral topographies shift dramatically toward localized visual scanning, approach, spatial restriction to the predictive locus, and tactile investigation.
• Consummatory Mode: Activated upon direct physical encounter with the food object itself, consisting of terminal, highly stereotyped motor programs such as grasping, pecking, chewing, swallowing, or salivating.

Viewed through the lens of Behavior Systems Theory, autoshaping is not a strange laboratory artifact or an associative paradox; it is the natural, logical activation of the avian feeding system’s focal search and consummatory modes. When the key-light illuminates, the pigeon’s central nervous system does not interpret the event as an abstract geometry problem. The predictive signal engages the feeding system, abruptly shifting the bird out of general search and instantly locking it into the focal search mode. Because the conditioned stimulus is a discrete, localized visual object resembling a seed or small spatial target, the pigeon directs its preformed, innate foraging and food-gathering motor programs precisely at the physical sign.

7.2 Ecological Validity and Naturalistic Significance

The ethological perspective reveals that sign-tracking is an extraordinarily adaptive evolutionary mechanism in natural, unconstrained environments. In the wild, primary biological resources such as seeds, insects, water pools, or prey items do not materialize out of thin air via automated electromechanical hoppers. In the natural world, a predictive sign is almost invariably physically contiguous with, or constitutes a structural part of, the goal itself.

For an avian forager in the wild, the bright visual glint of an insect shell, the distinctive morphological shape of a seed pod, or the localized movement of leaf litter are predictive conditioned stimuli that physically contain or immediately precede the primary reinforcer. An animal that possesses an innate, powerful associative architecture that compels it to physically track, approach, seize, and strike localized visual signals that predict nutrients will consistently out-compete an animal that waits passively for food to enter its mouth. Sign-tracking is the motor engine of active foraging: it forces the animal to close the spatial gap between its sensory apparatus and the predictive indices of biological sustenance.

The sterile, artificial geometry of the standard laboratory operant chamber creates an ecological anomaly by artificially bifurcating the predictive sign (the key on the panel) from the consummatory goal (the hopper on the floor). In the natural habitat of Columba livia, the sign and the goal are spatially unified: the visual appearance of the grain is the grain. Autoshaping exposes the underlying evolutionary wiring of an organism that evolved under ecological constraints where predictive cues demand immediate physical, predatory, or consummatory engagement. The laboratory setting does not create this behavior; it merely splits the naturalistic feedback loop, exposing the raw, unconditioned power of the sign-tracking instinct.

7.3 Stimulus Dimension and Distance Determinants

Behavior Systems Theory is powerfully substantiated by parametric manipulations of the sensory dimensions and spatial coordinates of the conditioned stimulus. The physical morphology of the autoshaped response is exquisitely sensitive to the structural properties of the CS, reflecting the precise environmental affordances that trigger specific predatory and foraging motor modules.

When the visual CS is a small, bright, discrete, and highly localized visual spot (such as a 2-centimeter illuminated circle), pigeons treat it as a seed-like target, exhibiting tight ballistic pecking with an open beak. However, if the experimental apparatus is modified such that the CS is a diffuse, non-localized environmental change—such as the entire conditioning chamber’s ambient ceiling lighting shifting from darkness to diffuse illumination—sign-tracking completely vanishes. A pigeon cannot peck “diffuse ambient light.” Instead, the identical predictive forward contingency produces pure goal-tracking: the bird immediately orients toward, approaches, and hovers directly over the dark food hopper, waiting for the grain to ascend. The focal search motor program requires a discrete, graspable spatial target; when denied this localized substrate, the feeding system shifts the motor output into anticipatory goal-site waiting.

Similarly, spatial distance and temporal proximity dictate whether sign-tracking or goal-tracking topographies dominate the behavioral landscape:

• Spatial Separation: Increasing the physical distance between the CS and the US locus strains the sign-tracking circuit. If the key is mounted several meters from the food hopper, the animal must calculate an energetic trade-off. While high-incentive sign-trackers continue to run to the cue, intermediate animals begin shifting toward goal-tracking topographies to minimize the risk of missing the brief reinforcement window.
• Temporal Duration: Exceedingly short CS durations (e.g., 2 to 4 seconds) maximize sign-tracking by compressing the focal search window into an immediate, urgent strike requirement. Conversely, prolonged CS durations (e.g., 30 to 60 seconds) cause the animal to initially engage in broad general search movements throughout the chamber during early cue exposure, only transitioning to focal sign-tracking as the moment of US delivery approaches, demonstrating the fluid, dynamic shifting of behavior systems along temporal and spatial gradients.

8. Neurobiological Substrates of Autoshaping and Sign-Tracking

8.1 Mesolimbic Dopaminergic Circuitry

The behavioral dissociation between sign-tracking and goal-tracking provided modern neuroscience with a profound experimental model to dissect the neurochemical and anatomical foundations of incentive motivation. Over two decades of intensive neurobiological investigation—spearheaded by researchers such as Terry Robinson, Kent Berridge, Shelly Flagel, and Paul Phillips—have definitively established that mesolimbic dopaminergic transmission within the nucleus accumbens (NAc) core is the absolute biological arbiter of sign-tracking, but is strikingly dispensable for goal-tracking.

