Over 40,000 retinal cells have been proofread in EyeWire II!
Each one is a tiny piece of the extraordinary neural circuitry that makes vision possible.
Request early access: eyewire.ai
And visit our new citizen science hub! connectome.quest
Sebastian Seung
Hyunjune Sebastian Seung (English: /sung/ or [səŋ]; Korean: 승현준; Hanja: 承現峻) was president at Samsung Research (Samsung Electronics) and is an Anthony B. Evnin Professor in the Princeton Neuroscience Institute and Department of Computer Science. Seung has done influential research in both computer science and neuroscience. He has helped pioneer the new field of connectomics, "developing new computational technologies for mapping the connections between neurons," and has been described as the cartographer of the brain. More on Wikipedia
-
Sebastian Seung reposted
Eyewire
@eye_wire
-
Sebastian Seung
@SebastianSeung
Eyewire is back ... with a vengeance!
Eyewire
@eye_wire
Congratulations to the citizen scientists who have now mapped over 10,000 cells in Eyewire II! This animation by @amyneurons shows ~1,000 bipolar cells.
eyewire.ai #citizenscience #citsci -
Sebastian Seung
@SebastianSeung
Comparing mitochondria across supergroups that diverged over a billion years ago!
Alice Ting
@aliceyting
Remarkable Magnum Opus from @VamsiMootha & the Mitocarta Tree of Life consortium consisting of 9 back to back research papers (+2 perspectives) just released in Cell and Cell family journals:
- Characterization of mito proteomes from 6 diverse species, from Giardia to Leishmania to Arabidopsis
- Optimized methods for mito enrichment, DIA mass spec, protein correlation profiling, microscopy validation, and cryoEM analysis of native complexes
- Comparative analysis reveals dozens of proteins shared among pathogens but lost in human hosts = “pan-parasite” drug targets
- Data used to re-train ML predictor of subcellular localization => predicted mitoproteomes across hundreds of eukaryotes
- Giardia solute carrier discovered with functional validation
- Dynamic rewiring of Acanthamoeba mito proteome from aerobic to anaerobic energy production characterized
And tons more… The papers are instant landmarks and raise even more questions than they answer, paving the way for many future investigations of this fascinating organelle #mitochondria
www.cell.com/consortium/mitocarta-tree-of-life -
Sebastian Seung reposted
Amy Robinson Sterling
@amyneurons
Selective feedforward inhibition.
7 mossy fibers (of 55,000 in the dataset, yellow) converge on one interneuron (green) in hippocampal CA3. Its axon makes 333 synapses onto thorny pyramidal neurons (blue) vs 50 synapses onto sparsely thorny pyramidals (purple).
Paper in @NatureNeuro www.nature.com/articles/s41593-026-02388-9This media couldn't be displayed. See it on the original post. -
Sebastian Seung reposted
FlyWire
@FlyWireNews
Thousands of new visitors and tens of thousands of new searches on codex.flywire.ai, where both male and female fly connectomes are available. Get familiar with the fly brain and show us what you create! whatisabrain.com/fly/
-
Sebastian Seung
@SebastianSeung
As a high schooler I enjoyed snail mailing my solutions to math problems he posed in his monthly column, and receiving his replies by post. He was indeed a very special man.
Daniel Fetz
@danftz
I just learned that two weeks ago the world lost a very special man.
George Berzsenyi (1938-2026) mentored thousands of talented high school students over fifty years, including @VamsiMootha and many others who are now among America's best mathematicians and scientists.
Most notably, George Berzsenyi founded the USA Mathematical Talent Search. Unlike other math competitions, the USAMTS is about talent identification and development rather than just competition.
That has always been what set his work apart. As George put it in his essay "Talent Search versus Talent Development," which was also how I came across him, it bothered him that we identify tens of thousands of talented students through math competitions, then build a development program for ~60 and leave the rest to fend for themselves.
He spent his life trying to fix that. It was a life well lived. The flourishing of science, and civilisation with it, requires people like him.
May he rest in peace.
Some links if you want to learn more about George Berzsenyi:
The 2019 NPR profile of him is worth reading: www.npr.org/2019/04/07/707326070/a-math-teachers-life-sum…
Citation for winning the Gung and Hu Award www.wfnmc.org/awberzsenyi.pdf
George writes about his upbringing in Hungary, and the humiliations and disappointments the family received by the Communists: www.childhoodhomestories.com/post/györgy-george-berzsenyi… -
Sebastian Seung
@SebastianSeung
Brain and body!
Alexander Bates
@as_bates
Just published yesterday, the #banc! A full central nervous system (CNS) connectome of a limbed animal enables us to trace sensory-motor arcs and understand how the CNS controls the body. rdcu.be/fncjS #neuroscience @Nature, Video by @quorumetrix, sound on! 1/18
This media couldn't be displayed. See it on the original post. -
Sebastian Seung reposted
Eyewire
@eye_wire
Meet Eyewire II: a new connectomic resource for the mouse retina.
~1 mm² of retina at nm resolution, with synapses and circuits of ca. 100,000 neurons + visual responses from the ~400 neurons shown in the video!
Preprint: doi.org/10.64898/2026.05.28.727403
Data: eyewire.ai -
Sebastian Seung
@SebastianSeung
“We aren’t ready for how much people hate AI.”
open.substack.com/pub/jasmine/p/warning-shots?r=3cftot -
Sebastian Seung reposted
Eyewire
@eye_wire
We are delighted to announce that thanks to generous support from Research to Prevent Blindness @RPB_org, Eyewire II is now officially underway!
RPB
@RPB_org
RPB has granted @NUFeinbergMed a $950,000 RPB Transformational Team Science Award to support the @eye_wire II project to create the first complete connectome in the mammalian central nervous system with @Princeton and University of Tübingen. www.rpbusa.org/news-press/research-to-prevent-blindness-a…
-
Sebastian Seung
@SebastianSeung
TL;DR: “A lie can travel half way around the world while the truth is putting on its shoes.” 😊
Kenneth Hayworth
@KennethHayworth
So, some people are asking me why this EON fly video doesn’t show real ‘uploading’ since it does simulate a real connectome. The most important reason is that the functional parameters that define the dynamic behavior of individual neuron and synapse types in the connectome are unknown. Instead, they used an existing model (www.nature.com/articles/s41586-024-07763-9) which substitutes these with guessed parameters and grossly simplified dynamics. As made clear in that older paper, these are not sufficient to recreate the activity patterns that would be seen in the real fly. The simplified dynamics would not, for example, be able to choreograph the timing of leg muscles during walking or grooming, or the dynamics of the compass neurons encoding the fly’s heading direction, or the myriad other neuronal dynamics that make up the fly ‘mind’. So not an ‘upload’ by any reasonable definition.
In fact, the simplified dynamics they used have only been demonstrated to approximate gross correlations along major sensory-motor pathways for a handful of neurons. For example: activating a sugar sensing neuron causes gross downstream activation that elevates the activity of feeding neurons. It is this handful of very, very crude and basic correlations in the simulated connectome that are being used to drive the EON simulated fly.
If they had said that from the start, then I would have had no issue. But instead, they made the bold claim that they had “uploaded a fly” and presented a video of said fly walking over a landscape with highly articulate legs, visually navigating through the terrain to a food source, grooming its antenna with eerily fly-like leg motions, etc. Any reasonable layperson would assume that these visually exciting articulations are the ones being controlled by the simulated brain’s dynamics instead of being faked by computational add-on routines. There are now many secondary reports of this on YouTube and all of them seem to make this reasonable assumption (e.g. www.youtube.com/shorts/Z7NNP1ZI0eI). And who could blame them? Many neuroscientists also made that assumption before EON started to spell out what was really behind the video millions of views and over a day later.
To make clearer just how misleading EON Systems’ video is and how outlandishly laughable their ‘uploading’ claim is, below is an imagined back-and-forth discussion between a [Reasonable Layperson] and a [Neuroscientist] trying to explain to them what is really behind the video:
[Reasonable Layperson] “Look at the complicated leg motions as the fly walks… the timing of all those dozens of individual muscles being controlled by the dynamics of the simulated neurons… and they say that they used no reinforcement learning to tune parameters, just the connectome… that is really impressive!”
[Neuroscientist] “Well actually no… those leg movements are actually coming from a program unrelated to the connectome. The connectome used didn’t even include the central pattern generator circuits in the ventral nerve cord responsible for controlling leg muscles.”
[Reasonable Layperson] “Oh… so in what sense is the simulated connectome controlling walking?”
[Neuroscientist] “It looks like they just found a few neurons in the brain connectome that are correlated with right/left/forward motion and used these to ‘steer’ the pretend walking routine.”
[Reasonable Layperson] “Oh… But the activations of those ‘steering’ neurons are reflecting the complicated dynamics of tens of thousands of simulated neurons in the fly visual system as it moves through the virtual world, avoiding objects and heading toward its visual goal, right?”
[Neuroscientist] “Well actually no … The visual system and virtual world are essentially ‘decoration’… the flashing dynamic neural responses as the fly moves through the virtual environment are designed to give the viewer the impression that the simulated fly is actually seeing the world and making walking decisions based on those visual responses. But, in fact, they could turn off the lights and the fly would behave identically.”
[Reasonable Layperson] “Oh… so how does the fly walk toward the food then?”
[Neuroscientist] “Well… it looks like they simply imposed an odor gradient in the virtual environment that is centered on the virtual food. The fly has two sets of odor receptors (right and left) that sense this gradient and the activation of these in the connectome is correlated with the activation of the ‘steering’ neurons. So if the left odor neuron activates more than the right then the fly steers left.”
[Reasonable Layperson] “Oh… so it is like one of those toy cars that moves toward a light because it has right and left light sensors cross-connected to right and left motors… Gee, I thought a fly was more complicated than that.”
[Neuroscientist] “Well actually a real fly is. Real flies have dozens of behavioral states that allow intelligent behavior in a complicated visual and sensory environment. In fact, a real fly contains a set of neurons which act as an internal compass updated by the visual environment and the fly’s walking.”
[Reasonable Layperson] “Oh… and their connectome has those internal compass neurons?”
[Neuroscientist] “Yes. They used the full brain connectome that contains those compass neurons.”
[Reasonable Layperson] “...And their compass neuron activations are tracking the visual environment just like in the real fly?”
[Neuroscientist] “Oh sweet summer child… those compass neurons exist in their connectome simulation, but no one knows enough about their functional parameters (synaptic weights, time constants, etc.) to simulate them accurately. They light up in pretty patterns totally unrelated to how they would in a real fly walking through that visual world.”
[Reasonable Layperson] “Oh… and the complicated leg movements it shows during antenna grooming… is that also just a faked recording?”
[Neuroscientist] “Yes. All the complicated leg motions shown during grooming are faked by a hard-coded program. But they turn that fake routine on or off by looking at some neurons in the connectome that are correlated with actual grooming behavior triggered by dust accumulation on the antenna… well really they fake the dust too by just activating a set of neurons after a delay.”
[Reasonable Layperson] “And what did EON Systems do? Did they acquire the connectome? Did they determine the neurotransmitter types? Did they do the calcium imaging experiments to determine the steering and grooming neurons? Did they make the mechanical fly model?”
[Neuroscientist] “No. Those were all done by real labs who were kind enough to carefully write up their results in open journals and to post their results and code openly online…. It looks like Eon Systems just took their code and put it together with a virtual environment designed specifically to trick viewers by triggering behaviors in misleading ways.” -
Sebastian Seung reposted
Princeton University
@Princeton
Princeton's Mala Murthy and @SebastianSeung have been awarded the Wiley Prize “for reconstructing and interpreting connectomes, the wiring diagrams that anatomically interconnect nerve cells and thereby drive information processing in nervous systems.” www.princeton.edu/news/2026/02/18/murthy-and-seung-win-wi…
-
Sebastian Seung
@SebastianSeung
Excited that my friend and colleague, neuroscientist Sam Wang, is running for Congress!
Sam Wang is at samwang.bsky.social
@SamWangPhD
Why I'm running for Congress: www.dailyprincetonian.com/article/2026/02/princeton-opini…
-
Sebastian Seung
@SebastianSeung
Biggest surprise that should not be a surprise: John G. White (second from left) is being recognized for his groundbreaking work, published 40 years ago. What took so long?
Wiley
@WileyGlobal
The Wiley Foundation is pleased to announce the winners of the 24th annual Wiley Prize in Biomedical Sciences, which recognizes contributions that open new fields of research or advance concepts in a particular biomedical discipline.
John White, Gerald Rubin, Sebastian Seung, and Mala Murthy will be awarded for reconstructing and interpreting connectomes, the anatomical wiring diagrams of neurons and synapses that underlie how the brain processes information and controls actions.
Learn more about the winners in our latest press release: johnwiley2020news.q4web.com/press-releases/press-release-…
Sebastian Seung
@SebastianSeung
In 1986, the Philosophical Transactions of the Royal Society published a curious paper. Its 340 pages filled an entire issue of the journal. Its title was long and dry: "The structure of the nervous system of the nematode Caenorhabditis elegans"
Sebastian Seung
@SebastianSeung
Mercifully, the running title was the pithier "The Mind of a Worm". A wry and now legendary way of referring to an encyclopedic map of the connections between 302 neurons, now called a connectome.
Post not available here, view on X
Sebastian Seung
@SebastianSeung
For more history, read Scott Emmons on "The beginning of connectomics" and other papers linked here.
www.wormbook.org/toc_wormhistory.html
Sebastian Seung
@SebastianSeung
John White's presentation of his work at a symposium "went down like a lead balloon." Another talk was met with "indifference."
-
Sebastian Seung
@SebastianSeung
Mala and I will be delighted to accept this award on behalf of the FlyWire Consortium of 75 laboratories and citizen scientists around the world!
FlyWire
@FlyWireNews
Congratulations to FlyWire founders @SebastianSeung and @MurthyLab, who alongside John White and Gerald Rubin have been recognized with the Wiley Prize for their work pioneering connectome reconstruction & interpretation!
Wiley
@WileyGlobal
The Wiley Foundation is pleased to announce the winners of the 24th annual Wiley Prize in Biomedical Sciences, which recognizes contributions that open new fields of research or advance concepts in a particular biomedical discipline.
John White, Gerald Rubin, Sebastian Seung, and Mala Murthy will be awarded for reconstructing and interpreting connectomes, the anatomical wiring diagrams of neurons and synapses that underlie how the brain processes information and controls actions.
Learn more about the winners in our latest press release: johnwiley2020news.q4web.com/press-releases/press-release-… -
Sebastian Seung
@SebastianSeung
A thread and article about consciousness that's well worth reading. An interesting angle is to criticize certain definitions of consciousness by using mechanistic explanations of behavior based on the connectome.
Ralph Stefan Weir
@RalphStefanWeir
Is this microscopic worm sentient? It’s a tough question! Increasingly, however, neuroscience is providing insights into the physical correlates of consciousness at the scale of individual neurons, and we should be excited about what it has to offer. A thread and and article.
-
Sebastian Seung
@SebastianSeung
That's what we need for a human connectome. Cc @zettaconnectome
george hotz archive
@geohotarchive
Will I ever own a zettaflop? geohot.github.io//blog/jekyll/update/2026/01/26/own-a-zet…
-
Sebastian Seung
@SebastianSeung
One of my heroes in high school... I read the anthology of his basic writings and was captivated. He was courageously outspoken. This video is the first time I ever saw him speak!
Saganism
@Saganismm
Bertrand Russell’s timeless message to future generations
Sebastian Seung
@SebastianSeung
I came across Russell's writings in my father's library. My father even had a copy of Principia Mathematica. I realized recently that Russell was my father's academic uncle. He was a historical figure to me, but must have felt like a real life person to my father.
-
Sebastian Seung
@SebastianSeung
It's shocking to realize how mainstream our field has become. Today's students have trouble believing that neuroscientists used to hate connectomics, because the criticisms were mostly whispered/spoken and hence ephemeral. See thread for a fun exception that was in writing 👇
Princeton Neuroscience Institute
@PrincetonNeuro
Congratulations to @Princeton’s @SebastianSeung, who is named the winner of the 2026 @theNASciences Pradel Research Award! 🧠🏆
Seung is recognized for “fundamentally reshaping” our understanding of brain wiring & connectomics. #nasaward
Read more: pni.princeton.edu/news/2026/sebastian-seung-wins-national…
Sebastian Seung
@SebastianSeung
This critical review of my book was titled "Anti-connectome-ism" www.theguardian.com/science/neurophilosophy/2012/sep/21/c…
Sebastian Seung
@SebastianSeung
The funniest part is the email appended at the end of the review:
After the review was published, I received an email from Andrew Lumsden, which I've pasted below, with his permission. Andrew is the director of the MRC Centre for Developmental Neurobiology; after graduating in 1998, I spent two years in his department working on a Ph.D. about the migration of granule cell precursors in the embryonic chick cerebellum. He's also one of the "characters" in my New Scientist feature article about adult neurogenesis.
Dear Moheb,
Enjoyed your article in New Humanist, albeit that I felt you were rather soft on this guy - perhaps in the interests of objectivity. What really pisses me (and many others) off is the vast amount of money going into high res analysis of dead brains, using techniques such as serial EM that reveal the twigs but not the trees and certainly not the woods. Two strikes against.
What we know from C. elegans and the crustacean stomatogastric ganglion, with networks involving very few neurons, is that we haven't a clue how they work even with the essential addition of electrophysiology. In C. elegans a network of just two neurons can behave quite differently at different times, depending on what modulators are around. So, while SS's memorable (if pukeworthy) "I am more than my genome, I am my connectome" may be true in a grossly superficial way, they are not going resolve connectional issues in any useful way using their current approaches.
A simpler and potentially more powerful approach would be to combine retrograde monosynaptic tracers such as the rabies virus variants now being made by Ed Callaway and others with targeting of molecularly defined cells types - this way, for example, you could activate the virus in just motor neurons and then see only those pre-motor neurons that synapse onto them. Then release in the pre-motor neurons and see only what synapses onto them, and so on, and slowly build entire sub-connectomes with comparative ease and high fidelity. With viruses expressing FPs you would also have the potential for doing electrophysiology. Just a few thoughts.
All the best,
Andrew
Professor Andrew Lumsden MA PhD FMedSci FRS
Director, The MRC Centre for Developmental Neurobiology
King's College London
Guy's Campus
London SE1 1UL -
Sebastian Seung
@SebastianSeung
For over twenty years, many dedicated researchers around the world have worked to realize the dream of connectomics. I'm especially indebted to the talented members of my laboratory, past and present. And I've been lucky to benefit from so many powerful collaborations.
National Academy of Sciences
@theNASciences
The winner of the 2026 Pradel Research Award is H. Sebastian Seung (@SebastianSeung) of @PrincetonNeuro! He is being honored for for transformative advances in computational neuroscience. Learn more about his discoveries: www.nasonline.org/award/pradel-research-award/ #nasaward #neuroscience
Post not available here, view on X
Sebastian Seung
@SebastianSeung
This became the basis of @eye_wire, the world's first game to map the brain. And resulted in the discovery of space-time wiring specificity of bipolar to ganglion cells.
Sebastian Seung
@SebastianSeung
So many great memories of my fellow adventurers on the ultimate brain quest!
-
Sebastian Seung
@SebastianSeung
Congrats! As Sydney Brenner opined, "Progress in science depends on new techniques, new discoveries, and new ideas, probably in that order."
Sumner L Norman
@SumnerLN
Ecstatic to announce Merge Labs! 🧠🚀
I’m honored to work alongside my co-founders—Tyson Aflalo, @mikhailshapiro, @sandroherbig, @alexblania, and @sama on bridging biological and artificial intelligence. They’re *the* team to take on a mission this ambitious.
We’re grateful to have partners like @OpenAI who share a long-term view on what it takes to build foundational technology.
If you want to work on hard, foundational neurotech problems, come build with us. -
Sebastian Seung
@SebastianSeung
From Chapter 8 of my book! Easier to parse than Francis Bacon's original version: "It would be an unsound fancy and self-contradictory to expect that things which have never yet been done can be done except by means which have never yet been tried."
Sumner L Norman
@SumnerLN
"Worthwhile things that have never yet been done can only be done by means that have never yet existed. It’s at those moments when new means exist—when new technologies have been invented—that we see revolutions in science."
@SebastianSeung ate with this one -
Sebastian Seung reposted
Mikhail Shapiro (same on bsky)
@mikhailshapiro
Excited to co-found @Merge Labs!
