Cognitive biases in human causal learning.

Area of Science:

• Cognitive Psychology
• Human Inference
• Causal Learning

Background:

• Existing theories often focus on either prior causal beliefs or covariation information, but not both.
• Understanding cognitive biases is crucial for explaining psychological processes in causal learning.
• Cue competition effects (overshadowing, blocking) highlight complexities in how humans weigh evidence.

Purpose of the Study:

• To investigate cognitive biases in human inference of causal relationships.
• To explore the psychological processes modulating causal learning.
• To determine how prior beliefs and new evidence interact in causal judgments.

Main Methods:

• Examined judgment frequency to study cue competition effects.
• Assessed the additive contribution of prior beliefs and covariation data.
• Investigated the role of belief and evidence reliability in modulating judgment weights.
• Tested for "inattentional blindness" in learning positive versus negative causal relationships.

Main Results:

• Prior beliefs and new covariation evidence contribute additively to causal judgments.
• The relative weight of beliefs and evidence depends on their respective reliability.
• "Inattentional blindness" was observed for negative/preventative causal relationships, but not positive/generative ones.
• A hierarchical architecture for causal learning is proposed, involving event codification, computation, and integration.

Conclusions:

• Human causal learning is complex, requiring the integration of both prior beliefs and covariation information.
• Cognitive biases, such as inattentional blindness, significantly impact causal inference.
• A three-level hierarchical model best explains the multifaceted nature of human causal learning.