On drivers of neural variability: " ...Such discoveries may lead us to see the vividly dynamic tapestry of neural activity as perhaps colored more by the organism’s own internal signals and events than by external stimuli or overt motor acts"

.In contrast, single trials of multi-neuron, multi-region simultaneous recordings suggest something starkly different: task-related neural dynamics appear to be more like a vibrant storm, with highly varying, internally-generated signals that have a brain-wide reach..."

On trial averaged vs single trial: "Trial-averaged population decoding of task-related signals has given us many fundamental insights. But, through averaging, it can make neural population dynamics seem like a calm, stately progression towards logical outcomes ...

On the drift diffusion model. "..it is remarkable that a simple one-dimensional behavioral model has proved so enduring .... first at the level of single neurons, then at the level of neural populations, and here, at the level of coordinated, brain-wide activity.”

Hello David, congratulations, cool work!. I wonder what are the connections with the in Darshan & Rivkind, 22?

End of thread
🧠 Read the full preprint here: www.biorxiv.org/content/10.1...

11/ Want to chat about this work or have questions? Reply or DM! 📬

10/ Big thanks to Semedo et al. (2019) for introducing communication subspaces—a pivotal concept that made this possible.

9/ Shout-out to our incredible collaborators & experimental inspiration—this project wouldn’t exist without the groundwork laid by recent advances in brain-wide recordings. 🧠❤️

8/ Why this matters:
This work bridges neuroanatomy, electrophysiology, and computational modeling to provide a unified theory of how cognitive networks interact dynamically.

7/ Additionally, he frontoparietal multiple-demand network displays a coexistence of stable and dynamic coding (see e.g., Jake P Stroud, John Duncan, and Máté Lengyel, 2024), suitable for top-down cognitive control.

6/ Our findings explain key cognitive processes:
🔹 How networks like DMN & DAN compete to focus attention
🔹 How the Multiple Demand Network (MDN) flexibly couples to these networks to decide what deserves attention
🔹 How salient stimuli drive global shifts in network activity

5/ Within-area connectivity includes symmetric, asymmetric, and random motifs, generating stable (attractor) or transient (sequential) dynamics.
Inter-areal connectivity is sparse, low-rank, and random, constrained by cognitive network activation maps.

4/Inspired by this idea we developed connectivity-constrained whole-cortex models for macaques and humans.

3/ Current recordings suggest long-range projections act as bottlenecks or "communication subspaces" (Semedo et al. 2019) that selectively route some neural activity while keeping others private. 🧠🔗
This inspired us to ask: Could these subspaces coordinate cognitive networks?

2/ Neocortex-wide neural activity is organized into distinct networks engaged in different cognitive processes. These networks flexibly reconfigure, but the mechanisms enabling this remain unclear. What’s the mechanism?

1/ What’s the big question? How does the neocortex reorganize activity across cognitive networks like the Default Mode Network (DMN) and Dorsal Attention Network (DAN)? 🤔 These networks control whether you're focused on the external world or internal thoughts.