The Higher-Order Conditioning Experiment – Ivan Pavlov

The dawn of twentieth-century behavioral physiology was characterized by a profound ontological and methodological rupture with classical nineteenth-century mentalism. At the epicenter of this paradigm shift stood the Russian physiologist Ivan Petrovich Pavlov, whose empirical investigations into the regulatory mechanisms of the mammalian digestive system inadvertently unlocked the physiological architecture of associative memory and learning. While Pavlov’s discovery of the basic conditional reflex—wherein a neutral sensory stimulus, through repeated temporal pairing with a biologically significant unconditioned event, acquires the capacity to evoke an anticipatory autonomic response—revolutionized naturalistic psychology, it was his subsequent exploration of higher-order conditioning that truly exposed the astonishing computational complexity of the cerebral cortex. This advanced dimension of reflexology demonstrated that an organism’s internal predictive model of the environment could extend far beyond immediate biological imperative, building multi-tiered, hierarchical cascades of conditional associations without the ongoing presence of primary reinforcement.

Higher-order conditioning represents an epistemological bridge linking elementary homeostatic reflex arcs to the expansive, abstract associative networks that govern advanced cognitive processes, human linguistic structures, evaluative judgment, and complex psychopathology. By demonstrating that an established conditional stimulus could itself function as an anchor of secondary reinforcement for a novel, unreinforced neutral cue, Pavlov dismantled the reductionist assertion that physiological associative learning was restricted to direct, primary visceral encounters with sustenance, pain, or thermal extremes. Instead, his laboratory uncovered a self-assembling architecture of cortical excitability: a biological mechanism through which organisms construct elaborate mental representations of temporal causality, mapping chains of predictive environmental signals across vast distances in space and time.

This treatise provides an exhaustive, multi-dimensional analysis of the higher-order conditioning experiment as conceived and executed within Pavlov’s Imperial Institute of Experimental Medicine in Saint Petersburg. Across twelve comprehensive structural dimensions, it traces the historical, philosophical, and methodological roots of Pavlovian reflexology; interrogates the mechanics of first-order classical conditioning; deconstructs the conceptual and operational paradigms of second- and third-order conditioning; analyzes the intricate cortical neurodynamics postulated by Pavlov; and resolves the fundamental operational paradoxes that emerge when unreinforced associations threaten to trigger experimental extinction or conditioned inhibition. Finally, it situates Pavlov’s foundational discoveries within the formal mathematical frameworks of modern computational neuroscience, contemporary cognitive learning theory, and clinical translational psychology.

1. Historical Context and Epistemological Foundations of Pavlovian Reflexology

1.1 The Evolution from Digestive Physiology to Objective Behavioral Science

The intellectual trajectory that culminated in the formulation of the higher-order conditional reflex began not within the discipline of psychology, but deep within the rigorous traditions of nineteenth-century European digestive physiology. Throughout the late 1880s and 1890s, Ivan Petrovich Pavlov directed his laboratory at the Institute of Experimental Medicine in Saint Petersburg toward unraveling the neural regulation of secretory phenomena within the mammalian gastrointestinal tract. Employing surgically sophisticated procedures that preserved the autonomic innervation of the gastrointestinal organs, Pavlov systematically isolated the secretory functions of the salivary glands, stomach, and pancreas, an endeavor that yielded the 1904 Nobel Prize in Physiology or Medicine. However, as his laboratory refined the collection and volumetric quantification of canine secretions, an persistent experimental anomaly disrupted their physiological baselines: laboratory canines routinely initiated salivary and gastric secretions prior to the physical contact of foodstuffs with the oral mucosa.

Initially dismissed as disruptive experimental noise or labeled anthropomorphically as “psychic secretions” (psikhicheskoe vydelenie), these anticipatory glandular outputs were provoked by contextual visual, auditory, and olfactory artifacts—such as the sight of the food dish, the sound of approaching laboratory assistants’ footsteps, or the appearance of white laboratory coats. Pavlov recognized that treating these phenomena as subjective mental events belonging to the domain of introspective psychology was an epistemological dead end. Subjective psychology, dominated by speculative introspection, lacked quantitative reproducibility and verifiable physical metrics. Consequently, Pavlov executed an uncompromising methodological transition: he abandoned mentalistic vocabulary and transformed “psychic secretions” into the rigorous empirical study of the acquired reflex, or more precisely, the conditional reflex (uslovnyi refleks).

This radical transformation was heavily indebted to the theoretical groundwork laid by Ivan Mikhailovich Sechenov, widely recognized as the father of Russian physiology. In his monumental 1863 work, Reflexes of the Brain (Refleksy golovnogo mozga), Sechenov advanced the audacious thesis that all acts of conscious and unconscious life, including complex psychological processes, were fundamentally reflex arcs mediated by the central nervous system. Sechenov asserted that psychical activity is initiated by external sensory stimulation, processed through cerebral connective paths, and culminated in an end-stage muscular or glandular response. Guided by Sechenov’s mechanistic materialism, Pavlov resolved to investigate these anticipatory secretions not as manifestations of an inaccessible canine “psyche,” but as purely objective, quantifiable neurophysiological responses to environmental signals operating under deterministic natural laws.

1.2 The Materialist Epistemology of the Russian Physiological Tradition

The ideological and methodological architecture of Pavlovian reflexology was deeply anchored in the materialist epistemology that swept through nineteenth-century Russian natural philosophy. Rejecting Cartesian mind-body dualism and the prevailing German introspectionist paradigms championed by Wilhelm Wundt, Pavlov and his contemporaries operated under the conviction that mental phenomena possessed no independent ontological status separate from biological tissue. The cerebral cortex was conceived as an exquisitely complex, self-regulating biological machine whose primary evolutionary function was the maintenance of dynamic equilibrium between the organism and its surrounding milieu. To invoke non-material forces or subjective conscious states to explain behavior was, in Pavlov’s view, an abdication of scientific responsibility.

Central to this epistemological stance was the principle of nervous determinism. Pavlov posited that every observable behavioral output was strictly determined by preceding external or internal material causes operating through the reflex arc: receptor stimulation, centripetal (afferent) transmission, central cortical processing and synthesis, centrifugal (efferent) transmission, and effector organ actuation. The ultimate objective of behavioral science was to map psychic phenomena directly onto the functional physiology of the higher nervous activity (vysshaya nervnaya deyatel’nost’). By focusing on the salivary gland—a purely autonomic effector completely divorced from the voluntary, conscious motor control of the animal—Pavlov secured an objective, uncontaminated physiological readout of central nervous processing.

The institutional vessel for this rigorous empirical program was the Institute of Experimental Medicine in Saint Petersburg, established in 1890. Funded through state patronage and private benefactors, the Institute provided Pavlov with unprecedented infrastructural resources. Within this facility, subjective interpretations were strictly prohibited; laboratory workers who utilized psychological vernacular—such as stating that a dog “remembered,” “wished,” or “expected” food—were subjected to formal laboratory fines and intellectual reprimands. The Institute operated as a tightly disciplined assembly of physiologists, surgeons, and technicians dedicated to translating the dynamic complexities of the animal mind into the precise, volumetric measurement of physiological fluid dynamics.

1.3 Initial Conceptualization of the Conditioned Reflex

The formalization of the conditioned reflex framework began with a categorical distinction between two discrete classes of physiological responses: the inborn, unconditioned reflex (bezuslovnyi refleks) and the acquired, conditional reflex (uslovnyi refleks). Unconditioned reflexes were conceptualized as phylogenetically hardwired, anatomically invariant, and biologically permanent pathways mediated primarily by subcortical structures and the brainstem. These reflexes guaranteed immediate survival through innate autonomic and somatic actions, such as swallowing, coughing, defensive withdrawal from nociceptive stimuli, and the unconditioned gustatory salivary reflex triggered when acidic substances or food entered the oral cavity.

In stark contrast, the conditional reflex represented an ontogenetic, plastic acquisition unique to the history of the individual organism. Mediated entirely by the cerebral neocortex, it was dependent upon environmental context and specific empirical contingencies. To forge a conditional reflex, a previously indifferent or neutral sensory stimulus had to be systematically coordinated in time with an unconditioned stimulus possessing direct biological significance. The unconditioned stimulus carried inherent survival value—such as energy replenishment via alimentary intake or bodily preservation via defensive avoidance. Without this primary biological anchor, the neutral stimulus could not engage the organism’s autonomic machinery.

Crucially, Pavlov recognized early on that conditional reflexes were defined by their fundamental malleability. Unlike rigid unconditioned connections, conditional reflexes were dynamic, temporary connections (vremennaya svyaz) established within the cortical substrate. If the environmental contingencies that birthed the association shifted—that is, if the conditional signal repeatedly manifested without the subsequent delivery of the biologically significant unconditioned reinforcement—the acquired reflex did not remain fixed. Instead, it underwent rapid attenuation, entering a state of temporary functional suspension through the active accumulation of internal cortical inhibition. This intrinsic plasticity, marked by the constant establishment, maintenance, and extinction of temporary neural links, formed the operational foundation upon which more complex, hierarchical associative paradigms would subsequently be discovered.

2. Theoretical Framework of First-Order Classical Conditioning

2.1 Constituent Elements of the Conditioning Paradigm

To comprehend the architectural leap demanded by higher-order conditioning, one must first dissect the foundational mechanics of first-order classical conditioning. The canonical conditioning paradigm consists of four core elements: the Unconditioned Stimulus (UCS), the Unconditioned Response (UCR), the Neutral Stimulus (NS), and the resultant Conditioned Stimulus (CS) and Conditioned Response (CR). The Unconditioned Stimulus represents a biologically potent, unlearned environmental event that reliably and automatically triggers an unconditioned physiological reaction without any prior training. In Pavlov’s alimentary experiments, the UCS typically comprised desiccated meat powder or diluted hydrochloric acid introduced directly into the animal’s oral cavity, invoking an obligatory, innate homeostatic adjustment.