Using in vivo microdialysis and fast-scan cyclic voltammetry (FSCV) in freely moving rodents undergoing lever-press autoshaping, Flagel et al. (2011) measured real-time, sub-second fluctuations in dopamine concentration within the NAc core. The empirical findings established a definitive neurochemical divergence between the phenotypes:

• Sign-Trackers: Early in training, dopaminergic neurons in the ventral tegmental area (VTA) fire phasic bursts of dopamine in the NAc core exclusively upon the delivery of the primary, unconditioned reward (the food pellet). However, as associative learning progresses, this phasic dopamine burst systematically shifts backward in time: dopamine release at the time of the food pellet wanes, and a massive, sharp spike of dopamine is evoked by the sudden presentation of the conditioned stimulus (the lever extension). For the sign-tracker, the cue itself evokes a massive dopaminergic signal, encoding the transfer of incentive salience onto the physical sign.
• Goal-Trackers: Goal-trackers learn the predictive contingency with equal speed and precision. Yet, fast-scan cyclic voltammetry revealed that throughout training, goal-trackers show minimal or non-significant phasic dopamine shifts to the cue within the NAc core. Dopamine transmission in the accumbens core does not transfer to the CS in goal-trackers; their predictive learning occurs completely independently of mesolimbic accumbens dopamine surges.

Causal pharmacology confirmed this foundational dissociation. When systemic or intra-accumbens dopamine receptor antagonists (such as the D1/D2 receptor blocker flupenthixol) were microinjected into subjects during autoshaping acquisition, sign-tracking was completely, catastrophically abolished. The animals were rendered entirely incapable of acquiring or expressing cue-directed approach and interaction. In stark contrast, the identical dopaminergic blockade had virtually zero effect on the acquisition or performance of goal-tracking. Goal-tracking animals continued to navigate toward the food hopper with undiminished accuracy. Mesolimbic dopamine is therefore not a general requirement for associative learning or reward anticipation; it is the specific, unique neurochemical currency through which a predictive sign is transformed into an attractive, consummatory motivational magnet.

8.2 Corticolimbic Regulation and Prefrontal Modulation

While the mesolimbic dopamine pathway provides the raw, unconditioned horsepower driving incentive salience attribution, this subcortical engine does not operate in isolation. It is embedded within an intricate corticolimbic network that exerts continuous top-down cognitive control and contextual gating over cue-elicited behavioral cascades. The primary hubs within this regulatory circuit include the medial prefrontal cortex (comprising the infralimbic and prelimbic cortices), the basolateral complex of the amygdala (BLA), and dynamic thalamocortical loops.

The basolateral amygdala is fundamentally critical for synthesizing sensory-specific conditioned representations and projecting them down into the ventral striatum. Functional disconnection lesions between the BLA and the nucleus accumbens core completely disrupt sign-tracking behavior, demonstrating that the BLA is required to transmit the associative CS-US representations that trigger the downstream phasic dopamine bursts in the accumbens. Without an intact BLA-NAc highway, the conditioned stimulus remains an emotionally inert sensory event incapable of driving somatic motor recruitment.

Conversely, the medial prefrontal cortex (mPFC)—specifically the prelimbic (PL) and infralimbic (IL) cortices—acts as the executive braking system, exerting top-down inhibitory modulation over subcortical appetitive impulses. Neuroimaging and immediate early gene (c-Fos) mapping studies reveal that goal-tracking phenotypes exhibit robust, coordinated engagement of prefrontal-to-striatal and prefrontal-to-thalamic pathways during cue exposure. Goal-trackers deploy the executive machinery of the mPFC to maintain top-down cognitive focus on the primary goal, suppressing the immediate visceral temptation to engage the physical sign. In marked contrast, sign-trackers display profound functional hypofrontality: their prefrontal cortices fail to mount adequate top-down inhibitory control over the hyperactive mesolimbic subcortical circuits, permitting the physical sign to seize control of the downstream premotor and somatic execution matrices.

8.3 Neurochemical and Genetic Determinants

The divergence between sign-tracking and goal-tracking is fundamentally biological, rooted in distinct neurochemical signatures, receptor kinetics, and differential gene expression profiles that distinguish the two phenotypes prior to any experimental conditioning experience.

Beyond mesolimbic dopamine, central cholinergic neuromodulation plays a decisive role in shaping the tracking phenotype. Cholinergic transmission within the prefrontal cortex, regulated by ascending projections from the basal forebrain (nucleus basalis of Meynert), is essential for sustained attentional processing and cue detection. Sarter and colleagues demonstrated that goal-trackers possess high-capacity, highly responsive prefrontal cholinergic release systems, allowing them to rapidly filter sensory cues, extract their informational content, and direct goal-oriented behavioral plans. Sign-trackers, however, exhibit chronically compromised prefrontal cholinergic capacity, resulting in poor attentional filtering and an inability to prevent subcortical motivational circuits from overwhelming cortical processing.

