In practice, a network model lets us ask interesting questions about brain organization. Which regions are particularly important? How easy would it be, or likelihood that, two regions communicate. Which regions should be grouped together based on shared network characteristics or connectivity?

Of course, the brain is not literally a tidy graph with perfectly behaved nodes and edges. Even though fundamentally the brain is a network of neurons, the networks I gather and interrogate are constructed from imperfect or (basically always) incomplete data. Every network representation that I deal with involves construction choices: parcellation, preprocessing, timescale, similarity metric, thresholding, and interpretation. Those choices matter. But that is part of what makes the work interesting.

The model is not the system; it is a compact entry point for probing organization and structure of the real system.

Much of my work circles around these tensions. I am especially interested in cases where a static summary (e.g. FC matrix) hides something important about time-varying network organization. Edge time series are one useful example because they shift attention from regional activity alone to the moment-to-moment behavior of relationships between regions.

This notes section will mostly be a place for short entries like this one: ideas about brain networks, dynamic connectivity, open tools, and the occasional methods thought.