The Unconditioned Response constitutes the hardwired autonomic or somatic output reliably evoked by the UCS. In the case of alimentary stimulation, the UCR manifests as copious, enzyme-rich salivary secretion alongside localized vascular dilation and masticatory motor movements. Conversely, the Neutral Stimulus is an environmental event that, prior to experimental manipulation, evokes no measurable change in the target physiological effector. To serve as a viable NS, the stimulus must meet rigorous criteria: it must possess sufficient physical intensity to exceed the sensory thresholds of the subject’s peripheral receptors, yet remain completely devoid of innate biological valence or prior associative meaning regarding the specific response system being monitored.

When an NS is systematically coupled with a UCS in accordance with precise temporal parameters, a structural transformation occurs. The NS sheds its neutrality, transforming into a Conditioned Stimulus (CS). Concurrently, the organism develops the capacity to emit an anticipatory, learned output—the Conditioned Response (CR)—in direct reaction to the solitary presentation of the newly minted CS. While the CR often shares structural similarities with the UCR, Pavlov’s careful physiochemical analyses demonstrated that it is not a direct carbon copy; the conditioned salivary output often differs from its unconditioned antecedent in terms of chemical composition, protein concentration, latency of onset, and total volumetric profile.

2.2 Temporal Dynamics and Reinforcement Contingencies

The successful establishment of an associative connection between a neutral stimulus and an unconditioned stimulus is exquisitely dependent upon the temporal dynamics orchestrating their co-occurrence. Pavlovian researchers identified four primary temporal conditioning paradigms: delay conditioning, trace conditioning, simultaneous conditioning, and backward conditioning. In *delay conditioning*, the onset of the neutral stimulus precedes the onset of the UCS by a variable interval, with the NS continuing to persist until the UCS terminates. When the interval is brief (typically fractions of a second to a few seconds), this paradigm proves overwhelmingly optimal for rapid associative acquisition, establishing an immediate, highly reliable predictive relationship.

In *trace conditioning*, an explicit temporal gap separates the termination of the neutral stimulus from the subsequent onset of the UCS. Under this arrangement, associative formation cannot rely on contiguous overlapping sensory inputs; rather, the central nervous system must rely on a persisting neurochemical or electrophysiological “memory trace” of the conditioned stimulus within the cortical sensory analyzer to bridge the temporal chasm. In *simultaneous conditioning*, the NS and UCS are presented concurrently, while in *backward conditioning*, the UCS precedes the NS. Crucially, empirical investigations established that both simultaneous and backward presentations yield remarkably weak, highly unstable, or entirely non-existent excitatory conditioning, often generating profound inhibitory properties instead.

These temporal dynamics underscored the profound theoretical debate between mere temporal *contiguity* and informational *contingency*. While Pavlov initially conceptualized conditioning primarily through the lens of temporal contiguity—the overlapping excitation of disparate cortical zones—the data inevitably pointed toward predictive informational value. The acquisition of associative strength follows an asymptotic trajectory characterized by the law of primary reinforcement: initial pairings yield rapid, non-linear surges in CR magnitude, which subsequently decelerate as the associative strength asymptotically approaches a physiological maximum dictated by the biological intensity of the UCS.

2.3 Extinction, Spontaneous Recovery, and External Disinhibition

A primary conditional reflex does not represent an indelible, permanent modification of the nervous system; rather, its phenotypic expression remains contingent upon the ongoing predictive accuracy of the CS. When an established CS is repeatedly presented in the complete absence of the reinforcing UCS, the magnitude of the Conditioned Response systematically attenuates across successive trials, eventually reaching a state of behavioral dormancy known as experimental extinction (ugashanie). Crucially, Pavlov established that extinction is not equivalent to passive forgetting, neural decay, or associative erasure; rather, it is an active, energetically demanding neurophysiological process termed internal inhibition (vnutrennee tormozhenie), which actively suppresses the overt manifestation of the learned response.

The non-destructive nature of extinction is empirically demonstrated by the robust phenomenon of spontaneous recovery. If an extinguished animal is removed from the experimental apparatus and returned after a period of temporal rest, the presentation of the extinguished CS will once again evoke a significant, quantifiable conditioned salivary output, despite the complete absence of any intermittent reinforcement pairings. The temporal hiatus allows the fragile, energetically vulnerable internal inhibition to dissipate more rapidly than the underlying excitatory trace, thereby unmasking the preserved core associative connection.

Furthermore, internal inhibition can be instantly disrupted through the presentation of an extraneous, novel sensory stimulus—such as a sudden acoustic click, a tactile vibration, or an illumination shift—delivered concurrently with the extinguished CS. This intervention, designated by Pavlov as disinhibition (rastormazhivanie), immediately restores the extinguished Conditioned Response. The novel extraneous stimulus evokes an innate investigatory or orienting reflex, which irradiates across the cortex and collides with the localized zone of internal inhibition, effectively neutralizing it and unmasking the underlying excitatory association. These precise phenomena proved that associative memory architectures persist within the nervous tissue even when entirely silenced at the behavioral surface.

3. Conceptualizing Higher-Order Conditioning: Theoretical Architecture

3.1 Defining Higher-Order Associative Architecture

Having systematically deciphered the physiological laws governing first-order classical conditioning, Pavlov and his research cadre confronted a fundamental theoretical question: Can an organism acquire novel behavioral adaptations without direct, unmediated contact with a biologically potent unconditioned stimulus? In natural ecological niches, direct encounters with primary biological events (such as visceral ingestion or predatory attack) are frequently preceded by sequences of distal, environmental signals arranged in complex temporal chains. If an organism could only learn through direct, simultaneous pairing with primary biological unconditioned events, its predictive horizons would remain severely restricted, limiting survival prospects in complex, dynamic environments.

Higher-order conditioning provides the physiological mechanism that resolves this ecological limitation. The fundamental mechanics of this paradigm involve the pairing of a novel, completely neutral stimulus (designated as a second-order conditional stimulus, or CS2) with a previously established, robustly reinforced first-order conditioned stimulus (CS1). Crucially, throughout this secondary pairing phase, the primary biological reinforcement—the Unconditioned Stimulus (UCS)—is entirely omitted from the experimental environment. Through repeated temporal coupling with CS1, the novel CS2 progressively acquires the capacity to evoke the conditioned response autonomously, despite having never experienced direct contiguous alignment with the primary biological reinforcer.

This process establishes a hierarchical associative chain. The motivational-affective properties and predictive potency originally anchored to the primary UCS are effectively transferred upstream, migrating from the primary reinforcement to CS1, and subsequently from CS1 to CS2. The operational distinction between these structural levels is defined by the underlying associative topology:

• Primary (First-Order) Conditioning: Direct associative coupling of an unconditioned stimulus (UCS) with a neutral stimulus, transforming it into a first-order conditioned stimulus (CS1 → UCS).
• Secondary (Second-Order) Conditioning: Coupling of a novel neutral stimulus with an established, reinforced first-order conditioned stimulus in the complete absence of the primary unconditioned stimulus (CS2 → CS1).
• Tertiary (Third-Order) Conditioning: Further hierarchical coupling of an additional neutral stimulus with an established second-order conditioned stimulus, absent any direct lower-tier reinforcement (CS3 → CS2).

3.2 Associative Transfer Models: S-R versus S-S Formulations

The theoretical interpretation of higher-order conditioning sparked an enduring debate within behavioral physiology and learning theory regarding the fundamental nature of the underlying neural substrate: Does the second-order stimulus link directly to the motor/autonomic effector program, or does it activate an intermediate cognitive/representational trace of the primary conditional stimulus? This debate crystallized into two competing theoretical formulations: the Stimulus-Response (S-R) model and the Stimulus-Stimulus (S-S) model. The S-R formulation, heavily favored by early American behaviorists, posited that during second-order pairing, the presentation of CS1 mechanically triggers the peripheral or central conditioned response (CR); consequently, CS2 becomes directly wired to this downstream autonomic response center, bypassing any active representational reliance on CS1.

Conversely, the S-S formulation, which aligned closely with Pavlov’s internal conceptualization of central cortical connectivity, asserted that higher-order conditioning represents the formation of a mediated central neural bridge between the sensory cortical analyzers of CS2 and CS1. Under the S-S model, the activation of CS2 does not directly trigger the final effector pathway; rather, it invokes an internal, central representation of CS1, which in turn cascades down the previously established neural path to evoke the conditioned salivary secretion. In this view, the animal does not merely acquire an isolated, mechanical glandular twitch; it forms a dynamic, multi-tiered representational model of external environmental contingency.

Testing these competing formulations empirically presented a profound challenge to early twentieth-century reflexology. The definitive empirical resolution required the post-conditioning devaluation of the first-order conditional stimulus. If the S-S hypothesis held true, devaluing CS1 (via intensive, isolated extinction trials to eliminate its associative capacity) should theoretically abolish the animal’s responsiveness to the unextinguished CS2, because the intermediate representational node has been functionally neutralized. Conversely, if S-R learning prevailed, CS2 should continue to evoke the conditioned response unabated, as its direct physical link to the autonomic motor/secretory center remains structurally intact. As subsequent twentieth-century research demonstrated, while some motor paradigms exhibit S-R properties, Pavlovian autonomic higher-order conditioning frequently exhibits robust S-S representational characteristics, validating Pavlov’s early cortical network hypotheses.

