Discovering Brain Mechanisms Using Network Analysis and Causal Modeling.

Area of Science:

• Neuroscience
• Philosophy of Science
• Computational Neuroscience

Background:

• Mechanistic philosophy investigates scientific strategies for causal discovery.
• Brain anatomy is crucial for many neuroscience discovery strategies.
• Network analysis and causal modeling in neuroscience lack philosophical exploration regarding anatomical integration.

Purpose of the Study:

• To clarify how network analysis and causal modeling incorporate brain anatomy for mechanism discovery.
• To distinguish between structural, functional, and effective connectivity in causal inference.
• To examine the roles of dynamic causal modeling and probabilistic graphical modeling in neuroscience research.

Main Methods:

• Analysis of dynamic causal modeling (DCM) and probabilistic graphical modeling (PGM) for causal discovery.
• Distinction between structural, functional, and effective brain connectivity.
• Examination of how anatomical information is used in quantitative causal discovery strategies.

Main Results:

• Dynamic causal modeling integrates brain anatomy to refine causal model parameter estimation.
• Probabilistic graphical modeling operates without anatomical data, defining conditions for causal inference from correlational data.
• The study contrasts the anatomical constraints applied by DCM versus PGM.

Conclusions:

• The necessity and utility of brain anatomy in constraining causal network inference remain an open question.
• Graphical modeling tools offer methods to address the role of anatomical findings in causal inference.
• Different causal discovery methods in neuroscience have varying relationships with anatomical data.