At the synaptic and molecular level, significant neurobiological divergences have been documented across the tracking phenotypes:

• Dopamine Receptor Expression: Sign-trackers exhibit significantly higher baseline levels of dopamine D1 receptor mRNA and lower levels of dopamine D2 autoreceptor mRNA within the striatum and VTA compared to goal-trackers. This profile renders their dopaminergic neurons resistant to inhibitory feedback and hyper-responsive to cue exposure.
• Synaptic Plasticity Regulators: Sign-trackers show altered AMPA-to-NMDA receptor ratios within the VTA and nucleus accumbens, alongside differential phosphorylation of GluA1 subunits, indicating an innate predisposition toward heightened long-term potentiation (LTP) within reward-encoding synapses upon initial exposure to appetitive cues.
• Epigenetic and Transcriptional Variations: Microarray profiling has identified differential expression of early growth response protein 1 (Egr1), brain-derived neurotrophic factor (BDNF), and cyclic AMP response element-binding protein (CREB) within the corticolimbic circuits of naive animals, demonstrating that the sign-tracking phenotype is an innate, genetically and epigenetically programmed biological variant that alters how an organism perceives, values, and reacts to environmental signs.

9. Comparative Analyses: Autoshaping Across Diverse Animal Taxa

9.1 Avian Implementations Beyond the Standard Pigeon Model

While the initial discovery of autoshaping was established in Columba livia, comparative psychologists quickly recognized that if sign-tracking represented a fundamental evolutionary adaptation, it should be replicable across diverse avian species possessing varied foraging ecologies, social structures, and neuroanatomical specializations. Subsequent research expanded the autoshaping paradigm across an expansive array of avian taxa, yielding critical evolutionary insights.

Studies conducted on Japanese quail (Coturnix japonica) revealed that autoshaping was not restricted to feeding systems, but operated with immense potency within sexual conditioning frameworks. When an illuminated visual placard or a distinct localized key was reliably paired with the subsequent release of a receptive female quail into the chamber, male quail rapidly acquired directed sign-tracking responses toward the inanimate visual cue. The male birds would approach, strut before, display courtship posturing toward, and even attempt to mount and copulate with the illuminated plastic key. This confirmed that autoshaping is a domain-general associative mechanism capable of engaging any survival-critical behavioral system, whether feeding, drinking, or reproduction.

Paradigms implemented in domestic chicks (Gallus gallus domesticus) demonstrated the role of thermal reinforcement. When naive chicks housed in cold environments were exposed to a visual cue that reliably predicted the activation of an overhead infrared heat lamp, the chicks rapidly developed sign-tracking pecks directed at the visual cue. The morphological structure of these pecks differed systematically from food pecks, consisting of soft, exploratory bill presses accompanied by distress-call cessation, reflecting thermal comfort-seeking modules.

Conversely, comparative experiments in corvids (such as crows, ravens, and scrub jays)—taxa renowned for exceptional forebrain development, complex tool use, and expansive executive control—revealed dramatically accelerated acquisition rates, accompanied by an enhanced capacity to switch flexibly between sign-tracking and goal-tracking topographies when environmental contingencies shifted. Corvids displayed an ability to inhibit cue-directed motor actions when subtle omission penalties were introduced, demonstrating that species-specific ecological specializations and relative encephalization profoundly modulate the degree of behavioral autonomy exerted by the conditioned sign.

9.2 Rodent Paradigms: Lever-Press Autoshaping

The conceptual migration of the autoshaping paradigm from avian subjects to laboratory rodents (Rattus norvegicus and Mus musculus) represented a monumental methodological breakthrough. It transformed what was initially viewed as an interesting avian key-peck curiosity into the primary experimental engine driving contemporary behavioral neuroscience, neuropsychopharmacology, and psychiatric modeling.

In standard rodent autoshaping (often referred to as Pavlovian Conditioned Approach, or PCA), the conditioned stimulus is typically a retractable, illuminated stainless-steel lever that mechanically inserts into the chamber panel for a discrete period (e.g., 8 seconds), accompanied by an illuminated cue light. Upon lever retraction, a food pellet (US) drops into an adjacent, recessed magazine. Just as pigeons peck the illuminated key, a substantial proportion of rats spontaneously develop vigorous, cue-directed sign-tracking behavior: they sprint across the chamber, seize the lever with their forepaws, gnaw on the metal bar, lick it, and depress it repeatedly with high frequency, despite the complete absence of any instrumental requirement to touch the lever.

The rodent lever-press model provided a profound structural equivalent to the avian key-peck, offering several immense technical advantages:

• It permitted the high-resolution, independent quantification of both sign-tracking (lever deflections, contacts, latencies) and goal-tracking (magazine head-entries, durations, latencies) concurrently within the exact same experimental session.
• It established a standardized quantitative index—the Pavlovian Conditioned Approach (PCA) Score—integrating response bias, latency score, and probability score into a continuous metric ranging from -1.0 (pure goal-tracker) to +1.0 (pure sign-tracker).
• It unlocked access to the vast armamentarium of modern rodent neuroscience, including stereotaxic cannulation, in vivo electrophysiology, optogenetics, chemogenetics, viral tract-tracing, and transgenic knockout technologies, cementing the rodent autoshaping preparation as the global gold-standard platform for interrogating the neural circuitry of incentive salience and compulsive behavior.