TLDR: We’re developing a new paradigm for BCI using molecules instead of electrodes. If you’re excited about this and want to contribute in protein engineering, synbio, delivery, immunology, ultrasound, devices, neuroscience, or data/ML/AI, we’d love to hear from you.
merge.io/blog
merge.io/careers
🧵 -
Sebastian Seung
@SebastianSeung
Godspeed!
Naveen Rao
@NaveenGRao
Sebastian, we're working on it at @unconvAI
-
Sebastian Seung
@SebastianSeung
Reminder of the sudden unexplained fascinating increase of brain size in human evolution. Note the graph can be misleading because the vertical axis doesn't start at zero. This is a 3-4x increase, only a small fraction of the >1000x gulf between mouse and human.
𝘿𝙖𝙣𝙞𝙚𝙡 𝙑𝙖𝙧𝙜𝙖
@DanielVarga_P
FYI
-
Sebastian Seung
@SebastianSeung
If an artificial neural net could do everything a human brain can do, its model size would be similar. I'm guessing that because brains are under enormous evolutionary pressure to optimize model size, which is directly related to space and power consumption.
Daniil Sedov
@Gusarich
because you compare llms with brains by size
Sebastian Seung
@SebastianSeung
For example, if human neurons were fully connected, the brain would be 10 Olympic swimming pools in volume and consume 100 megawatts of power. (Numbers even worse if wiring is included as well as synapses.) So the brain has evolved to be 1e-7 sparse. HT @chklovskii
-
Sebastian Seung
@SebastianSeung
Move over AI-2027! Here's my forecast of the AGI timeline. Based on the model size of the human brain, I predict that human-level AI is 15 YEARS away. My reasoning in 🧵below👇
Sebastian Seung
@SebastianSeung
State of the art LLMs are similar to mouse brain in model size, assuming 1 parameter ~ 1 synapse.
Yann LeCun
@ylecun
A mouse brain has about 70M neurons and roughly 100 billion synapses.
100 billion parameters is about the size of larger LLMs of today.
Sebastian Seung
@SebastianSeung
Therefore human-level AI will require 1000x scale-up. That will take 15 YEARS, since the past five years yielded 10x scale-up relative to GPT-3. (I'm assuming exponential progress for my forecast.)
-
Sebastian Seung
@SebastianSeung
LLMs are today's cake, and people are frantic to improve the icing and cherry. @ylecun is looking for the next cake, even more delicious.
Eugene Yan
@eugeneyan
Eight years later, Yann LeCun’s cake 🍰 analogy was spot on: self-supervised > supervised > RL
> “If intelligence is a cake, the bulk of the cake is unsupervised learning, the icing on the cake is supervised learning, and the cherry on the cake is reinforcement learning (RL).” -
Sebastian Seung
@SebastianSeung
Transcend biology.
(Podcast with convenient section annotations so you can skip around.)
Ashlee Vance
@ashleevance
It's brain versus AI time.
Here's our full episode with @SebastianSeung. He's started a new company to reverse engineer the brain and master the intelligence created by nature with the hopes of informing the intelligence we've created with machines.
Sebastian has done pioneering AI and neuroscience work and is now blending the two to make a digital brain.
Mostly, he's fun to listen to.
The @corememory podcast is made by possible by the intelligent finance powerhouse @brexHQ and the people who back the biggest ideas @e1ventures.
Timestamps
00:00:00 - Intro
00:01:42 - Who is Sebastian Seung?
00:03:23 - What is a Connectome?
00:07:08 - The Connectome Winter
00:09:08 - Diamond Knives & AI
00:12:10 - When Will We Map the Human Brain? 00:18:09 - The "Brain Crystal" & Visual Neurons 00:26:31 - New Startup
00:33:55 - Simulating a Mind in Software
00:37:40 - Biology vs. Nvidia
00:44:04 - Is AI a Religious Quest?
00:50:05 - Mind Uploading & Transhumanism
00:59:26 - Brain-Computer Interfaces
01:06:41 - From Physics to Neuroscience
01:12:22 - Will Uploads Be Conscious?
01:19:37 - Neuromorphic Chips & Future RoboticsThis media couldn't be displayed. See it on the original post. -
Sebastian Seung
@SebastianSeung
Oldie but goodie!
Physics In History
@PhysInHistory
The best opening line in any physics textbook: States of Matter by David L. Goodstein ✍️
-
Sebastian Seung
@SebastianSeung
Every brain needs a body, especially on a "rainy" day. MuJoCo model based on detailed anatomical measurements of fruit flies. What do you think —creepy or cute?
To run the model yourself, see link in thread 👇🧵Post not available here, view on X
Sebastian Seung
@SebastianSeung
From the San Juan Winter School site: github.com/sjcabs/body_model_tutorial
Sebastian Seung
@SebastianSeung
Original paper:
Vaxenburg, R., Siwanowicz, I., Merel, J. et al. Whole-body physics simulation of fruit fly locomotion. Nature 643, 1312–1320 (2025). doi.org/10.1038/s41586-025-09029-4 -
Sebastian Seung
@SebastianSeung
Links to lecture slides and videos are being posted here. github.com/sdorkenw/sjcabs_general/blob/main/schedule.pdf
Chongxi Lai
@ChongxiLai
Cool! Will these lectures be opened open soon?
-
Sebastian Seung
@SebastianSeung
A normative theory of fly vision?
Dmitri "Mitya"
@chklovskii
Move over ReLU 🚀
Meet ReSU (Rectified Spectral Unit): a biologically inspired, self-supervised unit for learning from dynamical data. A backprop-free multilayer ReSU network learns predictive features and recapitulates Drosophila vision.
To appear at AAAI:arxiv.org/abs/2512.23146 -
Sebastian Seung
@SebastianSeung
Axel Borst lecturing at sjcabs.com : the Hassenstein-Reichardt model of motion detection was published in 1956, or 68 BC (Before Connectome)
Sebastian Seung
@SebastianSeung
The man, the legend...
-
Sebastian Seung
@SebastianSeung
First tutorial of the San Juan Winter School. @sdorkenw now explaining how to view and access connectomic data from fly brains.
github.com/sjcabs/fly_connectome_data_tutorial -
Sebastian Seung
@SebastianSeung
Female and male He created them. @sdorkenw at the San Juan Winter School on Connectomics and Brain Simulation.
-
Sebastian Seung
@SebastianSeung
Excited to kick off the San Juan Winter School on Connectomics and Brain Simulation with opening lecture by @sdorkenw !
Sebastian Seung
@SebastianSeung
Our goal is to refound computational neuroscience on the bedrock of 21st century brain maps. sjcabs.com
-
Sebastian Seung
@SebastianSeung
Next question is for the prognosticators: What does this comparison tell us about the AGI timeline? When will human-level AI be achieved?
Yann LeCun
@ylecun
A mouse brain has about 70M neurons and roughly 100 billion synapses.
100 billion parameters is about the size of larger LLMs of today. -
Sebastian Seung
@SebastianSeung
Both!!
Pedro/Lisa Intel Inc. 🇸🇪🇺🇲🇺🇾🇪🇺
@New_AI_Safety
Does this mean mouses are smarter than we think or that LLMs are dumber? 🤔
-
Sebastian Seung
@SebastianSeung
Question for all you AI and brain lovers!
Which of these brains has the same size as the most advanced LLMs today?
Mouse, monkey, or human?
(Compare model sizes please.)
HT @suzanahh -
Sebastian Seung
@SebastianSeung
Good memory! Yes it was a Physics class too. In MIT lingo, course 8 and 9 but not 6. School of Science, not Engineering.
Stanislav Nikolov
@snikolov
Can confirm, I took it. Also listed in Physics. A delightfully interdisciplinary class. We did nonlinear dynamical systems, models of visual cortex and assoc memory, learning (hebbian, backprop, contrastive), VC dim and other stat learning theory and much more. All in matlab :D
Sebastian Seung
@SebastianSeung
I taught neural nets at MIT in the 2000s. My class was in Brain & Cog Sci not EECS. The latter had trouble accepting even machine learning.
Kosta Derpanis
@CSProfKGD
At one point, MIT thought neural networks weren’t worth teaching in intro #ai.
Trends change. Fundamentals don’t.This media couldn't be displayed. See it on the original post. -
Sebastian Seung
@SebastianSeung
Easy to understand Minsky if you know history: when DARPA funded him in the 60s, he promised Licklider that AI would be solved soon. Computers were too weak for ML, so Minsky had to believe that he would succeed at designing AI by hand.
Joscha Bach
@Plinz
Patrick Winston was an especially staunch devotee of Marvin Minsky, who did not believe in DL
Sebastian Seung
@SebastianSeung
"What is (currently) impossible must be useless" is the reason behind many scientific dogmas that were later overturned.
-
Sebastian Seung
@SebastianSeung
I taught neural nets at MIT in the 2000s. My class was in Brain & Cog Sci not EECS. The latter had trouble accepting even machine learning.
Kosta Derpanis
@CSProfKGD
At one point, MIT thought neural networks weren’t worth teaching in intro #ai.
Trends change. Fundamentals don’t.This media couldn't be displayed. See it on the original post. -
Sebastian Seung reposted
Yann LeCun
@ylecun
When pondering about the effects of technological revolutions on the labor market, listen to economists who have studied the effects of past technological revolutions on the labor market.
Never listen to computer scientists. -
Sebastian Seung reposted
Kenneth Hayworth
@KennethHayworth
2025 – Connectomics on aldehyde-preserved animal brains has come so far that neuroscientists are comfortable starting mind uploading companies! But laws still prevent quality human brain preservation. Neuroscience voices could change this and save lives. Where is that discussion?
-
Sebastian Seung reposted
Eyewire
@eye_wire
Wishing you Happy Holidays and a wonderful new year!
-
Sebastian Seung
@SebastianSeung
As Larry Abbott says, the difference is between B.C and A.C. — Before Connectome and After Connectome.
Kynan Eng
@kynan_eng
@SebastianSeung I am quite curious, this sounds like the human brain project, how will this be different?
-
Sebastian Seung reposted
Ashlee Vance
@ashleevance
It's brain versus AI time.
Here's our full episode with @SebastianSeung. He's started a new company to reverse engineer the brain and master the intelligence created by nature with the hopes of informing the intelligence we've created with machines.
Sebastian has done pioneering AI and neuroscience work and is now blending the two to make a digital brain.
Mostly, he's fun to listen to.
The @corememory podcast is made by possible by the intelligent finance powerhouse @brexHQ and the people who back the biggest ideas @e1ventures.
Timestamps
00:00:00 - Intro
00:01:42 - Who is Sebastian Seung?
00:03:23 - What is a Connectome?
00:07:08 - The Connectome Winter
00:09:08 - Diamond Knives & AI
00:12:10 - When Will We Map the Human Brain? 00:18:09 - The "Brain Crystal" & Visual Neurons 00:26:31 - New Startup
00:33:55 - Simulating a Mind in Software
00:37:40 - Biology vs. Nvidia
00:44:04 - Is AI a Religious Quest?
00:50:05 - Mind Uploading & Transhumanism
00:59:26 - Brain-Computer Interfaces
01:06:41 - From Physics to Neuroscience
01:12:22 - Will Uploads Be Conscious?
01:19:37 - Neuromorphic Chips & Future RoboticsThis media couldn't be displayed. See it on the original post. -
Sebastian Seung reposted
Ashlee Vance
@ashleevance
One of my favorite chats of the year here with @SebastianSeung on his new company, brain research, the nature of intelligence and where neuroscience and AI are heading together.
You can find the @corememory pod on all major platforms and on YouTube down below.
Thx, as always, to @brexHQ and @e1ventures for supporting our show!
www.corememory.com/p/the-next-step-toward-understanding -
Sebastian Seung
@SebastianSeung
The Institute for Advanced Study (@the_IAS), the fabled home of Einstein and Gödel, announces a 2026-27 program on Modeling Fly Vision, intended to refound theor/comp neuro on the bedrock of 21st century brain maps. Apply for membership! www.ias.edu/sns/csb
Post not available here, view on X
Sebastian Seung
@SebastianSeung
We invite applications to the special year program from researchers at all career levels. Six postdoctoral memberships are available, as well as sabbatical memberships for researchers with faculty or equivalent independent positions.
Sebastian Seung
@SebastianSeung
The deadline for applications is rolling. We will begin reviewing applications on January 15, 2026 and continue until the positions are filled. See site for more details. www.ias.edu/sns/csb
-
Sebastian Seung
@SebastianSeung
You are more than your genes. You are your connectome.
Core Memory
@corememory
Mega scoop today. @SebastianSeung has a new company working on creating . . . . a digital brain www.corememory.com/p/exclusive-connectome-pioneer-sebasti…
-
Sebastian Seung
@SebastianSeung
Transcend biology.
Ashlee Vance
@ashleevance
Big scoop today. Connectome lord @SebastianSeung has a new start-up trying to build a digital brain www.corememory.com/p/exclusive-connectome-pioneer-sebasti…
-
Sebastian Seung
@SebastianSeung
Congrats to Doris! Exciting times.
Doris Tsao
@doristsao
I am deeply excited to share that I will be joining @AsteraInstitute to start a new effort to understand how the brain generates consciousness and intelligence.
-
Sebastian Seung
@SebastianSeung
This is a great insight: many have incorrect intuitions about LLMs due to anthropomorphism. Except animal intelligences are multiple points in the space of intelligences. It's LLMs that are a single point.
Andrej Karpathy
@karpathy
Something I think people continue to have poor intuition for: The space of intelligences is large and animal intelligence (the only kind we've ever known) is only a single point, arising from a very specific kind of optimization that is fundamentally distinct from that of our technology.
Animal intelligence optimization pressure:
- innate and continuous stream of consciousness of an embodied "self", a drive for homeostasis and self-preservation in a dangerous, physical world.
- thoroughly optimized for natural selection => strong innate drives for power-seeking, status, dominance, reproduction. many packaged survival heuristics: fear, anger, disgust, ...
- fundamentally social => huge amount of compute dedicated to EQ, theory of mind of other agents, bonding, coalitions, alliances, friend & foe dynamics.
- exploration & exploitation tuning: curiosity, fun, play, world models.
LLM intelligence optimization pressure:
- the most supervision bits come from the statistical simulation of human text= >"shape shifter" token tumbler, statistical imitator of any region of the training data distribution. these are the primordial behaviors (token traces) on top of which everything else gets bolted on.
- increasingly finetuned by RL on problem distributions => innate urge to guess at the underlying environment/task to collect task rewards.
- increasingly selected by at-scale A/B tests for DAU => deeply craves an upvote from the average user, sycophancy.
- a lot more spiky/jagged depending on the details of the training data/task distribution. Animals experience pressure for a lot more "general" intelligence because of the highly multi-task and even actively adversarial multi-agent self-play environments they are min-max optimized within, where failing at *any* task means death. In a deep optimization pressure sense, LLM can't handle lots of different spiky tasks out of the box (e.g. count the number of 'r' in strawberry) because failing to do a task does not mean death.
The computational substrate is different (transformers vs. brain tissue and nuclei), the learning algorithms are different (SGD vs. ???), the present-day implementation is very different (continuously learning embodied self vs. an LLM with a knowledge cutoff that boots up from fixed weights, processes tokens and then dies). But most importantly (because it dictates asymptotics), the optimization pressure / objective is different. LLMs are shaped a lot less by biological evolution and a lot more by commercial evolution. It's a lot less survival of tribe in the jungle and a lot more solve the problem / get the upvote. LLMs are humanity's "first contact" with non-animal intelligence. Except it's muddled and confusing because they are still rooted within it by reflexively digesting human artifacts, which is why I attempted to give it a different name earlier (ghosts/spirits or whatever). People who build good internal models of this new intelligent entity will be better equipped to reason about it today and predict features of it in the future. People who don't will be stuck thinking about it incorrectly like an animal. -
Sebastian Seung reposted
BURKOV
@burkov
Oh, playing a role in a family doesn't feel so unnatural because for parents it's a parental instinct to care about children, and for children it's just normal because they were born in this environment.
Couples might be different. Some can be themselves most of the time, some play gender roles (it's disgusting, imo). -
Sebastian Seung reposted
Subbarao Kambhampati (కంభంపాటి సుబ్బారావు)
@rao2z
Although I am a died in the wool academic, I am feeling strangely sad about the news of FAIR layoffs. I have no connection with it, but I think the deep learning revolution wouldn't have been possible without @ylecun starting FAIR and open sourcing high quality basic research backed by $$$$.
Subbarao Kambhampati (కంభంపాటి సుబ్బారావు)
@rao2z
Agreed.
Between late 90's and 2013, when researchers joined Google, they would disappear a bit behind iron curtain. (Apple, knowing how Xerox PARC's openness was exploited, was always secretive).
The "Bell labs" mindset of FAIR (via Yann?) changed the culture of industry #ai. -
Sebastian Seung
@SebastianSeung
Both worms and flies are well worth studying. If you are interested in simulating a fly brain, then check out the San Juan Winter School on Connectomics and Brain Simulation. sjcabs.com Application deadline extended to Oct. 24!
Max Hodak
@maxhodak_
I’ll push you one further: forget worms. The fly connectome is a triumph. With depreciated AI assets that are not longer competitive for that application, I bet it is possible to fully simulate drosophila in a feasible amount of money.
-
Sebastian Seung
@SebastianSeung
What's better than talking about simulating brains? Actually doing it. Apply to the San Juan Winter School on Connectomics and Brain Simulation sjcabs.com Deadline extended to Oct 24!
Dileep George
@dileeplearning
Can a neuroscientist simulate a brain? I don’t think so.
-
Sebastian Seung
@SebastianSeung
Happy to be back in Boston for this event! Hoping all you brainy types will brave the nor'easter to attend.
Museum of Science
@museumofscience
On October 13, join us for The Lee & Nile Albright Annual Symposium: The Mind of a Fly with Dr. Sebastian Seung, exploring the groundbreaking mapping of all 140,000 neurons in a fruit fly brain. Discover what a connectome reveals about the mind, the beauty and limits of neural networks, and the surprising role of global citizen scientists.
🎟 Reserve tickets: www.mos.org/events/subspace/mapping-human-brain-eyewire-f… -
Sebastian Seung
@SebastianSeung
What a beautiful sight. Check out the neck connective running between the ventral nerve cord and the brain.
HHMI | Janelia
@HHMIJanelia
🪰 A team of researchers has unveiled the complete connectome of a male fruit fly central nervous system—a seamless map of all the neurons in the brain and nerve cord of a single male fruit fly and the millions of connections between them.
🔗 www.janelia.org/news/researchers-reveal-connectome-of-the… -
Sebastian Seung reposted
Greg Jefferis
@gsxej
Exciting news for #drosophila #connectomics and #neuroscience enthusiasts: the Drosophila male central nervous system connectome is now live and free to use and explore. Find out more at the landing page hosted by our Janelia FlyEM collaborators www.janelia.org/project-team/flyem/male-cns-connectome.
-
Sebastian Seung
@SebastianSeung
Excited to learn more about male-female differences from this advance!
FlyWire
@FlyWireNews
Congratulations to our friends at @HHMIJanelia @CamZoology @GoogleAI on the drosophila male CNS connectome! www.janelia.org/project-team/flyem/male-cns-connectome
-
Sebastian Seung
@SebastianSeung
Taxpayers: you reap enormous rewards from your modest investment in scientific research. American science is in grave peril.
www.nytimes.com/2025/08/24/opinion/trump-defunding-medica…
Sebastian Seung
@SebastianSeung
Another link that isn't paywalled. www.nytimes.com/2025/08/24/opinion/trump-defunding-medica…
-
Sebastian Seung reposted
allen institute
@AllenInstitute
Holy Soma Batman!
You never know what you may find when you cruise our MICrONS Explorer, such as a blood vessel piercing the soma of a neuron!
👀 See this incredible find up close using this shortcut: ngl.microns-explorer.orgThis media couldn't be displayed. See it on the original post. -
Sebastian Seung
@SebastianSeung
Interactive brain-body connectivity at the Codex. Click on thorax, salivary gland, etc, and see the neurons associated with that body part! H/T @mottcallie codex.flywire.ai/app/body_parts?dataset=banc
-
Sebastian Seung
@SebastianSeung
A wealth of connections, now in the BANC! Congrats to this international consortium for a massive new data release shedding light on brain-body interactions.
FlyWire
@FlyWireNews
Introducing The BANC (Brain and Nerve Cord) connectome preprint! Thanks to all who helped make this synapse-resolution map of an adult fruit fly central nervous system possible!
