Classical Conditioning: Mechanisms

Determinants of Effective Conditioned and Unconditioned Stimuli

  • Relativity of Stimulus Identification:

    • Pavlovian conditioning defines a Conditioned Stimulus (CSCS) as a stimulus that does not initially elicit the target response but comes to do so through association with an Unconditioned Stimulus (USUS).

    • The USUS is defined as a stimulus that elicits the target response from the outset unconditionally, without any prior training.

    • Because identification relies on comparing initial responses elicited before training, stimulus roles are relative rather than absolute:

      • A stimulus can serve as a CSCS relative to one event and as a USUS relative to another.

      • Empirical Example (Franklin & Hall, 2011): Sucrose-flavored food pellets act as a CSCS when paired with lithium-induced illness in a taste-aversion procedure, causing subjects to avoid the pellets. Conversely, the same sucrose pellets serve as a USUS in a sign-tracking experiment where inserting a response lever (CSCS) precedes pellet delivery, causing rats to approach and press the lever.

  • Novelty of the Conditioned Stimulus (The Latent-Inhibition / CS-Preexposure Effect):

    • Highly familiar stimuli produce less vigorous behavioral responses than novel stimuli.

    • Experimental Procedure: Conducted in two successive phases:

      • Phase 1 (Preexposure): The prospective CSCS is repeatedly presented alone without reinforcement, rendering it familiar and inconsequential.

      • Phase 2 (Conditioning): The familiarized CSCS is paired with a USUS using standard classical conditioning protocols.

    • Behavioral Outcome: Prior preexposure significantly retards or disrupts the subsequent rate of conditioned response acquisition compared to novel controls (De la Casa, Marquez, & Lubow, 2009).

    • Functional Role and Attentional Mechanisms: Described by Lubow (1959, 2011) as an adaptive mechanism protecting organisms from information overload by attenuating processing resources dedicated to irrelevant cues (Schmajuk, 2010).

    • Neurobiological and Clinical Significance: Latent inhibition is disrupted in individuals diagnosed with schizophrenia, who exhibit an inability to suppress attention to irrelevant sensory inputs. Both latent inhibition and schizophrenia share dopaminergic neurobiological substrates, making latent inhibition a core translational assay for studying cognitive dysfunctions.

  • Novelty of the Unconditioned Stimulus (The US-Preexposure Effect):

    • Conditioning develops substantially slower when the USUS is familiar prior to conditioning trials.

    • Empirical Investigation (Franklin & Hall, 2011):

      • Preexposure Phase: Rats received 8080 presentations per day for 55 consecutive days of either bacon-flavored or chocolate-flavored food pellets into a food cup, accumulating 400400 total preexposure presentations.

      • Conditioning Phase: Rats received 2020 Pavlovian trials where a 10 s10\,\text{s} presentation of a retractable lever was paired with bacon pellets, and another lever was paired with chocolate pellets.

      • Results: Acquisition of sign-tracking lever contacts was markedly faster for the lever paired with the novel food pellet than for the lever paired with the familiar food pellet.

    • Generality: Demonstrated across appetitive conditioning, fear conditioning, conditioned taste aversion, and drug conditioning preparations (Randich & LoLordo, 1979; Hall, 2009).

Rate of lever contacts across trials showing the US-preexposure effect
  • Stimulus Intensity and Salience:

    • Higher stimulus intensity yields more rapid learning and more vigorous conditioned responding (Bevins et al., 1997; Kamin, 1965; Scavio & Gormezano, 1974).

    • Stimulus Salience: Represents the perceptual noticeability or biological significance of an event (Pearce & Hall, 1980).

    • Methods of Increasing Salience:

      • Sensory Intensity: Increasing physical volume, brightness, or concentration.

      • Biological Need States: Salt deprivation increases attentiveness to saline tastes (Krieckhaus & Wolf, 1968); sodium-deficient rats acquire stronger conditioned taste aversions to salt than nondeficient controls (Sawa, Nakajima, & Imada, 1999).

      • Ecological Naturalism: Using biologically relevant cues over arbitrary stimuli.

    • Naturalistic CS in Sexual Conditioning (Cusato & Domjan, 1998, 2012; Domjan et al., 2004):

      • Male domesticated quail were conditioned with copulatory access to a female (USUS) paired with either an arbitrary terrycloth object or a naturalistic object containing the taxidermically prepared head and neck feathers of a female quail.

      • The naturalistic CSCS accelerated learning rates, engaged a wider spectrum of sexual response components, and resisted degradation when the CS−USCS-US temporal interval was lengthened.