3.3 The Evolutionary Significance of Second-Order Associations

From an evolutionary and teleological standpoint, the biological machinery sustaining higher-order conditioning represents an extraordinary adaptive innovation. In the wild, primary biological outcomes—such as the capture of prey, the ingestion of water, or mortal injury from a predator—are rarely preceded by immediate, instantaneous unconditioned sensations. Instead, the approach of a predator generates a complex, temporal-spatial cascade of distal cues: the subtle rustling of high grass (CS2), which reliably precedes the visual silhouette of the stalking carnivore (CS1), which ultimately precedes the lethal physical attack (UCS). An animal restricted exclusively to first-order conditioning could only initiate defensive or evasive maneuvers upon detecting the immediate proximal stimulus (CS1).

The survival advantage conferred by second-order conditioning lies in its predictive reach. By enabling the central nervous system to treat reliable predictors of predictors as behaviorally actionable signals, the organism gains critical temporal margins to enact preparatory autonomic adjustments, optimize metabolic expenditure, and initiate evasive locomotor maneuvers long before the primary threat or reward enters immediate physical proximity. The ecological validity of higher-order conditioning is thus evident in natural foraging strategies, territorial defense, mating rituals, and anti-predator avoidance, all of which operate across multi-tiered informational hierarchies.

However, this computational capacity incurs severe energetic costs and neurological demands. The establishment and maintenance of extensive higher-order predictive networks requires an immensely expanded neocortical surface, high-fidelity sensory analyzers, and complex cellular mechanisms capable of sustained metabolic activity. Moreover, higher-order predictive networks present an inherent biological risk: if the nervous system indiscriminately links every secondary environmental cue to primary homeostatic centers, it risks runaway cortical excitation, sensory overload, and the inappropriate expenditure of vital metabolic resources on tangential environmental noise. Consequently, natural selection had to evolve delicate homeostatic regulatory mechanisms—namely, powerful processes of cortical inhibition—to strictly delineate the operational boundaries of higher-order conditioning.

4. Pavlov’s Experimental Apparatus, Methodologies, and Protocol Design

4.1 The Architecture of the ‘Tower of Silence’

To isolate higher-order conditioned reflexes from confounding environmental noise, Pavlov orchestrated the design and construction of an unprecedented research facility at the Institute of Experimental Medicine: the famed “Tower of Silence” (Bashnya Molchaniya). Constructed between 1910 and 1914 with special funding, this three-story research citadel was engineered specifically to eliminate acoustic, vibrational, thermal, and olfactory contamination that could destabilize delicate cortical dynamics. The experimental chambers were isolated from external vibrations through double-foundation architecture, surrounded by deep trenches filled with vibration-absorbing materials such as sand, sawdust, and viscous fluids, and shielded by double-walled rooms separated by air gaps and dense insulation composed of straw and lead sheeting.

Within these hermetically sealed, windowless testing cells, the canine subject stood suspended in a padded, non-restrictive leather harness mounted upon an isolated wooden frame. The chamber was engineered to sever completely the physical and sensory presence of the human experimenter from the animal. Pavlov had identified that unintentional, unconscious experimenter cues—such as subtle shifts in respiratory patterns, micro-movements, changes in posture, or the odor of human sweat—inevitably functioned as extraneous conditional signals, generating pervasive laboratory artifacts that obscured genuine higher-order processes.

To eliminate this contamination, the Tower of Silence incorporated complex pneumatic, mechanical, and electrical automation systems. The experimenter was stationed entirely outside the sealed chamber in an adjoining control room, operating behind thick acoustic isolation panels and viewing the subject through specialized periscopic mirrors and double-glass inspection ports. All sensory stimuli were delivered via automated, remote pneumatic lines and low-voltage electrical relays. Desiccated meat powder was delivered directly into the animal’s feed container through pneumatic tubes powered by pressurized air bulbs, allowing the operational sequence to unfold with clockwork precision, totally devoid of human sensory intrusion.

4.2 Surgical Preparation and Quantification of Salivary Secretion

The foundation of Pavlov’s quantitative methodology rested upon his pioneering, minimally traumatic chronic surgical protocols. Pavlov staunchly rejected acute vivisectional methods that involved severe surgical trauma, extensive anesthesia, or acute tissue damage, recognizing that physiological baseline functions under such conditions were fundamentally pathological. Instead, his subjects underwent sterile, aseptic surgical exteriorization of the salivary ducts, primarily the parotid gland (situated near the ear) or the submaxillary gland (located beneath the jaw). By freeing the terminal end of the salivary duct from the internal mucosal lining of the cheek and translocating it through an incision outward to the external epidermal surface, Pavlov transformed the flow of saliva into an externally accessible, quantifiable phenomenon.

Once healed, the exteriorized fistula allowed saliva to discharge directly on the outside of the animal’s cheek, entirely unmixed with oral secretions, masticated food residues, or gastric fluids. Over this external fistula, a small, calibrated glass funnel or specialized hemispherical glass collection cup was hermetically affixed using a specialized adhesive compound composed of zinc oxide, resin, and beeswax. The collection funnel channeled the fluid directly into a graduated manometer tube or a high-precision mechanical drop-recording apparatus.

The quantification of the secretory output was exceptionally rigorous. The measurement parameters extended beyond mere volumetric yield to track three primary variables: the *latency of onset* (the precise interval, in fractions of a second, elapsed between the onset of the conditional stimulus and the initial emergence of fluid); the *total volume* of saliva secreted (measured in fractions of a cubic centimeter or precise drop counts); and the *rate of secretory flow* over extended temporal intervals (drops per five-second or ten-second epoch). Baseline homeostatic stability was meticulously established prior to every experimental session; if an animal exhibited spontaneous, irregular salivation due to hunger, systemic inflammation, or emotional agitation, the trial was immediately aborted to preserve data integrity.

4.3 Standardization and Calibration of Auditory, Visual, and Tactile Stimuli

To construct a multi-tiered higher-order associative network, Pavlov required sensory stimuli that were not only intensely distinct from one another, but also thoroughly calibrated in terms of their physical characteristics. The Saint Petersburg laboratory developed an extraordinary array of standardized sensory delivery apparatuses spanning auditory, visual, and somatosensory modalities. In the auditory domain, precision was achieved through mechanical metronomes calibrated to tick at precise, unwavering frequencies (e.g., 60, 100, or 120 beats per minute), electric bells mounted at calibrated acoustic distances, specialized tuning forks set to specific acoustic pitches, and compressed-air organ pipes capable of emitting pure auditory tones across the canine hearing spectrum.

Visual stimuli were engineered to engage the canine visual analyzer through distinct geometrical and kinetic configurations. These included backlit glass panels featuring illuminated geometrical shapes (such as squares, triangles, or circles), rotating black-and-white spiral discs, mechanical shields that abruptly descended to expose illuminated surfaces, and synchronized flashes of varying luminance. Tactile and somatosensory stimuli were operationalized using the taktator, a mechanical scratching and thermal apparatus developed directly within the Institute. The taktator used small, pneumatically driven blunt needles or thermal brass pads applied directly to the shaved skin of the animal’s trunk, thigh, or paw, allowing the experimenters to vary both the spatial location and frequency of cutaneous mechanical pressure or thermal shifts (ranging from ice-cold water to elevated warmth).

Every stimulus was subjected to exacting standardization: its physical duration (routinely fixed at 15 to 30 seconds), its physical intensity (measured in decibels for acoustic cues, candlepower for visual signals, and pneumatic pressure for tactile activators), and the inter-trial intervals separating presentations. Inter-trial intervals were typically maintained between five and fifteen minutes to ensure the complete dissipation of trace excitation and to prevent physiological fatigue of the glandular secretory cells, establishing a clean neurochemical slate for each successive empirical iteration.

5. The Primary Empirical Sequence: Establishing Second-Order Conditioning

5.1 Phase One: Establishing Robust First-Order Conditioning (CS1-UCS)

The systematic realization of second-order conditioning required the flawless execution of a tripartite empirical sequence. Phase One focused exclusively on the establishment of an immutable, asymptotic first-order conditional reflex. In a typical canonical protocol within Pavlov’s laboratory, an acoustic stimulus was selected as the first-order conditional stimulus (CS1)—frequently a mechanical metronome ticking at a rigid frequency of 120 beats per minute. The canine, positioned within the isolated testing chamber, was subjected to repeated forward delay-conditioning trials. The metronome was activated; after a strictly maintained interval (typically 15 to 22 seconds of isolated acoustic stimulation), a small quantity of meat powder (the UCS) was delivered pneumatically into the feed bowl, prompting the animal to consume the food while the metronome continued to tick for an additional five to ten seconds.

This coupling protocol was repeated dozens, sometimes hundreds, of times across multiple consecutive days. Pavlov insisted on strict criterion testing to verify that the first-order reflex had attained structural stability. The animal was required to exhibit stable asymptotic salivary output: upon the isolated activation of the metronome (CS1), a latency of initial secretion ranging between one and three seconds had to be reliably observed, followed by a continuous, profuse outflow of saliva (often reaching 30 to 60 drops per 30-second epoch) prior to any food presentation. Furthermore, the structural stability of CS1 was confirmed by testing its resistance to spontaneous day-to-day fluctuations, ensuring that the cortical representation of the metronome was firmly anchored to the subcortical unconditioned alimentary reflex center.