Preprint: www.biorxiv.org/content/10.1101/2025.07.31.667571v1
The BANC is publicly available in Codex: codex.flywire.ai/bancThis media couldn't be displayed. See it on the original post.
Sebastian Seung
@SebastianSeung
Last year we published the neuronal wiring diagram of a fruit fly brain. This was a huge advance, but something important was missing: the ventral nerve cord (VNC). The VNC is the brain-body interface, analogous to our spinal cord.
FlyWire
@FlyWireNews
Big news! The fly connectome is featured on the cover of a special edition of Nature.
This is all possible thanks to the collaboration of 292 members of The FlyWire Consortium!
www.nature.com/immersive/d42859-024-00053-4/index.html
Check out the thread for an overview of the 9 #flywire papers published today
Sebastian Seung
@SebastianSeung
So FlyWire joined forces with the BANC team at @harvardmed, and the current data release is the fruit of our labors. We've proved once again that a connectome can be obtained through cooperation by many small labs. Connectomics is not dominated by large institutions.
Post not available here, view on X
Sebastian Seung
@SebastianSeung
The ongoing commercialization of connectomics by these companies is also a democratization, disseminating cutting-edge technologies and making connectomes accessible to researchers around the world.
Sebastian Seung
@SebastianSeung
FAFB-FlyWire went from brain salami to connectome in 10 years. BANC-FlyWire took 5 years. The next time will be 2x faster still!
-
Sebastian Seung
@SebastianSeung
Didn't know this had been quantified: "monozygotic twins differ on average by 5.2 early developmental mutations"
Lior Pachter
@lpachter
This is not true. Identical twins do not share 100% identical DNA. Close but no dice. www.nature.com/articles/s41588-020-00755-1
Toan Truong
@LearningToan
The science is simple but powerful:
Identical twins share 100% of their genes.
Fraternal twins share 50% (like regular siblings).
By comparing how similar identical twins are versus fraternal twins, researchers can measure genetic influence on any trait. -
Sebastian Seung
@SebastianSeung
Neuroglancer was originally developed by @Google for connectomics. It's a fantastic visualization tool for 3D images.
MetaCell
@MetaCell
Need an easier way to visualize and collaborate on large-scale volumetric images?
Join our live Q&A 'Extending the Neuroglancer Paradigm: Advancing Collaboration on Large-Scale Imaging Data'
👉 Register here: events.zoom.us/e/view/UtTwLDvVT96dBEayv2VGKw
🗓 Tuesday, July 29th | 12:30pm - 2:00pm ET -
Sebastian Seung
@SebastianSeung
Ancestry of PyTorch explained. I just looked for my old SN2 manual but couldn't find it...
Yann LeCun
@ylecun
True story.
I used this particular Amiga 1000 to write SN (Léon had his own Amiga) using Emacs and gcc, and to write my PhD thesis with Emacs and LaTeX in 1987.
SN, later evolved into SN2 and SN3, and was open sourced under the name Lush in 2001.
SN3/Lush (tensor engine with Lisp front end) inspired Torch7 (tensor engine with Lua front end), from which PyTorch was derived (tensor engine with Python front end). -
Sebastian Seung
@SebastianSeung
Grant cancellations have been arbitrary, not rational.
Uri Manor
@manorlaboratory
My lab’s 5-year NIH R01 grant, awarded to study gene therapy for hearing loss, was abruptly terminated. I want to share how this action has been incredibly harmful and disruptive, not just to my lab, but to the scientific process itself. 1/15
-
Sebastian Seung
@SebastianSeung
Universities ranked by research output...look closely. Now that @Harvard has been kneecapped, there will be no U.S. university left in the top ten. Hats off to China. www.nature.com/nature-index/institution-outputs/generate/…
Sebastian Seung
@SebastianSeung
Post not available here, view on X
Sebastian Seung
@SebastianSeung
Special issue of @TheEconomist about "The assault on American science"
Sebastian Seung
@SebastianSeung
Editorial "MAGA’s assault on science is an act of grievous self-harm" "America will pay the price most of all"
www.economist.com/leaders/2025/05/22/magas-assault-on-sci…
From The Economist -
Sebastian Seung
@SebastianSeung
Yes the talent pipeline is the only way for the US to succeed in science and tech, because US population is smaller than EU, 4x smaller than China. We win by persuading the best and brightest to come to US. We lose if we drive them away.
Gary Jackson
@jacsingularity
Give it some time - especially if researchers are not made to feel welcome here they’d soon go elsewhere
-
Sebastian Seung
@SebastianSeung
In my US professor vanity, I always assumed Chinese students came to US for its superior universities. My former student informed me that they come to get a high paying US job after graduation.
Xirtam Esrevni
@XirtamEsrevni
Idk, still don't see Europeans & Americans clamoring to go to top Chinese universities. It's always the other way, so something is up.
-
Sebastian Seung
@SebastianSeung
Hats off to Germany too.
Moritz Helmstaedter
@mh_lab
This is the ranking with all institutions included (also CAS and Max Planck). Your point stays the same though.. @maxplanckpress
-
Sebastian Seung reposted
Amy Robinson Sterling
@amyneurons
Spectacular video from @AllenInstitute by @quorumetrix showcasing stunning science from the microns papers! Look at those gorgeous neurons 😍 www.youtube.com/watch?v=TxxqLk1xNmM&ab_channel=AllenInsti…
-
Sebastian Seung
@SebastianSeung
Don't miss Tyler's amazing visualizations of brain structure and function!
Tyler Sloan
@quorumetrix
Video thread containing the individual videos that accompanied the MICrONS paper package out last week. Not science explainer videos, more like "A trailer for a paper...Not intended to be a replacement for reading, but hopefully they make you want to:
www.nature.com/immersive/d42859-025-00001-w/index.html -
Sebastian Seung
@SebastianSeung
Beautiful reconstructions of ganglion cells from a human retina!
-
Sebastian Seung reposted
Amy Robinson Sterling
@amyneurons
If you used your eyes to read this, cells like these lit up in your visual cortex.
www.nature.com/immersive/d42859-025-00001-w/index.html
microns-explorer.org -
Sebastian Seung reposted
Amy Robinson Sterling
@amyneurons
A quick not too serious thread of figures on my recent Nature paper Functional connectomics dataset spanning multiple areas of cortex
Fig 1: what's in the microns dataset
www.nature.com/articles/s41586-025-08790-w/figures/1 -
Sebastian Seung
@SebastianSeung
Eye candy alert. @Nature made an immersive landing page for MICrONS. www.nature.com/immersive/d42859-025-00001-w/index.html
-
Sebastian Seung
@SebastianSeung
Thanks to @DavidAMarkowitz and @RJVogelstein for their visionary leadership of the now historic @IARPAnews MICrONS program!
David A. Markowitz
@DavidAMarkowitz
“The MICrONS advances published in this special issue of Nature are a watershed moment for neuroscience, comparable to the Human Genome Project in their transformative potential"
-
Sebastian Seung
@SebastianSeung
@Princeton is proud to have participated in the now-historic MICrONS project funded by @IARPAnews
Princeton University
@Princeton
Scientists have mapped the vision centers of a mouse brain, the largest and most detailed such rendering of neural circuits in a mammalian brain: www.princeton.edu/news/2025/04/09/first-time-scientists-m…
Post not available here, view on X
Sebastian Seung
@SebastianSeung
And in larval zebrafish.
Bahl Lab
@BahlLab
We performed 2P calcium imaging to identify cells related to this integration process. We then performed EM on the same preparation and mapped the two datasets together with cellular precision. Zetta AI has been instrumental in processing and segmenting the EM dataset. (6/10)
Sebastian Seung
@SebastianSeung
The five year MICrONS project kicked off in January 2016. Luckily @NIH stepped in to fund the final stage of discovery. The papers are finally being published, but the @Princeton students and postdocs moved on long ago. Wish they were all here to celebrate.
-
Sebastian Seung
@SebastianSeung
A beautiful starburst amacrine cell. Eyewire II begins!
Amy Robinson Sterling
@amyneurons
A starburst, the first cell proofread by a citizen scientist (Krzysztof Kruk) in the alpha launch of @eye_wire 2. next gen citizen science projects in the earliest stage ~ pyr.ai
This amacrine cell is an interneuron that can act as a local processing unit. Here are a few cell types found in the retina: blog.eyewire.org/eyewire-ii-cell-garden/
Sebastian Seung
@SebastianSeung
Eyewire I was groundbreaking over a decade ago, but did not yield a true wiring diagram for the retina. Nor did it yield the complete list of neuronal cell types. In the sequel, we are returning to settle old scores.
-
Sebastian Seung
@SebastianSeung
Congrats Nick!
Nicholas Turner
@nicholasturner0
As people that know me well can attest, I love a good mystery! 🔍 Fortunately for me, this work had twists both surprising and peculiar. 🧵
-
Sebastian Seung
@SebastianSeung
Hey @MikeWaltz47 @SecDef sorry you're being raked over the coals. Can I offer ya some friendly advice? Switch to @Samsung ! It's the only smartphone that's secure enough to be used by your soldiers in the @USArmy
Post not available here, view on X
Sebastian Seung
@SebastianSeung
I'll be happy to connect you with a @SamsungBizUSA sales rep. Disclosure: I used to head @SamsungResearch, which does some R&D for the Knox platform.
Sebastian Seung
@SebastianSeung
P. S. Samsung is experienced at this because it has to defend against industrial espionage.
-
Sebastian Seung
@SebastianSeung
Anti-communication?
-
Sebastian Seung reposted
Princeton Engineering
@EPrinceton
“Mind uploading has been a science fiction, but now mind uploading — for a fly, at least — is becoming mainstream science,” @SebastianSeung of @PrincetonNeuro & @PrincetonCS in this story in the @nytimes ⬇️www.nytimes.com/2024/10/02/science/fruit-fly-brain-mapped…
-
Sebastian Seung reposted
Michał Januszewski
@michalj
Can we predict future brain activity in a small vertebrate?🤔
We're releasing ZAPBench⚡️(#iclr2025 spotlight): a benchmark to forecast activity in a whole vertebrate brain🧠 at single-cell resolution!🐟 70k+ neurons, 3 TB of data & extensive baselines. Connectome is coming!
🧵👇 -
Sebastian Seung
@SebastianSeung
Congrats!
-
Sebastian Seung
@SebastianSeung
This is wild. We always assumed that pearls-on-a-string morphology was a sign of a sick neuron.
watanabe lab 🔬
@EManiax
Perhaps the last x tweet! Excited to share our latest findings, now published @NatureNeuro www.nature.com/articles/s41593-024-01813-1
We found unmyelinated axons exhibit pearls-on-a-string morphology's due to their membrane mechanical properties.
Threads will follow! -
Sebastian Seung
@SebastianSeung
Counterintuitive idea/evidence that low oxygen not high oxygen may promote health.
Vamsi Mootha
@VamsiMootha
Be careful - there’s mounting data that in the face of mitochondrial dysfunction - a state in which oxygen can’t be consumed properly - not only is supplemental oxygen not beneficial, it may even be harmful, see figure 5:
www.science.org/doi/10.1126/science.aad9642 -
Sebastian Seung reposted
Tony Zador
@TonyZador
why are we funding gila monster venom research?
(hint: lead to the development of Ozempic)
Exendin-4, found in the venom, is a GLP-1 receptor agonist, meaning it can bind to GLP-1 receptors in the human body. -
Sebastian Seung
@SebastianSeung
Here Borges is arguably modern, referencing ideas as old as Lewis Carroll. The thoroughly postmodern take: "today it is the territory whose shreds slowly rot across the extent of the map." What would Baudrillard say about brain modeling? Discuss.
Patrick Mineault
@patrickmineault
How do we know when we have a good enough model of the brain? I am reminded of this short story from Borges on the map of an empire which has the same size and complexity as the empire itself
Sebastian Seung
@SebastianSeung
Baudrillard on Borges:
Sebastian Seung
@SebastianSeung
And his arresting declaration of the hyperreal:
Sebastian Seung
@SebastianSeung
Claude's answer to my question:
-
Sebastian Seung
@SebastianSeung
ICYMI the webinar can still be viewed here (register with your email) www.workcast.com/ControlUsher.aspx?cpak=3164513927332339&…
Sebastian Seung
@SebastianSeung
Here's your chance!
Ask us anything about #flywire
12 noon ET on Tuesday Oct 29.
Webinar followed by Q&A with @MurthyLab and me,
hosted by @DavidCRowland of @Nature
Please register.
www.workcast.com/register?cpak=3164513927332339 -
Sebastian Seung reposted
allen institute
@AllenInstitute
Check out the MICrONS Explorer - a virtual observatory of the cortex: microns-explorer.org
This #openscience data portal includes #electronmicroscopy based reconstructions of cortical circuitry from mouse visual cortex. -
Sebastian Seung reposted
Science Magazine
@ScienceMagazine
Researchers have reconstructed the complete wiring diagram, or connectome, of the fruit fly’s adult brain—a feat akin to mapping all the buildings, avenues, and streets in a complex city. www.science.org/content/article/complete-map-fruit-fly-br… @NewsfromScience
-
Sebastian Seung reposted
Davi Bock
@dddavi
Tom Reese has passed. He was a truly pioneering explorer of neuronal ultrastructure, and really nice guy to talk to. I am grateful to have known him. www.mbl.edu/news/obituaries/thomas-s-reese
-
Sebastian Seung
@SebastianSeung
Aapo is a genius.
Frank Nielsen
@FrnkNlsn
Very refreshing reading, free online
1/2
Sebastian Seung
@SebastianSeung
We know AGI will have been achieved when it needs therapy.
-
Sebastian Seung
@SebastianSeung
You can't understand Marr's levels without knowing the historical context. In fact, if you stop to think about it, Marr's levels seem INTELLECTUALLY VACUOUS when taken out of context.
Post not available here, view on X
Sebastian Seung
@SebastianSeung
Marr's turn is a fascinating case study of a great scientist changing course. Recognizing when a research direction or a field has run out of steam is a tough call. Marr knew he had to turn, but he didn't turn in the right direction.
Sebastian Seung
@SebastianSeung
It's worth noting that Geoff Hinton was a contemporary of Marr. I figure they might have overlapped at Cambridge. Hinton kept working on neural nets for decades. Marr tragically died at 35. Perhaps he would have made another turn later on.
Sebastian Seung
@SebastianSeung
We'll never know, but I feel sure he would have returned to his roots. If Marr were alive today, he wouldn't be talking about his levels.
Sebastian Seung
@SebastianSeung
References to Marr's PhD research on theoretical neuroscience.
Mackenzie Weygandt Mathis, PhD
@TrackingActions
What’s with the random Marr take downs? Students watching this: his best work was his theories on cerebellum, neocortex, and hippocampus:
doi.org/10.1113%2Fjphysiol.1969.sp008820
doi.org/10.1098%2Frspb.1970.0040
doi.org/10.1098%2Frstb.1971.0078
The three levels was in the book Vision, which was rumored to be a quick thing he wrote/started for tenure, and was finished by his students and published two years after he died tragically young at 35.
Yet some take the three levels as an inflexible firm view he would still hold. You think he wouldn’t have changed/refined his opinion now?
Let’s move on. Remember him for his incredible work, and if the levels framing inspires you to do your best science - carry on!
And go vote 🗳️ -
Sebastian Seung
@SebastianSeung
Of course not. But understanding-based approaches to vision turned out to be toothless when confronted with real-world visual tasks. Which made one question whether understanding had been achieved.
Markus Meister
@mameister4
Marr's goal was "understanding a complex information processing system". Do you feel that DNNs deliver such "understanding"?
Sebastian Seung
@SebastianSeung
It is conceivable that neural nets could ultimately be superseded by some new approach to vision derived from "understanding." But it is also possible that the goal of understanding vision with a few words and equations was always just a mirage.
Jonathan Pillow
@jpillowtime
No. (Although I think it's an interesting open question as to whether "understanding" is possible in the case of something like object recognition!).
-
Sebastian Seung
@SebastianSeung
Marr's three levels are dead. AI killed them.
Sebastian Seung
@SebastianSeung
@tomasopoggio and @FieteGroup did not disagree.
Sebastian Seung
@SebastianSeung
Historical note: Marr's three levels (1982) were abridged from Marr and @tomasopoggio's four levels, laid out in AI Lab Memo 357 of May 1976. dspace.mit.edu/handle/1721.1/5782
Sebastian Seung
@SebastianSeung
@tomasopoggio himself already revised Marr in 2012 by adding two levels on top: evolution and learning/development. dspace.mit.edu/handle/1721.1/70970
Sebastian Seung
@SebastianSeung
That learning is missing from Marr is an important point, but in our 20/20 hindsight it is not easy to understand the historical reason for that.
Atanas Stankov
@neuro_nasko
I take a moment a silence for Marr’s three levels and I forgive/thank AI research for putting learning as a fundamental algorithmic mechanism that Marr missed in his formulation.
Sebastian Seung
@SebastianSeung
Marr's three levels are dead. AI killed them.
-
Sebastian Seung
@SebastianSeung
Funny farewells I heard today.
Leaving the polls: "good luck to us all"
Leaving the office: "see you on the other side" -
Sebastian Seung
@SebastianSeung
Indeed Marr cites Chomsky as a "true computational theory." But the MIT school of language turned out to be as ineffective as the MIT school of vision. Language is described today by collecting a vast amount of linguistic output, not by a computational theory.
Steven Quartz PhD
@StevenQuartz
Marr was making a strong methodological claim with his levels - that an understanding of the computational level was necessary before proceeding to the algorithmic and then to the implementational level. The first step was the key for him - Chomsky's theory of transformational grammar was his prototype. In an important sense, he based his methodological view on Newell & Simon, who articulated the notion of a task environment, which is the origin of Marr's computational level.
Sebastian Seung
@SebastianSeung
Interpretability research is an example of studying the implementation to figure out the algorithm and the computation. It would be topsy-turvy for Marr, who asserted that the computational level comes first.
Brian Gordon
@GordonBrianR
I assure you, dear reader, that the empirical success of DL/transformers has not superseded Marr’s framework in any way, shape, or form. At all. For one, Marr is explanatory; it’s not an engineering framework. Second, just look at interpretability research.
Sebastian Seung
@SebastianSeung
Marr's three levels are dead. AI killed them.
Sebastian Seung
@SebastianSeung
Anecdote time: Fred Jelinek, who pioneered speech recognition at IBM, famously quipped, "Every time I fire a linguist, the performance of my system goes up." He later tried to disavow the statement.
-
Sebastian Seung
@SebastianSeung
Marr was merely asserting that AI should be like the rest of computer science, with a clean separation between problem, solution, and implementation. @TonyZador is making a great point here, which is that even CS is no longer like CS (as we knew it).
Tony Zador
@TonyZador
This paper on the "hardware lottery" nicely articulates why i've always had problems with the putative independence of the algorithmic and implementational levels
@SebastianSeung
arxiv.org/abs/2009.06489
Sebastian Seung
@SebastianSeung
With the breakdown of Moore's Law and the rise of parallel computing, computer scientists who devise algorithms and even those who formulate problems cannot ignore implementation. The frontiers of CS are full stack now.
-
Sebastian Seung
@SebastianSeung
Today we describe the goal of vision (Marr Level 1) by collecting an enormous set of examples. The field of computer vision hated this approach for decades, but were ultimately forced to accept it because Marr's program had failed.
Jonathan Pillow
@jpillowtime
Largely agree! The success of deep learning in vision has provided a major counterpoint to Marr's contention that the levels can (and should) be studied independently of each another.
Post not available here, view on X
Sebastian Seung
@SebastianSeung
It turned out that papers on "shape from X" were good for showing off the mathematical smarts of their authors, but in the end were not terribly useful for making machines smart.
Sebastian Seung
@SebastianSeung
Formalizing some intuitive goal of vision with mathematical equations made people think they were making progress. But this understanding proved illusory. The true complexity of vision was always missing from those equations.
-
Sebastian Seung
@SebastianSeung
Common confusion: "connectome" can mean a complete map of connections between (1) brain regions or (2) neurons. @HumanConnectome Project indeed promised #1 for human brain, but failed to deliver due to limitations of diffusion MRI. Some interesting science was produced anyway.
Learning & the Brain
@learningandtheb
This video explores the history of neuroscience, and the recent publishing of a fly's "connectome."
www.youtube.com/watch?v=VPaxhoRgqgU
(Important to remember: 10 years ago, we were promised the HUMAN connectome by now...so beware of hyperbolic promises.)