Arbitrary terrycloth versus naturalistic quail head CS objects
  • CS–US Relevance or Belongingness:

    • Garcia and Koelling Experiment (1966):

      • Hypothesis: Animals possess genetic predispositions associating specific sensory cues with specific visceral or physical consequences based on ecological survival demands.

      • Conditioning Phase: Rats drank from a spout containing flavored water (salty or sweet). Each lick concurrently triggered an audiovisual stimulus (a click and a flash of light), creating a compound [taste+audiovisual][\text{taste} + \text{audiovisual}] CSCS. One group received peripheral pain via brief electric foot-shock (USUS); the other group received gastric illness (USUS) via ionizing radiation or lithium chloride injection.

      • Testing Phase: Conditioned suppression of drinking was evaluated by presenting taste cues alone (flavored water without lights/clicks) or audiovisual cues alone (unflavored water with lick-activated lights/clicks).

      • Data and Findings: Shock-conditioned rats suppressed drinking significantly more to the audiovisual cue than to the taste cue (∼0.6 licks/min\sim 0.6\,\text{licks/min} for audiovisual vs. ∼3.0 licks/min\sim 3.0\,\text{licks/min} for taste). Sickness-conditioned rats exhibited the exact inverse, suppressing drinking profoundly to the taste cue (∼0.7 licks/min\sim 0.7\,\text{licks/min} for taste vs. ∼2.4 licks/min\sim 2.4\,\text{licks/min} for audiovisual).

    • Genetic and Evolutionary Preparedness:

      • Selective association occurs in rat pups at 1 day1\,\text{day} of age (Gemberling & Domjan, 1982), demonstrating that prior postnatal learning is not necessary.

      • Documented in human food aversions (Logue et al., 1981; Pelchat & Rozin, 1982).

      • Primate Fear Acquisition (Öhman & Mineka, 2001; Mineka & Öhman, 2002): Rhesus monkeys and humans acquire fear conditioning faster to evolutionarily fear-relevant stimuli (photos of snakes) than to fear-irrelevant stimuli (photos of flowers or mushrooms). This advantage occurs without conscious perception (Öhman et al., 2007). Human infants aged 8−14 months8-14\,\text{months} orient selectively and rapidly toward snake pictures over flower pictures (LoBue & DeLoache, 2010).

Licks per minute demonstrating selective association in Garcia and Koelling experiment

Learning Without an Unconditioned Stimulus

  • Higher-Order Conditioning:

    • Conditioning occurring in the absence of a primary biological USUS, operating across successive associative tiers:

      • First-Order Conditioning: A conditioned stimulus (CS1CS_1) is repeatedly paired with a primary USUS (CS1→USCS_1 \rightarrow US) until robust conditioned responding is acquired.

      • Second-Order Conditioning: A novel neutral cue (CS2CS_2) is paired directly with CS1CS_1 in the absence of the USUS (CS2→CS1CS_2 \rightarrow CS_1). Consequently, CS2CS_2 acquires the capacity to elicit the conditioned response (CRCR).

      • Third-Order Conditioning: CS2CS_2 is used to condition a third stimulus (CS3CS_3).

    • Clinical Case Illustration (Wolpe, 1990): A woman developed an incapacitating crowd phobia (CS1CS_1), likely originating from physical jostling and injury (USUS) in crowded spaces. To evade crowds, she visited movie theaters exclusively during empty daytime screenings. On one occasion, a theater suddenly filled with students (CS1CS_1). The theater environment (CS2CS_2) became associated with crowds (CS1CS_1), producing a conditioned fear of movie theaters despite never experiencing physical injury (USUS) within a theater.

    • Second-Order Conditioning vs. Conditioned Inhibition:

      • Both protocols present CS1→USCS_1 \rightarrow US paired with nonreinforced compound presentations (CS1/CS2→no USCS_1/CS_2 \rightarrow \text{no } US).

      • Number of Nonreinforced Trials: A small number of nonreinforced trials favors second-order excitatory conditioning to CS2CS_2; extensive nonreinforced presentations transition CS2CS_2 into a conditioned inhibitor (Yin, Barnet, & Miller, 1994).

      • Stimulus Timing: Sequential presentations (CS2→CS1CS_2 \rightarrow CS_1) facilitate second-order excitation; simultaneous presentation ([CS1+CS2][CS_1 + CS_2]) promotes conditioned inhibition (Stout, Escobar, & Miller, 2004).

