The tight coupling between category and causal learning

Area of Science:

• Cognitive Psychology
• Machine Learning
• Causal Inference

Background:

• Category induction is a fundamental cognitive process.
• Causal learning involves understanding cause-and-effect relationships.
• The interplay between categorization and causal reasoning is not fully understood.

Purpose of the Study:

• To investigate how category induction interacts with causal model learning.
• To examine whether learned categories are transferred across different causal structures.
• To determine if learners prioritize category consistency over optimal causal prediction.

Main Methods:

• A two-phase learning procedure was employed.
• Participants learned causal relations presented as chains or common-cause models.
• Uncategorized exemplars were used, with category labels induced implicitly.

Main Results:

• Participants spontaneously induced categories that maximized predictability in the first phase.
• In the second phase, participants predominantly transferred existing categories, even if suboptimal.
• Category transfer occurred across both causal chain and common-cause structures.

Conclusions:

• Category induction significantly influences causal model learning.
• Learners exhibit a bias towards maintaining category consistency across related tasks.
• The findings suggest that cognitive heuristics for categorization can impact causal reasoning.