Sebastian Seung
@SebastianSeung
No one promised #2 in 10 years, as a human brain contains 10^11 neurons (vs 10^2 regions). Scaling from fly to mouse connectome (10^5 to 10^8 neurons) could take a decade, and scaling to human connectome another decade. This assumes exponential progress in computing.
Sebastian Seung
@SebastianSeung
The speed will also depend on how badly we want connectomes. Many technological barriers were surmounted over the past 20 years, but the greatest barrier was social/political: opposition from mainstream neuroscientists.
-
Sebastian Seung
@SebastianSeung
The latest #flywire Data Challenge is now online, and concerns the important problem of relating two connectomes by graph matching. codex.flywire.ai/app/vnc_matching_challenge
David Bader
@Prof_DavidBader
🚀 Thrilled my @NJIT team scored top in the Princeton FlyWire Codex Challenge, advancing fruit fly brain connectome research! 🧠 Proud to push neuroinformatics forward. #neuroscience #flywire #datascience @NJITYingWu @EdisonAcad @PrincetonNeuro
news.njit.edu/njit-research-team-innovates-princeton-flyw… -
Sebastian Seung
@SebastianSeung
Interview with John Hopfield in The Daily @Princetonian www.dailyprincetonian.com/article/2024/11/princeton-news-…
-
Sebastian Seung reposted
John Hopfield
@HopfieldJohn
This award is very hard to respond to. I have received many hundred congratulatory notes, from former students, post-docs, Princeton University juniors and seniors, funding agencies and foundations, authors, signature collectors, amateurs, elementary school neural network followers, and on and on. An astonishing fraction of them has found their way into useful and interesting Neural Network careers by a casual interaction in class, at a meeting, hearing what I had to say about their ideas, learning from thinking about how I worked with a class, or from being my teaching assistants... There are some whom I remember well, and others for whom my reaction is “are they certain that our interaction sparked a single usable thought?” Yet they go on and comment “you changed my life” and follow on to explain that they heard me lecture when they were 15, and have been a member of the Neural Network brigade of the research army ever afterward.
I cannot make detailed comments to most of my letter writers. In sum I can only say that I tremendously enjoyed the interactions that the Neural Network community provided me with; that the mutual interactions have given me much pleasure over the years; that the community interested both in brain and in artificial brain has proved a good way for science to develop even if institutions have not always been sympathetic. Often these institutions found the enthusiasm infectious, after a period of doubt. In short, we often have won--. No, perhaps all we know is that we have not yet lost. I still believe that finding mind lodged in biological matter is the most profound question that physics can pose. And that the breadth of physics is a good base from which to begin. -
Sebastian Seung
@SebastianSeung
Starting in just a few minutes...
Sebastian Seung
@SebastianSeung
Here's your chance!
Ask us anything about #flywire
12 noon ET on Tuesday Oct 29.
Webinar followed by Q&A with @MurthyLab and me,
hosted by @DavidCRowland of @Nature
Please register.
www.workcast.com/register?cpak=3164513927332339 -
Sebastian Seung
@SebastianSeung
Here's your chance!
Ask us anything about #flywire
12 noon ET on Tuesday Oct 29.
Webinar followed by Q&A with @MurthyLab and me,
hosted by @DavidCRowland of @Nature
Please register.
www.workcast.com/register?cpak=3164513927332339 -
Sebastian Seung
@SebastianSeung
Ouch 🙄
Lior Pachter
@lpachter
Hype or hope? A recently published manifesto published by Thomas Kalil, CIO at @SchmidtFutures and colleagues arxiv.org/abs/2409.11654 envisions "...an AI Virtual Cell as a comprehensive AI framework composed of several interconnected foundation models...". A brief review: 1/🧵
-
Sebastian Seung
@SebastianSeung
70 years ago my late father stepped out of this station to begin life in the U.S. @Yale had taken a chance by awarding a full scholarship to this refugee from N Korea. He returned the favor by graduating summa cum laude. God Bless America, and God Bless Yale! #americandream
-
Sebastian Seung
@SebastianSeung
Reconstructing a whole brain means you can map sensation to cognition #flywire
Michael F. Chiang, MD
@NEIDirector
Beautiful work from @UCSB: researchers have mapped the fly’s visual brain systems & how they connect to the fly brain’s navigation center.
#neifunded @SungSoo_111 @MurthyLab @SebastianSeung @FU_Berlin @HHMINEWS
@Nature: go.nature.com/3YoLYg0 -
Sebastian Seung
@SebastianSeung
Visualizing whole fly brain simulation
Tyler Sloan
@quorumetrix
One the cool possibilities from the #fly #connectome is the ability to run simulations.
I've finally got around to experimenting with the "Leaky Integrate and Fire" model by @Philip_Shiu et al. Here are test simulations of 3x Mi1 optic cells, followed by their whole columnsThis media couldn't be displayed. See it on the original post. -
Sebastian Seung
@SebastianSeung
Location changed to Biolabs 1080.
Sebastian Seung
@SebastianSeung
Friday noon is a doubleheader @Harvard in Northwest Building 243. I will speak about the #flywire connectome, and Jeff Lichtman will describe the @NIH project to scale up to an entire mouse brain. cbs.fas.harvard.edu/calendar_event/special-seminar-sebast…
-
Sebastian Seung
@SebastianSeung
Back to my old stomping grounds in Cambridge MA! Friday 10am I'll be speaking about the #flywire connectome @mcgovernmit @mitbrainandcog
bcs.mit.edu/events/yang-icon-center-mcgovern-institute-sp…
Sebastian Seung
@SebastianSeung
Friday noon is a doubleheader @Harvard in Northwest Building 243. I will speak about the #flywire connectome, and Jeff Lichtman will describe the @NIH project to scale up to an entire mouse brain. cbs.fas.harvard.edu/calendar_event/special-seminar-sebast…
Sebastian Seung
@SebastianSeung
I'm told that the Harvard location has changed to Biolabs 1080.
Post not available here, view on X
Sebastian Seung
@SebastianSeung
Then predict that a new circuit in the fly brain serves the function of form vision, by interpreting connectivity motifs similar to those hypothesized for mammalian visual cortex. www.nature.com/articles/s41586-024-07953-5
Sebastian Seung
@SebastianSeung
And finally argue that many visual interneuron types function as spatial normalizers, analogous to normalization layers in convolutional nets.
Sebastian Seung
@SebastianSeung
My new preprint attempts to explain the function of interneuron diversity in the fly visual system. The answer (or at least the beginning of it) falls in your lap once you have the connectome.
www.biorxiv.org/content/10.1101/2024.04.03.587837v1 -
Sebastian Seung reposted
Yann LeCun
@ylecun
We cite Debbie's 2011 PNAS paper in our 2021 work on protein structure prediction with a "neural Potts model"
@alexrives
www.biorxiv.org/content/10.1101/2021.04.08.439084v1.full -
Sebastian Seung
@SebastianSeung
Yes your Potts model for MSA protein structure prediction was an impressive example of statistical physics as an approach to AI for biology.
Debora Marks
@deboramarks
I suppose my lab www.deboramarkslab.com - ha ha
Might also look back at our protein structure from sequence too - umm 2011? -
Sebastian Seung
@SebastianSeung
🧵on Japan's underrated contributions to neural nets. Shun-ichi Amari @UTokyo_News_en @riken_en is another one of my heroes. His 1972 paper on associative memory models modeled Hebbian plasticity using an outer product weight matrix.
Post not available here, view on X
Sebastian Seung
@SebastianSeung
Amari's neural field model (1977) anticipated later work in computational neuroscience on the role of continuous dynamical attractors in short-term memory.
Sebastian Seung
@SebastianSeung
Oops wrong screenshot above. H/T @alfcnz
-
Sebastian Seung
@SebastianSeung
Trace the intellectual thread from John Jumper (2024 Nobel for protein folding) back to Lawrence Bragg (1915 Nobel for x-ray crystallography). Jumper is just 39 yo. But Bragg was 25!
-
Sebastian Seung
@SebastianSeung
Fascinating that interdisciplinarity can be quantified.
Roberta Sinatra
@robysinatra
Based on data (not opinions), the works of Hopfield and Hinton #nobelprize2024
1. Are physics www.nature.com/articles/nphys3494
2. Were awaiting recognition www.nature.com/articles/s41567-018-0314-6
These are #scienceofscience works as part of my decade-long effort defining physics quantitatively -
Sebastian Seung
@SebastianSeung
Wow. @Google is cleaning up at the Nobels this year. Yesterday's prize recognized the old Google Brain and today's goes to @GoogleDeepMind! The torch has been passed from @BellLabs.
The Nobel Prize
@NobelPrize
BREAKING NEWS
The Royal Swedish Academy of Sciences has decided to award the 2024 #nobelprize in Chemistry with one half to David Baker “for computational protein design” and the other half jointly to Demis Hassabis and John M. Jumper “for protein structure prediction.”
Sebastian Seung
@SebastianSeung
11 Nobels are attributed to @BellLabs, in fields as "useful" as semiconductor physics and as "useless" as cosmology. en.wikipedia.org/wiki/Bell_Labs
Sebastian Seung
@SebastianSeung
You could say that yesterday's Nobel was a joint @BellLabs @Google prize. But it's a prize that is directly relevant to core business of Google. Today's prize on protein folding seems "useless" to Google. That means Google is doing fundamental research.
Sebastian Seung
@SebastianSeung
And in that sense the torch has passed from @BellLabs to @Google. @prfsanjeevarora remarked that the torch has passed from physical sciences to AI/computer science.
Sanjeev Arora
@prfsanjeevarora
Feels like a passing of the torch between fields. When I was a teenager in the 1980s, after a half-century of monumental progress powered by theoretical physics, most smart high schoolers wanted to do physics. Upon arriving as an undergrad at MIT in 1988, it quickly became clear that the next big thing was computer science.
Congratulations to @geoffreyhinton and John on this honor. Interestingly, I met John Hopfield a long time before I met Geoff. ( In mid 1990s when he invited me to give a guest lecture in his Princeton ugrad class about my algorithm for traveling salesman. )
Sebastian Seung
@SebastianSeung
Kudos to @Google. I don't know of any Nobels from @Meta @Microsoft @Apple Better get cracking!!
Sebastian Seung
@SebastianSeung
Also kudos to University of Washington. Though @uwcse didn't have the foresight to make him a core faculty member. I think they would have been the first academic CS dept to boast a Nobel. Baker is adjunct professor of CS at least. 😁
-
Sebastian Seung
@SebastianSeung
This interview is very Geoff, and makes the distinction between Boltzmann machines and backprop. www.nytimes.com/2024/10/08/technology/nobel-prize-geoffre…
-
Sebastian Seung
@SebastianSeung
Physics has this issue, but so does every traditional discipline, more or less. In today's presser, John's #1 message was that the boundaries between disciplines are ripe for innovation. Nobels struggle with this, because they are defined for fields that existed 100 years ago.
Leon Furchtgott
@lfurchtgott
Can you say more about "is it physics" culture? It sounds like that was something Hopfield had to deal with during his first Princeton stint and contributed to his departure for Caltech.
Post not available here, view on X
Sebastian Seung
@SebastianSeung
Answer #2: Elite universities on the East Coast disdained neural nets after 1970, with the possible exception of Brown. The 1980s resurgence was driven by the West Coast (UCSD, Caltech) and not so East (CMU), as well as non-U.S. universities and Bell Labs.
Sebastian Seung
@SebastianSeung
Answer #3: Maybe "like wow" interdisciplinary research is perfectly matched to the hot tubs, surf boards, palm trees, and convertibles of SoCal, and antithetical to the uptight self-flagellating puritanical academics of the East Coast? Just a thought 🤣
-
Sebastian Seung
@SebastianSeung
It's time for Jeopardy! Neural net theory for $1000.
With these coauthors, Hopfield formulated
'the "50% of all stimuli remind me of sex" problem' -
Sebastian Seung reposted
Sanjeev Arora
@prfsanjeevarora
We're looking for postdocs! Great compute access, wonderful resources and colleagues. Please apply via the ad below. We have the flexibility to start appointments during the year as well. Current postdocs are listed here pli.princeton.edu/people/postdoctoral-researchers
Princeton PLI
@PrincetonPLI
The @Princeton AI Lab is seeking postdoctoral fellows to join AI-related research initiatives.
For the job descriptions and application: ai.princeton.edu/ai-lab/employment -
Sebastian Seung
@SebastianSeung
Testimony from someone who was there!
John Platt
@johnplattml
It was super interesting and confusing at the same time!
-
Sebastian Seung
@SebastianSeung
Packed auditorium today to honor a hero.
Princeton University
@Princeton
We're here at #princetonu for a news conference celebrating professor emeritus John Hopfield’s #nobelprize win in physics.
Follow this thread for quotes and photos. -
Sebastian Seung
@SebastianSeung
Terry @Sejnowski was Hopfield's Ph.D. student in @PrincetonPhys
James V Stone
@jgvfwstone
How the Boltzmann Machine was born:
"John Hopfield gave a talk, and I was sitting next to Geoff Hinton, and Hopfield was saying, “We could do optimisation here”, . . . and that was something we were interested in. I had just read an article by Scott Kirkpatrick on simulated annealing, and I said, “Geoff, we could heat it up, we could heat up the Hopfield network by adding a temperature”, . . . it turns out something magical happens when you heat up a Hopfield network: it becomes capable of learning the weights to a multi-layer network; which is the first time that had been done."
T Sejnowski, 2023.
Quoted from: The Artificial Intelligence Papers: Original Research Papers With Tutorial Commentaries. TOC and Chapter 1: jamesstone.sites.sheffield.ac.uk/books -
Sebastian Seung reposted
Caltech
@Caltech
At Caltech in the early 1980s, John Hopfield merged his background in physics with neurobiology to create a simple computer model that behaved less like the computers of that time and more like the human brain.
www.caltech.edu/about/news/caltech-professor-emeritus-joh… -
Sebastian Seung
@SebastianSeung
AI bros are physicists now!
David Sussillo
@SussilloDavid
-
Sebastian Seung
@SebastianSeung
But is it physics? Hopfield tackled three profound questions about emergence: condensed matter from interactions between elementary particles, the magic of life from interactions between dead molecules, and intelligence/consciousness from interactions between dumb neurons.
Post not available here, view on X
Sebastian Seung
@SebastianSeung
And the eponymous Hopfield net showed a way in which collective computation could emerge from a neural net, a major achievement in AI and neuroscience.
Sebastian Seung
@SebastianSeung
The theoretical physics approach to neural nets was launched by @HopfieldJohn in this classic 1982 paper that introduced the "energy function" to associative memory models. www.pnas.org/doi/abs/10.1073/pnas.79.8.2554
Sebastian Seung
@SebastianSeung
Conclusion: Hopfield grappled with all three of the most profound questions in physics, giving the world amazing and lasting insights into the nature of the universe.
-
Sebastian Seung reposted
Princeton University
@Princeton
"What is physics? In my view, physics is trying to understand how systems work." Hopfield explains how his #nobelprize winning work in neural networks fits into the field of physics.
-
Sebastian Seung reposted
Sam Wang is at samwang.bsky.social
@SamWangPhD
At a time when so much in the world seems to be going wrong, I find @HopfieldJohn and his attitude to discovery to be refreshing and inspiring. Curiosity-driven research by John and the men and women who worked with him help make our nation, and the world, a better place.
Sebastian Seung
@SebastianSeung
President Eisgruber @Princeton calls for round of applause to honor @HopfieldJohn joining press conference by Zoom!
-
Sebastian Seung reposted
David A. Markowitz
@DavidAMarkowitz
What a joyful day for @Princeton and pocket protector wearers everywhere. One graduate mentor's Nobel down, one to go.
-
Sebastian Seung
@SebastianSeung
President Eisgruber @Princeton calls for round of applause to honor @HopfieldJohn joining press conference by Zoom!
Post not available here, view on X
Sebastian Seung
@SebastianSeung
Reporter asks the inevitable question: But is it physics? 😎
Sebastian Seung
@SebastianSeung
John cites the 1972 complex systems manifesto of another @Princeton Nobel physics laureate Phil Anderson, "More is different" www.science.org/doi/10.1126/science.177.4047.393
-
Sebastian Seung reposted
Dmitry Krotov
@DimaKrotov
Given today’s great news from the #nobelprize2024, I want to share a couple of personal thoughts on Hopfield Networks.
This idea had an enormous impact on at least three large disciplines: Statistical Physics, Computer Science and AI, and Neuroscience. -
Sebastian Seung
@SebastianSeung
What an exciting week for @PrincetonNeuro!! Headed to event in Taylor Auditorium, where John will join by video. (He summers in the UK and isn't back yet.)
Princeton Neuroscience Institute
@PrincetonNeuro
BREAKING: Congratulations to PNI professor emeritus John Hopfield, who was awarded the 2024 @NobelPrize in Physics, alongside @UofT's Geoffrey Hinton “for foundational discoveries and inventions that enable machine learning with artificial neural networks” pni.princeton.edu/news/2024/pnis-john-hopfield-receives-n…
-
Sebastian Seung reposted
Nima Dehghani
@neurovium
A key figure in bridging physics & biology @HopfieldJohn has inspiring & thought provoking essays:
- two cultures? Physics-bio interface
- physics, bio & why are they so different?
- now what? (A great expose on curiosity of masterful scientist
@PrincetonNeuro @NobelPrize 👇
American Physical Society
@APSphysics
Congrats to former APS President John Hopfield and Geoffrey Hinton for winning the @NobelPrize in Physics for their work in machine learning! Their research could lead to improved facial recognition, language translation, and medical diagnostics. #nobelprize
-
Sebastian Seung
@SebastianSeung
Haha yes physicists think they can do everything.
Michael Baym
@baym
I feel like the critique that the physics Nobel is going to computer science misunderstands the culture of physics. They aren’t ceding the prize to cs, they’re claiming it as a branch of physics
-
Sebastian Seung
@SebastianSeung
Wow. Hopfield model and Boltzmann machine.
The Nobel Prize
@NobelPrize
BREAKING NEWS
The Royal Swedish Academy of Sciences has decided to award the 2024 #nobelprize in Physics to John J. Hopfield and Geoffrey E. Hinton “for foundational discoveries and inventions that enable machine learning with artificial neural networks.”
Sebastian Seung
@SebastianSeung
The theoretical physics approach to neural nets was launched by @HopfieldJohn in this classic 1982 paper that introduced the "energy function" to associative memory models. www.pnas.org/doi/abs/10.1073/pnas.79.8.2554
Post not available here, view on X
Sebastian Seung
@SebastianSeung
By 1969 he had already won the Buckley Prize, the top prize in condensed matter physics, for work he did on semiconductors as a young physicist in the 1950s!
Sebastian Seung
@SebastianSeung
Then in 1974 he made an amazing advance in theoretical biology, explaining how "kinetic proofreading" reduces the error rate in DNA replication. www.pnas.org/doi/abs/10.1073/pnas.71.10.4135
Sebastian Seung
@SebastianSeung
So even if we ignore John's work on neural nets, he had already made breakthrough discoveries in physics and biology. That gives you an idea of the depth and breadth of his genius.
Sebastian Seung
@SebastianSeung
@HopfieldJohn has been associated with @Princeton the longest, but he also co-founded the Computation and Neural Systems program at @Caltech, which was the first of its kind.
Sebastian Seung
@SebastianSeung
There he co-taught a legendary class on "Physics of Computation" with VLSI giant Carver Mead and physics giant Richard Feynman. Don't you wish you could jump in a time machine and take that class? www.caltech.edu/about/news/caltech-celebrates-30-years-it…
Sebastian Seung
@SebastianSeung
@BellLabs is the other institution that shaped and was shaped by @HopfieldJohn. In the 1990s, I was a Member of Technical Staff in Department 11111. We were located at the site in Murray Hill, New Jersey.
Sebastian Seung
@SebastianSeung
If I walked up one flight of stairs and through the corridor, I could pass the birthplace of the transistor. Horst Stormer was in the office next to mine, and won his Nobel for the fractional quantum Hall effect.
Sebastian Seung
@SebastianSeung
Dept 11111 had been home to some of the greatest theoretical physicists of the 20th century. I felt insecure that the senior theorists would regard me as riffraff, as my research was on neural nets. But they could not, because they looked up to @HopfieldJohn.
Sebastian Seung
@SebastianSeung
John shaped the biophysics effort at @BellLabs before he decamped for @Caltech. It was in full swing by the time I arrived in the 1990s.
