Epistemic trust: modeling children's reasoning about others' knowledge and intent

Area of Science:

• Developmental Psychology
• Cognitive Science
• Computational Modeling

Background:

• Children's development relies on learning from others (indirect experience or testimony).
• Testimony may not always be accurate, requiring children to assess informant trustworthiness.
• Understanding how children infer trustworthiness is key to understanding knowledge acquisition.

Purpose of the Study:

• To develop a computational model of epistemic trust in learners.
• To investigate how learners infer informant helpfulness and knowledgeability.
• To explain children's abilities in assessing trustworthiness and developmental changes.

Main Methods:

• Developed a computational model of epistemic trust.
• The model reasons about informant helpfulness and knowledgeability.
• Model performance was compared to observed child competencies.

Main Results:

• The model successfully captured children's ability to use informant accuracy to infer trust.
• It explained children's use of recent accuracy, informant consensus, and detection of mal-intent.
• Developmental shifts in trust between ages 3 and 4 were explained by changing default assumptions.

Conclusions:

• Learners, including young children, can computationally infer informant trustworthiness.
• Epistemic trust development involves evaluating accuracy, consensus, and informant intent.
• Model provides a framework for understanding social learning and cognitive development.