ScholarGate
Get started
Process / pipelineNeuroimagingTime-varying network analysisPipeline

Dynamic Functional Connectivity

Also known as: dFC, time-varying connectivity, sliding window connectivity

OriginatorRyan M. HutchisonYear2013Sources2Related methods8

Dynamic Functional Connectivity (dFC) is an analytical framework that tracks changes in functional connectivity between brain regions over time, rather than averaging connectivity across an entire scanning session. Systematized by Hutchison and colleagues in 2013, dFC reveals how brain networks reorganize moment-to-moment, providing insights into transient brain states and cognitive flexibility.

Key highlights

  • Captures real-time changes in brain connectivity, revealing a richer picture than static connectivity
  • Identifies distinct connectivity states without a priori hypothesis; unsupervised approaches discover recurrent patterns
  • High sensitivity to brain state transitions; useful for detecting attention lapses, sleep transitions, and altered consciousness
  • Generalizable to any connectivity metric (correlation, coherence, phase locking) and any imaging modality

Intuition

This section is available to Pro members. Upgrade to Pro

How it works

This section is available to Pro members. Upgrade to Pro

When to use it

dFC is appropriate when interest is in transient brain states, when monitoring how connectivity changes during cognitive or behavioral transitions, and in clinical contexts where dynamic reorganization differs from healthy controls. Use dFC for tasks with known cognitive state transitions (e.g., task engagement vs. disengagement). Avoid dFC for very short scans (<5 minutes) where window-based estimates are unreliable, or when static connectivity patterns are the primary question.

Strengths & limitations

Strengths
  • Captures real-time changes in brain connectivity, revealing a richer picture than static connectivity
  • Identifies distinct connectivity states without a priori hypothesis; unsupervised approaches discover recurrent patterns
  • High sensitivity to brain state transitions; useful for detecting attention lapses, sleep transitions, and altered consciousness
  • Generalizable to any connectivity metric (correlation, coherence, phase locking) and any imaging modality
Limitations
  • Window length is arbitrary; different window choices produce different dFC patterns; sensitivity to this choice unclear
  • Statistically underpowered: each window contains fewer time points than full-session analyses, reducing reliability of individual connectivity estimates
  • Circular dependencies: neighboring windows overlap heavily, violating statistical independence assumptions
  • Interpretation ambiguous: is dFC a true reflection of changing neural dynamics, or an artifact of windowing?

Common pitfalls

This section is available to Pro members. Upgrade to Pro

Applications

This section is available to Pro members. Upgrade to Pro

Frequently asked

What window length should I use for dFC?

No consensus exists. Common choices are 44 seconds (50 TRs at TR=0.88s), 60 seconds (44 TRs at TR=1.36s), or flexible approaches using wavelets. Sensitivity to window length is a known limitation. Best practice: test multiple window lengths and report results for the most interpretable choice. Biological validity of short-window estimates (<1 minute) remains debated.

How do I identify connectivity states in dFC?

Common approaches include k-means clustering (divides dFC matrices into k states), hidden Markov models (probabilistic state transitions), and principal component analysis (extracts dominant dFC patterns). Each method has assumptions; clustering assumes discrete states, while continuous models better capture state transitions. Choose based on your hypothesis.

Does dFC reflect true neural dynamics or preprocessing artifacts?

Active debate. Some studies show dFC correlates with behavior (supporting neurobiological validity), while others argue preprocessing steps (filtering, smoothing) create spurious dFC. Consensus: some dFC is neurobiologically meaningful, but proportion of true vs. artifact signal remains unknown. Control for head motion carefully.

Can I use dFC to predict clinical outcomes?

Yes, but with caveats. dFC features (state entropy, transition probability, dwell time) show promise for predicting schizophrenia, autism, and depression severity. However, effect sizes are often modest, replication rates are mixed, and mechanistic interpretations are speculative. Use dFC for hypothesis generation; verify with independent samples.

Sources

  1. 1.
    Hutchison, R. M., Womelsdorf, T., Allen, E. A., et al. (2013). Dynamic functional connectivity: promise, problems, and perspectives. NeuroImage, 80, 360–378.
    https://scholar.google.com/scholar?q=Dynamic+functional+connectivity%3A+promise%2C+problems%2C+and+perspectives+Hutchison
  2. 2.
    Calhoun, V. D., Miller, R., Pearlson, G., & Adalı, T. (2014). The chronnectome: time-varying connectivity networks as the next frontier in fMRI data discovery. Neuron, 84(2), 262–274.
    DOI 10.1016/j.neuron.2014.10.015

You have read it. What now?

Ask about this methodSimilar methods

Cite this page

ScholarGate. (2026, June 3). Dynamic Functional Connectivity. ScholarGate. https://scholargate.app/neuroimaging/dynamic-functional-connectivity

Dynamic Functional Connectivity | ScholarGate