Outcome predictability biases learning.

Area of Science:

• Cognitive Psychology
• Associative Learning
• Human Behavior

Background:

• Contemporary associative learning research primarily examines cue history's impact on learning.
• The influence of outcome history on human associative learning remains under-explored.
• Existing paradigms often overlook the prior predictability of outcomes.

Purpose of the Study:

• To investigate how an outcome's prior predictability affects subsequent human associative learning.
• To extend the 'learned irrelevance' paradigm to human learning scenarios involving outcome history.
• To determine if prior outcome predictability biases learning of cue-outcome relationships.

Main Methods:

• Three experiments adapted the 'learned irrelevance' paradigm.
• Participants learned relationships between cues and outcomes with varying prior predictability.
• The study analyzed how prior outcome predictability influenced new cue-outcome associations.

Main Results:

• All experiments demonstrated that learning is biased by an outcome's prior predictability.
• Previously predictable outcomes were readily associated with novel predictive cues.
• Previously unpredictable outcomes were more readily associated with novel nonpredictive cues.

Conclusions:

• The associative history of outcomes significantly impacts human associative learning.
• Findings support the 'certainty matching' effect, where learning aligns with prior outcome predictability.
• Multistage learning designs must consider both cue and outcome associative histories for accurate interpretation.