    • Ecological and Economic Applications: Money represents a powerful higher-order CS1CS_1 backed by secondary reinforcers. In consumer advertising, an established positive stimulus (CS1CS_1) is paired with a novel commercial brand (CS2CS_2) to transfer positive affective value (Schachtman, Walker, & Fowler, 2011).

  • Sensory Preconditioning:

    • Associations established between two initially benign, non-reinforced stimuli that elicit only weak orienting reactions.

    • Protocol Sequence:

      • Phase 1: Two neutral stimuli are repeatedly paired together (CS2→CS1CS_2 \rightarrow CS_1) without any unconditioned reinforcer (e.g., pastry containing both vanilla and cinnamon flavors).

      • Phase 2: CS1CS_1 is paired with a primary USUS (CS1→USCS_1 \rightarrow US; e.g., cinnamon paired with gastric food poisoning).

      • Test Phase: Presentation of CS2CS_2 alone (vanilla) elicits a conditioned response (flavor avoidance), despite CS2CS_2 never having direct temporal contiguity with the USUS.

    • Contrast with Higher-Order Conditioning: In sensory preconditioning, the stimulus-stimulus pairing (CS2−CS1CS_2 - CS_1) precedes first-order training (CS1−USCS_1 - US), whereas in higher-order conditioning, first-order training precedes the CS2−CS1CS_2 - CS_1 pairings.

Determinants of the Conditioned Response

  • The Unconditioned Stimulus as a Determinant:

    • Conditioned responses closely mimic the morphology of the unconditioned response elicited by the paired USUS.

    • Pigeon Autoshaping Experiment (Jenkins & Moore, 1973; Klein, LaMon, & Zeigler, 1983):

      • Pigeons conditioned with grain (USUS) directed rapid, forceful pecks at the key light (CSCS) with their beaks held open upon contact, matching the consummatory motor pattern of eating.

      • Pigeons conditioned with water (USUS) pecked the key light with slower, gentler movements, keeping their beaks closed upon impact and executing rhythmic swallowing actions, matching the consummatory motor pattern of drinking.

    • Stimulus Substitution Model (Pavlov):

      • Conditioning establishes direct neural pathways connecting the CS cortical representation to the US center, turning the CS into an artificial surrogate for the US.

      • Empirical Shortcomings: The model fails when the CRCR differs drastically from the URUR. Foot-shock causes rodents to leap, run, and vocalize (URUR), whereas an auditory CSCS predicting foot-shock elicits behavioral immobility and freezing (CRCR).

  • The Conditioned Stimulus as a Determinant:

    • Live Rat CS Experiment (Timberlake & Grant, 1975):

      • Design: Food-deprived experimental rats received food pellets (USUS). The CSCS consisted of introducing a live, harness-restrained stimulus rat into the chamber on a movable motorized platform through a flap door 10 s10\,\text{s} prior to food delivery.

      • Theoretical Predictions: Stimulus substitution predicted that the experimental rat would treat the stimulus rat as a food surrogate, directing biting, gnawing, and consummatory grasping toward it.

      • Observed Response: The experimental rats displayed explicit social-affiliative responses: approaching, orienting, sniffing, grooming, and social paw contacts with the stimulus rat. Unpaired control groups showed no such behaviors.

    • Sign-Tracking Variations: When the CSCS for food is an inanimate mechanical lever, rodents exhibit biting and gnawing directed at the lever rather than social bonding (Holland, 1984; Kim et al., 1996).

Apparatus for presenting a live rat CS on a movable platform
  • The CS–US Interval as a Determinant:

    • Short temporal intervals promote active, immediate coping reactions; extended intervals promote distant preparatory actions.

    • Aversive Conditioning: Short CS−USCS-US intervals generate conditioned panic or rapid flight; long CS−USCS-US intervals produce prolonged conditioned anxiety (Waddell, Morris, & Bouton, 2006; Esmorís-Arranz, Pardo-Vázquez, & Vázquez-Garcia, 2003).

    • Quail Sexual Conditioning (Akins, 2000):

      • Male quail were conditioned with a female access USUS using either a 1 min1\,\text{min} or a 20 min20\,\text{min} CS−USCS-US interval.

      • Findings: The 1 min1\,\text{min} interval produced focal sign-tracking (subjects spent nearly 70%70\% of their time positioned adjacent to the CSCS and exhibited zero elevated locomotion). The 20 min20\,\text{min} interval yielded virtually zero focal CSCS approach (<5%< 5\%) but drove general locomotor pacing (>10 chamber crossings> 10\,\text{chamber crossings}).