5.2 Phase Two: Introducing and Coupling the Second-Order Stimulus (CS2-CS1)

Once Phase One satisfied these rigorous physiological benchmarks, the experimental protocol transitioned into Phase Two: the introduction and associative pairing of the second-order stimulus. For the CS2, Pavlov typically selected a sensory stimulus operating within a completely distinct sensory analyzer to prevent cross-modal sensory confusion or stimulus generalization. A preferred CS2 was a silent visual stimulus, such as the sudden illumination of a black square on a translucent glass screen positioned directly within the dog’s forward visual field.

The pairing protocol in Phase Two required exacting temporal coordination and, fundamentally, the total, absolute omission of the primary unconditioned stimulus. The meat powder was completely excluded from the chamber. The trial sequence commenced with the presentation of the novel visual stimulus (CS2). After an isolated presentation of CS2 lasting approximately five to ten seconds, the acoustic metronome (CS1) was activated, typically overlapping with the visual stimulus for a brief duration, or initiated immediately following the termination of CS2 (a trace/short-delay configuration). After both stimuli completed their paired operational window (e.g., a total epoch of 15 to 20 seconds), the stimuli ceased simultaneously, and no food was delivered.

During the initial presentations of Phase Two, the experimenter observed the emergence of the unconditioned investigatory reflex (the “what-is-it?” reflex, or orientirovochnyi refleks). Upon the visual appearance of the black square, the animal abruptly lifted its head, pricked its ears, dilated its pupils, and visually fixated upon the novel object. Because the orienting reflex generates a localized focus of cortical excitation that can induce momentary external inhibition over other cerebral processes, Phase Two pairing could only proceed effectively once this orienting response had naturally habituated through repeated non-reinforced exposures.

5.3 Phase Three: Verification and Behavioral Manifestation of the CS2 Reflex

The definitive empirical test arrived in Phase Three: the isolated presentation of the novel secondary stimulus (CS2) completely detached from both the original primary stimulus (CS1) and the unconditioned food reinforcement (UCS). The black square was illuminated within the silent, isolated chamber. The experimenters, monitoring the exteriorized salivary fistula through the glass viewing port, recorded the biological readout. Within seconds of the visual onset of the black square, fluid actively moved through the graduated manometer: the parotid and submaxillary glands were actively secreting saliva.

The quantitative profile of this newly minted second-order conditioned response exhibited fascinating structural divergences from the primary reflex. While the first-order reflex (CS1) generated immediate, copious secretion (e.g., 45 drops within 30 seconds, with an onset latency of 1 to 2 seconds), the second-order reflex (CS2) elicited a more modest, yet undeniably significant output (typically 5 to 15 drops within the same temporal window), accompanied by a distinctly prolonged onset latency (often 4 to 8 seconds). The salivary flow was physiologically genuine, characterized by high-viscosity, enzyme-bearing droplets typical of anticipatory alimentary activation.

Beyond the autonomic glandular output, Pavlov and his colleagues documented non-salivary somatic conditioned behaviors. Upon the isolated activation of the black square, the animal exhibited postural adjustments, moved its snout toward the food receptacle, licked its lips repeatedly, and engaged in localized masticatory movements. Crucially, empirical controls verified that this salivary production was not an unconditioned artifact of the visual stimulus itself. Prior to Phase Two pairing, the black square had been presented to the naive animal multiple times, yielding zero drops of saliva. The capacity of this visual pattern to unlock the animal’s autonomic machinery was exclusively the consequence of its associative coupling with the acoustic metronome, proving that associative excitation had successfully migrated across disparate sensory modalities in the cerebral cortex.

6. Neurophysiological Hypotheses: Cortical Dynamics and Excitation Waves

6.1 The Theory of Cortical Mosaic and Topographical Representation

To explain the neurobiological mechanics underpinning higher-order conditioning, Pavlov formulated a comprehensive theoretical model of the central nervous system: the theory of the cortical mosaic (korkovaya mozaika). In Pavlov’s architecture, the neocortex of the canine was conceptualized as a vast, continuous sensory receptive surface comprising specialized cortical “analyzers” (analizatory). Each analyzer consisted of a peripheral receptor organ (e.g., the retina, the cochlea, the cutaneous mechanoreceptors), centripetal nerve conduits, and their corresponding cortical projection zones. The cerebral mantle was thus envisioned as a dynamic, shifting mosaic composed of millions of discrete functional points, each capable of alternating between states of excitation (vozbuzhdenie) and inhibition (tormozhenie).

Every specific environmental stimulus projected a precise spatial and temporal pattern of excitation upon this cortical mosaic. When a sound of a specific frequency was perceived, it activated a localized cluster of neurons within the acoustic analyzer of the temporal cortex; when a geometric shape was viewed, it cast an isomorphic focus of excitation across the visual analyzer in the occipital cortex. The unconditioned food stimulus, operating through subcortical taste buds and cranial nerves, activated a powerful, phylogenetically hardwired alimentary representation within the subcortical and cortical gustatory centers.

According to Pavlov, classical conditioning was the physical establishment of a temporary connection—a real, electrophysiological and structural bridge (vremennaya svyaz)—uniting these distinct cortical foci. Pavlov postulated that a zone of intense, persistent excitation acts as an electrophysiological attractor, drawing toward itself weaker waves of nervous excitation occurring elsewhere in the cortex. Thus, the intense excitation produced by food in the alimentary center established an open neural pathway that funneled energy from the activated acoustic analyzer, physically linking the metronome to the mechanism of salivary discharge.

6.2 Irradiation and Concentration of Nervous Processes

Pavlov maintained that nervous activity within the cortical mosaic is governed by two fundamental, opposing physiological laws: the law of irradiation (irradiatsiya) and the law of concentration (kontsentratsiya). When a sensory analyzer receives stimulation, the localized focus of excitation does not remain static; instead, it immediately spreads outward across the surrounding cortical tissue like a wave across the surface of water—a process designated as irradiation. Following this initial expansive phase, the wave of excitation recedes, focusing and condensing back into its specific anatomical locus—a process termed concentration.

In the context of second-order conditioning, these cortical wave dynamics operate with immense kinetic complexity. When the second-order visual stimulus (CS2) is activated, a wave of excitation erupts within the visual cortex and irradiates across the neocortex. Because the first-order acoustic stimulus (CS1) is presented in tight temporal proximity, its corresponding focus in the acoustic cortex is simultaneously transitioning through its own cycle of excitation and concentration. The irradiated excitation originating from the CS2 visual focus collides with, and is gravitationally captured by, the active focus of the CS1 acoustic center.

This interaction is further sculpted by the wave dynamics of cortical induction (induktsiya). According to the principle of reciprocal induction (a concept adapted from Charles Sherrington’s work on spinal reflexes), an active focus of excitation automatically generates a surrounding zone of negative induction—a halo of deep inhibition that suppresses extraneous neural firing in adjacent cortical sectors. In higher-order conditioning, this delicate interplay of positive and negative induction prevents the excitation from degenerating into generalized seizure-like activity, channeling the electrical and neurochemical cascade along defined cortical tracks between the visual analyzer, the acoustic analyzer, and the subcortical autonomic nuclei.

6.3 The Neural Mechanism of Secondary Trace Activation

Pavlov synthesized these concepts to postulate the exact neural wiring diagram supporting higher-order associative transfer. He maintained that second-order conditioning created a closed, intracortical functional loop directly linking the secondary sensory analyzer to the primary sensory analyzer. In the case of visual-auditory-alimentary conditioning, the primary training (Phase One) had already established a functional neural conduit running from the acoustic analyzer directly into the unconditioned subcortical-cortical salivary center. When Phase Two successfully forged a secondary link between the visual analyzer and the acoustic analyzer, the structural chain was complete:

Visual Analyzer (CS2) → Acoustic Analyzer (CS1) → Subcortical Salivary Nuclei → Parotid Glands

Under this physiological architecture, the activation of the visual center (CS2) generates an efferent discharge that travels horizontally across the neocortical neuropil to re-ignite the trace excitation of the acoustic analyzer (CS1), which subsequently routes the command down the established efferent trunk to trigger the secretion of saliva. Pavlov fiercely defended this intracortical model against reductionist claims that the reflex was mediated merely through peripheral sensory receptors or purely subcortical loops.

Modern neurobiology and neuroanatomy have substantially reinterpreted, refined, and validated the core intuition of Pavlov’s hypotheses. While Pavlov lacked the microelectrode recording techniques, patch-clamp electrophysiology, and neuroimaging modalities required to verify cellular configurations, his macroscopic predictions were remarkably prescient. Contemporary neuroscience has confirmed that higher-order conditioning is heavily mediated by complex corticocortical and cortico-subcortical circuits involving high-level polymodal association cortices, the thalamic reticular nucleus (acting as an attentional gating mechanism), the basolateral amygdala complex, and the ventral striatum. What Pavlov visualized as “waves of irradiation across the mosaic” is now recognized as dynamic, synchronized oscillations of neuronal ensembles operating across theta and gamma bands, establishing long-term potentiation (LTP) across distributed cortical and subcortical networks.

7. The Paradox of Extinction and Conditioned Inhibition in Higher-Order Trials

7.1 The Threat of Simultaneous First-Order Extinction

The empirical pursuit of higher-order conditioning thrust Pavlov’s laboratory into direct confrontation with a formidable neurophysiological paradox: the inescapable threat of simultaneous first-order extinction. To establish a second-order reflex, the experimenter is fundamentally obligated to present the first-order conditioned stimulus (CS1) paired with the second-order stimulus (CS2) without the primary unconditioned stimulus (UCS). If meat powder were to be introduced during Phase Two trials, the protocol would instantly collapse into concurrent first-order conditioning, wherein CS2 would become directly and independently conditioned to the food, completely obliterating the higher-order nature of the experiment.