9.3 Primate and Non-Mammalian Models

To establish whether the sign-tracking phenomenon represents an ancestral behavioral mechanism deeply conserved across evolutionary time, comparative researchers deployed autoshaping paradigms across non-human primates, diverse mammalian species, and non-mammalian lineages extending back to teleost fish and invertebrates.

In non-human primates, including rhesus macaques (Macaca mulatta) and common marmosets (Callithrix jacchus), automated autoshaping paradigms utilizing high-resolution touchscreens were implemented. Naive monkeys presented with an arbitrary visual icon on a touchscreen that consistently predicted the delivery of sweet fruit puree or sucrose pellets rapidly acquired direct manual sign-tracking. The primates did not simply wait for the food to emerge from the delivery tube; they lunged forward, repeatedly touching, tapping, slapping, and licking the visual icon on the display screen. Furthermore, non-human primates exhibited marked individual variations directly paralleling the rodent sign-tracker/goal-tracker phenotypes, with sign-tracking monkeys displaying pronounced elevations in systemic cortisol output, heightened behavioral reactivity to acute stressors, and marked impulsivity profiles during working memory tasks.

Even more remarkably, autoshaping analogues have been successfully verified within phylogenetically ancient lineages:

• Teleost Fish: Goldfish (Carassius auratus) exposed to a localized visual light spot that reliably predicted the delivery of brine shrimp food pellets exhibited rapid acquisition of sign-tracking, swimming directly to the light cue and attempting to bite and suck the illuminated glass panel.
• Reptiles: Autoshaping has been documented in various lizard species, which systematically orient toward and launch predatory tongue strikes against localized artificial predictive cues paired with cricket delivery.
• Invertebrates: Associative tracking phenomena sharing fundamental characteristics with autoshaping have been identified in cephalopods (such as cuttlefish) and selected terrestrial arthropods (such as honeybees and crickets), demonstrating that the neurological architecture driving an organism to project appetitive motor programs directly onto predictive environmental signals is an evolutionarily ancient, deeply conserved biological imperative that spans hundreds of millions of years of vertebrate and invertebrate divergence.

10. Clinical Relevance: Animal Models of Addiction, Compulsion, and Impulse Disorders

10.1 Vulnerability to Substance Use Disorders (SUD)

Perhaps the most transformative translational development stemming from autoshaping research is the discovery that the sign-tracking endophenotype provides an extraordinary preclinical window into human vulnerability to Substance Use Disorders (SUD). It has long been recognized that only a subpopulation of individuals who experiment with drugs of abuse transition from initial recreational use to compulsive, intractable addiction. The biological markers mediating this selective vulnerability remained maddeningly elusive until researchers began screening outbred animals for sign-tracking prior to drug exposure.

A staggering body of research—spearheaded by Terry Robinson, Shelly Flagel, Barry Everitt, and David Belin—has established that sign-tracking animals are vastly more vulnerable to every defining hallmark of addiction across multiple drug classes, including cocaine, amphetamines, opioids, and nicotine. When given access to drug self-administration, sign-trackers exhibit accelerated acquisition, escalating their drug intake more rapidly, working harder under progressive ratio schedules of reinforcement, and continuing to seek the drug even when delivery is paired with noxious aversive foot-shocks, demonstrating compulsive drug-seeking in the face of negative consequences.

Most critically, sign-trackers show catastrophic vulnerability to cue-induced relapse. In addiction paradigms, drug-associated cues (such as a needle, a pipe, a distinct room, or packaging) acquire immense power to trigger intense physiological craving. Because sign-trackers are biologically hardwired to attribute excessive incentive salience to conditioned stimuli, drug-paired cues become irresistible motivational magnets. Following prolonged periods of extinction or forced abstinence, re-exposing a sign-tracker to the non-contingent illumination of a drug-associated cue unleashes massive, compulsive reinstatement of drug-seeking actions, modeling with exquisite fidelity the cue-reactivity, attentional bias, and relapse crises observed in human clinical populations suffering from substance dependence.

10.2 Pathological Gambling and Behavioral Addictions

The clinical footprint of the sign-tracking model extends seamlessly beyond chemical addictions into the realm of behavioral addictions, most notably Pathological Gambling (Gambling Disorder). A defining characteristic of modern electronic gaming machines—such as digital slot machines and video lottery terminals—is the pervasive deployment of hyper-stimulating, multi-sensory conditioned cues: flashing neon lights, rhythmic celebratory chimes, dynamic graphic displays, and escalating mechanical counters that reliably precede, accompany, and predict payout distributions.

