Dynamic Causal Modeling in Neuroscience

Dynamic Causal Modeling (DCM) is a Bayesian framework for inferring directed interactions among neural populations from neuroimaging and electrophysiological data. Rather than measuring mere statistical dependencies, DCM employs generative models that link hidden neuronal states to observed signals, thereby estimating the strength and sign of causal connections—termed effective connectivity. Central to DCM is the specification of neuronal state equations coupled with a biophysical forward model, typically describing the haemodynamic response in fMRI or the equivalent mechanisms in EEG and MEG. Model inversion yields posterior estimates of connection strengths and their uncertainties, while model comparison exploits approximations to the model evidence—often implemented via variational Bayes or free-energy optimisation—to adjudicate among competing hypotheses about network architecture. Over the past decade, DCM has evolved to accommodate large‐scale resting‐state networks, refine neural mass formulations, and incorporate hierarchical group analyses through Parametric Empirical Bayes. These developments have expanded the scope of DCM from small, task‐based paradigms to clinical and cognitive investigations of psychiatric disorders, neuromodulation effects and developmental trajectories. By enabling explicit tests of mechanistic hypotheses, DCM bridges theoretical neuroscience and practical applications, informing neuromodulatory interventions, biomarker discovery and personalised models of brain dysfunction.

Research from Nature Portfolio

No recent Nature Portfolio content available.

Research from all publishers

A recent methodological contribution has presented a comprehensive primer on Variational Laplace, demonstrating how variational Bayesian inference can be deployed across a broad class of DCMs without bespoke derivations for each model. This approach streamlines the inversion of static and dynamic generative models, supports estimation of log model evidence for rigorous model selection and provides open-source code to facilitate adoption in both academic and clinical settings.

Complementing these theoretical advances, applications of spectral DCM to resting-state fMRI have shed light on clinical interventions and network neurobiology. One study used high-frequency transcranial magnetic stimulation in traumatic brain injury patients to reveal that effective connectivity within cortico-limbic circuits is modulated by treatment, with inhibitory coupling strengthening and excitatory links weakening in key regions such as the dorsal anterior cingulate cortex. Another investigation employed a triple-network spectral DCM to characterise mood side effects of oral contraceptives, finding that altered directed interactions among default mode, salience and executive control networks predict individual variability in mood lability during treatment.

Dynamic Causal Modeling in Neuroscience publication trend

The graph below shows the total number of articles in dynamic causal modeling in neuroscience across all publications each year (not limited to Nature Index journals).

Technical terms

Dynamic Causal Modeling (DCM): A Bayesian framework for inferring directed (causal) interactions among neural populations based on generative models of brain activity.

Effective connectivity: The influence that one neural region exerts over another, estimated within a specified model of neuronal and measurement dynamics.

Generative model: A mathematical formulation that maps hidden neuronal states and parameters to observed neuroimaging or electrophysiological data.

Spectral DCM: A variant of DCM that parameterises neuronal fluctuations in the frequency domain, enabling efficient estimation of resting-state effective connectivity.

Parametric Empirical Bayes (PEB): A hierarchical Bayesian approach for group-level DCM analyses that uses subject-specific estimates as empirical priors to test hypotheses about between-subject effects.

References

1. Neural mechanisms of emotional health in traumatic brain injury patients undergoing rTMS treatment. Molecular Psychiatry (2023).
2. Triple network model of brain connectivity changes related to adverse mood effects in an oral contraceptive placebo-controlled trial. Translational Psychiatry (2023).
3. A primer on Variational Laplace (VL). NeuroImage (2023).
4. A guide to group effective connectivity analysis, part 2: Second level analysis with PEB. NeuroImage (2019).
5. Dynamic causal modelling revisited. NeuroImage (2017).
6. Large-scale DCMs for resting-state fMRI. Network Neuroscience (2017).
7. Effective connectivity: Influence, causality and biophysical modeling. NeuroImage (2011).
8. Comparing Dynamic Causal Models using AIC, BIC and Free Energy. NeuroImage (2011).