However, the absolute omission of the UCS activates an immutable physiological law that Pavlov had already codified: the law of experimental extinction. Every time CS1 is presented in the absence of primary reinforcement, it sustains an incremental loss of associative strength, accumulating internal cortical inhibition. Consequently, the experimenter is trapped in a treacherous physiological race against time:

• The pairing trials (CS2-CS1) are essential to build associative strength between the secondary and primary cortical analyzers.
• Yet every non-reinforced pairing trial simultaneously erodes the excitatory potency of CS1, pushing it closer toward total experimental extinction.
• If CS1 extinguishes before CS2 has consolidated its associative link, the entire hierarchical cascade collapses, and both stimuli become functionally inert.

To overcome this operational crisis, Pavlov’s researchers devised specialized reinforcement and maintenance schedules. They determined that pairing trials could not be administered in dense, unbroken blocks. Instead, the experimental protocol required the strict, judicious interleaving of “booster” or maintenance trials. For every one or two unreinforced CS2-CS1 pairings, the experimenter was required to reintroduce isolated, fully reinforced CS1-UCS trials. This schedule periodically re-established the asymptotic associative strength of CS1, clearing away accumulated internal inhibition and preserving its capacity to serve as an effective proxy for primary reinforcement.

7.2 Conditioned Inhibition (CI) as a Competing Neural Process

An even more insidious physiological challenge discovered during higher-order trials was the sudden emergence of Conditioned Inhibition (uslovnoe tormozhenie). Under certain precise temporal configurations, the presentation of a novel stimulus alongside an established CS1 without primary reinforcement does not transform the novel stimulus into an excitatory second-order trigger; instead, it transforms it into an active, inhibitory brake that suppresses the conditioned response. The central nervous system, rather than interpreting CS2 as a harbinger of CS1 and its downstream consequences, interprets CS2 as a definitive signal that food will not arrive.

Pavlov conducted extensive empirical trials to isolate the procedural variables that dictate whether a protocol yields second-order excitation or conditioned inhibition. He discovered that the temporal relationship between CS2 and CS1 is the critical structural determinant:

• Forward Sequential Pairing (CS2 precedes CS1): When CS2 is presented first, terminates, or shortly overlaps before CS1 is initiated, the nervous system preferentially interprets CS2 as an early predictive warning signal for CS1, reliably establishing Second-Order Conditioning.
• Simultaneous or Overlapping Pairing (CS2 and CS1 presented together): When CS2 and CS1 are introduced simultaneously, or when CS2 is presented concurrently throughout the entire duration of CS1 without clear temporal antecedence, the compound stimulus signals an anomalous, non-reinforced state. The novel stimulus is branded by the cortex as an inhibitor, producing Conditioned Inhibition.

To definitively prove that an ambiguous stimulus had acquired conditioned inhibitory properties rather than second-order excitatory properties, Pavlov developed two empirical validation tests that remain foundational to modern behavioral neuroscience: the *summation test* and the *retardation-of-acquisition test*. In the summation test, the putative inhibitory stimulus is presented alongside an entirely different, highly reinforced first-order conditioned stimulus (e.g., a tactile scratcher); if it actively suppresses the salivary secretion normally evoked by that independent stimulus, its inhibitory identity is mathematically proven. In the retardation test, the experimenter attempts to convert the stimulus into a standard excitatory CS by directly pairing it with meat powder; if acquisition of the CR is significantly delayed compared to a naive neutral stimulus, the persistence of active internal inhibition is empirically confirmed.

7.3 The Delicate Operational Window for Second-Order Acquisition

These dual threats—extinction and conditioned inhibition—mean that the physiological operational window for successfully establishing and observing second-order conditioning is extraordinarily narrow and fragile. In canine subjects, the second-order response curve does not follow a stable, indefinitely sustaining plateau. Rather, it manifests as a fleeting inverted U-shaped function across trials.

During the first few Phase Two pairings (typically trials 1 through 5), the conditioned salivary output to CS2 steadily rises, achieving its peak excitatory expression. However, if the experimenter attempts to push the protocol further without introducing primary booster reinforcements for CS1, performance immediately deteriorates. By trials 8 through 15, the unreinforced presentations overwhelm the fragile excitatory link, internal inhibition cascades through the cortical network, and the salivary output drops abruptly to zero. The second-order reflex dissolves, frequently converting into a conditioned inhibitor.

Moreover, Pavlov documented the extreme vulnerability of second-order reflexes to external auditory, visual, and internal physiological distractions. While a robust first-order conditional reflex can easily withstand minor laboratory perturbations, a second-order reflex is instantly abolished by the faintest extraneous noise, minor ambient temperature fluctuations, or mild internal visceral tensions (such as a full bladder). This profound fragility highlighted the fundamental physiological truth that higher-order associations represent the most energetically demanding, functionally delicate, and plastic evolutionary achievements of the higher nervous system.

8. Limits of the Paradigm: Third-Order Conditioning and Species Constraints

8.1 Pavlov’s Attempts at Third-Order Conditioning in Canines

Driven by the conceptual ambition to discover how far these associative hierarchies could extend throughout the higher nervous system, Pavlov attempted to establish third-order conditioning in his canine subjects. The methodological protocol was logically rigorous: having successfully stabilized a second-order conditional reflex (CS2 → CS1), the researchers introduced a third, entirely novel neutral stimulus (CS3)—such as a distinct olfactory scent or a unique tactile vibration on a novel skin locus. In Phase Three trials, CS3 was forwardly paired with the validated CS2, completely omitting both the primary food reinforcement (UCS) and the first-order conditional stimulus (CS1).

The experimental outcome was an unequivocal, systematic failure. Across hundreds of meticulously calibrated attempts spanning multiple canine subjects, Pavlov was entirely unable to establish a stable, statistically significant third-order conditional reflex in the canine alimentary system. The presentation of CS3, despite exhaustive pairing with CS2, failed to reliably elicit parotid or submaxillary salivary secretion.

Pavlov formulated deep neurophysiological hypotheses to account for this biological barrier. He argued that each subsequent tier of higher-order conditioning imposes an exponential increase in internal cortical inhibition. By the time a third-order stimulus is introduced, the cortical circuits are deprived of any proximal relationship with primary biological reinforcement. The prolonged absence of biological reinforcement induces a protective state termed transmarginal inhibition (predel’noe tormozhenie)—a protective biological shutdown wherein cortical cells, threatened by energetic exhaustion and non-functional excitation, actively plunge their local networks into deep functional dormancy. Pavlov concluded that in the canine alimentary reflex system, a hard physiological ceiling existed: the higher nervous activity of dogs was fundamentally incapable of sustaining associative chains beyond the second order.

8.2 Cross-Species Variations: Third-Order Conditioning Beyond Canines

Subsequent comparative psychologists and neurobiologists quickly sought to determine whether this ceiling was a universal law of biological learning or merely an artifact of canine physiology and alimentary response systems. The empirical record revealed that the hierarchical depth of higher-order conditioning varies profoundly across phylogenetic taxa, heavily dictated by the organism’s evolutionary encephalization quotient, the complexity of its neocortical architecture, and the functional demands of its ecological niche.

In human subjects, higher-order conditioning extends far beyond the biological ceiling identified in canines. Employing motor paradigms (such as finger withdrawal or eyeblink conditioning) and sophisticated verbal and semantic conditioning protocols, researchers routinely establish robust third-order, fourth-order, and quaternary associative hierarchies. In humans, the semantic and linguistic systems act as powerful cortical amplifiers, utilizing symbolic representations to bridge temporal and biological chasms that cause purely visceral autonomic reflexes in non-human animals to collapse into extinction.

In standard laboratory rodents (rats and mice), researchers exploring higher-order fear conditioning uncovered remarkable associative hierarchies. In these paradigms, the conditioned response is not quantified through salivary fluid dynamics, but through somatic freezing behavior, ultrasonic vocalizations, and the potentiated startle reflex. In fear conditioning, rodents reliably display robust second-order conditioning and, under specific parameter optimizations, detectable third-order associations. Even in invertebrate models with drastically streamlined nervous systems, such as the marine mollusk Aplysia californica and the honeybee (Apis mellifera), researchers utilizing proboscis extension reflexes and gill-withdrawal mechanics have successfully demonstrated rudimentary second-order conditioning, demonstrating that the basic cellular mechanisms supporting multi-tiered learning evolved remarkably early in biological history.

8.3 Defense Mechanisms: Aversive versus Appetitive Conditioning Differences

The stark disparity between Pavlov’s failures to achieve third-order conditioning in canines and the successful demonstrations of deep associative hierarchies in other contexts is largely resolved by examining the fundamental evolutionary divergence between appetitive and aversive (defensive) conditioning paradigms. Pavlov’s higher-order research rested almost exclusively on the alimentary system, an appetitive behavioral domain. In appetitive learning, the primary unconditioned stimulus (food) represents an energetic gain; however, missing an isolated feeding opportunity rarely results in immediate mortality.

In defensive, aversive, and nociceptive conditioning, the evolutionary calculus is radically inverted. The primary unconditioned stimulus (such as a predatory strike, physical tissue damage, or an electric shock) carries the immediate, irreversible penalty of death or catastrophic bodily compromise. Consequently, evolutionary pressure forged defensive neural circuits that prioritize extreme speed of acquisition, profound resistance to extinction, and deep associative branching. In aversive paradigms, the cost of a false negative (ignoring a distal warning cue that heralds danger) is lethal, whereas the cost of a false positive (unnecessary anticipatory fear) is merely a transient expenditure of energy.