Translational research utilizing human eye-tracking, galvanic skin conductance, and functional neuroimaging (fMRI) paradigms has revealed striking human parallels to the sign-tracker endophenotype. Individuals diagnosed with pathological gambling, alongside high-risk recreational gamblers, display severe attentional capture directed specifically toward the predictive peripheral cues of the gambling apparatus. When interacting with experimental gambling tasks, sign-tracking humans show sustained, involuntary visual fixations on the spinning reels, flashing peripheral icons, and prediction banners, fundamentally neglecting optimal probability matrices.

This attentional fixation is driven by a hyper-sensitized mesolimbic dopamine system that attributes excessive incentive salience to the signs of impending reward. In the human sign-tracker, the near-miss light pattern or the rhythmic escalation of the auditory chime ceases to function merely as an informational progress indicator; it becomes an active incentive trigger that drives the somatic motor execution of the next bet. The individual is caught in an evolutionary feedback trap identical to that of the Williams and Williams pigeon: the environmental sign compels immediate motor approach and engagement, overriding higher-order cognitive evaluations of long-term economic ruin.

10.3 Obsessive-Compulsive and Neuropsychiatric Pathologies

Beyond addiction, the autoshaping framework provides vital mechanistic insights into the pathophysiology of Obsessive-Compulsive Disorder (OCD), Tourette syndrome, and broader impulse control spectrum pathologies. At its core, OCD is characterized by intrusive, distressing thoughts (obsessions) that trigger repetitive, stereotyped, non-teleological motor or cognitive rituals (compulsions) aimed at neutralizing distress or preventing an imagined catastrophe, despite the patient consciously recognizing the irrationality of the behavior.

The negative automaintenance paradigm provides an astonishingly precise preclinical analogue for the compulsive motor sequences seen in OCD. In negative automaintenance, the animal recognizes at some cognitive level the arrival of the food, yet it cannot prevent the execution of a maladaptive, motor ritual (the key-peck) that directly results in negative outcomes. The behavior represents a complete failure of top-down prefrontal inhibitory architecture to assert veto power over elicited, cue-driven striatal motor routines.

Neurobiologically, both sign-tracking animals and human clinical cohorts suffering from OCD and Tourette syndrome exhibit profound structural and functional abnormalities within cortico-striatal-thalamo-cortical (CSTC) loops. Hyperactivity within the ventral striatum and motor regions of the basal ganglia, paired with impaired functional connectivity from the orbitofrontal cortex (OFC) and anterior cingulate cortex (ACC), leaves the organism defenseless against cue-triggered action releases. Understanding that sign-tracking reflects an innate, biological imbalance between subcortical incentive attribution and prefrontal cognitive gating provides critical translational avenues for developing targeted, individualized therapeutics—ranging from deep brain stimulation (DBS) targeting the nucleus accumbens core to personalized cognitive-behavioral extinction paradigms designed to defuse the motivational charge of conditioned signs.

11. Methodological Variations, Modern Paradigms, and Computational Models

11.1 Parametric Manipulations of Temporal Dynamics

The speed, magnitude, and phenotypic distribution of autoshaping are deeply governed by the temporal architecture of the conditioning environment. Decades of parametric psychophysical studies have illuminated how the temporal relationship between the conditioned stimulus duration and the inter-trial interval dictates whether sign-tracking or goal-tracking topographies emerge.

The foundational metric governing associative acquisition speed in autoshaping is the cycle-to-trial ratio (often formalized as the $C/T$ ratio), where $T$ represents the absolute duration of the Conditioned Stimulus (the trial time), and $C$ represents the total duration of the conditioning cycle (the Inter-Trial Interval plus the trial duration). Pioneering psychophysical work by Peter Balsam and Charles Gallistel demonstrated that acquisition rate is an invariant function of this ratio:

$$\text{Acquisition Speed} propto \frac{C}{T}$$

When the $C/T$ ratio is exceedingly high—meaning the predictive cue ($T$) is brief (e.g., 5 seconds) relative to a prolonged, quiet inter-trial interval ($C$, e.g., 200 seconds)—the conditioned stimulus provides an extraordinarily high degree of information relative to the background context. Under high $C/T$ conditions, acquisition occurs with lightning rapidity, often within 10 to 20 trials, and the behavioral output heavily favors intense, high-frequency sign-tracking. The cue shines like a blazing informational and motivational beacon against a silent background.

Conversely, as the $C/T$ ratio is compressed—meaning the trial duration ($T$) is prolonged (e.g., 60 seconds) relative to a short ITI ($C$, e.g., 30 seconds)—the information rate carried by the CS collapses. Under low $C/T$ dynamics, autoshaping acquisition is severely retarded or entirely prevented; the animals fail to develop sign-tracking, shifting instead into lethargic goal-tracking or generalized chamber pacing. Temporal Difference (TD) reinforcement learning models incorporate these dynamics, demonstrating that the backpropagation of reward prediction error signals is critically dependent upon the sharp temporal boundaries of the predictive CS.

