Decomposing causality into its synergistic, unique, and redundant components.

Area of Science:

• Complex systems analysis
• Information theory
• Causal inference

Background:

• Causality is fundamental to scientific understanding but challenging to infer.
• Existing causal inference methods struggle with nonlinearities, stochasticity, self-causation, colliders, and exogenous factors.
• No single method comprehensively addresses these multifaceted challenges.

Purpose of the Study:

• To develop a novel framework for causal inference that overcomes limitations of existing methods.
• To introduce Synergistic-Unique-Redundant Decomposition (SURD) for quantifying causality.
• To provide a non-intrusive method applicable to diverse investigations, including those with scarce data.

Main Methods:

• SURD quantifies causality by measuring increments of redundant, unique, and synergistic information.
• The approach analyzes information gained about future events from past observations.
• Formulation is non-intrusive, suitable for computational and experimental settings.

Main Results:

• SURD was benchmarked in challenging causal inference scenarios.
• The method demonstrated superior reliability in quantifying causality compared to prior approaches.
• SURD effectively handles complex interactions like nonlinear dependencies and stochasticity.

Conclusions:

• SURD provides a robust and integrated approach to causal inference.
• The method offers a more reliable quantification of causality, especially in complex systems.
• SURD's applicability to scarce data and diverse investigations enhances its utility.