Inferring correlations associated to causal interactions in brain signals using autoregressive models

Area of Science:

• Neuroscience
• Computational Neuroscience
• Systems Neuroscience

Background:

• Neuronal network connectivity shapes signal dynamics.
• Understanding directed relationships between neuronal nodes is crucial.
• Existing methods like Granger causality (GC) lack mechanistic insights into connection properties.

Purpose of the Study:

• To extend Granger causality (GC) to differentiate positive and negative influences in neuronal connections.
• To analyze the specific impact of transmitter nodes on receiver nodes.
• To validate the extended GC method in a neuronal population model.

Main Methods:

• Developed an extension of Granger causality (GC).
• Analyzed the correlation between transmitter and receiver node activity.
• Incorporated the concept of positive (excitatory) and negative (inhibitory) G-causal links.
• Validated the method using a neuronal population model.

Main Results:

• The extended GC method successfully infers positive and negative couplings.
• The approach correctly identifies differential connectivity effects of excitatory and inhibitory neurons.
• The method provides additional characterization of G-causal connections.

Conclusions:

• The proposed extended Granger causality (GC) offers enhanced insights into neuronal connectivity.
• This method can help elucidate the mechanisms underlying brain circuit interactions.
• Differentiating positive and negative influences is key to understanding neural communication.