11.2 Extinction, Renewal, and Reinstatement Paradigms

The stability and resilience of autoshaped responses have made them a premier laboratory model for investigating the neurobiology of behavioral extinction, relapse, and contextual recovery. When an animal that has acquired robust sign-tracking is subjected to an extinction protocol—wherein the conditioned stimulus is repeatedly presented in the chamber but the primary reinforcer (food or water) is permanently withheld—the behavioral trajectory diverges sharply across tracking phenotypes.

Goal-tracking responses typically extinguish with marked rapidity. Because goal-trackers treat the cue primarily as an epistemic, informational signal, the sudden omission of the reward quickly updates their cognitive expectancy matrices: the prediction error is recognized, the cognitive map is revised, and hopper-checking behavior drops to baseline within a handful of sessions. Sign-tracking responses, however, exhibit immense resistance to extinction. Because the sign-tracker has imbued the physical cue with intrinsic incentive salience, the cue has become an independently rewarding stimulus. The animal continues to approach, grasp, bite, and peck the cue for prolonged periods long after it has ceased to deliver food, demonstrating the profound tenacity of conditioned incentive stimuli.

Furthermore, autoshaped responses serve as foundational platforms for modeling relapse configurations pioneered by Mark Bouton:

• ABA Renewal: When an animal undergoes autoshaping in Context A, undergoes extinction in Context B, and is subsequently placed back into Context A without any food delivery, the presentation of the CS instantly produces massive renewal of sign-tracking. The extinguished behavior was not erased; it was merely inhibited in Context B.
• ABC Renewal: Demonstrates that moving the animal to an entirely novel, neutral Context C following extinction in Context B is sufficient to unleash the conditioned sign-tracking response, confirming that extinction is deeply context-dependent, whereas initial conditioning generalizes widely.
• Reinstatement: Exposing an extinguished subject to a single, non-contingent presentation of the primary reinforcer (a free food pellet) completely outside the trial structure instantly reinstates vigorous sign-tracking upon the subsequent presentation of the CS, illustrating how non-contingent stress or reward primes can shatter previously learned extinction memory networks.

11.3 Advanced Computational and Neural Network Models

The theoretical richness of autoshaping has made it a fertile testing ground for modern computational neuroscience, formal reinforcement learning algorithms, and deep neural network architectures seeking to simulate biological action selection.

Classical Temporal Difference (TD) learning algorithms, such as the standard SARSA or Q-learning models, successfully capture the basic shift of reward prediction errors from the US back to the CS. However, standard TD models historically failed to account for the segregation between sign-tracking and goal-tracking, as they assume a unified, monolithic value function that merely drives an abstract “state-value.”

To resolve this limitation, computational theorists have turned to multi-system Actor-Critic Architectures and Bayesian inference frameworks:

• Model-Free vs. Model-Based Dissociation: Modern computational models conceptualize sign-trackers as relying predominantly on a Model-Free reinforcement learning system. Model-free systems calculate caching values based strictly on sensory-motor state-action pairings, driving rapid, automatic, habit-like and cue-driven behaviors that are impervious to changes in outcome value. Goal-trackers, conversely, are modeled as relying heavily on a Model-Based system, which constructs an internal forward model of the world—a cognitive spatial map of transitions between states—allowing flexible, goal-directed navigation directly to the reward hopper.
• Bayesian Predictive Inference: Bayesian models incorporate precision-weighted prediction errors. In these simulations, an individual’s prior beliefs regarding environmental uncertainty and cue reliability dictate how dopamine prediction errors are distributed. In agents configured with high prior uncertainty regarding environmental transitions, precision weights shift associative value directly onto the immediate, tangible sensory cue (the sign), mathematically simulating the emergence of sign-tracking.
• Neural Network Simulations of Behavior Systems: Deep artificial neural networks integrating simulated spatial reinforcement fields, sensory visual inputs, and motor execution outputs have successfully reproduced the spatial navigation dynamics of sign-tracking. By programming the network with basic ethological constraints mirroring Timberlake’s behavior systems, the networks naturally evolve sign-tracking spatial attraction loops when presented with localized visual inputs, proving that sign-tracking is the inevitable mathematical and computational emergence of an evolutionary system balancing focal search efficiency against spatial distance.

12. Legacy, Theoretical Evolution, and Future Directions in Autoshaping Research

12.1 Long-Term Impact on Learning Theory and Cognitive Psychology

More than half a century after Paul Brown and Herbert Jenkins published their modest five-page paper in the Journal of the Experimental Analysis of Behavior, the legacy of their experiment remains profound. Autoshaping served as the theoretical crowbar that helped pry open the dogmatic, monolithic edifice of mid-twentieth-century radical behaviorism, facilitating the transformative “cognitive revolution” across comparative psychology and behavioral neuroscience.

By decisively demonstrating that skeletal motor repertoires could be elicited by stimulus-reinforcer contingencies alone, autoshaping forced science to discard the crude view of the organism as an empty mechanical vessel whose actions are merely selected by environmental reinforcement. It established that conditioning is an active, dynamic biological process wherein an organism continuously generates internal cognitive representations, builds predictive models of environmental relationships, and deploys pre-existing, evolutionarily honed behavioral modules to interact with those representations. The illuminated key-peck of Columba livia revealed that the mind of an animal is dense with evolutionary architecture, compelling behavioral science to forever integrate ethology, cognitive psychology, and associative learning theory.

