Comprehensive Study Notes on Classical Conditioning and Predictive Learning

Classical conditioning is a learning process that involves associating a neutral stimulus with an unconditioned stimulus (US) to elicit a conditioned response (CR).

Contingency Learning: The relationship between a Conditioned Stimulus (CS) and Unconditioned Stimulus (US).
- Positive Contingency: When the CS is consistently followed by the US. This creates a strong excitatory conditioning effect.
- Negative Contingency: When the CS is associated with the absence of the US, demonstrating inhibitory conditioning, meaning the subject predicts no US will occur.
CS-: A stimulus that predicts that a US will not occur.
- Example provided: The difficulty of learning a previously significant CS- as a CS+ is due to prior learning.

CS variables: An earlier learned CS- (say “x”) takes longer to learn as a CS+ (predictive) because of its previously learned inhibitory role.
New stimuli (e.g., “y”) have no prior meaning, enabling quicker associations as they could mean anything to the subject.
When both x and y are presented together, the concern arises if they produce a new, unrelated effect—like producing an entirely new sound.

Animal drinking behavior was used to illustrate latency in drinking time in response to predictive sounds.
Longer Latency: Indicates fear or increased caution due to prediction from the stimulus (e.g., predicted shocks). Longer drinking times indicated less fear, while shorter times indicate anticipation of negative consequences.

Learning Novel Stimuli: Novel or neutral stimuli like y allow for easier associations as there are no prior associations influencing learning.
Blocking Effect: Illustrates how previously learned associations inhibit new learning due to full predictability from the established CS.
- Example: If a tone and a light predict a shock, a new introduced CS not changing the outcome (the tone’s prediction of shock) results in the introduction failing to elicit a CR.

Further Developments in Conditioning: Discussed further effects such as blocking, explaining that redundancy in prediction prevents additional learning.
Conditioning Interference: The learning of a new CS can be blocked if a perfect predictor is already established.

Contingency: The correlation between CS and US. Ideal learning involves understanding the predictability of each stimulus.
- Zero Contingency: Occurs when there’s no connection between CS and US; learned attributions become significant (e.g., shoe size and intelligence).

Higher Order Conditioning: Involves predicting predictors; seeing one CS (like a leash) predicts another CS (the act of going for a walk).
Sensory Preconditioning: Prior associations are made between neutral stimuli before any US relationship is formed, leading to unexpected CR aversions without revival of the US.
- Example: Linking peanut butter with jelly; if one becomes poisonous, both may evoke an aversion.

Experiment Description: Two groups, control and pretrained, show how prior learning blocks new learning opportunities, leading to predictable retrieval from existing associations.
- Control Group: Learns prediction newly; Experimental Group: Prior learning inhibits learning of new cues.

Response Theory: Two major theories exist regarding how learning occurs in predictive environments:
- SS Learning (Stimulus-Stimulus): When a CS predicts a US, both become linked neurologically.
- SR Learning (Stimulus-Response): Predictive actions become habitual regardless of the US’s presence; habitual automatic responses align to CS without requiring continuous association with the US.

The debate over SS vs. SR learning theories revolves around the dependency on US importance post-learning.
- Evidence Evaluation: Both SS and SR theories have supporting evidence, yet SS learning tends to provide a more comprehensive explanation of phenomena like sign tracking or auto shaping (pigeons responding to stimulus beyond just food).

Learning Methodology: The focus is shifted from mere associations to environmental recognition and predictability spanning several stimuli.
- Animals attend more to their overall environment rather than ignoring predictable predictors, indicating sensitivity to their surroundings.

Classical conditioning serves as a model for understanding behavioral predictions, teaching the importance of stimulus relationships and interactive learning in dynamic environments, where the understanding of both predictive effectiveness and redundancy leads to nuanced behavioral responses.