CCN 2026 | Tutorial: State-Space Modelling for Human Neuroscience

Value of the Information & Strength of the Argument

The tutorial provides substantial value by offering a clear, step-by-step introduction to state-space modeling, a powerful but often underutilized approach in cognitive neuroscience. The argumentation is solid: the presenters justify the use of SSMs by highlighting their ability to model latent dynamics, handle noise, and provide interpretable parameters. They contrast SSMs with simpler methods like PCA and HMMs, explaining why the added complexity is beneficial. The presentation is well-structured, moving from theory to practical implementation, and includes concrete examples from EEG and fMRI. The emphasis on understanding the math behind the methods, even when using high-level packages, strengthens the educational value. The tutorial also acknowledges limitations, such as the point-estimate nature of EM, and encourages a Bayesian perspective where possible.

Scientific Rigor, Source Quality, Title Accuracy

The tutorial demonstrates scientific rigor by grounding the methods in established literature, referencing key papers and researchers in the field. The presenters are credible experts, and the content aligns with current best practices in computational neuroscience. The title accurately reflects the content, and the tutorial’s structure is logical and well-paced. The inclusion of practical resources, such as GitHub notebooks and references, enhances its reliability. However, as a tutorial, it does not present new empirical findings, and the reliance on the presenters’ own packages introduces a potential bias, though they acknowledge this. Overall, the sources are appropriate and the content is trustworthy.

The title accurately reflects the content: a tutorial on state-space modeling applied to human neuroscience.

8/10

The tutorial is presented by leading researchers in computational neuroscience, with a clear pedagogical structure and references to established methods. The content is technically accurate and aligns with current literature, though it is an introductory tutorial and not a peer-reviewed study.

Markers derived by PSI from the transcript: the creator did not define chapters.

• 00:00Introduction and overview of tutorial structure
• 05:00Definition of state-space models and their components
• 10:00Motivation: why use state-space models in neuroscience
• 15:00Introduction to Bayesian filtering and expectation maximization
• 20:00Explanation of the Kalman filter and its derivation
• 25:00Linear dynamical systems: equations and interpretation
• 30:00Simulation and fitting of LDS models
• 35:00Introduction to switching linear dynamical systems
• 40:00Application to HCP resting-state fMRI data
• 45:00Discussion of results and implications

• CCN 2026 Keynote and Tutorial Language: Putting Dynamics First — Official conference page for the tutorial, providing context and resources.

• A unifying review of linear Gaussian models — Referenced in the tutorial as a key review for understanding the family of linear Gaussian models.
• Dynamax (Python library) — Mentioned as a tool for implementing state-space models.