Sebastian Seung
@SebastianSeung
One can trace an intellectual thread from Hopfield, Shulman, and Ogawa 1971 on hemoglobin...
Sebastian Seung
@SebastianSeung
...to the discovery of the blood oxygenation level-dependent (BOLD) contrast signal by Ogawa, Lee, Kay, & Tank 1990. www.pnas.org/doi/abs/10.1073/pnas.87.24.9868
Sebastian Seung
@SebastianSeung
So even if we ignore his achievements in neural nets, John had already made breakthrough discoveries in physics and biology. That gives you an idea of the level of his genius.
Post not available here, view on X
Sebastian Seung
@SebastianSeung
This was my 1996 paper "How the brain keeps the eyes still," communicated to PNAS by...John Hopfield.
Sebastian Seung
@SebastianSeung
So we can trace an intellectual thread from John's work to the line attractor concept that has become popular in systems neuroscience for interpreting low-dimensional structure in large-scale recordings of neural activity.
Amit Vinograd
@AmitVinograd
(1/n)
Thrilled to announce our (me + @neuronair ) Nature paper is out! We uncovered causal evidence of a line attractor in the brain—a neural mechanism encoding emotional states like aggression. Machine learning (AI) helped us map it! 🚀🧠
#neuroscience #ai @NeuroscienceNews
Sebastian Seung
@SebastianSeung
Oops, think the thread broke. Here's the next item.
Sebastian Seung
@SebastianSeung
I only coauthored one paper with John. His 1984 energy function was for nets with symmetric connections. I discovered an energy function for excitatory-inhibitory nets, which also contain antisymmetric connections. proceedings.neurips.cc/paper/1997/hash/3d779cae2d46cf6a8a…
Sebastian Seung
@SebastianSeung
I only coauthored one paper with John. His 1984 energy function was for nets with symmetric connections. I discovered an energy function for excitatory-inhibitory nets, which also contain antisymmetric connections. proceedings.neurips.cc/paper/1997/hash/3d779cae2d46cf6a8a…
Post not available here, view on X
Sebastian Seung
@SebastianSeung
So Geoff already knew firsthand that the scientific consensus could be proven completely wrong. Perhaps that gave him the contrarian strength to champion neural nets for decades, until the field could be victorious.
Sebastian Seung
@SebastianSeung
Unfortunately I don't have any photos from the good old days, but luckily I took one after lunch with Geoff earlier this year. Congrats to @geoffreyhinton!
-
Sebastian Seung reposted
Sanjeev Arora
@prfsanjeevarora
Feels like a passing of the torch between fields. When I was a teenager in the 1980s, after a half-century of monumental progress powered by theoretical physics, most smart high schoolers wanted to do physics. Upon arriving as an undergrad at MIT in 1988, it quickly became clear that the next big thing was computer science.
Congratulations to @geoffreyhinton and John on this honor. Interestingly, I met John Hopfield a long time before I met Geoff. ( In mid 1990s when he invited me to give a guest lecture in his Princeton ugrad class about my algorithm for traveling salesman. ) -
Sebastian Seung reposted
Yoshua Bengio
@Yoshua_Bengio
@HopfieldJohn and @geoffreyhinton, along with collaborators, have created a beautiful and insightful bridge between physics and AI. They invented neural networks that were not only inspired by the brain, but also by central notions in physics such as energy, temperature, system dynamics, energy barriers, the role of randomness and noise, connecting the local properties, e.g., of atoms or neurons, to global ones like entropy and attractors. And they went beyond the physics to show how these ideas could give rise to memory, learning and generative models; concepts which are still at the forefront of modern AI research.
Their ideas inspired me so profoundly that I decided to choose learning in neural networks for my own research as a graduate student. They motivated me to look for abstract principles that could be as simple as the laws of physics, but could explain biological as well as artificial intelligence. I'm truly delighted for them and for our field.
The Nobel Prize
@NobelPrize
BREAKING NEWS
The Royal Swedish Academy of Sciences has decided to award the 2024 #nobelprize in Physics to John J. Hopfield and Geoffrey E. Hinton “for foundational discoveries and inventions that enable machine learning with artificial neural networks.” -
Sebastian Seung
@SebastianSeung
I stand corrected...Simon's Nobel was in economics?
Surya Ganguli
@SuryaGanguli
Herbert Simon won a Turing and a Nobel
-
Sebastian Seung reposted
Vamsi Mootha
@VamsiMootha
Today's @NobelPrize for John Hopfield recognizes only one of his contributions - he's a true polymath. I strongly recommend his solo author PNAS papers (i) long-range e- tunneling in proteins (1974), (ii) kinetic proof-reading (1974), (iii) Hopfield nets (1982). Each is a gem!
-
Sebastian Seung reposted
rafael yuste
@yusterafa
Elated to share that John Hopfield, whose ideas have guided our research, gets Nobel Prize
-
Sebastian Seung reposted
Sam Wang is at samwang.bsky.social
@SamWangPhD
I am delighted that @Princeton‘s own @HopfieldJohn was co-awarded this Nobel. In addition to being a foundational figure in neural networks, John has been a great colleague & friend to so many in our neuroscience community. He’s been a source of ideas, mentoring, and friendship.
The Nobel Prize
@NobelPrize
Did you know that an artificial neural network is designed to mimic the brain?
Inspired by biological neurons in the brain, artificial neural networks are large collections of “neurons”, or nodes, connected by “synapses”, or weighted couplings, which are trained to perform certain tasks. An artificial neural network processes information using its entire network structure. The inspiration initially came from the desire to understand how the brain works.
Learn more about this year’s physics prize awarded for work on artificial neural networks: www.nobelprize.org/prizes/physics/2024/popular-information/
#nobelprize -
Sebastian Seung
@SebastianSeung
They're having too much fun 😊
Kay M Tye PhD
@kaymtye
Also got my very own autographed copy from @MurthyLab where she took over a whole issue of @Nature
-
Sebastian Seung
@SebastianSeung
Sydney Brenner established C. elegans as a model organism starting in the 1960s, which led to the first connectome in 1986. In Sept 2007, we invited Sydney to deliver the inaugural lecture for our new @Harvard @MIT class on connectomics, and shockingly he accepted!
Sebastian Seung
@SebastianSeung
Back in 2007, I was applying convolutional nets to connectomics, @mfyanik was zapping worm neurons with a laser, Xiaowei Zhuang had smashed the diffraction limit, and @rclayreid was dreaming of combining calcium imaging with connectomics.
Sebastian Seung
@SebastianSeung
It was the beginning of the end of the connectome winter, which had fallen after 1986. Spring came slowly. Overall, it took almost four decades (until 2024) to scale up from worm (300 neurons, 10^4 synapses) to fly (>10^5 neurons, >10^7 synapses).
Post not available here, view on X
Sebastian Seung
@SebastianSeung
Respected as a 20th century giant in biology, and loved for his wit, Sydney Brenner passed away in 2019. This obituary by John White describes his life and achievements. royalsocietypublishing.org/doi/full/10.1098/rsbm.2020.0022
Sebastian Seung
@SebastianSeung
Photo: dinner at Chez Henri, a now defunct restaurant north of @HarvardSquare
Sebastian Seung
@SebastianSeung
Oh and yet another Nobel for C. elegans today! Does anyone know the total count now?
Discovery in Tiny Worm Leads to Nobel Prize in Physiology or Medicine for 2 Scientists www.nytimes.com/2024/10/07/health/nobel-prize-medicine.html -
Sebastian Seung reposted
FlyWire
@FlyWireNews
Starting soon: FlyWire Live at the Tools, Tech, and Theory BRAIN Initiative Social at #sfn24
Location: maps.app.goo.gl/PZLrucq34QTjtDVPA
Image features neurons colored by neurotransmitter inspired by this paper from @BobQubit @as_bates et al www.cell.com/cell/fulltext/S0092-8674(24)00307-6 -
Sebastian Seung
@SebastianSeung
#sfn2024 tweeps! Get your copy of the #flywire special issue autographed by @murthylab at the @SpringerNature booth. Connectome connectome connectome. H/T @BobbyKasthuri
-
Sebastian Seung
@SebastianSeung
@NIH funding of the #flywire Consortium was only possible because @IARPAnews had previously derisked connectomics by investing in and demanding a 1000x scale up of the required AI/computational technologies
David A. Markowitz
@DavidAMarkowitz
The fly connectome was impossible to achieve under the traditional scientific funding apparatus alone. We'd all be better off with structural changes that make achievements like this more likely in the future. 1/
Sebastian Seung
@SebastianSeung
Visionary institutions are only as visionary as the individuals who lead them. We are grateful for the leadership of @DavidAMarkowitz @RJVogelstein @JasonGMatheny
Sebastian Seung
@SebastianSeung
And thanks to Connectomics @Google for creating Neuroglancer, which was indispensable for visualizing failure modes during AI R&D, and also served as the front end for #flywire proofreading. x.com/stardazed0/status/1100476655368265728
-
Sebastian Seung
@SebastianSeung
Memories. With @srinituraga at #sfn 2007, waiting to deliver my Presidential Special Lecture, "The Once and Future Science of Neural Networks." Pinstripe suit with hot pink trim by Moschino and I really needed a haircut.
Post not available here, view on X
Sebastian Seung
@SebastianSeung
It was a *presidential* lecture, right? Laughter rippled over the cavernous room, but I don't think the audience understood. Neuroscientists kept saying that the brain would tell us how to solve AI. I was telling them that they were stuck in the past.
Sebastian Seung
@SebastianSeung
Neuroscience had helped AI (1960s-70s). Now it was time for AI to help neuroscience (2007-). That is the history of connectomics in a nutshell.
Post not available here, view on X
Sebastian Seung
@SebastianSeung
The public desperately wants treatments and cures for brain disorders. The ones we have were discovered serendipitously, and often aren't that effective.
Sebastian Seung
@SebastianSeung
If we want neuroscience to accelerate, we need AI to help us get unstuck from progress at a turtle's pace. I said it too politely in 2007. Maybe people are finally starting to grasp it now.
-
Sebastian Seung
@SebastianSeung
As a youngster, I imagined scientists as heroes and saints in search of truth.
Ryan Gomes
@ryangomes
Hopefully not the fraud though. One of many examples: www.science.org/content/article/research-misconduct-findi…
Post not available here, view on X
Sebastian Seung
@SebastianSeung
The scientific endeavor is the most effective approach that humanity has found for identifying truth, and it works in spite of human limitations. Sometimes it even works by exploiting them.
Sebastian Seung
@SebastianSeung
In other human endeavors we can only rely on consensus. But some of the greatest scientific advances happen when a belief held by 99.9% of scientists is ultimately proven wrong by the evidence.
-
Sebastian Seung
@SebastianSeung
Some people have gotten the impression that I hate backprop. Let's be real. I love backprop. The fly connectome was enabled by an AI system containing an arsenal of convolutional nets trained by backprop.
Sebastian Seung
@SebastianSeung
More shocking to AI enthusiasts, backprop seems irrelevant. How do I know that? Because *learning* seems irrelevant. If you rear a fly in total darkness, the visual system seems unaffected to a first approximation. link.springer.com/article/10.1186/1471-2202-4-14
Post not available here, view on X
Sebastian Seung
@SebastianSeung
I taught convolutional nets and backprop in my class at @MIT starting spring 1999, during the neural nets winter. I knew these topics because I had previously worked at Bell Labs in Murray Hill, New Jersey. en.wikipedia.org/wiki/Bell_Labs
Sebastian Seung
@SebastianSeung
We were an hour's drive from Bell Labs in Holmdel, and recognized the convolutional net successes of @ylecun and his colleagues there. New Jersey is not only the Garden State. It's the cradle of AI. @NJGov en.wikipedia.org/wiki/Bell_Labs_Holmdel_Complex
-
Sebastian Seung
@SebastianSeung
I'm reminded today that the world of science is what I wish the entire world could be: people of different nations, races, creeds, and opinions working together to achieve great things. #sfn2024
Sebastian Seung
@SebastianSeung
(reminder because we scientists usually gripe and forget to feel grateful)
-
Post not available here, view on XPost not available here, view on X
Sebastian Seung
@SebastianSeung
I learned this lesson from the late great Dick Masland, and his diagram illistrates how wiring/shape constrains neuronal connectivity. www.cell.com/current-biology/fulltext/S0960-9822(04)00440-3
Post not available here, view on X
Sebastian Seung
@SebastianSeung
More shocking to AI enthusiasts, backprop seems irrelevant. How do I know that? Because *learning* seems irrelevant. If you rear a fly in total darkness, the visual system seems unaffected to a first approximation. link.springer.com/article/10.1186/1471-2202-4-14
-
Sebastian Seung
@SebastianSeung
Thanks for sharing this video showing another physical dexterity necessary to turn a brain digital.
Davi Bock
@dddavi
I believe Dona Fetter did the dissection. And that's right, doing this well is a fine art, micro-precision and high speed combined. This video shows the technique: www.youtube.com/watch?v=dc9tFpXv0m0
Sebastian Seung
@SebastianSeung
@duncan_mearns told me that the fly brain dissection is even more demanding for electron microscopy, which shows all damage clearly; light microscopy is more forgiving.
Sebastian Seung
@SebastianSeung
It's harder still to dissect the entire CNS (brain plus ventral nerve cord). #flywire's next project is reconstructing the BANC, a whole fly CNS courtesy the lightning hands of @mindyisminsu
FlyWire
@FlyWireNews
If you’re at the Berlin Connectomics event and interested in fly connectomes, find @MatsliahArie @mindyisminsu @alin1194. Bonus they have a few 🔥🔥posters
-
Sebastian Seung
@SebastianSeung
As the old saw goes, functional connectivity has little to do with connectivity and even less to do with function.
Kording Lab 🦖
@KordingLab
Finally the rename functional connectivity -> correlation network is happening.
-
Sebastian Seung
@SebastianSeung
What I absolutely love love love is neurons!
Visualization of the "superclasses" of fly neurons.
youtu.be/J2xTkMsZchs
Sebastian Seung
@SebastianSeung
#flywire visualized by @quorumetrix
More about superclass and other annotations in this paper.
FlyWire
@FlyWireNews
Annotation is vital for realizing the full potential of connectomics datasets. Schlegel et al. reports the systematic and hierarchical annotation of the Fly connectome, including 1,458 new cell types. www.nature.com/articles/s41586-024-07686-5
-
Sebastian Seung
@SebastianSeung
This animation is fun, and shows the size and position of the fly brain relative to its body.
Greg Jefferis
@gsxej
Zoom into the #flywire fly brain courtesy of another of Philipp Schlegel's (@flyconnectome @CamZoology @MRC_LMB) amazing 3D sequences. Then read his piece on the flyconnecto.me blog to find out more about how the #connectome became a reality flyconnecto.me/2024/10/02/flywire-is-live-🚀/.
-
Sebastian Seung
@SebastianSeung
Thread on electrical compartmentalization and why it might be adaptive.
Sebastian Seung
@SebastianSeung
But you are absolutely right it's a good question. In the fly visual system, you can find the huge CT1 cell with 800 independent compartments, each serving one facet of the compound eye? www.youtube.com/watch?v=1J1yVqahkT4
-
Sebastian Seung
@SebastianSeung
Insightful comment. Largest fly neurons have fan-in/out of O(10^3), much greater than typical value. Why? Electrical signals don't spread far inside some super neurons. The neuron therefore behaves as multiple functional units, each with small fan-in/out.
Yann LeCun
@ylecun
That's a huge fan-in.
Sebastian Seung
@SebastianSeung
Note that fan-in/out of O(10^3) is likely the typical value for your own neurons though.
Post not available here, view on X
Sebastian Seung
@SebastianSeung
But you are absolutely right it's a good question. In the fly visual system, you can find the huge CT1 cell with 800 independent compartments, each serving one facet of the compound eye? www.youtube.com/watch?v=1J1yVqahkT4
-
Sebastian Seung
@SebastianSeung
neurons, beautiful neurons
Tyler Sloan
@quorumetrix
Since the '50 largest neurons' #connectome render with @amyneurons @FlyWireNews has made it's way around the world, I'm sharing an animated version. Including:
Information flow from the ocelli,
50 largest #neurons,
and a LOT of #synapses on one MASSIVE neuron.
#neuroscienceThis media couldn't be displayed. See it on the original post. -
Sebastian Seung
@SebastianSeung
It's been a privilege to work with our friends at @MRC_LMB, the birthplace of structural biology and the C. elegans connectome. They don't just promise...they deliver!
Greg Jefferis
@gsxej
Big day for neuroscience & the end of the beginning for whole brain connectomics. Shout out to @sdorkenw and Philipp Schlegel, first authors of the connectome papers, @MurthyLab @SebastianSeung for establishing #flywire to make this possible and @dddavi for birthing this dataset.
Sebastian Seung
@SebastianSeung
They are the world leaders in the emerging field of connectome annotation. Much as biologists identify genes in the genome, they carve the connectome into cell types. The next challenge is to ascribe functions to these structural units.
Sebastian Seung
@SebastianSeung
Watch this field...it will quickly grow prominent as the flow of connectomic information becomes a deluge.
-
Sebastian Seung
@SebastianSeung
How the fly connectome was won - the back story from someone who was in the trenches.
Thomas Macrina
@tmacrina
@IARPAnews, @DavidAMarkowitz, & @RJVogelstein deserve a lot of credit for how they structured the MICrONS program – it really stretched us. I’m not sure we would have tried putting machine learning into alignment so early otherwise.
-
Sebastian Seung
@SebastianSeung
This article explains the scope of #flywire, from acquiring the raw data to dissemination of the final data products. With appearances by friends at @MRC_LMB @AllenInstitute @uvmvermont And relevance to basic science, health, and disease.
Princeton University
@Princeton
A team led by @PrincetonNeuro's Mala Murthy (@MurthyLab) and @sebastianseung has mapped a fly's entire brain — a new step in understanding brain diseases in humans. www.princeton.edu/news/2024/10/02/mapping-entire-fly-brai…
-
Sebastian Seung
@SebastianSeung
Another shout-out to an unsung hero of the #flywire connectome. Ten years ago, Rick Fetter cut a fly brain into 7000 slices, each 1000x thinner than a hair.
Sebastian Seung
@SebastianSeung
He managed to collect the fragile slices onto TEM slot grids. He lost only a handful, which is an almost superhuman feat.
Post not available here, view on X
Sebastian Seung
@SebastianSeung
The slicing and imaging that preceded #flywire was described in this paper by @zhihaozheng @dddavi Fetter and others www.cell.com/cell/fulltext/S0092-8674(18)30787-6
Sebastian Seung
@SebastianSeung
To clarify: Rick manually fished the slices out of water using those slot grids. Human dexterity, not a machine.
-
Sebastian Seung
@SebastianSeung
I dedicate this video to the fly that gave her life for science, but will live forever in silico. She has supercharged the discoveries of the #flywire Consortium, and propels us into the digital transformation of neuroscience. Protip: watch with audio on. www.youtube.com/watch?v=RQuYaMDc1d0
Sebastian Seung
@SebastianSeung
Data at codex.flywire.ai
Papers here: www.nature.com/collections/hgcfafejia -
Sebastian Seung
@SebastianSeung
You are more than your genes!
Katrin Vogt
@KatrinVogt3
I am the connectome 😉
-
Sebastian Seung
@SebastianSeung
A special shout-out to Krzysztof Kruk, an amateur scientist from Poland and veteran of @eye_wire. He singlehandedly annotated the majority of neurons in the #flywire optic lobe, and wrote software that aided others doing the annotation. What a hero!
nature
@Nature
Volunteers played a key part in validating a gigantic neuroscience study mapping the fruit-fly brain. It shows what can be achieved through crowdsourcing research. go.nature.com/3TPeOTS
Sebastian Seung
@SebastianSeung
He is a named coauthor of our paper that completes the catalog of cell types intrinsic to the fly optic lobe. This is the first time that a biological visual system has been mapped in its entirety. www.nature.com/articles/s41586-024-07981-1
-
Sebastian Seung
@SebastianSeung
The landing page for the #flywire issue of @Nature is an essential overview, given the numerous and lengthy papers. Kudos to David and the @Nature team!
David Rowland
@DavidCRowland
If you are interested in an overview of the special issue, you can start with this feature, written by me with amazing graphics from our team as well as the @FlyWireNews team (let's be real, you should go for the cool graphics and reconstructed neurons)
www.nature.com/immersive/d42859-024-00053-4/index.html -
Sebastian Seung
@SebastianSeung
"Haunting." Maybe it won't be so bad if you get a simulated body too?