As a result, higher-order aversive conditioning exhibits vastly superior structural stability compared to appetitive reflexology. In defensive fear conditioning, a second-order stimulus paired with a shock-predictive CS1 acquires immense excitatory potency after only a minimal number of pairings. Furthermore, these second-order aversive associations display pronounced resistance to extinction, retaining their motivational power long after the primary UCS has ceased. This fundamental survival imperative explains why defensive conditioning can occasionally break through the physiological barriers that rigorously constrain appetitive autonomic secretions.

9. Rigorous Differentiation: Higher-Order Conditioning versus Sensory Preconditioning

9.1 Procedural Inversion: The Architecture of Sensory Preconditioning

Within the theoretical landscape of associative learning, higher-order conditioning is frequently confused with another complex multi-stimulus paradigm: sensory preconditioning. While both protocols produce a conditioned response to a stimulus that was never directly paired with a primary biological unconditioned stimulus, their underlying chronological procedures, cognitive requirements, and neural dynamics are radically divergent. The procedural architecture of sensory preconditioning was first systematically demonstrated by W. J. Brogden in 1939, offering a profound challenge to simplistic reflex mechanics.

The critical difference between the two paradigms lies in the temporal sequence of the associative phases:

• Higher-Order Conditioning Sequence:
  1. Phase 1: CS1 is systematically paired with the UCS until it becomes a biologically potent conditioned stimulus (CS1 → UCS).
  2. Phase 2: A novel neutral stimulus (CS2) is paired with the already meaningful CS1 in the absence of the UCS (CS2 → CS1).
  3. Phase 3: CS2 is presented alone to confirm its newly acquired associative potency (CS2 → CR).
• Sensory Preconditioning Sequence:
  1. Phase 1: Two completely neutral, biologically indifferent stimuli (NS2 and NS1) are paired together repeatedly in the absence of any primary reinforcement (NS2 → NS1). At this stage, neither stimulus possesses any motivational valence or capacity to evoke a response.
  2. Phase 2: NS1 is subsequently isolated and directly paired with a primary unconditioned stimulus (UCS), transforming it into a robust first-order conditioned stimulus (CS1 → UCS).
  3. Phase 3: The unreinforced NS2 is presented alone for the first time since Phase 1, testing whether the associative value newly acquired by CS1 cascades backward through the latent memory trace formed in Phase 1 (NS2 → CR).

9.2 Mechanistic and Cognitive Distinctions

The mechanistic and theoretical implications of this procedural inversion are profound. In higher-order conditioning, the pairing in Phase 2 occurs at a moment when CS1 is already saturated with conditioned motivational, affective, and physiological significance. CS1 acts as an acquired biological surrogate; the animal experiences an explicit emotional and physiological reaction during the CS2-CS1 pairing event. Higher-order conditioning is thus fundamentally driven by the transfer of acquired reinforcement value.

Sensory preconditioning, in stark contrast, is impossible to explain through the transfer of primary or acquired reinforcement during the initial pairing phase. During Phase 1 of sensory preconditioning, neither NS2 nor NS1 triggers any significant autonomic or emotional response; both are completely mundane sensory inputs. The animal forms an association between them silently, without any immediate biological utility, overt behavioral change, or autonomic perturbation. This represents an empirical validation of latent learning—the quiet, unreinforced construction of an internal cognitive map reflecting environmental correlations, aligning with the theoretical paradigms of Edward Tolman rather than strict, mechanical stimulus-response reflexology.

Furthermore, these two paradigms diverge dramatically in their susceptibility to post-conditioning devaluation of the unconditioned stimulus. If, after completing the conditioning sequences, the primary UCS is devalued (e.g., by pairing the food with lithium chloride to induce visceral illness, or through absolute satiety), sensory preconditioning and higher-order conditioning exhibit divergent response profiles. In sensory preconditioning, the CR to NS2 is immediately attenuated or extinguished, confirming that the association is heavily mediated by an active, dynamic representation of the UCS. In higher-order conditioning, the CR to CS2 frequently persists completely unabated despite the devaluation of the primary UCS, proving that the second-order stimulus has achieved functional, autonomous independence from the original biological anchor.

9.3 Empirical Diagnostic Matrix for Discriminating Paradigms

To provide unambiguous empirical separation between higher-order conditioning and sensory preconditioning, contemporary cognitive neuroscientists utilize a diagnostic matrix comparing functional attributes across both learning paradigms:

Functional Dimension	Higher-Order Conditioning (HOC)	Sensory Preconditioning (SPC)
Initial Pairing Valence	CS1 possesses established, acquired motivational valence.	Both NS1 and NS2 are strictly neutral and biologically indifferent.
Reinforcement Mechanism	Driven by acquired secondary reinforcement from CS1.	Latent associative encoding without reinforcement.
Effect of CS1 Extinction	Minimal or delayed effect; CS2 often retains its response capacity autonomously (S-R or autonomous affective habit).	Massive, immediate effect; extinguishing CS1 generally abolishes the behavioral response to NS2.
Effect of UCS Devaluation	CS2 response is frequently immune to post-conditioning UCS devaluation.	NS2 response is highly vulnerable and drops immediately following UCS devaluation.
Underlying Neuroanatomy	Heavily relies upon the basolateral amygdala (BLA) and dopamine prediction error signaling.	Critically dependent upon the hippocampus, perirhinal cortex, and broader parahippocampal structures.

10. Modern Cognitive and Computational Formulations of Higher-Order Learning

10.1 The Rescorla-Wagner Model and Prediction Error Adjustments

In the latter half of the twentieth century, the intuitive neurophysiological models of Pavlov were formalized into rigorous mathematical algorithms. The definitive benchmark for classical conditioning was established in 1972 by Robert Rescorla and Allan Wagner with the formulation of the Rescorla-Wagner Model. This model deconstructed associative learning into a mathematical function of prediction error, asserting that the change in associative strength ($\Delta V$) that occurs when an organism encounters a stimulus configuration is proportional to the difference between the maximum biological reinforcement supported by the outcome ($lambda$) and the current aggregate associative strength of all cues present ($\sum V$):

$\Delta V_A = \alpha_A \beta (\lambda – \sum V)$

Where $\alpha_A$ represents the salience of the conditioned stimulus, $\beta$ is the learning rate parameter determined by the unconditioned stimulus, and $(\lambda – \sum V)$ represents the mathematical prediction error. In a standard first-order conditioning trial, the primary unconditioned stimulus is present, meaning that $lambda > 0$. As the stimulus is repeatedly paired with the UCS, $\sum V$ progressively approaches $lambda$, driving the prediction error toward zero as the associative learning curve reaches its stable asymptote.

However, when the Rescorla-Wagner model is applied strictly to higher-order conditioning, it encounters a profound mathematical crisis. In Phase Two of higher-order conditioning (CS2-CS1), the primary unconditioned stimulus is entirely omitted; therefore, by mathematical definition, $lambda = 0$. If $lambda = 0$, and CS1 enters the trial with significant positive associative strength acquired in Phase One ($V_{CS1} > 0$), the prediction error calculation yields an inescapable negative value:

$lambda – sum V = 0 – (V_{CS2} + V_{CS1}) < 0$

Under these standard assumptions, the model mathematically predicts that CS2 can never acquire positive excitatory associative strength. Instead, it mandates that the trial must generate a negative increment ($Delta V < 0$), predicting t\hat CS2 must become a conditioned inhibitor. This mathematical contradiction forced modern theorists to formulate vital modifications to the standard Rescorla-Wagner architecture. To account for higher-order learning, computational models must incorporate dynamic associative assignment, wherein an established CS1 can temporarily step in to supply an internal, acquired$lambda$ parameter ($\lambda_{acquired} > 0$), allowing positive prediction error calculations to unfold until the onset of experimental extinction resets the computational baseline.

10.2 SOP and Sometimes-Opponent-Process Theories

To resolve the limitations inherent in purely algebraic prediction error models, Allan Wagner advanced the “Sometimes-Opponent-Process” (SOP) model of associative memory. SOP conceptualizes memory representations not as monolithic scalar values, but as distributed networks of informational nodes capable of residing in one of three distinct, quantifiable activation states:

• State A1 (Primary Active State): A high-energy, short-latency, transient operational state induced exclusively by the immediate, physical presentation of a sensory stimulus.
• State A2 (Secondary Active State): A lower-energy, decaying, protracted state entered either through the natural metabolic decay of activity from A1, or through indirect, associative retrieval by another conditioned cue.
• State I (Inactive State): The baseline, quiescent metabolic state wherein the memory node remains biologically silent.

The SOP framework provides an elegant, highly sophisticated resolution to the paradox of higher-order conditioning versus conditioned inhibition. Wagner demonstrated that the operational outcome of a pairing trial is governed entirely by the specific activation states of the interacting neural nodes. Excitatory associative connections are formed exclusively when a novel stimulus residing in state A1 coincides with another stimulus representation that is simultaneously occupying state A1.

In higher-order conditioning, when CS2 is initiated prior to CS1 in a forward sequential design, the activation of CS2 triggers an immediate A1 state that transitions into an A2 state just as CS1 bursts into state A1. Wagner’s model shows that if CS2 can engage the representations while maintaining specific temporal overlaps between these decaying active states, an excitatory higher-order bond is forged. Conversely, if CS2 and CS1 are introduced simultaneously, their mutual, concurrent transitions directly into state A2 without appropriate forward processing dynamics engages an inhibitory learning rule, mathematically dictating the emergence of conditioned inhibition. SOP thus successfully links the behavioral phenomena first cataloged in Pavlov’s Tower of Silence to the kinetic activation states of modern cognitive memory theory.