CS approach and general locomotion as a function of CS-US interval in quail
  • Behavior Systems Theory:

    • Formulated by Timberlake (2001), Domjan (1997), and Rau & Fanselow (2007), positing that learning superimposes an artificial stimulus schedule onto innate, evolved behavior systems designed for fundamental ecological challenges (foraging, mating, defense).

    • Linear Sequence of Response Modes:

      1. General Search Behavior: Nondirected locomotion and spatial scanning when a resource is unlocalized.

      2. Focal Search Behavior: Spatially directed approach and manipulation once resources or danger signals are identified.

      3. Consummatory Behavior: Fixed motor patterns executing terminal actions (chewing, copulating, striking).

    • Temporal Integration Mechanism: The CS−USCS-US interval dictates the precise temporal point at which the CSCS becomes embedded within the behavioral sequence:

      • A short interval introduces the CSCS into the focal search node, generating localized tracking and approach.

      • A long interval embeds the CSCS into the general search node, generating non-directed exploration and locomotor pacing.

Sequence of sexual behavior system response modes and CS-US interval integration

S–R Versus S–S Learning

  • Theoretical Distinctions:

    • Stimulus–Response (S−RS-R) Learning: Conditioning establishes an unmediated connection between the neural representation of the CSCS and the motor circuit executing the CRCR. The status or current value of the USUS is irrelevant to CRCR elicitation once the connection is formed.

    • Stimulus–Stimulus (S−SS-S) Learning: Conditioning creates a mnemonic association between the CSCS and a central cognitive representation of the USUS. Responding to the CSCS occurs because the CSCS activates this internal USUS representation, whose current motivational status determines CRCR execution.

  • The US Devaluation Paradigm (Holland & Rescorla, 1975):

    • Phase 1: Food-deprived rats received pairings of an auditory tone (CSCS) with food pellets (USUS), establishing high rates of conditioned ambulatory activity.

    • Phase 2: The experimental group was given continuous ad libitum access to food until fully satiated, thereby devaluing the internal representation of the food USUS. The control group maintained baseline food deprivation.

    • Testing Phase: Presenting the tone CSCS without food produced a significant decrement in conditioned responding in the devalued experimental group compared to the control group.

    • Theoretical Conclusion: Because devaluing the USUS post-training impairs responding to the un-manipulated CSCS, the conditioned response is governed by an intact S−SS-S neural association rather than an immutable, direct S−RS-R reflex pathway.

Conditioned Modification of the Unconditioned Response

  • Conditioned Diminution of the UR:

    • A primary evolutionary function of Pavlovian conditioning is preparing the organism to interact with the impending USUS, altering unconditioned reflex dynamics.

    • Experimental Designs:

      • Common Testing Design: Group 1 receives paired A→USA \rightarrow US; Group 2 receives unpaired A/USA / US. Both groups are tested with A→USA \rightarrow US, measuring the unconditioned response during the USUS.

      • Common Training Design: Subjects receive paired A→USA \rightarrow US and unpaired B/USB / US. The URUR is measured following either stimulus AA or stimulus BB.

    • Human Eyeblink Conditioning (Kimble & Ost, 1961): Pairing a light CSCS with a corneal air puff USUS causes the conditioned eyelid blink to preemptively attenuate the amplitude and physical force of the unconditioned blink to the air puff.

  • Conditioned Analgesia:

    • Exposure to painful foot-shocks triggers the endogenous opioid system, producing internal analgesia.

    • Conditioned stimuli repeatedly paired with foot-shock elicit conditioned release of endogenous opioids, systematically diminishing perceived pain across training sessions (Zelikowsky & Fanselow, 2011).

  • Conditioned Drug Tolerance:

    • Tolerance develops when repeated administration of a pharmacological agent produces progressively weaker physiological effects.

    • Pavlovian Compensatory Model (Siegel, 1999, 2005, 2008):

      • Environmental context, drug paraphernalia, tastes, and administration rituals serve as the CSCS.

      • Direct physiological disruptions caused by the drug constitute the pharmacological USUS.

      • Opponent-Process Interaction: The drug triggers a primary biological disruption (aa process). In response, the body recruits an automated homeostatic compensatory response (bb process) to counteract the drug.

      • Conditioning enables the predictive administration CSCS to elicit the compensatory bb process prior to drug absorption, generating an attenuated net physiological response (A−BA - B).

    • Conditioned Overdose (Siegel et al., 2000; Box 4.1):

      • Addicted individuals establish profound tolerance within their customary drug-taking environments because strong conditioned compensatory responses counteract high drug doses.

      • If the drug is consumed in a novel geographical setting or in the absence of familiar paraphernalia, the predictive CSsCSs are missing.

