Basketball teams do not run every play on the same timer. Some possessions are a blur - catch, cut, score. Others turn into a slow half-court chess match with one player milking the shot clock while everyone else pretends this was definitely the design. This paper argues your brain works like that too: different neural regions play on different internal clocks, called intrinsic neural timescales or INTs. Some areas update fast. Others hold onto information longer.

A Playbook With Weirdly Important Pauses

The new study by Kim and colleagues asks a sneaky question: can you infer those tempos from the brain's wiring itself? Not just from watching activity, but from the connectome - the map of which brain regions are structurally linked to which others - plus network control theory. Control theory is the branch of math that asks, "If I want a system to get from here to there, what inputs do I need, and how much effort will that take?" Engineers use it for machines. Neuroscientists use it for brains, because of course they do.

Older versions of this framework assumed every brain region had the same built-in decay rate. Tidy, yes. Also a bit like assuming every player has the same stamina and reaction time. Kim et al. dropped that assumption and let different regions have different decay rates, which correspond to different intrinsic timescales. Then they optimized those rates so simulated brain activity would move through realistic resting-state patterns more efficiently (Kim et al., 2025).

Why This Is More Than Fancy Spreadsheet Acrobatics

When the researchers allowed regions to have different timescales, the model needed less "control energy" to move between brain states. Translation: the simulated brain got from one pattern of activity to another with less mathematical shoving. If the wiring diagram is real and the timing is realistic, transitions should look more natural. This version did.

The model-based timescales also lined up with empirical timescales measured from resting-state fMRI, and with neurobiological patterns such as gene expression and inhibitory cell-type distributions. So these timescales are not just a modeling artifact with good posture. They seem to track real biology across humans and mice.

There was also a cognition angle. When the team fit the model to individual people's connectomes, the optimized version improved out-of-sample prediction of behavioral performance relative to the uniform one. Not mind reading. More like a better bridge between a person's brain wiring, neural pacing, and task performance.

The Geometry of Thought, With Fewer Straight Lines Than You'd Expect

This matters because it pushes against a very old temptation in neuroscience: treating the brain like a uniform blob with localized job titles. In reality, it looks more like a layered geometry problem. Some regions integrate over short windows, some over long ones, and cognition seems to emerge from that hierarchy rather than from one magical "thinking spot."

Recent work has been leaning the same way. Reviews in Nature Reviews Neuroscience and Neuroscience and Biobehavioral Reviews argue that neural processing depends on multiple interacting timescales, with implications ranging from multisensory integration to psychiatry (Cao et al., 2024; Ibanez and Northoff, 2024). A 2024 Nature Communications paper in rhesus macaques found that neural timescales shift with behavioral demands, and a 2024 Nature Protocols paper laid out the broader network-control pipeline behind this kind of analysis (Manea et al., 2024, PMCID: PMC10925022; Parkes et al., 2024).

Why You Should Care, Even If You Do Not Spend Friday Nights Ranking Interneurons

If this line of work keeps holding up, it could help explain why some brains switch states smoothly and others get stuck. That matters for cognitive flexibility, psychiatric disorders, and eventually brain stimulation. In principle, if you know which regions operate on slower or faster internal clocks, you may get better at predicting where to nudge the system and when. Think less "zap random cortex and hope for the best," more "target the part of the network that actually runs the tempo."

That said, the paper is not a final boss battle victory screen. The model still depends on tractography, which is useful but imperfect. It validates against fMRI, which is powerful but temporally blunt compared with electrophysiology. And the control setup still uses many more regions than a real stimulation device would target. So no, nobody has built a remote control for consciousness.

Still, this is a sharp piece of work. It takes a vague intuition - different brain regions run on different clocks - and turns it into a quantitative framework tied to structure, dynamics, and behavior. In math terms, it is a proof of concept that the brain's timing is not noise around the edges. It is part of the design.

References

Kim JZ, Betzel RF, Beyh A, et al. Inferring intrinsic neural timescales using optimal control theory. Nature Communications. 2025;16:article pending in PubMed record. DOI: 10.1038/s41467-025-66542-w. PubMed: PMID 41298426.

Parkes L, Kim JZ, Stiso J, et al. A network control theory pipeline for studying the dynamics of the structural connectome. Nature Protocols. 2024;19(12):3721-3749. DOI: 10.1038/s41596-024-01023-w. PubMed: PMID 39075309.

Manea AMG, Maisson DJ-N, Voloh B, et al. Neural timescales reflect behavioral demands in freely moving rhesus macaques. Nature Communications. 2024;15(1):2151. DOI: 10.1038/s41467-024-46488-1. PubMed: PMID 38461167. PMCID: PMC10925022.

Ibanez A, Northoff G. Intrinsic timescales and predictive allostatic interoception in brain health and disease. Neuroscience and Biobehavioral Reviews. 2024;157:105510. DOI: 10.1016/j.neubiorev.2023.105510. PubMed: PMID 38104789.

Cao Y, Gross J, Kayser C, et al. Multi-timescale neural dynamics for multisensory integration. Nature Reviews Neuroscience. 2024;25:625-642. DOI: 10.1038/s41583-024-00845-7.

Spitmaan M, Seo H, Lee D, Soltani A. Multiple timescales of neural dynamics and integration of task-relevant signals across cortex. Proceedings of the National Academy of Sciences of the United States of America. 2020;117(36):22522-22531. DOI: 10.1073/pnas.2005993117. PubMed: PMID 32839338. PMCID: PMC7486728.

Disclaimer: The image accompanying this article is for illustrative purposes only and does not depict actual experimental results, data, or biological mechanisms.