10.3 Modern Neuroimaging and Neurochemical Correlates

Twenty-first-century neuroscience has dismantled the historical notion that higher-order conditioning is merely an abstract psychological construct, uncovering the exact neuroanatomical substrates, cellular mechanisms, and neurochemical cascades that sustain it. Foremost among these anatomical structures is the basolateral amygdala (BLA). Contemporary lesion and optogenetic studies in mammalian models demonstrate that the BLA is strictly required for the acquisition and behavioral expression of second-order associative learning. While basic first-order conditioning can frequently survive restricted localized damage to specialized subcortical or cerebellar circuits, the multi-tiered architecture of higher-order conditioning collapses completely if the basolateral amygdala is functionally compromised.

At the neurochemical level, higher-order learning is driven by dynamic dopaminergic prediction error signaling within the ventral tegmental area (VTA) and its functional projections to the ventral striatum (nucleus accumbens) and the orbitofrontal cortex (OFC). Utilizing advanced fiber photometry and genetically encoded fluorescent biosensors, modern researchers have visually tracked dopamine releases in real time during higher-order trials. During Phase One, dopamine neurons initially fire in response to the presentation of the primary unconditioned reward (UCS). As learning progresses, this dopaminergic burst migrates backward in time, firing precisely at the onset of the first-order cue (CS1).

Crucially, during Phase Two of a higher-order conditioning protocol, when the novel CS2 is successfully paired with CS1, a second backward migration occurs: the dopaminergic burst migrates from the onset of CS1 to the onset of CS2. The basolateral amygdala, operating in direct coordination with the prefrontal cortex, actively gates this transfer, arbitrating whether the neural ensemble establishes an excitatory second-order trace or plunges the network into extinction. Using modern optogenetic dissection, scientists can now illuminate specific, light-sensitive channelrhodopsin proteins embedded within specific BLA-to-striatum pathways, selectively erasing or synthetically igniting higher-order associative memories with millisecond precision, directly confirming the multi-tiered cortical and subcortical dynamic networks that Pavlov inferred over a century ago.

11. Contemporary Behavioral Manifestations and Translational Applications

11.1 Etiology and Maintenance of Complex Human Phobias

The clinical utility of higher-order conditioning is profoundly evident within psychopathology, particularly regarding the etiology, propagation, and maintenance of complex anxiety disorders and phobias. For decades, traditional behavioral models struggled to explain why millions of clinical phobic patients suffered from debilitating, terrifying avoidance reactions toward stimuli with which they had never experienced a single traumatic, nociceptive, or physically painful event. A purely reductionist, first-order conditioning framework mandates that for a stimulus to elicit visceral panic, it must have been directly contiguous with an unconditioned, traumatic event.

Higher-order conditioning provides the missing epidemiological link, illuminating how phobic networks construct sprawling, multi-tiered architectures of terror completely detached from direct trauma. Consider a representative clinical trajectory:

• Tier 1 (First-Order Conditioning): An individual survives a horrific, violent automobile collision (UCS), an event characterized by catastrophic physical trauma, acute pain, and sympathetic autonomic overload (UCR). The physical act of driving the motor vehicle on high-speed highways becomes an intense, primary first-order conditioned stimulus (CS1), capable of eliciting profound anticipatory panic and somatic terror (CR).
• Tier 2 (Second-Order Conditioning): In an effort to avoid this intense distress, the individual ceases driving entirely. However, secondary cues routinely associated with the automobile—such as the mechanical jingle of car keys, the distinctive smell of gasoline, the acoustic roar of distant traffic, or even viewing a travel agency advertisement—are repeatedly paired with the cognitive and emotional representation of driving (CS1). Through second-order conditioning, these entirely benign secondary stimuli (CS2) acquire the autonomous capacity to trigger severe anxiety, visceral gastrointestinal distress, and elevated heart rates, completely absent any physical vehicle.
• Tier 3 (Third-Order Semantic Conditioning): In highly verbal human subjects, the associative chain propagates into abstract, symbolic tiers. Merely reading words such as “commute,” “transit,” or “collision” on a printed page (CS3) can evoke autonomic panic, generating a profound, multi-layered phobic condition that paralyzes the patient’s occupational and social functioning.

This multi-tiered architecture presents profound challenges for standard clinical interventions. If a psychotherapist utilizes systematic desensitization, virtual reality exposure, or in vivo exposure therapy targeting exclusively the primary trigger (CS1), the intervention may fail to eradicate the broader clinical syndrome. If the myriad of secondary and tertiary cues (CS2s and CS3s) are left untreated, they retain their autonomous capacity to trigger profound distress. Furthermore, through the mechanisms of spontaneous recovery and reinstatement, these intact higher-order associative nodes can actively reignite the extinguished lower-tier connections, precipitating a full clinical relapse.

11.2 Evaluative Conditioning and Modern Consumer Psychology

Beyond the domain of clinical psychopathology, higher-order conditioning operates as the primary cognitive and behavioral engine driving modern advertising, commercial brand design, and evaluative consumer psychology. Evaluative conditioning refers to a variant of classical conditioning wherein the perceived valence—the subjective positive or negative emotional evaluation—of an environmental stimulus shifts as a direct consequence of its pairing with an emotionally charged antecedent. Modern corporate marketing rarely relies upon crude, first-order conditioning linking a commercial product directly to primary biological unconditioned rewards (such as the immediate physiological relief of hunger or thirst).

Instead, commercial brand architectures are engineered as sophisticated, highly calculated systems of second-order and third-order evaluative conditioning:

• The Primary Associative Anchor: A marketing enterprise selects an established cultural icon, an internationally adored celebrity, or an evocative aesthetic movement that already commands intense, positive affective valence across the target demographic. This cultural entity operates as a powerful first-order conditioned stimulus (CS1), firmly tethered to feelings of status, sexual attraction, physical vitality, or cultural belonging.
• The Second-Order Associative Pairing: The commercial brand logo, a novel geometric visual pattern, or an otherwise meaningless brand name (CS2) is systematically and repeatedly paired with this cultural icon (CS1) through high-production media campaigns. The primary reinforcement (the actual physiological reality of biological pleasure or achievement) is completely absent; the corporate logo feeds exclusively upon the acquired emotional valence of the celebrity or cultural trope.
• Autonomous Secondary Valuation: Once the second-order association has consolidated, the corporate logo (CS2) autonomously evokes positive affective shifts, elevated subjective evaluation, and autonomic approach behavior in the consumer. The consumer experiences a measurable surge in positive valuation when viewing the product on a supermarket shelf, completely divorced from any analytical or rational assessment of its physical utility, chemical composition, or economic merit.

Empirical consumer studies demonstrate that these second-order evaluative associations are remarkably resistant to cognitive counter-evidence. Even when consumers are explicitly presented with rational, empirical documentation demonstrating that a lower-priced, generic competitor possesses superior physical quality and performance characteristics, the second-order emotional valence anchored to the branded CS2 consistently overrides analytical decision-making, compelling purchase behavior through automated, conditioned affective habits.

11.3 Substance Use Disorders, Environmental Cue Triggers, and Relapse Dynamics

Perhaps the most devastating translational manifestation of higher-order conditioning is observed in the neurobiology of substance use disorders, environmental cue reactivity, and chronic addiction relapse dynamics. Chemical substances of abuse—such as opioids, psychostimulants, alcohol, and nicotine—operate as intensely powerful, pharmacologically unconditioned stimuli (UCS), driving massive, supra-physiological surges of dopamine within the mesolimbic reward system and activating profound, hardwired physiological adjustments.

Through initial first-order conditioning, the physical paraphernalia immediately contiguous with drug consumption—the hypodermic syringe, the glass pipe, the rolling papers, or the distinctive smell of combusted plant matter—rapidly transforms into potent first-order conditioned stimuli (CS1). These cues acquire the neurochemical capacity to elicit intense psychological craving, anticipatory physiological compensation, and autonomic withdrawal symptoms. However, the etiology of addiction relapse extends far deeper into the environment through the insidious mechanics of second-order associative networks:

Pharmacological Drug Infusion (UCS) → Drug Paraphernalia (CS1) → Ambient Environmental Cues (CS2)

The secondary tier incorporates ambient, seemingly benign environmental contexts that routinely precede or accompany the preparation of the drug paraphernalia. These second-order stimuli (CS2) include specific geographical locations (a particular street corner, an alleyway, or an apartment building), distinct social networks (former associates or specific voices), precise temporal windows (dusk, midnight, or specific days of the week), atmospheric lighting conditions, or unique ambient acoustic backgrounds. Through sustained second-order pairing with the paraphernalia (CS1), these complex environmental contexts acquire the autonomous power to trigger severe mesolimbic dopamine surges, autonomic arousal, and overwhelming compulsive drug cravings, completely absent the physical presence of the drug or the paraphernalia itself.

This multi-tiered dynamic explains the tragic phenomenon of clinical relapse occurring months or years after an individual has successfully navigated primary detoxification and withdrawal within an isolated rehabilitation facility. While residing within the clinical center, the patient successfully extinguishes their responsiveness to direct, first-order cues. However, the moment the individual is discharged and steps back into their former geographical, social, and acoustic environment, they are instantly inundated with a barrage of intact, non-extinguished second-order conditioned stimuli (CS2s). These secondary cues instantly reactivate downstream craving networks, triggering profound physiological distress that frequently precipitates an immediate return to drug-seeking behavior.