      • Without anticipatory conditioned compensatory responses, the lethal chemical impact of the identical dosage is unattenuated, precipitating fatal overdoses.

Primary drug reaction, conditioned compensatory response, and net attenuated tolerance response
  • Conditioned Reproduction and Fertility:

    • Blue Gourami Fish Experiment (Hollis et al., 1997):

      • Male gouramis received 1818 trials where a 10 s10\,\text{s} light CSCS signaled visual access to a female for 5 min5\,\text{min} behind a barrier (USUS). An unpaired group received non-contingent presentations.

      • Mating Test: The barrier was permanently removed after a 10 s10\,\text{s} CSCS exposure, allowing continuous social interaction over a 6 day6\,\text{day} breeding cycle.

      • Results: Paired males exhibited drastically reduced aggressive biting (∼7 bites\sim 7\,\text{bites} vs. ∼26 bites\sim 26\,\text{bites} for unpaired), built more nests (∼200 bouts\sim 200\,\text{bouts} vs. ∼45 bouts\sim 45\,\text{bouts}), displayed higher clasp frequencies (∼6.5 clasps\sim 6.5\,\text{clasps} vs. ∼1.5 clasps\sim 1.5\,\text{clasps}), and produced over 1,000 offspring1{,}000\,\text{offspring} compared to zero offspring in the unpaired control group.

    • Quail Fertility (Domjan & Akins, 2011; Domjan, Mahometa, & Matthews, 2012): Presenting a sexual CSCS elevates female copulatory receptivity, accelerates male mounting latency, enhances ejaculatory cloacal contractions, increases transferred sperm counts, and elevates egg fertilization rates.

Behavioral and reproductive outcomes in male blue gourami paired versus unpaired with CS

The Blocking Effect

  • Empirical Demonstration (Kamin, 1968, 1969):

    • Demonstrated in rat conditioned emotional response (CER) paradigms that temporal contiguity between a CSCS and a USUS is insufficient for associative learning.

    • Experimental Protocol:

      • Phase 1: The experimental group receives training pairing stimulus AA with a USUS (A→USA \rightarrow US) until conditioned responding reaches asymptote. The control group receives presentations of stimulus AA and the USUS in an explicitly unpaired fashion (A/USA / US).

      • Phase 2: Both groups receive identical compound conditioning trials where stimulus AA and a novel stimulus BB are presented simultaneously followed by the USUS ([A+B]→US[A + B] \rightarrow US).

      • Testing Phase: Target stimulus BB is presented alone.

    • Outcome: The experimental group shows negligible conditioned responding to stimulus BB, whereas the control group displays strong responding to stimulus BB. Pretraining stimulus AA blocks the conditioning of stimulus BB.

  • Human Geometric Learning (Prados, 2011):

    • Undergraduate students navigated virtual floor plans where triangle shapes indicated target locations containing food.

    • Blocking Group: Phase 1 trained triangle AA alone (+ reinforcement+\text{ reinforcement}). Phase 2 presented triangle AA conjoined with triangle BB (+ reinforcement+\text{ reinforcement}).

    • Control Groups: Control 1 received only the compound [A+B][A + B] in Phase 2; Control 2 received only triangle BB in Phase 2.

    • Test Accuracy on Triangle B: Control 2 achieved 92%92\% correct; Control 1 achieved 79%79\% correct; the Blocking Group achieved only 20%20\% correct.

  • Educational Reading Interference (Singh & Solman, 1990; Box 4.2):

    • Children with intellectual disabilities were taught vocabulary words (e.g., knife, lemon, radio, stamp, chalk).

    • Teaching words alongside familiar pictorial depictions produced the slowest rate of word acquisition in all 88 participants.

    • Six of eight children learned written words fastest when the words were presented without pictures. Preexisting semantic mastery of the picture blocked processing of the novel printed word.

  • Theoretical Rationale (Kamin's Surprise Hypothesis):

    • A unconditioned stimulus must be surprising to drive associative learning. When a USUS is fully predicted by a previously conditioned cue (AA), it lacks surprise, eliciting minimal mental processing or error signaling to establish new associations to newly added cues (BB).

