Adiabatic dynamic causal modelling.

Area of Science:

• Computational Neuroscience
• Systems Neuroscience
• Biophysics

Background:

• Understanding brain activity requires modeling dynamic changes in biophysical parameters.
• Neuronal states exhibit fast fluctuations influenced by slower underlying processes.
• Phase transitions in brain activity, such as seizures, are linked to alterations in these parameters.

Purpose of the Study:

• Introduce adiabatic dynamic causal modeling (a DCM) for inferring slow biophysical parameter changes.
• Develop a method to link slow parameter dynamics to fast neuronal state fluctuations.
• Apply the method to understand seizure activity and synaptic plasticity.

Main Methods:

• Utilize an adiabatic approximation to summarize fast neuronal state fluctuations using second-order statistics (complex cross-spectra).
• Employ Bayesian model reduction to compare models of slowly changing parameters generating empirical cross-spectra.
• Incorporate slow fluctuations in spectral power as priors for synaptic parameter changes.

Main Results:

• Demonstrate the efficiency and biophysical interpretability of the adiabatic DCM approach.
• Validate the model's face validity through simulations.
• Provide an illustrative application to a chemoconvulsant animal model of seizure activity.

Conclusions:

• Adiabatic DCM offers a powerful framework for inferring slow biophysical changes in neural systems.
• The method captures the circular causality between synaptic parameters and neuronal activity.
• This approach has significant implications for understanding brain dynamics and neurological disorders.