To address this clinical vulnerability, contemporary addiction medicine has developed advanced Cue Exposure Therapy (CET) protocols. CET incorporates multi-tiered extinction hierarchies, utilizing immersive virtual reality environments to systematically expose recovering individuals to multi-layered arrays of second-order and third-order environmental contexts. By forcing the central nervous system to confront these complex higher-order associations repeatedly in the absence of chemical reinforcement, therapists facilitate the systematic accumulation of internal cortical inhibition across every tier of the associative network, substantially reducing the probability of environmental relapse.

12. Epistemological Assessment, Methodological Critique, and Pavlov’s Legacy

12.1 Methodological Limitations and Replication Challenges in Pavlov’s Original Work

Viewed through the rigorous epistemological and statistical standards of contemporary psychological science, the monumental experimental output of Pavlov’s Saint Petersburg laboratory reveals several methodological vulnerabilities and historical limitations. Foremost among these was the total reliance upon single-subject or small-sample experimental designs, paired with the complete absence of inferential aggregate statistics. Pavlov and his contemporaries did not utilize randomized controlled trials, factorial analyses of variance, or double-blind protocols. Instead, experimental data were typically communicated through exhaustive chronological narrative case studies detailing the physiological readouts of isolated, highly trained, individual canine subjects.

This reliance on illustrative single-subject paradigms created profound challenges regarding cross-subject generalizability and experimental replication. Canines within the Institute were often categorized into specific, qualitative “temperamental typologies”—classifications adapted from the ancient Galenic temperaments (sanguine, choleric, phlegmatic, and melancholic)—based upon their perceived balance of cortical excitation and inhibition. When an individual animal failed to demonstrate successful second-order conditioning, or when an attempted higher-order sequence collapsed prematurely into experimental extinction or conditioned inhibition, the failure was frequently attributed to the animal’s specific temperamental infirmity or an atypical imbalance in its higher nervous activity, rather than an underlying limitation or invalidation of the core theoretical model.

Furthermore, early twentieth-century physiological instrumentation, despite its astonishing ingenuity, was plagued by physical limitations. While the “Tower of Silence” was heralded as an impenetrable fortress of sensory isolation, modern acoustic and structural engineers note that complete isolation was an absolute physical impossibility given the materials of the era. Low-frequency ground vibrations, structural micro-resonances, barometric fluctuations, and subtle human scent trails inevitably penetrated the testing chambers. Western functionalist and behaviorist critics, such as John B. Watson and later American operationalists, frequently disputed Pavlov’s sweeping, literal topographical assertions regarding the cerebral cortex, arguing that his concepts of “cortical mosaics,” “irradiation waves,” and “concentration foci” were speculative, unobservable mechanical metaphors masquerading as concrete neuroanatomy.

12.2 Ethical Paradigms in Historical Experimental Physiology

No comprehensive academic assessment of Pavlov’s higher-order conditioning experiments can bypass a critical evaluation of the historical ethical paradigms that governed early twentieth-century Russian physiology. The surgical exteriorization of parotid and submaxillary ducts, the creation of chronic gastric and intestinal fistulas, and the protracted immobilization of canine subjects within testing harnesses for hours at a time represent interventions that stand in stark, irreconcilable conflict with modern institutional animal care and use standards.

Canine subjects within the Institute of Experimental Medicine endured permanent, disfiguring surgical modifications. While Pavlov maintained exceptional standards of sterile surgical asepsis for his era—correctly recognizing that systemic infection, chronic pain, and post-operative inflammation utterly corrupted physiological baselines—the animals nevertheless lived within confined institutional kennels, subjected to endless cycles of physical restraint, food deprivation, and invasive physiological monitoring. In certain extreme non-alimentary protocols exploring traumatic neurosis, cortical pathology, and defensive conditioning, animals were subjected to distressing nociceptive shocks, high-pressure water inundation, and deliberately induced nervous breakdowns (experimental neuroses).

Pavlov himself was not blind to this moral burden. He repeatedly documented his profound personal ambivalence and psychological distress regarding the physical sacrifice demanded of his experimental animals. In his published lectures and personal correspondence, he passionately defended the absolute necessity of animal experimentation, framing it as a tragic, yet irreplaceable moral trade-off: the immediate physical sacrifice of a finite cohort of laboratory animals was, in his view, the only viable path to illuminate the immutable natural laws governing the mammalian nervous system, thereby alleviating human neurological suffering, curing psychiatric disease, and advancing the rational enlightenment of civilization. The modern transition to non-invasive functional magnetic resonance imaging (fMRI), magnetoencephalography (MEG), and human psychophysiological testing stands as the ethical fulfillment of Pavlov’s original vision, enabling the continuous exploration of higher-order cortical dynamics without invasive surgical intervention.

12.3 Enduring Influence on Behavioral Science and Theoretical Neurobiology

Despite inevitable methodological critique and historical ethical distance, Ivan Petrovich Pavlov’s discovery and theoretical formalization of higher-order conditioning stands as one of the most enduring, transformative achievements in the history of science. His work served as the absolute intellectual bedrock upon which twentieth-century behavioral science was constructed. In the United States, John B. Watson seized upon Pavlovian reflexology as the foundational scaffolding for classical Behaviorism, utilizing the conditioned reflex to eradicate Cartesian mentalism from mainstream academic psychology. Later, B. F. Skinner, while diverging into the domain of operant conditioning, recognized Pavlov’s profound debt, categorizing classical conditioning as respondent learning and integrating its principles into the broader science of behavioral analysis.

In theoretical neurobiology and contemporary cognitive science, Pavlov’s higher-order conditioning paradigm anticipated the foundational principles of modern connectionist networks, artificial neural networks, and predictive processing models by more than half a century. Contemporary computational models of the brain—such as Karl Friston’s Free Energy Principle and hierarchical predictive coding architectures—mirror the core architectural intuition that Pavlov laid bare in Saint Petersburg: the brain is an anticipatory biological inference engine whose primary functional imperative is the continuous minimization of prediction error through the multi-tiered construction of internal models mapping external environmental causality.

Higher-order conditioning conclusively proved that learning is not a passive, singular, reactive twitch, but a dynamic, self-assembling, multi-tiered hierarchy. By showing how a simple salivary gland could be marshaled to respond to complex, distal environmental patterns completely divorced from immediate biological sustenance, Pavlov expanded our understanding of physiological adaptability. He established, with permanent empirical finality, that the complex, ethereal phenomena of associative memory, emotional transfer, and behavioral foresight are firmly rooted in the physical, quantifiable, and beautiful mechanics of the living brain.

---

The legacy of Ivan Pavlov’s higher-order conditioning experiment transcends the boundaries of classical digestive physiology, standing as a monumental milestone in humanity’s quest to comprehend the physical basis of mind and behavior. By subjecting the dynamic, multi-layered complexities of associative learning to the rigorous precision of materialist science, Pavlov uncovered an operational architecture that unites the lowest autonomic reflex arc with the highest reaches of abstract cognitive processing. His work demonstrated that the central nervous system does not merely endure environmental contingencies—it actively, systematically deconstructs them, forging multi-tiered predictive networks that permit the living organism to anticipate the future, navigate an ever-shifting world, and maintain the delicate, sacred equilibrium of life.

References

• Brogden, W. J. (1939). Sensory pre-conditioning. Journal of Experimental Psychology, 25(4), 323–332. https://psycnet.apa.org/record/1939-04149-001
• Friston, K. (2010). The free-energy principle: a unified brain theory?. Nature Reviews Neuroscience, 11(2), 127–138. https://doi.org/10.1038/nrn2787
• Gewirtz, J. C., & Davis, M. (2000). Using conditioned inhibition to investigate the nature of higher-order conditioning: A role for the basolateral amygdala. The Journal of Neuroscience, 20(11), 4331–4339. https://doi.org/10.1523/JNEUROSCI.20-11-04331.2000
• Holland, P. C. (1981). Third-order conditioning with appetitive unconditioned stimuli. Animal Learning & Behavior, 9(2), 209–216. https://doi.org/10.3758/BF03212044
• Pavlov, I. P. (1906). The scientific investigation of the psychical faculties or processes in the higher animals. Science, 24(620), 613–619. https://doi.org/10.1126/science.24.620.613
• Pavlov, I. P. (1927). Conditioned Reflexes: An Investigation of the Physiological Activity of the Cerebral Cortex (G. V. Anrep, Trans.). Oxford University Press. https://psychclassics.yorku.ca/Pavlov/
• Rescorla, R. A. (1980). Simultaneous and Successive Secondary Conditioning. Animal Learning & Behavior, 8(4), 567–576. https://doi.org/10.3758/BF03199648
• 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. https://www.sciencedirect.com/science/article/pii/B9780121774509500078
• Sechenov, I. M. (1965). Reflexes of the Brain (S. Belsky, Trans.). The M.I.T. Press. (Original work published 1863). https://archive.org/details/reflexesofbrain00sech
• Todes, D. P. (2014). Ivan Pavlov: A Russian Life in Science. Oxford University Press. https://global.oup.com/academic/product/ivan-pavlov-9780199925193
• Wagner, A. R. (1981). SOP: A model of automatic memory processing in animal behavior. In N. E. Spear & R. R. Miller (Eds.), Information Processing in Animals: Memory Mechanisms (pp. 5–47). Lawrence Erlbaum Associates. https://psycnet.apa.org/record/1982-12792-001
• Windholz, G. (1990). Pavlov and the concept of higher-order conditioning. The Psychological Record, 40(2), 273–282. https://doi.org/10.1007/BF03395932