The Rescorla–Wagner Model

  • Mathematical Formalization:

    • Formulated by Robert Rescorla and Allan Wagner (1972) as an error-correction mechanism where learning on trial nn is proportional to the degree of surprise of the USUS:

ΔV=k(λ−V)\Delta V = k(\lambda - V)

*   *Variable Definitions:*
    *   ΔV\Delta V represents the change in associative strength of a conditioned stimulus on a given trial.
    *   kk is a combined scaling constant reflecting the salience of the CSCS and the intensity of the USUS (k=αβk = \alpha \beta, where 0≤k≤10 \le k \le 1).
    *   λ\lambda represents the asymptote of associative value supported by the delivered USUS (its actual magnitude).
    *   VV represents the total aggregate associative value of all stimuli present during the trial (V=∑ViV = \sum V_i).
    *   (λ−V)(\lambda - V) represents the prediction error or degree of surprise.
  • Theoretical Curve and Dynamic Predictions:

    • Early Training: Associative expectations (VV) are low, making (λ−V)(\lambda - V) large, resulting in large gains in associative strength (ΔV\Delta V).

    • Late Training: As VV approaches λ\lambda, the prediction error (λ−V)(\lambda - V) approaches zero, causing associative growth to plateau at asymptote.

  • Mathematical Account of the Blocking Effect:

    • Phase 1: Stimulus AA is reinforced alone until its associative value matches the full asymptote of the USUS: VA=λV_A = \lambda.

    • Phase 2: Stimulus BB is introduced simultaneously with AA. The total expectation is calculated by summing all present cues: V=VA+VBV = V_A + V_B.

    • At the start of Phase 2, VB=0V_B = 0, so V=λ+0=λV = \lambda + 0 = \lambda.

    • Applying the delta rule to stimulus BB yields:

ΔVB=k(λ−[VA+VB])=k(λ−λ)=k(0)=0\Delta V_B = k(\lambda - [V_A + V_B]) = k(\lambda - \lambda) = k(0) = 0

*   Because the prediction error is zero, stimulus BB gains no associative value, correctly predicting the blocking effect.
  • The Overexpectation Effect:

    • Phase 1: Cues AA and BB are conditioned separately to asymptote using a single food pellet reinforcer (λ=1\lambda = 1), establishing VA=1V_A = 1 and VB=1V_B = 1.

    • Phase 2: Cues AA and BB are presented simultaneously as a compound followed by the identical single food pellet reinforcer (λ=1\lambda = 1).

    • Calculations: The organism sums the individual expectancies: V=VA+VB=1+1=2V = V_A + V_B = 1 + 1 = 2. However, the delivered USUS is only λ=1\lambda = 1.

    • Applying the error equation yields:

(λ−V)=(1−2)=−1(\lambda - V) = (1 - 2) = -1

*   The prediction error is negative, forcing both cues to lose associative strength despite continuous pairings with the reinforcer until VA+VB=1V_A + V_B = 1.
*   This counterintuitive outcome has been confirmed across appetitive and aversive preparations (Kehoe & White, 2004; Lattal & Nakajima, 1998; Sissons & Miller, 2009).
  • Conditioned Inhibition:

    • Reinforced excitatory trials (CS+→USCS+ \rightarrow US) establish a positive associative value (VCS+>0V_{CS+} > 0).

    • Nonreinforced compound trials ([CS++CS−]→no US[CS+ + CS-] \rightarrow \text{no } US) yield zero delivered reinforcer (λ=0\lambda = 0).

    • To satisfy the condition that the combined associative strength matches the outcome (VCS++VCS−=0V_{CS+} + V_{CS-} = 0), the value of VCS−V_{CS-} must become negative.

    • Thus, conditioned inhibition is formally defined as negative associative strength.

  • Theoretical Failures of the Rescorla–Wagner Model:

    • Extinction of Conditioned Inhibition: The model predicts that repeatedly presenting an unreinforced conditioned inhibitor (CS−CS-) alone will extinguish its negative associative value toward zero (ΔV=k(0−[−V])=k(+V)\Delta V = k(0 - [-V]) = k(+V)). In laboratory tests, presenting a CS−CS- alone does not eliminate inhibition, and can even enhance it (Zimmer-Hart & Rescorla, 1974; DeVito & Fowler, 1987).

    • Extinction of Conditioned Excitation as Unlearning: The model treats extinction as reversing acquisition (V→0V \rightarrow 0). Behavioral evidence indicates that extinction represents new inhibitory learning that superimposes over the intact original excitatory association (Chapter 9).

    • Unlearning Inhibitory Properties: Empirical inhibition of a CS−CS- is eliminated by extinguishing the excitatory CS+CS+ that accompanied it during training, an outcome not predicted by the Rescorla–Wagner model (Best et al., 1985; Lysle & Fowler, 1985).