Today, the autoshaping paradigm is universally embedded within standard university curricula, undergraduate textbooks, and advanced doctoral seminars worldwide. It stands alongside Pavlov’s salivating dogs and Skinner’s operant chambers as one of the foundational, canonized experimental paradigms in the history of psychology—an enduring monument to the profound scientific revelations that can emerge from a single, beautifully controlled behavioral anomaly.

12.2 Contemporary Debates and Unresolved Questions

Despite over fifty years of intensive global scrutiny, autoshaping continues to generate lively, contentious theoretical debates at the cutting edge of contemporary behavioral science. One persistent controversy revolves around the precise boundary separating habitual actions from sign-tracking execution. While many neuroscientists conceptualize sign-tracking as a pure manifestation of Pavlovian incentive salience, others argue that extended sign-tracking gradually recruits dorsal striatal habitual machinery, transitioning from a purely elicited Pavlovian motivational response into an entrenched, automated motor habit.

Another profound unresolved frontier involves the challenge of translational human validity. While rodents and birds demonstrate clean, binary phenotypic segregations in simple operant chambers, human behavior is mediated by massive, layered prefrontal cortices, intricate linguistic structures, and dense socio-cultural landscapes. Designing translational human laboratory paradigms that can cleanly isolate sign-tracking from complex operant task-demands remains technically challenging. Researchers are currently utilizing immersive virtual reality (VR) environments, mobile eye-tracking headsets, and high-density electroencephalography (EEG) to establish whether human sign-tracking correlates reliably with specific psychiatric genetic matrices in real-world clinical contexts.

Finally, intense scientific inquiry is directed at understanding the molecular and environmental mechanisms governing resilience. Why do some outbred individuals raised under identical environmental conditions maintain an unyielding goal-tracking phenotype, completely resistant to the siren song of the conditioned sign? Deciphering the genetic, maternal, and developmental factors that confer this neurobiological protection against incentive over-attribution holds the key to developing preventative interventions against addiction and impulse disorders before pathological behaviors take root.

12.3 Emerging Frontiers: Optogenetics, Chemogenetics, and Precision Psychiatry

As autoshaping research moves deep into the twenty-first century, it is being propelled by an astonishing technological revolution in molecular neuroscience. The deployment of cell-type-specific optogenetics and chemogenetics (Designer Receptors Exclusively Activated by Designer Drugs, or DREADDs) has allowed researchers to move beyond correlational measurements to execute exquisite, real-time causal manipulation of the precise neural circuits underlying sign-tracking.

Using optogenetic stimulation, researchers can now shine targeted laser light through intracranial optical fibers to selectively silence or excite specific sub-populations of dopamine neurons within the ventral tegmental area with millisecond precision. Studies have demonstrated that optogenetically mimicking the phasic dopamine burst at the precise moment a conditioned cue appears can instantly transform an innate goal-tracking rat into a vigorous sign-tracker, causally proving that the sub-second temporal firing of dopaminergic synapses is the precise molecular trigger for incentive salience attribution.

Simultaneously, chemogenetic technologies are being deployed to pioneer potential therapeutic reversals of maladaptive sign-tracking. By infusing inhibitory DREADDs targeting the prelimbic-to-accumbens projection pathways, neuroscientists can pharmacologically reactivate top-down prefrontal cognitive control, selectively muting excessive sign-tracking while leaving normal, adaptive goal-directed behavior completely intact.

These breakthroughs herald the dawn of an era of precision psychiatry. In the clinical paradigms of tomorrow, an individual’s behavioral tracking endophenotype—measured through standardized computational tasks or eye-tracking metrics—could serve as a definitive biomarker. Patients identified as possessing high sign-tracking profiles could be routed away from treatments reliant purely on conscious cognitive willpower, and instead matched with targeted pharmacological, neuromodulatory, and behavioral extinction therapies specifically engineered to recalibrate the hyper-active incentive circuits that bind the human mind to the signs of our modern world.

Conclusion

The journey that began in 1968 with Paul Brown and Herbert Jenkins observing a small group of pigeons pecking an automatically illuminated piece of translucent plastic has profoundly reshaped the landscape of scientific psychology. What was initially conceived as an automated method for training an animal—an “autoshaping” curiosity—revealed itself to be an intellectual fissure that toppled the monolithic divide separating classical Pavlovian conditioning from operant learning. The pigeon’s key-peck was not a passive habit shaped by post-hoc consequences, nor was it a random superstition; it was the raw, unconditioned emergence of an evolutionarily conserved behavior system, driven by the profound psychological power of incentive salience.

