Oscar Uzdelewicz - Computational Neuro & BCIs | Ep. 1
Oscar Uzdelewicz — Computational Neuroscience Researcher
Oscar Uzdelewicz, an NYU mathematics and physics student, on computational neuroscience: what brain-computer interfaces can read today, the geometry of neural activity, and where it leads.
Key Takeaways
Model the brain, don't just describe it
Biology teaches facts about a system; maths and physics let you model it. He expects physics-based methods to come to dominate neuroscience. 3:27
A foundation model for brain signals
Synchron plans one model to interpret anyone's neural activity, which could cut the long calibration every implant needs today. 7:01
Implants read a tiny sample
Today's interfaces record from thousands of neurons out of tens of billions, and some only average the activity nearby. 10:21
Repair first, diagnosis second
Interfaces are built to restore function, such as walking after paralysis. Spotting disease in the recordings would be a by-product. 15:01
Neural activity has a shape
Give each neuron an axis and a stimulus traces out a surface. The lab's example: a dog and a cat land on separate spheres. 23:41
“In the long term, these physics-based methods are going to, if not completely replace, dominate the more traditional ways of understanding the brain.”
“As long as you're doing something computational, I'd say, you have a pretty good chance to contribute.”
Chapters
- 0:00Intro1 min
- 0:47What computational neuroscience is2 min
- 2:42Why switch to maths and physics3 min
- 5:42Research worth knowing about1 min
- 6:24Synchron's foundation model3 min
- 9:19Can we read the brain accurately?3 min
- 12:04Beyond electrical signals2 min
- 14:24Could implants diagnose disease?6 min
- 20:13A summer at the Flatiron Institute7 min
- 26:56The geometry, restated3 min
- 29:36Medicine in 20 years5 min
- 34:23Imaging and the MRI2 min
- 36:25Spending a billion dollars7 min
- 43:45A question for the next guest2 min
- 45:15Outro1 min
Mentioned in This Episode
About this Guest
Oscar Uzdelewicz
Computational Neuroscience Researcher
Computational neuroscience researcher who has worked at the Flatiron Institute on the geometry of neural representations, brain-computer interfaces, and the precision of neural data.