Neurobiology of Fear Conditioning and the Amygdala

  • Anatomical Circuitry of Fear Conditioning (Box 4.3):

    • Sensory Convergence in the Basolateral Amygdala (BLA):

      • Information regarding discrete conditioned stimuli (CSCS) reaches the lateral nucleus of the amygdala via two anatomical routes: a direct thalamic pathway (low latency, low-resolution) and an indirect transcortical pathway (slower latency, highly detailed feature representation).

      • Complex contextual and spatial cues reach the basolateral complex from the hippocampus via the perirhinal and entorhinal cortices.

      • Aversive USUS inputs converge onto the lateral amygdala via parallel somatic pathways projecting from the posterior thalamus, parabrachial nucleus, and locus coeruleus.

    • Synaptic Mechanisms: Plasticity within the lateral/basolateral amygdala requires the activation of NN-methyl-DD-aspartate (NMDANMDA) receptor channels. Strong depolarizing currents from the shock USUS remove magnesium blockades from the NMDANMDA channel, enabling long-term synaptic potentiation that links the CSCS input to downstream defensive networks. Infusions of NMDANMDA receptor antagonists into the BLA completely block fear conditioning.

  • Efferent Outputs of the Central Nucleus of the Amygdala (CeA):

    • Periaqueductal Gray (PAG):

      • Dorsolateral PAG (dlPAG): Triggers active defensive behaviors (jumping, flight, running, circa-strike reactions).

      • Ventrolateral PAG (vlPAG): Coordinates behavioral immobility (conditioned freezing) and opioid-mediated analgesia.

    • Autonomic and Endocrine Targets:

      • Lateral Hypothalamus to Rostral Ventrolateral Medulla: Induces tachycardia and elevated blood pressure.

      • Parabrachial Nucleus: Accelerates respiration and panting.

      • Caudal Pontine Nucleus of the Reticular Formation: Mediates fear-potentiated startle.

      • Ventral Tegmental Area and Paraventricular Hypothalamus: Drives behavioral arousal and initiates hypothalamic-pituitary-adrenal (HPAHPA) release of adrenocorticotropic hormone (ACTHACTH) and corticosteroids.

  • Neural Computation of the Prediction Error Signal (vlPAG-BLA Loop):

    • The ventrolateral periaqueductal gray (vlPAGvlPAG) integrates ascending sensory pain signals (λ\lambda) with descending projections from the central nucleus of the amygdala (CeACeA) representing anticipated danger (VV).

    • Descending inputs stimulate local endogenous opioid release within the vlPAGvlPAG, muting ascending pain signals in proportion to CSCS expectancy.

    • The resulting neural output represents an empirical error signal (ΔV=λ−V\Delta V = \lambda - V), which projects via the midline and intralaminar thalamus back to the dorsomedial prefrontal cortex (dmPFCdmPFC) and basolateral amygdala to calibrate synaptic plasticity.

    • Administering opioid antagonists (e.g., naloxone) directly into the vlPAGvlPAG blocks this negative feedback, restoring prediction error and unblocking learning to an expected USUS (McNally, Johansen, & Blair, 2011).

Neural circuit diagram and periaqueductal gray error-prediction architecture for fear conditioning

Alternative Models: Attention, Timing, and Information Theory

  • Attentional Models of Conditioning:

    • Contrast with the Rescorla–Wagner model by emphasizing fluctuations in CSCS processing (associability) rather than variations in USUS effectiveness (Mackintosh, 1975).

    • Pearce–Hall Model (1980): The attentional processing allocated to a CSCS on trial nn is proportional to how surprising the USUS was on trial n−1n-1. Once a USUS is completely predicted, attention to the CSCS drops, halting further associative learning.

    • Tripartite Attentional Taxonomy (Hogarth, Dickinson, & Duka, 2011):

      1. Looking for Action: High attention commanded by well-trained, reliable predictive cues requiring minimal cognitive effort (Mackintosh, 1975).

      2. Looking for Learning: Attentional scanning directed toward ambiguous cues whose predictive outcomes are uncertain (Pearce & Hall, 1980).

      3. Looking for Liking: Attention directed toward stimuli based on acquired emotional or hedonic valence.

    • Failure to Explain One-Trial Blocking: Attentional models assume that variations in attention occur prospectively across trials. In one-trial blocking preparations, the introduction of a previously trained cue (AA) blocks acquisition to cue BB on its very first presentation (Dickinson, Nicholas, & Mackintosh, 1983; Azorlosa & Cicala, 1986).

  • Temporal Coding Hypothesis:

    • Organisms encode the exact temporal point at which the USUS occurs relative to the onset of the CSCS, learning when events happen in addition to whether they occur (Balsam, Drew, & Gallistel, 2010; Williams et al., 2008).