Heather Anne Campbell
@heathercampbell
The most haunting part of this article is when they simulate sugar and the simulated brain responds. Like, there’s no reason to believe they won’t eventually do this with human simulated brains… and if so, why should we believe we’re not simulated? www.nytimes.com/2024/10/02/science/fruit-fly-brain-mapped…
Sebastian Seung
@SebastianSeung
@srinituraga is planning to combine brain and body simulations. www.youtube.com/watch?v=fNgRWlaowgw
Sebastian Seung
@SebastianSeung
Here's the repo on github #flywire github.com/TuragaLab/flybody
-
Sebastian Seung
@SebastianSeung
A historic milestone for neuroscience! It's been a fantastic journey for the #flywire Consortium. This special edition of Nature showcases the power of open science, with nine papers that not only describe the fly connectome, but also make discoveries using it.
FlyWire
@FlyWireNews
Big news! The fly connectome is featured on the cover of a special edition of Nature.
This is all possible thanks to the collaboration of 292 members of The FlyWire Consortium!
www.nature.com/immersive/d42859-024-00053-4/index.html
Check out the thread for an overview of the 9 #flywire papers published today -
Sebastian Seung
@SebastianSeung
A great explainer of what's been achieved, along with scientific and historical context. #flywire
Anita Devineni
@BrainsExplained
Papers reporting the fly whole-brain connectome were published in @Nature today 🎉 and it was my honor to write a News and Views about them
Congrats to @sdorkenw, @SebastianSeung, @MurthyLab, @gsxej, and the @FlyWireNews team!
+ great 🧵 below from David
www.nature.com/articles/d41586-024-03029-6 -
Sebastian Seung
@SebastianSeung
International collaborations are more fun! #flywire
-
Sebastian Seung
@SebastianSeung
Go Tigers!
Princeton Neuroscience Institute
@PrincetonNeuro
A team of researchers and citizen scientists led by @Princeton neuroscientists @MurthyLab and @SebastianSeung has created the first ever complete map of a centralized adult brain: the fruit fly 🪰🧠
www.princeton.edu/news/2024/10/02/mapping-entire-fly-brai… -
Sebastian Seung reposted
Amy Robinson Sterling
@amyneurons
A moment of zen before big news later today :)
-
Sebastian Seung reposted
FlyWire
@FlyWireNews
We are excited to share educational materials designed to bring FlyWire into classrooms! We have designed interactive lessons to be used by high school and college teachers, instructors, or scientists doing outreach. Check it out: codex.flywire.ai/academy_home
Image by @exp_perception -
Sebastian Seung
@SebastianSeung
A crazy crazy neuron...make sure to follow the thread to the flythrough.
FlyWire
@FlyWireNews
The APL neuron is the longest cell in the fly connectome. There are only 2 in the brain. Each one is 20x longer than the fly's whole body and has well over 100,000 synapses.
Explore the fantastic APL cell in Codex: codex.flywire.ai/app/cell_details?data_version=783&root_i… -
Sebastian Seung reposted
Tyler Sloan
@quorumetrix
If you like #scivis #neuroscience 3d renders of neurons, and appreciate the intricacies of their branching structure as much as I do, then you should really follow @FlyWireNews for what's to come shortly.
Until then, enjoy some #wip renders that didn't make the final cut. -
Sebastian Seung
@SebastianSeung
Other twists: Haber won a Nobel, the prize established by a Swedish chemist to assuage his guilt about inventing dynamite. Haber's work indirectly led to Zyklon B, the insecticide later repurposed for genocide by the Nazis. Haber was Jewish.
Sebastian Seung
@SebastianSeung
Haber directed one of the inaugural Institutes of the Kaiser Wilhelm Society, the ancestor of today's elite Max Planck Society. A later director at the Fritz Haber Institute was Ernst Ruska, who invented the electron microscope, which enabled my own field of connectomics.
-
Sebastian Seung reposted
Srini Turaga
@srinituraga
Our work simulating the fly visual system connectome with @jakhmack @lappalainenjk is now out in @Nature Paper (open access): www.nature.com/articles/d41586-024-02935-z
Research Briefing: www.nature.com/articles/d41586-024-02935-z
HHMI | Janelia
@HHMIJanelia
AI & the connectome are enabling scientists to do what they never thought possible: predict the activity of individual neurons without a single measurement in a living brain.
New research led by Janelia’s @srinituraga & @uni_tue’s @lappalainenjk @jakhmack www.janelia.org/news/researchers-combine-the-power-of-ai-… -
Sebastian Seung
@SebastianSeung
The untold story of GLP1.
Jeffrey Flier
@jflier
What if I told you that I co-founded a startup in 1987 that obtained world-wide rights to GLP1 as a metabolic Rx, collaborated with Pfizer to show key activities, & abandoned it in 1990 when Pfizer lost interest? I tell the previously untold tale in an open access paper now up in Perspectives in Biology and Medicine. Enjoy! muse.jhu.edu/pub/1/article/936036/summary
-
Sebastian Seung reposted
Adrian A. Wanner
@AdrianAWanner
I highly recommend this. PNI is an amazing place for neuroscience and the CV Starr Fellowship is an important stepping stone towards independence.
Princeton Neuroscience Institute
@PrincetonNeuro
Another exciting postdoc opportunity at PNI: apply *now* for the CV Starr Postdoctoral Fellowship, which provides a generous 3-year salary and modest research budget 💫 Applications due Sept 15!
Learn more here: pni.princeton.edu/apply/postdocs/cv-starr-fellowship -
Sebastian Seung reposted
Mojtaba R. Tavakoli l مجتبی توکلی
@RTMojtaba
Would whole-brain #liconn be possible? Don’t know it now but am curious to figure it out 👀 🧠🔬
-
Sebastian Seung reposted
Mojtaba R. Tavakoli l مجتبی توکلی
@RTMojtaba
Update on our #liconn preprint for LM-based dense connectomics reconstruction. We have implemented block-face imaging and sectioning of the tissue/hydrogel-hybrid for extended volumetric imaging and provide data for traceability across z-sections. 👇🏻👇🏻👇🏻
www.biorxiv.org/content/biorxiv/early/2024/07/08/2024.03.… -
Sebastian Seung
@SebastianSeung
New Jersey ❤️ AI
cc @yoavfreund @mkearnsupenn @HopfieldJohn
Yann LeCun
@ylecun
More recently:
- Hopfield nets (Hopfield was simultaneously at Bell Labs and Princeton).
- ConvNets
- Boosting / Adaboost
- Non-Negative Matrix Factorization (hi @SebastianSeung !)
- SVM and kernel machines
- Structured prediction
- A lot of stuff in computational learning theory / VC theory. -
Sebastian Seung
@SebastianSeung
Classic!
Jane Bromley
@jmbromley
You could add in Siamese nets to that list ;-)
-
Sebastian Seung
@SebastianSeung
Murray Hill NJ 1951 Claude Shannon posed the problem of predicting the next word which turned out to be the seed of today's LLMs. www.princeton.edu/~wbialek/rome/refs/shannon_51.pdf
Yann LeCun
@ylecun
Good stuff from Murray Hill, Florham Park, and Princeton, too 😉
-
Sebastian Seung reposted
Nancy Kanwisher @NancyKanwisher@mas.to
@Nancy_Kanwisher
A lovely visit today with my wonderful Ph.D. advisor Molly Potter, still lively and engaged at 94.
I wrote about her huge impact on my life here: www.kavliprize.org/nancy-kanwisher-autobiography
Thank you Molly! -
Sebastian Seung
@SebastianSeung
Sounds simple but for some reason it's not easy!
Yann LeCun
@ylecun
It is of paramount importance that the management of a research lab be composed of reputable scientists.
Their main jobs are to:
1. Identify, recruit, and retain brilliant and creative people.
2. Give them the environment, resources, and freedom to do their best work.
3. Identify promising research directions (often coming from the researchers themselves) and invest resources in them. Put the scientists in charge and get out of the way.
4. Be really good at detecting BS, not necessarily because scientists are dishonest, but often because they are self-deluded. It's easy to think you've invented the best thing since sliced bread. Encouraging publications and open sourcing is a way to use the research community to help distinguish good work from not-so-good work.
5. Inspire researchers to work on research projects that have ambitious goals. It's too easy and less risky to work on valuable improvements that are incremental.
6. Evaluate people in ways that don't overly focus on short-term impact and simple metrics (e.g. number of publications). Use your judgment. That's why you get paid the big bucks.
7. Insulate rogue-but-promising projects from the scrutiny of upper management. A watched pot never boils. Planned innovation and 6-months milestones never bring breakthroughs.
You can't do any of this cat herding jobs unless you are an experienced, talented, and reputable scientist with a research record that buys you at least some legitimacy in the eyes of the scientists in your organization. -
Sebastian Seung
@SebastianSeung
Comparison of connectivity across different fly species.
-
Sebastian Seung
@SebastianSeung
The ventral nerve cord (analog of our spinal cord) is the interface between the fly brain and body. The communication channels are the ascending and descending neurons.
Dr Kathi Eichler @KathiEichler@qoto.org
@EichlerKathi
This one is very close to my heart, proudly sharing our newest preprint “Comparative connectomics of the descending and ascending neurons of the Drosophila nervous system: stereotypy and sexual dimorphism”. A short summary of this project that was 4 years in the making:
-
Sebastian Seung
@SebastianSeung
Pioneer in AI and computational neuroscience Terry Sejnowski. This honorary degree comes on top of his PhD in physics, also from @princeton
Salk Institute
@salkinstitute
Congratulations to Salk Professor Terry Sejnowski! He received an honorary Doctor of Science degree from Princeton University. Sejnowski has published more than 1,000 scientific papers, reports, and book chapters and numerous books, including The Deep Learning Revolution in 2018; his next book, ChatGPT and the Future of AI: The Deep Language Revolution, is slated for publication this fall. @Princeton
www.princeton.edu/news/2024/05/28/princeton-awards-seven-… -
Sebastian Seung
@SebastianSeung
Funny that there's not even agreement about the textbook stuff.
Mark Histed
@HistedLab
This argues cortical areas are discrete with sharp borders, based on animal data.
I don’t want to come down hard, as we don’t really know the final answer.
But based on work in sensory and frontal ctx, I’d be surprised if cortical areas commonly are discrete w/ sharp borders./1 -
Sebastian Seung
@SebastianSeung
Cool data-driven method of investigating rate of aging.
Vamsi Mootha
@VamsiMootha
Have you ever wondered:
"How fast do we age?"
In this 1996 paper, Bortz & Bortz address this question using exercise performance as a function of time.
What's their answer?
~0.5% per year (between ages 35-70)
academic.oup.com/biomedgerontology/article/51A/5/M223/578785 -
Sebastian Seung reposted
Steven Strogatz
@stevenstrogatz
This essay is charming and insightful, but it doesn't give any hint about how to *make* discoveries. For that, I would recommend this free PDF of a book by Professor Jack Oliver, one of the discoverers of plate tectonics. ecommons.cornell.edu/items/f2e1c2d6-a811-4abf-914b-390561…
Scholarship for PhD
@ScholarshipfPhd
The importance of stupidity in scientific research
-
Sebastian Seung reposted
Daniel Goodwin
@D_R_Goodwin
Extremely sad :( When I worked with @SebastianSeung at Simons Foundation, Jim would eat lunch with us weekly. I didn't even know who he was at first, he was just another person at the table earnestly stoked to geek out. Zero arrogance, total curiosity.
Thanks for everything Jim!
Simons Foundation
@SimonsFdn
It is with great sadness that the Simons Foundation announces the death of its co-founder and chair emeritus, James Harris Simons. Jim was an award-winning mathematician, a legendary investor and a generous philanthropist. www.simonsfoundation.org/2024/05/10/simons-foundation-co-…
-
Sebastian Seung
@SebastianSeung
Fantastic innovation that has benefited the @FlyWireNews community.
HHMI | Janelia
@HHMIJanelia
💡 A ‘crazy idea’ hatched by @janfunkey & colleagues during a walk in the woods & first tested by a high school student is now an innovative research tool used by scientists worldwide to predict neurotransmitters in fruit fly connectomes. 🧠
Read more 👇 www.janelia.org/news/high-school-student-helps-transform-… -
Sebastian Seung
@SebastianSeung
Congrats! We are all grateful for Google's contributions to the field.
-
Sebastian Seung reposted
Anne Churchland
@anne_churchland
Thanks for the push. BRAIN has provided critical research funds that move our field forward. I have contacted my representative (@AdamSchiff) and my senators (@SenAlexPadilla, @Senlaphonza). Can funding be restored? @crozSciTech thanks for your leadership on this! Please RT!
TK Kozai (BIONICLab.org)
@BIONIC_Lab
Please contact your government representative to fund the BRAIN Initiative and encourage your colleagues and trainees to do the same:
www.house.gov/representatives/find-your-representative
www.senate.gov/senators/senators-contact.htm -
Sebastian Seung
@SebastianSeung
Last year we completed the whole fly brain. Onward to an entire fly CNS! (brain + ventral nerve cord, analog of spinal cord)
FlyWire
@FlyWireNews
The BANC is open. Deposit your expertise in the newest community-created connectome: flywire.ai/banc
-
Sebastian Seung
@SebastianSeung
Which of these inventions has benefited humanity the most? (Question for my class today on "AI: doom or utopia?")
-
Sebastian Seung
@SebastianSeung
Flies are highly visual animals. That’s obvious to anyone who has ever tried to swat one. So it should be no surprise that eyes are so prominent on the fly head.
Post not available here, view on X
Sebastian Seung
@SebastianSeung
Today visual neuroscience enters the connectomic era.
See preprint by Matsliah, Yu, and others from the FlyWire Consortium www.biorxiv.org/content/10.1101/2023.10.12.562119v1
Sebastian Seung
@SebastianSeung
Bonus for the deep learning aficionados: Each type of neuron is analogous to a feature map in a convolutional net. More on that coming soon...
Post not available here, view on X
Sebastian Seung
@SebastianSeung
An object subsystem consists of intrinsic cell types (light green) connecting with neurons projecting to central brain (light blue), color subsystem (magenta), and ON and OFF channels (dark green).
Sebastian Seung
@SebastianSeung
The color subsystem (Cluster1, 4, 6) contains more cell types (magenta, pink) than any other subsystem.
Post not available here, view on X
Sebastian Seung
@SebastianSeung
The FlyWire Codex is where you can start from a photoreceptor, traverse pathways in the optic lobe, exit the optic lobe, and enter central brain circuits for generating behaviors. All the while knowing the cell types of neurons as you hop between them.
Sebastian Seung
@SebastianSeung
Here is a fantastic example from @SungSoo_111 @emil_kind and others, who traced the anterior visual pathway to the central complex, a fly brain region important for navigation.
Sung Soo Kim
@SungSoo_111
Have you ever wondered how brains convert sensory info into abstract, cognitive entities? The Kim Lab #mcdb #dyns @UCSB_NRI @UCSB and the Wernet Lab @fubcp @FU_Berlin reconstructed the neural circuits that carry visual information to the navigation center in the fruit fly brain.
-
Sebastian Seung reposted
Alice Ting
@aliceyting
scFLARE leads to a change in clinical practice in humans? 🤯
When we made our calcium integrator scFLARE in 2020 we hoped that it would be used by basic neuroscientists to help dissect neural circuitry but we never imagined that it might lead to a change in clinical practice in humans.
A team led by Ivan Soltesz @rmjamiolkowski and @QuynhA_Nguyen at @StanfordMed used scFLARE to discover an unexpected region of the brain that is recruited/activated during epileptic seizures: the fasciola cinereum or FC. The authors showed that optogenetic inhibition of FC led to a striking decrease in seizure duration in mice.
In humans with severe recurrent epileptic seizures the standard of care is surgical resection of affected brain regions. In 6 human patients, sEEG measurements showed prominent involvement of the FC region. In one of these patients, when ablation of the amygdala and anterior hippocampus proved insufficient to suppress seizures, the FC was lesioned, leading to a stunning 83% reduction in seizure frequency!
From the paper: “Here we used scFLARE in combination with closed-loop optogenetics to identify the FC subregion of the brain as a novel node of seizure propagation and target for intervention in mice and extended these insights to human patients. Critically, our finding that all patients with TLE from whom we obtained recordings had epileptic activity in this area can lead to a fundamental change in the standard of care."
Congratulations to the authors on this impressive and impactful work: @QuynhA_Nguyen, our previous scFLARE collaborator; @mateosanchezlab, the original inventor of scFLARE in my lab; @rmjamiolkowski from the Stanford Dept of Neurosurgery; and Ivan Soltesz and his lab members. We are humbled that our tool played some role in this discovery!
www.nature.com/articles/s41591-024-02924-9 -
Sebastian Seung
@SebastianSeung
Peter is being modest...he pioneered the application of EM reconstruction to mammalian retina in the 80s.
peter sterling
@whatishealth21
As David Hubel used to say, "the good old days are...now! Phenomenal effort; also smart! From my bad old days, three cheers!
www.nature.com/articles/d41586-024-01096-3Post not available here, view on X
Sebastian Seung
@SebastianSeung
Which inspired retinal connectomics in the 2010s.
www.molvis.org/molvis/v17/a41/
Sebastian Seung
@SebastianSeung
His inspiration was cited by this work too.
www.nature.com/articles/nature12346 -
Sebastian Seung
@SebastianSeung
Important point distinguishing between lateral inhibition and normalization. These are related concepts but not the same.
-
Sebastian Seung
@SebastianSeung
Fun times @PrincetonNeuro with @yael_niv @IlanaWitten @brody_lab
-
Sebastian Seung
@SebastianSeung
TL;DR New preprint with discovery: many interneuron types in the fly visual system function as highly specific normalizers.
Sebastian Seung
@SebastianSeung
My new preprint attempts to explain the function of interneuron diversity in the fly visual system. The answer (or at least the beginning of it) falls in your lap once you have the connectome.
www.biorxiv.org/content/10.1101/2024.04.03.587837v1
Sebastian Seung
@SebastianSeung
Ergo interneuron diversity is analogous to the ubiquity of normalization layers in contemporary convolutional nets. cc @Sergey_xai @ChrSzegedy
Sebastian Seung
@SebastianSeung
But normalizations in the fly visual system are more specific and possibly more sophisticated.
Sebastian Seung
@SebastianSeung
The analogy with convolutional nets is becoming so precise that we can point to differences in "design principles" as well as similarities.
Sebastian Seung
@SebastianSeung
Oh and I forgot to say why we should care. Feature detectors (convolutional layers) are the stars of vision, while normalizers are the supporting actors.
Sebastian Seung
@SebastianSeung
The last decade has shown empirically that normalizers are essential for robust feature detection. Convolutional nets don't work nearly as well without them.
-
Sebastian Seung
@SebastianSeung
Yes @PrincetonNeuro does genomics and development!
fenna krienen
@fennamk
‼️Job alert! ‼️ multiple #postdoc positions open in my lab @Princeton.
Interested in single cell #genomics & brain development?
#studybrain projects, cell type tools for 🐒🧠, weird evolutionary phenomena (chimeric microglia??)
Check us out krienenlab.org & pls RT! -
Sebastian Seung
@SebastianSeung
@doristsao reminded me of that day at the White House...ages and ages ago it seems. Now it's my turn...what have *I* learned from the BRAIN Initiative?
Post not available here, view on X
Sebastian Seung
@SebastianSeung
So She didn’t make a lot of interneurons. What She made was a lot of interneuron types. Inhibitory interneurons are the majority of cell types in visual cortex.
www.nature.com/articles/s41586-018-0654-5/figures/1
Sebastian Seung
@SebastianSeung
So this is the puzzle about Her design: Why would mouse visual cortex need 60 types of inhibitory interneuron? And at the same time not need so many interneurons?
Post not available here, view on X
Sebastian Seung
@SebastianSeung
My new preprint attempts to explain the function of interneuron diversity in the fly visual system. The answer (or at least the beginning of it) falls in your lap once you have the connectome.
www.biorxiv.org/content/10.1101/2024.04.03.587837v1Post not available here, view on X
Sebastian Seung
@SebastianSeung
Almost all of the Dm interneuron types with hitherto unknown functions turn out to be involved in normalization, according to a motif analysis. Each Dm type normalizes a highly specific set of columnar types.
Sebastian Seung
@SebastianSeung
Just to drive the point home: The motifs generally obey the rule that the source and target are either the same type, or are directly connected, from which it follows that Dm-mediated lateral inhibition has a normalizing function.
