The development of causal categorization.

Area of Science:

• Cognitive Psychology
• Developmental Psychology
• Artificial Intelligence (AI) and Machine Learning

Background:

• Categorization is a fundamental cognitive process enabling efficient information processing.
• Understanding how causal relationships influence categorization is crucial for cognitive development and AI models.
• Previous research has explored feature-based categorization, but the role of causal structure requires further investigation.

Purpose of the Study:

• To investigate the impact of causal relations between features on categorization in children and adults.
• To differentiate the effects of causal coherence, causal status, and relational centrality on category learning.
• To evaluate the applicability of the generative model (Rehder, 2003a) to causal categorization across age groups.

Main Methods:

• Two experiments were conducted with 5- to 6-year-old children and adults.
• Participants learned artificial categories with causally related and noncausal features.
• Classification patterns and logistic regression analysis were used to assess feature influence.

Main Results:

• Adults' categorization was primarily driven by causal coherence in deterministic scenarios and by coherence plus causal status in probabilistic scenarios.
• Children's categorization relied mainly on causal coherence, irrespective of deterministic or probabilistic causal links.
• Relational centrality did not significantly affect categorization in either children or adults.

Conclusions:

• Causal coherence plays a significant role in categorization for both children and adults.
• Adults exhibit more flexible use of causal information, adapting to probabilistic relationships.
• The generative model effectively explains causal categorization processes in both developmental and adult populations.