    • Temporal mappings acquired in early phases integrate directly into compound conditioning, second-order associations, and blocking protocols (Savastano & Miller, 1998; Amundson & Miller, 2008).

  • The Intertrial Interval / Trial Duration Ratio (I/T Ratio):

    • Conditioned responding is governed by the relative relationship between two intervals: the intertrial interval (II, measured between successive USUS presentations) and the trial duration (TT, the duration of the CSCS).

    • Holland (2000) Parametric Rat Study:

      • Food-deprived rats were conditioned with auditory white noise paired with food delivery into a cup, using trial durations of either T=10 sT = 10\,\text{s} or T=20 sT = 20\,\text{s}.

      • Intertrial intervals varied from 15 s15\,\text{s} to 960 s960\,\text{s}, producing I/TI/T ratios ranging from 1.51.5 to 48.048.0.

      • Findings: The percentage of time spent goal-tracking during the CSCS mapped directly onto the I/TI/T ratio. Absolute trial durations of 10 s10\,\text{s} and 20 s20\,\text{s} yielded equivalent response rates as long as the relative I/TI/T ratio was held constant.

Percentage of time spent nosing the food cup as a function of the I/T ratio
  • The Relative-Waiting-Time Hypothesis:

    • Proposes that conditioned responding reflects a comparison between the duration an organism must wait for the USUS in the presence of the CSCS (TT) versus the duration it must wait during the background intertrial interval (II) (Jenkins, Barnes, & Barrera, 1981).

    • When the I/TI/T ratio is high, the CSCS signals a substantial reduction in the time to the next reinforcer, providing high informational value and driving strong responding (Gallistel & Gibbon, 2000; Balsam & Gallistel, 2009).

    • When the I/TI/T ratio is low, waiting time during the CSCS matches waiting time during the intertrial interval, rendering the CSCS uninformative and generating minimal conditioned responding.

The Comparator Hypothesis

  • Core Conceptual Principles:

    • Formulated by Ralph Miller and associates (Miller & Matzel, 1988; Stout & Miller, 2007) as a model of performance rather than a model of learning.

    • Rejects the assumption that learning failure accounts for phenomena like blocking, asserting instead that associations are acquired normally but are masked during behavioral retrieval.

    • Assumes the nervous system forms only excitatory associations with the USUS. Whether responding manifests as conditioned excitation or conditioned inhibition depends on a computational comparison between competing associative structures at the time of testing.

  • The Three Associative Links:

    1. Link 1 (Target CS–US Association): Direct excitatory association between the target conditioned stimulus (XX) and the USUS.

    2. Link 2 (Target CS–Comparator Association): Within-compound association linking the target CSCS to all other environmental stimuli present during training (discrete cues and background context).

    3. Link 3 (Comparator–US Association): Direct excitatory association connecting the comparator cues to the USUS.

Associative structure of the comparator hypothesis showing direct and indirect links to the US representation
  • The Comparator Balance:

    • When target stimulus XX is presented at test, it activates the cognitive representation of the USUS through two routes:

      • Direct Route: Activated directly via Link 1.

      • Indirect Route: Activated through the compound pathway of Link 2 and Link 3.

    • Excitation: If the direct Link 1 activation exceeds the indirect Link 2/3 pathway, the balance favors excitatory responding.

    • Inhibition: If the indirect Link 2/3 pathway exceeds the direct Link 1 activation, the balance tips in favor of conditioned inhibition.

  • Post-Conditioning Revaluation Effects:

    • Unmasking Blocked Stimuli: In a blocking experiment (A→USA \rightarrow US, then [A+B]→US[A + B] \rightarrow US), the target cue BB has a strong Link 2 association to comparator AA, and cue AA has a strong Link 3 association to the USUS. The indirect activation suppresses responding to BB.

    • Extinguishing the blocking stimulus AA alone following compound training removes its excitatory Link 3 association with the USUS.

    • At testing, target cue BB now faces minimal indirect comparator competition, unmasking robust conditioned responding to cue BB (Blaisdell, Gunther, & Miller, 1999; Boddez et al., 2011).

    • Extinction of Conditioned Inhibition: In standard conditioned inhibition (CS+→USCS+ \rightarrow US, [CS++CS−]→no US[CS+ + CS-] \rightarrow \text{no } US), the CS+CS+ serves as the comparator cue for target CS−CS-. Extinguishing the CS+CS+ eliminates the Link 3 association, removing the indirect activation that drove the comparison balance negative, which eliminates conditioned inhibition to the CS−CS- (Best et al., 1985; Lysle & Fowler, 1985).