By demonstrating that a neutral predictive sign can be transformed into a motivational magnet—a surrogate for the biological goal itself—autoshaping illuminated the deep neural wiring that connects perception, expectation, and motor execution. The paradigm revealed that within the central nervous system of vertebrate life, signs do not merely point; they pull. The conditioned stimulus acquires the power to seize attention, summon spatial navigation, and compel consummatory engagement, a biological imperative that operates with relentless autonomy across animal phyla, from goldfish and pigeons to laboratory rodents, primates, and human beings.

Today, as autoshaping continues to inform the frontiers of neurobiology, computational psychiatry, and addiction medicine, it stands as a testament to the supreme importance of empirical observation over theoretical dogma. Brown and Jenkins did not invent the pigeon’s key-peck; they simply designed an experiment pure enough to let the organism reveal its true nature. In doing so, they provided experimental psychology with one of its most enduring, intellectually fertile, and transnationally vital paradigms—a timeless window into the mechanics of desire, the architecture of associative memory, and the ancient biological forces that dictate the behavior of living organisms.

References

• Balsam, P. D., & Gallistel, C. R. (2009). Temporal maps and associative learning. Trends in Neurosciences, 32(2), 73–78. https://doi.org/10.1016/j.tins.2008.10.004
• Berridge, K. C., & Robinson, T. E. (1998). What is the role of dopamine in reward: Hedonic impact, reward learning, or incentive salience? Brain Research Reviews, 28(3), 309–369. https://doi.org/10.1016/S0165-0173(98)00019-8
• Boakes, R. A. (1977). Performance on learning to associate a stimulus with positive reinforcement. In H. Davis & H. M. C. Hurwitz (Eds.), Operant-Pavlovian Interactions (pp. 67–97). Lawrence Erlbaum Associates.
• Bouton, M. E. (2004). Context and behavioral processes in extinction. Learning & Memory, 11(5), 485–494. https://doi.org/10.1101/lm.78804
• Breland, K., & Breland, M. (1961). The misbehavior of organisms. American Psychologist, 16(11), 681–684. https://doi.org/10.1037/h0040090
• Brown, P. L., & Jenkins, H. M. (1968). Auto-shaping of the pigeon’s key-peck. Journal of the Experimental Analysis of Behavior, 11(1), 1–8. https://doi.org/10.1901/jeab.11-1
• Flagel, S. B., Clark, J. J., Robinson, T. E., Mayo, L., Czuj, A., Willuhn, I., Akers, C. A., Clinton, S. M., Phillips, P. E. M., & Akil, H. (2011). A selective role for dopamine in the acquisition of an approach response to a conditioned stimulus. Nature, 469(7328), 53–57. https://doi.org/10.1038/nature09588
• Hearst, E., & Jenkins, H. M. (1974). Sign-Tracking: The Stimulus-Reinforcer Relation and Directed Action. Psychonomic Society.
• Jenkins, H. M., & Moore, B. R. (1973). The form of the auto-shaped response with food or water reinforcers. Journal of the Experimental Analysis of Behavior, 20(2), 163–181. https://doi.org/10.1901/jeab.1973.20-2
• Pavlov, I. P. (1927). Conditioned Reflexes: An Investigation of the Physiological Activity of the Cerebral Cortex (G. V. Anrep, Trans.). Oxford University Press.
• Rescorla, R. A., & Wagner, A. R. (1972). A theory of Pavlovian conditioning: Variations in the effectiveness of reinforcement and nonreinforcement. In A. H. Black & W. F. Prokasy (Eds.), Classical Conditioning II: Current Research and Theory (pp. 64–99). Appleton-Century-Crofts.
• Robinson, T. E., & Berridge, K. C. (1993). The neural basis of drug craving: An incentive-sensitization theory of addiction. Brain Research Reviews, 18(3), 247–291. https://doi.org/10.1016/0165-0173(93)90013-P
• Sarter, M., & Phillips, P. E. M. (2018). The neuroscience of sign-tracking and goal-tracking: Attentional, motivational and neurochemical mechanisms. Neuropharmacology, 137, 240–248. https://doi.org/10.1016/j.neuropharm.2018.04.025
• Skinner, B. F. (1938). The Behavior of Organisms: An Experimental Analysis. Appleton-Century.
• Skinner, B. F. (1948). ‘Superstition’ in the pigeon. Journal of Experimental Psychology, 38(2), 168–172. https://doi.org/10.1037/h0055873
• Timberlake, W. (1993). Behavior systems and reinforcement: An integrative approach. Journal of the Experimental Analysis of Behavior, 60(1), 105–128. https://doi.org/10.1901/jeab.1993.60-105
• Timberlake, W., & Lucas, G. A. (1989). Behavior systems and learning: From misbehavior to general principles. In S. B. Klein & R. R. Mowrer (Eds.), Contemporary Learning Theories: Instrumental Conditioning Theory and the Impact of Biological Constraints on Learning (pp. 237–275). Lawrence Erlbaum Associates.
• Williams, D. R., & Williams, H. (1969). Auto-maintenance in the pigeon: Sustained pecking despite contingent non-reinforcement. Journal of the Experimental Analysis of Behavior, 12(4), 511–520. https://doi.org/10.1901/jeab.1969.12-511