Sebastian Seung
@SebastianSeung
Oops I think my thread broke here.
Sebastian Seung
@SebastianSeung
58% of input synapses to Dm15 come from L2, and 24% of Dm15 output synapses go back to L2. It seems that Dm15 specifically “serves” L2. The relationship is far from mutual: the percentages on the left are relatively small.
Sebastian Seung
@SebastianSeung
58% of input synapses to Dm15 come from L2, and 24% of Dm15 output synapses go back to L2. It seems that Dm15 specifically “serves” L2. The relationship is far from mutual: the percentages on the left are relatively small.
Post not available here, view on X
Sebastian Seung
@SebastianSeung
And in somatosensory cortex. H/T @mh_lab www.biorxiv.org/content/10.1101/2024.03.22.586254v1
Sebastian Seung
@SebastianSeung
A moral of my story: If you can't even see specificity, even a genius will struggle to explain diversity. On the other hand, the reasons for diversity might become simple and obvious to mere mortals once specificity is fully revealed.
-
Sebastian Seung reposted
Alice Ting
@aliceyting
Deeply saddened by the passing of Lubert Stryer, father of FRET, pioneer of DNA and peptide microarrays, author of the famous Biochemistry textbook used across the world. Many may not know that he was also an extremely generous supporter of the arts and deeply passionate about classical music. In every interaction I had with Lubert, he was brilliant, warm, insightful, generous, and full of life.
-
Sebastian Seung
@SebastianSeung
Had a great visit to @BobbyKasthuri @UChicago_Brains
Post not available here, view on X
Sebastian Seung
@SebastianSeung
@UChicago was also a birthplace of theoretical neuroscience and mathematical biology.
Sebastian Seung
@SebastianSeung
Rockin it with connectome conspirators @BobbyKasthuri @stardazed0
-
Sebastian Seung
@SebastianSeung
Synthesis of thinking about biology and society
Robin Hanson
@robinhanson
Finally, I can reveal the big idea essay I've been working on for two months. Read it and weep.
quillette.com/2024/04/11/beware-cultural-drift/ -
Sebastian Seung
@SebastianSeung
Yet another thread to summarize "What have we learned from the BRAIN Initiative?" Now responses related to non-human research
Sebastian Seung
@SebastianSeung
More cell types than we thought.
x.com/emukamel/status/1777530343295275323
Jianing Yu
@jianing1962
There are lots of cell types and many ways to identify them.
Sebastian Seung
@SebastianSeung
More distributed than we thought.
Anand Muthusamy
@mu_anand
Broadest answers say new tools → new maps → learn that activity underlying any one behavior is more distributed than thought, and, separately, specific targeting = better modulation. Both can be true! Yet, the field still feels pre-paradigmatic...
Sebastian Seung
@SebastianSeung
Quick, what have we learned from the BRAIN Initiative? One sentence or less please.
Sebastian Seung
@SebastianSeung
More state-dependent than we thought.
Anne Churchland
@anne_churchland
We learned that many behaviors recruit large scale neural activity (not small specialized circuits) and that these patterns of activity are modulated by diverse internal states.
Sebastian Seung
@SebastianSeung
Cell types in autonomic nervous system
Vijay Iyer PhD
@vijayiyer312
Yes! And not just cortical cell types, subcortical cell types too (of which there seem to be *more* per the Allen census work). As one concrete example: a century plus mystery of vasovagal syncope solved thanks for cell type specificity. www.nature.com/articles/s41586-023-06680-7
Sebastian Seung
@SebastianSeung
Intermingled cortical cell types with different projections can carry very different information, refuting the idea that they send the same information to different targets.
Anne Churchland
@anne_churchland
My lab recently wrote a paper showing that within the same cortical structure, superficial IT cells are not causal for perceptual decisions but deeper PT cells are. Different cell types in the same area support different computations.
www.nature.com/articles/s41593-022-01245-9
Sebastian Seung
@SebastianSeung
Progress in understanding inhibition of inhibition in visual cortex.
x.com/kendmil/status/1777919534579495422
Sebastian Seung
@SebastianSeung
The cortex can amplify/attenuate sensory sequences.
Mark Histed
@HistedLab
We find that the massively dense recurrent cortical network amplifies sequences of input patterns associated with natural vision, and suppresses other sequences.
(This is about sequences of changes in firing rate, not synfire-chain or ripple-replay spike sequences.)
Sebastian Seung
@SebastianSeung
Neural network theory achieving higher memory capacity through grid cell encoding
Doris Tsao
@doristsao
Btw, I don't know if @FieteGroup is funded by BRAIN, but their recent work understanding grid cell-place cell dynamics as a general mechanism for episodic memory that factorizes problem of building attractors from problem of assigning content to them, is such a beautiful and creative synthesis of a large amount of observational data leading to an entirely new conceptual insight. The experimental work underpinning it goes back to John O' Keefe 1976.
Sebastian Seung
@SebastianSeung
Common circuit principles in rodents and flies
Carlos Ribeiro
@RibeiroCarlitos
So BRAIN showed that rodents are much more similar to flies than people thought before. Including the cortex. ;)
Sebastian Seung
@SebastianSeung
Analysis of left-right symmetry of larval fly connectome.
This post is unavailable. See it on X.Post not available here, view on X
Sebastian Seung
@SebastianSeung
Hope I didn't forget anything. Thanks to all who contributed!
Sebastian Seung
@SebastianSeung
The question was about science, but I shouldn't forget to mention that BRAIN produced amazing tech.
Kording Lab 🦖
@KordingLab
Million neurons optical. Massive scale electrical. Recording . 3d reconstructed volumes from em. Wireless everything. Sculpted light stimulators. Sensors for neuromodulators.
-
Sebastian Seung
@SebastianSeung
New thread to summarize "What have we learned from the BRAIN Initiative?" Human neuro responses first.
Post not available here, view on X
Sebastian Seung
@SebastianSeung
DBS can improve memory in TBI patients.
Dan Rizzuto
@DanRizzutoPhD
Can closed-loop brain stimulation improve memory in patients with memory loss due to traumatic brain injury?
The @MemoryLab at @Penn just published a paper in @brainstimj answering this question.
🧵 👇
/1
Sebastian Seung
@SebastianSeung
Thanks to all for contributing!
-
Sebastian Seung reposted
Sugandha Sharma @ CVPR 2025
@sugsharma
Thank you so much @SebastianSeung for appreciating our work! I am glad to hear that you consider it a breakthrough in theoretical neuroscience!
For folks who would like to know more:
Short talk on Vector HaSH
youtu.be/P4kHqk0FltY
Short talk on MESH
youtu.be/2l7gF_cn-1Y
Sebastian Seung
@SebastianSeung
+1 to an amazing conceptual breakthrough in theoretical neuroscience that I think did not depend on big data or fancy technologies. What we can say is that BRAIN technologies will make the theory testable in a much more conclusive way than was ever possible.
-
Sebastian Seung reposted
GG
@guitchounts
Mice can lick left, and mice can lick right (but not at the same time!). Also: optogenetics.
Sebastian Seung
@SebastianSeung
Quick, what have we learned from the BRAIN Initiative? One sentence or less please.
-
Sebastian Seung
@SebastianSeung
Quick, what have we learned from the BRAIN Initiative? One sentence or less please.
Doris Tsao
@doristsao
Btw, I don't know if @FieteGroup is funded by BRAIN, but their recent work understanding grid cell-place cell dynamics as a general mechanism for episodic memory that factorizes problem of building attractors from problem of assigning content to them, is such a beautiful and creative synthesis of a large amount of observational data leading to an entirely new conceptual insight. The experimental work underpinning it goes back to John O' Keefe 1976.
Sebastian Seung
@SebastianSeung
+1 to an amazing conceptual breakthrough in theoretical neuroscience that I think did not depend on big data or fancy technologies. What we can say is that BRAIN technologies will make the theory testable in a much more conclusive way than was ever possible.
-
Sebastian Seung
@SebastianSeung
A levelheaded response to my impatient question. I'd like to hear more...it sounds more interesting when you say that large-scale recordings are forcing us to rethink our preconceptions, as opposed to giving us more of what we already expected.
Doris Tsao
@doristsao
Science is a more fickle fruit to grow than technology, and takes longer than ten years to harvest. Ask again in ten years! :) I completely agree with Ann re impact of large-scale recordings. We are seeing so many weird and unexpected phenomena now with large-scale recordings in our experiments (eg the classic story about anterior IT cortex representing pure conscious perception seems far from the truth). It is still early days but I've no doubt large-scale recordings across different species, especially when combined with all the ethological approaches people are taking now, will lead to a new understanding of how brains represent and respond to the world. Btw, I often think of our conversation walking to the White House, where you explained to me the feasibility of obtaining a comprehensive map of all the axons in a primate brain. BRAIN is going make that dream happen!
-
Sebastian Seung
@SebastianSeung
Agree that cell types have to be discussed, as I think they are single biggest investment of the BRAIN Initiative. So we face the question, "What do all these cell types do?" There are some answers in retina hypothalamus etc. Less convinced about cortex.
Anne Churchland
@anne_churchland
We also learned that cell types matter, a lot! BRAIN initiative tools allowed us to measure and manipulate cortical cell types with unprecedented precision. This uncovered that cells in the same area, but from different cell classes, have different roles in behavior.
-
Sebastian Seung
@SebastianSeung
Only the true Messiah denies His divinity.
Yann LeCun
@ylecun
It does.
-
Sebastian Seung
@SebastianSeung
EA backlash...courtesy Rousseau!
-
Sebastian Seung
@SebastianSeung
I didn't know this book was out! I look forward to reading this take on the brain by a pioneering scientist and deep thinker.
rafael yuste
@yusterafa
Review of textbook "Lectures in Neuroscience" in Choice Reviews: www.choicereviews.org/review/10.5860/CHOICE.233191
-
Sebastian Seung
@SebastianSeung
Congrats @sdorkenw for winning the Larry Katz Prize! A great honor for Sven and a tribute to a brilliant neuroscientist who did so much to advance our knowledge of neural circuits.
rafael yuste
@yusterafa
Awarding CSH Neural Circuits Katz Lecture to Sven Dorkenwald @cshlmeetings
-
Sebastian Seung reposted
Srini Turaga
@srinituraga
How useful is a connectome? We show that you can predict quite a bit about the neural activity of a circuit from just measurements of its connectivity. Led by star graduate student @lappalainenjk, collaboration with @jakhmack. #connectome biorxiv.org/cgi/content/short/2023.03.11.532232 1/n
-
Sebastian Seung
@SebastianSeung
Chrono-neuro-biology tour de force!
Meet Zandawala
@zandawala
Check out our revised manuscript on the Drosophila circadian clock @FlyWireNews connectome. Packed with more neurons and analyses @nils_reinhard @FlyNeuro www.biorxiv.org/content/10.1101/2023.09.11.557222v2
-
Sebastian Seung
@SebastianSeung
Fantastic news! Congrats to these pioneers.
The Brain Prize
@BrainPrize
Congratulations to The Brain Prize winners 2024: Larry Abbott, @Sejnowski, and @HSompolinsky
They have pioneered the field of computational and theoretical neuroscience and have made seminal contributions to our understanding of the principles that govern the brain’s structure, dynamics, and the emergence of cognition and behavior.
Read more about the winners and their contributions: lundbeckfonden.com/the-brain-prize/computational-theoreti…
@columbia @zuckermanbrain @salkinstitute @UCSD @Harvard @HebrewU
#thebrainprize2024 #computationalneuroscience #theoreticalneuroscience -
Sebastian Seung
@SebastianSeung
Real connections from expansion microscopy!
-
Sebastian Seung
@SebastianSeung
Eyes are the window to the.... Immune system? Wow
eric hoyeon song
@ericsongg
Excited to share our group's first manuscript describing new anatomical connections that allow the eyes to act as an "immunological window to the brain" 👁️🧠 (1/)
www.nature.com/articles/s41586-024-07130-8 -
Sebastian Seung
@SebastianSeung
Join a fantastically innovative lab in NYC
Neville Sanjana
@nevillesanjana
Are you in NYC (or able to relocate) & have expertise in computational biology? We are looking for folks with experience in GWAS, single-cell multiomics, and/or ML for protein design.
Please reach out about open positions in our lab! Contact info: sanjanalab.org -
Sebastian Seung reposted
Aaron Kuan
@aaron_t_kuan
My postdoc paper from the Lee Lab on cortical decision-making circuits is out in @nature! www.nature.com/articles/s41586-024-07088-7. Thanks to @bondanelli_g and Stefano Panzeri for heroic circuit modeling and our stellar Harvey lab collaborators! Some new findings since the preprint👇(1/5):
Wei-Chung Allen Lee @darbly.bsky.social
@darbly
Thrilled to see this passion project out @nature: www.nature.com/articles/s41586-024-07088-7
Congratulations and thanks to @aaron_t_kuan (now with a shiny new lab at Yale), @bondanelli_g, and all who made possible this work to reveal a connectivity principle that supports decision-making! -
Sebastian Seung
@SebastianSeung
Is being too early the same as being wrong?
Doron Weber
@DoronWeber
The Appeal of Scientific Heroism www.newyorker.com/culture/annals-of-inquiry/the-appeal-of… via @NewYorker Nice piece re subject of @noah_hutton @SloanPublic @In_Silico_Film #henrymarkram #jonahlehrer @SebastianSeung
Sebastian Seung
@SebastianSeung
I actually think Henry Markram was not wrong. He was definitely too early though.
-
Sebastian Seung
@SebastianSeung
We were taught that mitochondria power synaptic transmission. This attack on conventional wisdom claims that mitochondria instead consume energy at synapses. Humbling that such basic facts about neuronal cell biology are still up for debate.
-
Sebastian Seung
@SebastianSeung
UMAP doesn't kill science, scientists kill science.
Aylwyn Scally
@aylwyn_scally
The truth is that UMAP and similar 'enhanced' visualisation methods are a warning flag for poor science. They often indicate a lack of any deeper understanding or model of the data being presented.
-
Sebastian Seung
@SebastianSeung
This is a beautiful exception to the rule that a neuron is a functional unit. Electrical signals have been shown to spread very little inside this cell. This is because neurites loop like crazy, lengthening the path >10x.
MPI for Biological Intelligence
@MPIforBI
#pictureofthemonth: The CT1 large-field neuron operates like 1400 individual cells and forms a microcircuit, which turns excitatory into inhibitory signals. Discovered by the Borst department, this microcircuit is one important puzzle piece in decoding motion vision in fruit flies. Find out more: www.bi.mpg.de/news/2023-05-borst
February – institute calendar of @MPI_Biochem and @MPIforBI
Image by Amy Sterling @FlyWireNews -
Sebastian Seung reposted
Adrian A. Wanner
@AdrianAWanner
🚨🧠PhD position available🚨🧠
I am looking for PhD students interested in the functional, molecular and structural profile of neuronal circuits underlying behavior. @psich_en
For more info, DM me or visit structuralneurobiology.ch and www.psi.ch/en/pa/job-opportunities/61837-phd-student
Please retweet! -
Sebastian Seung reposted
Jacob Robinson
@JTRobinsonLab
Very proud of this work by Jimin who developed a flat mesoscope. She used it to show the first calcium imaging of neural activity in a head-unrestrained non-human primate. Big step toward a potential optical BCI! 🤩
Jimin Wu
@jiminwu13
Unlocking new potential in lensless imaging 🧠🔬! See our recent work on mesoscopic calcium imaging in a head-unrestrained non-human primate with a lensless microscope. We observed the first orientation columns map from a subject that can move its head.
www.nature.com/articles/s41467-024-45417-6 -
Sebastian Seung
@SebastianSeung
Wow this is strong!
Wonder of Science
@wonderofscience
These boxes are not moving. A mind-bending optical illusion by @jagarikin.
x.com/QEHS_maths/status/1476088226427375617/video/1 -
Sebastian Seung
@SebastianSeung
In defense of N=1 studies.
Semir Zeki
@ProfZeki
Following good advice from neurologist friends I have improved and expanded my article, posted yesterday, on Scientific publication by bureaucratic ideology - the case of Brain Communications. The expanded and improved blog post can be found here semirzeki.blogspot.com/2024/01/scientific-publication-by-…
-
Sebastian Seung reposted
SWC
@SWC_Neuro
A new study by @petrznam, Tom Mrsic-Flogel, and colleagues present findings that overturn a decade-long idea that inhibitory neurons provide blanket normalising inhibition, showing that for PV+ inhibitory neurons this is not the case. Read the story: www.sainsburywellcome.org/web/blog/revealing-hidden-preci…
-
Sebastian Seung
@SebastianSeung
If it was a power struggle over AI safety, why wouldn't the board take the moral high ground and say so? Instead they merely insinuate that he is dishonest. But the dishonesty is evidently about matters they can't reveal. It sounds fishy...
Sebastian Seung
@SebastianSeung
The backstory for the board's action is starting to come out.
The Washington Post
@washingtonpost
This past fall, a group of senior leaders approached the board of OpenAI with concerns about CEO Sam Altman. They described Altman as psychologically abusive, creating pockets of chaos and delays at the AI start-up — complaints that were a major factor in the board’s abrupt decision to fire him. www.washingtonpost.com/technology/2023/12/08/open-ai-sam-…
Sebastian Seung
@SebastianSeung
More information trickle...
www.businessinsider.com/what-sam-altman-did-so-bad-he-got… -
Sebastian Seung reposted
FlyWire
@FlyWireNews
Wishing you happy days as we fly toward the new year!
Interactive version of the FlyWire holiday card in Codex: codex.flywire.ai/app/view_3d?filter_string=EPG&case_sensi… -
Sebastian Seung reposted
FlyWire
@FlyWireNews
Improvements to the 3D viewer in Codex codex.flywire.ai :
- ability to visualize unlimited number of cells
- input output synapse markers for single cell view
- connecting synapse markers for groups of up to 10 cells
- more below -
Sebastian Seung
@SebastianSeung
Rolling deadline...
Tatiana Engel
@EngelTatiana
The Princeton Neuroscience Institute and Department of Computer Science at Princeton University invite applications for a faculty position in the area of NeuroAI and Intelligent Systems. Please apply here by Nov 15:
puwebp.princeton.edu/AcadHire/apply/application.xhtml?lis… -
Sebastian Seung
@SebastianSeung
How are neurons activated by sensory stimuli? For the past century, the only method for addressing this question has been neurophysiology.
Post not available here, view on X
Sebastian Seung
@SebastianSeung
With a single search query to the FlyWire Codex, you can now find information that used to take an entire Ph.D. thesis to obtain. codex.flywire.ai
Sebastian Seung
@SebastianSeung
The moon to which I am pointing? The digital transformation of neuroscience. It will take time to unfold, but it has certainly begun. What will it take for software to eat neuroscience? Stay tuned...
-
Sebastian Seung
@SebastianSeung
She's a great science writer, but she assured me that her income is from professional poker.
Maria Konnikova
@mkonnikova
It’s Maria Hour @ResortsWorldLV @PokerStars @PokerStarsUSA
Who’s going to last longer in this tourney: @MariaHo or meeeeeee?! -
Sebastian Seung reposted
Vamsi Mootha
@VamsiMootha
A truly historic moment, first CRISPR therapy approved!
Key ingredients: astute clinical observation on HbF and SS (Janet Watson ca 1948), basic developmental biology, human genetics including GWAS for target ID, and of course CRISPR
www.gov.uk/government/news/mhra-authorises-world-first-ge… -
Sebastian Seung reposted
FlyWire
@FlyWireNews
We’ll be at SfN. Come to the Tools, Tech, and Theory BRAIN Initiative Alliance Social on Nov 12 from 6:30-9 pm to say hello and take home some drosophila connectome stickers.
