LPI Topic 6 Flashcards
Negative Patterning
A learning process where individual stimuli (A and B) predict an outcome, but their combination (AB) predicts no outcome.
Example:
- A → Reward (+)
- B → Reward (+)
- AB → No Reward (-)
Problem for Elemental Models
The Rescorla-Wagner model predicts AB should lead to an outcome, but animals and humans learn AB = no outcome.
Solution (Configural Models)
- Pearce (1987): Treats AB as a unique stimulus rather than a sum of A and B.
- Rescorla-Wagner’s Expansion: Introduced an additional “configural cue” (X) that signals AB as a separate entity.
Generalisation
Applying learned responses to new, similar stimuli.
Factors Affecting Generalisation
- Stimulus Similarity – More similar, stronger generalisation.
- Experience – Exposure refines generalisation.
- Categorisation (Humans) – Use learned rules to generalise.
Models of Generalisation
- Elemental (Rescorla-Wagner): Generalisation depends on summed associative strengths.
- Configural (Pearce, 1987): Treats new stimuli holistically.
Peak Shift
After discrimination training, the peak response shifts away from S- and is biased towards a new stimulus.
Example (Hansen, 1959 – Pigeons)
- Control group pecked at trained light (550 nm, S+).
- Experimental group shifted preference away from S- (560 nm) to 540 nm.
Peak Shift Explanation
- Relational Learning Theory: Learns “more/less than” relationships.
- Associative Learning: Overlapping elements of S+ and S- lead to peak shift.
Human Application
- Occurs with unfamiliar or artificial stimuli.
- Familiar categories prevent peak shift (Mackintosh, 1997).
Extinction
Gradual reduction of a conditioned response when the CS is presented without the US.
Key Extinction Effects
- Renewal Effect
- Spontaneous Recovery
- Reinstatement
- Rapid Reacquisition
Spontaneous Recovery
Extinguished response re-emerges after a time delay.
Reinstatement
- US presented alone after extinction restores the CR.
- Example: After therapy, a dog phobia re-emerges when encountering an aggressive dog.
Rapid Reacquisition
Re-learning occurs faster if CS-US pairing is reintroduced.
Renewal Effect
- Extinguished response returns in a different context.
- ABA Renewal: Learning in A, extinction in B, return to A = response reappears.
- ABC Renewal: Learning in A, extinction in B, tested in C = response reappears.
Blocking Effect (Kamin, 1969)
Definition
- Prior learning prevents new learning about an additional stimulus.
Example
- Phase 1: A → Shock (A becomes predictive).
- Phase 2: A + B → Shock (B is ignored).
- Test: B alone → No response (no learning about B).
Explanation
- Surprise is essential for learning.
- Blocking occurs when an existing CS fully predicts the US, preventing learning about new stimuli.
Unblocking
- Changing the outcome (e.g., two shocks instead of one) can lead to new learning.
Cue Competition and Summation
When multiple stimuli are paired with the same outcome, they compete for associative strength.
Example (AB+ Trials)
- AB+ (compound training):
- A and B share associative strength (e.g., 0.5 each).
- C+ (single training):
- C receives full associative strength (1.0).
- Learning about A and B is weaker due to competition.
Acquired Equivalence and Distinctiveness
Prior experiences affect how we group or differentiate stimuli.
Key Studies
- Honey & Hall (1989): Flavour aversion – pairing two stimuli makes them more similar.
- Mackintosh, Kaye & Bennett (1991): Pre-exposure to stimuli improves discrimination.
Implication
- Exposure increases discriminability.
- Pairing stimuli with the same outcome makes discrimination harder.
- Pairing stimuli with different outcomes improves discrimination.
Context-Dependent Learning
Learning is better remembered when tested in the same environment.
Examples
- Underwater Learning: Easier recall if tested underwater.
- State-Dependent Learning: Learning under stress is recalled better under stress.
