Predictive and diagnostic learning within causal models: asymmetries in cue competition.

Area of Science:

• Cognitive Psychology
• Learning Theory
• Causal Inference

Background:

• Recent claims suggest higher-order learning (e.g., categorization, causal induction) can be explained by lower-order associative learning.
• These claims cite cue competition in higher-order learning, analogous to blocking in classical conditioning.

Purpose of the Study:

• To investigate the relationship between predictive learning (causes of an effect) and diagnostic learning (effects of a cause).
• To test whether these two types of reasoning are symmetrical and reducible to associative principles.

Main Methods:

• Three experiments were conducted where participants learned cue-response associations.
• Cues were defined as either potential causes of a common effect (predictive learning) or potential effects of a common cause (diagnostic learning).

Main Results:

• Results demonstrated asymmetry between predictive and diagnostic learning, contradicting associationistic models.
• Cue competition was observed in predictive learning (multiple causes competing), but not in diagnostic learning (multiple effects did not compete).

Conclusions:

• Diagnostic and predictive reasoning are not symmetrical and cannot be fully reduced to associative learning.
• Findings support causal-model theories over purely associationistic explanations for higher-order learning.