I’m sure Bluesky will have brilliant suggestions for people or labs doing exciting work on modeling subjective experience in the broadest sense. If anything (or anyone!) comes to mind, I’d love to hear from you—always up for a chat and curious to explore new directions!

Thanks in advance!

I enjoy building big-brain models and simulations to uncover the underlying mechanisms of cognitive functions and large-scale neural dynamics. Along the way, I’ve developed strong skills in hardcore maths, machine learning, whole-cortex simulations, neuroimaging, and neuroanatomy.

Cool question!

Yes, absolutely, we actually used the full structural connectivity matrix from Kennedy's lab. We don't model thalamic input directly, but we include many cortico-cortical pathways.

pubmed.ncbi.nlm.nih.gov/23010748/

and special thanks to xiao-jing wang for his guidance, and to @seanfw.bsky.social for his kind mentoring, support and hard work 🙌!

many thanks to @jorgefmejias.bsky.social, daniel bliss, @theodoni.bsky.social for the help on the modeling work, to meiqi niu, lucija rapan & nicola palomero-gallagher for the experimental data, and to @clairesergent.bsky.social & @standehaene.bsky.social for bringing consciousness to the model!

But I can already hear people writing letters, so here’s a quick disclaimer!

We’re talking about the neural correlates of conscious access, not the correlates of specific experiences—that’s a whole other (super cool) story.

Feel free to reach out if you want to discuss it!

So to sum up:

✔ We built a biophysical model of the cortex, grounded in real data
✔ Explored the unknown parameter space
✔ Uncovered a cool underlying mathematical mechanism
✔ Experimentally validated model predictions

Time to test it!👨‍🔬

We teamed up with Nicola Palomero's lab to measure the NMDA/AMPA ratio across the cortex (after all the work we put in building the model, it would be rude of the data not to match our predictions).

And Voilà!🧠✨Monkeys' brains agree with us.

Very polite

Wait a second... this makes a testable prediction!🤯

Since sensory regions get more feedback projections, they should have more NMDA receptors than associative areas.

Which is counterintuitive—past studies emphasized NMDA in association areas, not sensory regions!

So, what makes this parameter space special?

✅ AMPA receptors strongly mediate feedforward pathways—so sensory info propagates fast
✅ Feedback pathways are relatively more NMDA-driven

Outside this sweet spot, we either lose bifurcation entirely or ignite the whole cortex

And here’s the cool part:

The model reproduces several experimental findings, including by @pieterroelfsema.bsky.social , @clairesergent.bsky.social & @standehaene.bsky.social, but only if we stick within this region of parameter space.

No scary math (it’s in the paper 👀), here’s the idea:

Stimuli shift the nullclines, pushing the system toward the excited state—or back to rest.

Then, a network of associative regions together switches between two states:
😴Resting(low activity)
🤩 Excited(high activity)

But of course, not everything in the cortex has been measured yet 🤷‍♀️. So, we explored the unknown parameter space.

And in a very specific region of this space, something cool happens...

🔥 A dynamic bifurcation! 🔥

To reproduce these dynamics, we built biophysical simulations of the macaque cortex—a species we have tons of detailed neural data for. 🧠🐵

This way, we can simulate neuronal activity across the whole cortex at the millisecond scale.

Let’s start broad,

Visual information enters the cortex through V1, crawls through the visual streams like a wave🌊, reaching associative regions.

Sometimes, the sensory wave🌊dies off. Other times, it triggers a cortex-wide ignition🔥.

What's behind this magic?

Haha, just like MESEC, acronyms which don't work are the best <3