Classical Conditioning Part 1

Ivan Pavlov (physiologist studying digestion) noticed dogs salivated before food reached the mouth; salivation became linked to arbitrary cues.
Discovery: Neutral cues can acquire the ability to trigger reflexive, biologically based responses after repeated pairings with food.

Unconditioned Stimulus (US)
- Biologically potent; elicits reflex without prior learning.
- Pavlov’s meat powder.
Unconditioned Response (UR)
- Reflex evoked by the US.
- Dog’s salivation.
Conditioned Stimulus (CS)
- Initially neutral; gains predictive value through pairing.
- Bell, tone, light, touch, odor, etc.
Conditioned Response (CR)
- Response elicited by CS after learning.
- Salivation to the bell.
Terminology logic
- “Unconditioned” = occurs in all conditions, no special procedure.
- “Conditioned” = requires special pairing conditions to evoke response.

Acquisition
- CS–US pairings strengthen CS→CR link.
- Probe trials (CS alone) reveal increasing response probability until a plateau.
Extinction
- CS presented without US repeatedly.
- CR probability declines across trials.
Spontaneous Recovery
- After a rest (e.g., one week), presenting CS alone revives CR above last-extinction level.
- Implies extinction is new, competing learning—not erasure of CS–US bond.

Species: pigeons (also rats, quail, fish).
Setup
- US = food grain; UR = eating.
- CS = illuminated key/light.
- CR = pecking the light.
Long-box video demo
1. Light (opposite end) illuminates.
2. Pigeon walks, pecks light, then returns to hopper for food.
3. Pecking is unnecessary for food delivery yet emerges reliably.
Key observations
- CR differs topographically from UR (peck vs. eat).
- CR frequency rises with trials; measurable as number of pecks.

Purpose: model fear/anxiety learning; quantify freezing.
Setup
- CS = tone; US = brief foot shock or loud noise.
- UR = startle/jump; CR = freezing (incompatible with lever pressing).
Measurement: lever-press suppression ratio $\text{CER ratio}=\frac{B}{A+B}$
- $A$ = presses in pre-CS period (tone absent).
- $B$ = presses during CS.
- $0.5$ → no conditioning; $0$ → maximal fear (complete suppression).
Example progression
- Early training: $\frac{25}{25+25}=0.5$ (no suppression).
- Mid training: $\frac{7}{25+7}\approx0.28$ (partial).
- Late training: $\frac{0}{25+0}=0$ (complete suppression).

Groups
1. I (Inhibitory): Tone (+) → shock; Light (–) → no shock.
2. R (Random): Tone (+) → shock; Light randomly paired/unpaired.
Probe: present Tone + Light without shock.
Results (CER ratio)
- Group I showed higher ratios (less suppression) → light signaled safety.
- Group R showed smaller increase → light carried weak/no safety value.
Conclusion: Animals learn cues for absence of danger.

Phase 1
- Control: Green key (+) → food; Red key (–) → none.
- Experimental: Green (+) → food; Green + white line (–) → none.
Phase 2: New CS = white line on black background (+) for both groups.
Findings
- Control pigeons learned new CS+ in <10 trials (7/8 birds pecking).
- Experimental pigeons showed >20-trial delay; only few birds pecking.
Interpretation: Prior inhibitory history retards later excitatory learning.
Generalization
- Inhibitory conditioning appears with aversive (shock) and appetitive (food) USs, across birds, rodents, and humans.

Learning depends on informational value: difference between $P(US|CS)$ and $P(US|\neg CS)$ .
- Excitatory when $P(US|CS) > P(US|\neg CS)$ .
- Inhibitory when $P(US|CS) < P(US|\neg CS)$ .
- No learning when probabilities equal (random relationship).
Bonds
- Positive (excitatory) or negative (inhibitory) associations assumed to form.

All groups: $P(US|CS)=0.4$ (tone predicts shock 40 % of the time).
Varied $P(US|\neg CS)$
- Group 0: $0$
- Group 0.1: $0.1$
- Group 0.2: $0.2$
- Group 0.4: $0.4$ (random)
Phase 2: Extinction (no shocks at all).
Outcomes
- Group 0: strong suppression, gradual extinction.
- Increasing $P(US|\neg CS)$ → weaker initial suppression & faster extinction.
- Group 0.4 showed virtually no conditioned suppression.
Supports idea that informational value governs strength of learning.

Group 0.4 still shows latent inhibition (delayed acquisition with same CS later) → some learning occurred despite “no information.”
Asymmetric effects when manipulating only CS vs. only US frequencies contradict simple information metric.
Motivated development of cognitive theories (expectancy, comparator models) covered in subsequent lecture.

Fear-conditioning paradigms inform treatments for anxiety, PTSD, phobias (e.g., exposure therapy must account for spontaneous recovery).
Drug-tolerance research leverages context-specific inhibitory/excitatory cues.
Ethical use of shock: minimized intensity/duration; alternatives (loud tones) explored.
Autoshaping illustrates how cues can drive maladaptive behavior—a parallel to human cue-induced craving or habitual actions.