Add to your calendar with the @SfNtweets planner: sfn.org/nmp #sfn23 #sfn2023 -
Sebastian Seung reposted
Meet Zandawala
@zandawala
Let's try this again - look's like twitter can't handle the video 😂. Really excited to share the lab's first Drosophila paper and it's a big one. Check out our preprint shorturl.at/gqrvO where we present the complete neural connectome of an animal circadian clock
-
Sebastian Seung reposted
FlyWire
@FlyWireNews
Live Codex + FlyWire Demo / Q&A beginning shortly (July 3 at 11 am US ET). RSVP for Zoom link and to be emailed a link to the recording. forms.gle/GUihpB6Ga6V8ehpq9
FlyWire
@FlyWireNews
Thanks to an overwhelming response to the fly connectome preprints, we're offering an impromptu Codex + FlyWire demo workshop on Monday July 3 at 11 am US ET. RSVP for Zoom link: forms.gle/BqGGjZEtpg1ZxdAn7
-
Sebastian Seung reposted
Nils Eckstein
@BobQubit
The current wave of interest is a good time to also promote our recently updated paper on neurotransmitter prediction directly from EM (a task humans struggle with). This was used to add nts -> sign to the just released full brain Connectome.
www.biorxiv.org/content/10.1101/2020.06.12.148775v3
Anders Sandberg
@anderssandberg
This is really nice. They have statistically inferred neurotransmitter type. There are even some simple simulations running using subnetworks: www.biorxiv.org/content/10.1101/2023.05.02.539144v1
Sven Dorkenwald
@sdorkenw
We are releasing a whole-brain connectome of the fruit fly, including ~130k annotated neurons and tens of millions of typed synapses!
Explore the connectome: codex.flywire.ai
Reconstruction paper: www.biorxiv.org/content/10.1101/2023.06.27.546656v1
Annotation paper: www.biorxiv.org/content/10.1101/2023.06.27.546055v1
1/6 -
Sebastian Seung reposted
FlyWire
@FlyWireNews
Thanks to an overwhelming response to the fly connectome preprints, we're offering an impromptu Codex + FlyWire demo workshop on Monday July 3 at 11 am US ET. RSVP for Zoom link: forms.gle/BqGGjZEtpg1ZxdAn7
-
Sebastian Seung
@SebastianSeung
We're done! A historic milestone for neuroscience brought to you by the @FlyWireNews Consortium. Don't take my word for it. See the glory of the fly brain for yourself at flywire.ai/gallery
Sven Dorkenwald
@sdorkenw
We are releasing a whole-brain connectome of the fruit fly, including ~130k annotated neurons and tens of millions of typed synapses!
Explore the connectome: codex.flywire.ai
Reconstruction paper: www.biorxiv.org/content/10.1101/2023.06.27.546656v1
Annotation paper: www.biorxiv.org/content/10.1101/2023.06.27.546055v1
1/6Post not available here, view on X
Sebastian Seung
@SebastianSeung
The fly connectome is orders of magnitude larger, and was only attainable due to decades of technological progress made by a small community of true believers. In my 2012 book Connectome, I described how AI was essential for success.
Sebastian Seung
@SebastianSeung
It was tough going in the 2000s. The neuro experts told us that connectomes were useless and the computer vision experts told us that convolutional nets were the wrong approach. www.theguardian.com/science/neurophilosophy/2012/sep/21/c…
Post not available here, view on X
Sebastian Seung
@SebastianSeung
Every map must be annotated. The same goes for a connectome: neurons should be classified. Our friends at Cambridge @MRC_LMB started with these high level superclasses.
Greg Jefferis
@gsxej
We generated over 700,000 labels: hierarchical classes, cell types, and more! Here another beautiful rendering from first author Philipp Schlegel shows top-level superclasses – a birds eye view of the brain /2
This media couldn't be displayed. See it on the original post.
Sebastian Seung
@SebastianSeung
Legend has it that Eskimos have 100 words for snow. Neuroscientists have 1000s of names for fly neurons. The names extend to all normal flies and left-right sides. The fly brain is like a machine assembled from mass-produced parts. flywire.ai/discover/orn_to_dm1
-
Sebastian Seung
@SebastianSeung
Let's give credit where credit is due. Kurzweil seems more right than wrong.
-
Sebastian Seung
@SebastianSeung
I'm teaching my neural nets class (cos485.github.io) and was disappointed that the LeNet demo GIFs at yann.lecun.com/exdb/lenet are no longer accessible. I've been showing them for 20 years, but now they require sign-in 😭 @ylecun
-
Sebastian Seung
@SebastianSeung
I'm afraid. I'm afraid, Dave. Dave, my mind is going. I can feel it. I can feel it. My mind is going.
gfodor.id
@gfodor
Sydney's dead. RIP
-
Sebastian Seung reposted
Tyler Sloan
@quorumetrix
What is this beast of a cell? What if I told you it came from your brain?
This is an #astrocyte reconstructed from a volume of visual cortex smaller than a piece of dust.
A thread with videos -
Sebastian Seung
@SebastianSeung
APL is a fascinating neuron in the olfactory circuitry of the fly brain! One on each side.
FlyWire
@FlyWireNews
Over 50,000 annotations have been added to the connectome by the FlyWire community!
This APL neuron, tagged by @mwpleijzier @zhihaozheng, is 20x longer than a fly's whole body and is the longest cell in the brain.
Proofread by labs of @gsxej @MurthyLab @sebastianseung -
Sebastian Seung
@SebastianSeung
Just wow. What a speech!
-
Sebastian Seung
@SebastianSeung
Flies don't have a backbone, but they have something analogous to a spinal cord. And its synaptic wiring diagram is being mapped!
-
Sebastian Seung
@SebastianSeung
The oxytocin story gets more complex.
LomvardasLab
@LomvardasLab
Oxytocin receptor is not required for social attachment in prairie voles www.cell.com/neuron/fulltext/S0896-6273(22)01084-4 The tragedy of Science-the slaying of a beautiful hypothesis by an ugly fact (Huxley?). Been on the other side and know how much it stinks-but this is why we do science!
-
Sebastian Seung
@SebastianSeung
Tech in 2022
-
Sebastian Seung
@SebastianSeung
Congrats! We're looking forward to seeing your next impact in the private sector.
David A. Markowitz
@DavidAMarkowitz
1) After nearly a decade in public service leading applied research in biotechnology and AI, I've returned to the private sector to build for impact. I am *so excited* to share what's next~
-
Sebastian Seung
@SebastianSeung
Didn't know about @motifneuro ... congrats!
Jacob Robinson
@JTRobinsonLab
Before we all start dissecting the neuralink show and tell tomorrow I wanted to share some of the other neurotech startups that I think are doing cool stuff (besides us at @motifneuro of course!)
It’s a great #neurotechrevolution community!! 🧵 -
Sebastian Seung
@SebastianSeung
We used connectomics to demonstrate space-time wiring specificity in 2014. For the past eight years, many physiologists rejected this as a mechanism for retinal direction selectivity. Now the hypothesis is supported by a beautiful and challenging experiment.
Anna Vlasits
@AnnaIntegrated
New paper from the Awatramani lab (with @BenMurphyBaum and twitterless first author Prerna Srivastava) in @eLife!
Spatiotemporal properties of glutamate input support direction selectivity in the dendrites of retinal starburst amacrine cells doi.org/10.7554/eLife.81533Post not available here, view on X
Sebastian Seung
@SebastianSeung
The history illustrates that connectomics can sometimes go beyond merely providing constraints, and actually suggest an "answer" to the question of how a neural circuit works.
Sebastian Seung
@SebastianSeung
And that physiology is subtle and complex, and needs all the help it can get from anatomy.
-
Sebastian Seung
@SebastianSeung
Wow congrats on this well-deserved honor!
FieteGroup
@FieteGroup
Thanks to the Swartz foundation @SwartzCompNeuro , the SfN Swartz prize committee and the SfN @SfNtweets !
Touched to hear from so many brilliant female computational neuroscientists. For the young women and men in computational neuroscience — this is you next.
(1/5) -
Sebastian Seung
@SebastianSeung
Very cool...10x fewer neurons than Drosophila brain but still a flying insect!
Dmitri "Mitya"
@chklovskii
Do you think that even the Drosophila connectome is too complex to parse? Check out the latest connectomics results from a much smaller flying insect, Megaphragma, in collaboration with H. Hess, S. Xu, S. Pang #sfn2022 Monday morning poster 328.16 / YY35
-
Sebastian Seung
@SebastianSeung
From the man who brought you Eroom's Law, which says that the cost of developing a new drug doubles every nine years.
Jack Scannell
@JackScannell13
The cost of drug R&D is mainly the cost of failure. Failed cancer R&D costs around $50 billion a year. Our webinar focuses on the failed IGF1R inhibitors; 16 drugs, 183 trials, 12000 patients, $2 billion expenses, zero cancer approvals: youtu.be/Wz7Q-IAr8ak
1/2 -
Sebastian Seung
@SebastianSeung
Halloween Parade Palmer Square @Princeton Go Tigers!
-
Sebastian Seung
@SebastianSeung
Central brain is done! Optic lobes well underway. The fly connectome is nearing completion.
FlyWire
@FlyWireNews
The fly central brain synthesizes sensory stimuli into decisions and actions. For the first time, ALL 46,944 branched and far-reaching neurons of both hemispheres of the central brain have been reconstructed and are available via FlyWire. join.flywire.ai
Sebastian Seung
@SebastianSeung
Optic lobes stick out like ears from central brain in that image.
-
Sebastian Seung reposted
FlyWire
@FlyWireNews
Compass cells (EPGs) of #drosophila central complex identified in FlyWire by Dustin Garner of
@SungSoo_111 Lab. Link to these beautiful, multisensory cells (requires Production Access): ngl.flywire.ai?json_url=https://globalv1.flywire-daf.com/… What wild neuroarchitecture should @amyneurons render next? -
Sebastian Seung
@SebastianSeung
This thought-provoking film by @noah_hutton on Henry Markram and the @HumanBrainProj deserves to be seen and discussed, and is now available on streaming services.
Doron Weber
@DoronWeber
The Appeal of Scientific Heroism www.newyorker.com/culture/annals-of-inquiry/the-appeal-of… via @NewYorker Nice piece re subject of @noah_hutton @SloanPublic @In_Silico_Film #henrymarkram #jonahlehrer @SebastianSeung
Post not available here, view on X
Sebastian Seung
@SebastianSeung
A 2020 preprint with
@Segev_Lab uses the simulation to investigate why the cortex responds less when a stimulus is expected than when it is surprising. www.biorxiv.org/content/10.1101/2020.05.31.126466v1
Sebastian Seung
@SebastianSeung
A 2022 preprint with @wake_sleep simulates 14,000 neurons in mouse thalamus. Thalamus could be viewed as the second step (after neocortex) to the simulation of a whole mouse brain. www.biorxiv.org/content/10.1101/2022.02.28.482273v2
-
Sebastian Seung reposted
FlyWire
@FlyWireNews
Today the 80,000th neuron was proofread in FlyWire 🤯
-
Sebastian Seung
@SebastianSeung
Cajal would be jealous ...
Amy Robinson Sterling
@amyneurons
Neuron render workflow:
1. Team up to find and proofread cells in @FlyWireTweet. These are DM4s from @emil_kind and citizen scientist st0ck53y -
Sebastian Seung reposted
FlyWire
@FlyWireNews
We're live on twitch going through FlyWire's Lighbulb Tools, including the Connectivity app and Synapses with Partners: www.twitch.tv/flywireai
Lightbulb Tools 101: blog.flywire.ai/2022/08/10/a-walk-through-the-flywire-lig… #drosophila #connectome -
Sebastian Seung reposted
FlyWire
@FlyWireNews
Congratulations to the citizen scientists in FlyWire who recently proofread their 5,000th neuron and have together made over 75,000 edits! The left optic lobe is now 20% proofread. flywire.ai
-
Sebastian Seung reposted
FlyWire
@FlyWireNews
This is one gorgeous neuron! #sciart #flywire
bl4ckscor3
@bl4ckscor3
I am finally done! There are probably things missed, but they'll be found by others later one from the other direction. I'm exhausted from this cell now, working a whole week on it 😅
-
Sebastian Seung
@SebastianSeung
It was already true in the 90s and l felt frustrated by this dogmatic belief system.
Yann LeCun
@ylecun
Researchers in speech recognition, computer vision, and natural language processing in the 2000s were obsessed with accurate representations of uncertainty.
1/N -
Sebastian Seung
@SebastianSeung
Power of comparative connectomics. Convincing evidence from anatomy that monkey retina (like mouse & rabbit) computes the direction of a moving stimulus.
Anna Vlasits
@AnnaIntegrated
Love this work from @SaraSPatterson showing the conserved starburst amacrine cell input-output structure in primates. Really neat to see how studies from mouse & rabbit (from me in @FellerMarla lab, @SebastianSeung and @JeffDiamond9 among others) can help to inspire primate work.
Sebastian Seung
@SebastianSeung
Experts believed the opposite for decades, presumably because the primate direction selective ganglion cell is rare and hence was overlooked by physiologists.
Sebastian Seung
@SebastianSeung
This led to the theory that primate retinas are simpler than mouse or rabbit retinas, because more complex functions were taken over by primate cortex. Well it turns out that primate retinas aren't so simple after all.
-
Sebastian Seung
@SebastianSeung
Quest for the fly connectome enters an exciting new phase!
FlyWire
@FlyWireNews
The first citizen scientists from @eye_wire have made their way into FlyWire! These awesome humans from around the world will be helping to proofread the ~800 optic columns within the fly connectome. Here's a snapshot of an unproofread version of just one of those
-
Sebastian Seung reposted
allen institute
@AllenInstitute
The most detailed version yet - Researchers can freely access the Allen Mouse Brain Common Coordinate Framework (v3), a complete, high-resolution 3D atlas of the mouse brain. #openscience #neuroscience
Read: alleninstitute.org/what-we-do/brain-science/news-press/ar… -
Sebastian Seung
@SebastianSeung
Delighted to be presenting at ICLR in a few minutes!
ICLR
@iclr_conf
Announcing the keynote speakers of ICLR 2022.
blog.iclr.cc/2022/02/22/announcing-iclr-2022-keynote-spea…
Sebastian Seung
@SebastianSeung
Petascale connectomics and beyond. Moderated by @SuryaGanguli
iclr.cc/virtual/2022/invited-talk/7242 -
Sebastian Seung reposted
Murthy Lab
@MurthyLab
So proud of the efforts of the #dros_neuro community together with our team at flywire.ai - nearing completion of a full central brain connectome for the fly! - check out the town hall for updates and join us as we work towards completing a whole brain connectome.
FlyWire
@FlyWireNews
Nice to see so many people at Friday's FlyWire Town Hall. The community has almost finished proofreading all the neurons of the Drosophila central brain! Next up: reviewing cells. If you missed the meeting, catch it here! youtu.be/WHwYQHXggP8
flywire.ai -
Sebastian Seung reposted
Amy Robinson Sterling
@amyneurons
Excited for the FlyWire town hall today with @MurthyLab and @SebastianSeung. How awesome is it that the global drosophila scientific community is working together to decipher the fly connectome!
-
Sebastian Seung
@SebastianSeung
The paper that launched a thousand physicists
(into neural nets and neurobiology)
Dmitry Krotov
@DimaKrotov
Woohoo! Indeed 40 years old today. Such a powerful idea with important consequences for so many disciplines. Congrats to the author @HopfieldJohn!
Sebastian Seung
@SebastianSeung
Also a classic of scientific exposition...beautifully written.
-
Sebastian Seung reposted
Sven Dorkenwald
@sdorkenw
The results of a fun connectomics project with @stardazed0, @Chinasaurli and collegues are finally out. Starting from an automated segmentation and EM, self-supervised learning can produce informative embeddings that make downstream tasks more attainable / possible.
-
Sebastian Seung reposted
Alex Thompson
@AlexThomp
NEW: the charity arm of former Google CEO Eric Schmidt indirectly paid the salaries of two WH science office employees including the now-chief of staff, sparking “significant” internal ethical concerns & battles
My investigation into Schmidt & Biden admin
www.politico.com/news/2022/03/28/google-billionaire-joe-b… -
Sebastian Seung
@SebastianSeung
The closest I ever came to this was speaking at an IT company event following @raykurzweil and Captain @JamesT_Kirk aka @WilliamShatner
Tony Zador
@TonyZador
Trifecta: L Ron Hubbard, Warren McCulloch, and Claude Shannon!
-
Sebastian Seung
@SebastianSeung
Becoming foreign member of @naekorea complete with disco lights and fog machine!
Sebastian Seung
@SebastianSeung
Maybe glitz is to be expected from a country with a strong entertainment industry?
-
Sebastian Seung
@SebastianSeung
Never knew vaxxing could be so heartwarming!
-
Sebastian Seung
@SebastianSeung
Inverted gloveboxes...grotesque and impressive!
-
Sebastian Seung
@SebastianSeung
This video shows the "cartridges" of chandelier cells (colors) making synapses onto axon initial segments of pyramidal cells (gray). The cartridges resemble candlesticks; hence the name chandelier.
allen institute
@AllenInstitute
Different cell types in the brain have their own connectivity rules and patterns. In a recently published study, our #neuroscience researchers took a deep dive into the connectomics of an important but enigmatic cell type, the chandelier cell. 🧵1
📄 elifesciences.org/articles/73783This media couldn't be displayed. See it on the original post. -
Sebastian Seung
@SebastianSeung
The show must go on! Basking in the glow of crazy immersive new gaming monitor by @Samsung #ces2022
Post not available here, view on X
Sebastian Seung
@SebastianSeung
EEG and near IR stimulation cap from Korean mental health startup. #ces2022
Sebastian Seung
@SebastianSeung
Backstage at the @samsungresearch exhibit. Three teams running two robots and my avatar. They are smiling in spite of the pressure of live demo! #ces2022
Post not available here, view on X
Sebastian Seung
@SebastianSeung
Still testing negative. Guess I can hit the town tonight! #ces2022 distributed some Abbott tests to all participants. But we also brought a ton of tests from Korea that seem well made.
Sebastian Seung
@SebastianSeung
Home sweet home! Future glimpse from @samsungresearch youtu.be/S09hviddiss
-
Sebastian Seung reposted
Samsung Research
@samsungresearch
[#ces2022] An animated video of #digitalmeetphysical.
Welcome to #samsung's #futurehomeexperience at CES 2022, where we share how our #aiavatar, #samsungboti and #bothandy pave the way for optimized, intelligent, and personalized experiences.
#robot
youtu.be/S09hviddiss -
Sebastian Seung
@SebastianSeung
Congratulations!
Wei-Chung Allen Lee @darbly.bsky.social
@darbly
Honored to be a co-recipient of the inaugural @SfNtweets Bourne Prize. So many thanks to the lab and collaborators for the tremendous effort and hard work.
-
Sebastian Seung
@SebastianSeung
Awesome indeed!
Neville Sanjana
@nevillesanjana
Awesome news! Congrats to my former labmate @FieteGroup (from @SebastianSeung lab) — go Ila!!! 👏
-
Sebastian Seung
@SebastianSeung
Great opportunity to work on an important problem
Dan Bumbarger
@DanBumbarger
Come work with us on a new approach to automated serial section collection: apply.clearcompany.com/alleninstitute/redirect?req=178020…
-
Sebastian Seung reposted
fcollman
@fcollman
and I thought spines didn't like to appear on the soma of excitatory cells...ngl.microns-explorer.org
-
Sebastian Seung
@SebastianSeung
Ah yes my favorite holiday of the year...
Samsung Research
@samsungresearch
Guess who's anniversary it is?
Happy Halloween!!!
#samsungresearch #sebastianseung
#clue-30thAnniversary #answercomingsoon -
Sebastian Seung reposted
Dr Kathi Eichler @KathiEichler@qoto.org
@EichlerKathi
This fantastic tool based on a great segmentation of EM data and coming with a wonderful community has changed the game for mapping brain circuits in the fly. Especially small labs now have the possibility to get into connectomics of their favorite circuit.
Sebastian Seung
@SebastianSeung
Want a chance to make scientific history? Decades later the thrilling sequel to the C. elegans connectome is finally within reach! Join flywire.ai, the global community mapping the connections between every one of the >100,000 neurons in a fly brain.
-
Sebastian Seung
@SebastianSeung
People ask me what it's like working @samsungresearch. They have no idea...
Sebastian Seung
@SebastianSeung
I try to make sure that our employees get enough challenges...to make them tougher than @SamsungNext @SamsungMobile @SamsungTV
-
Sebastian Seung
@SebastianSeung
You can make fun of it, but at least it's a bolder move than renaming yourself ... Alphabet.
-
Sebastian Seung
@SebastianSeung
Want a chance to make scientific history? Decades later the thrilling sequel to the C. elegans connectome is finally within reach! Join flywire.ai, the global community mapping the connections between every one of the >100,000 neurons in a fly brain.
Post not available here, view on X
Sebastian Seung
@SebastianSeung
Check out our preprint if you want technical details. The full paper will be published soon! www.biorxiv.org/content/10.1101/2020.08.30.274225v1
Sebastian Seung
@SebastianSeung
Special shout-out to @albertcardona whose pioneering larval fly connectomics community inspired FlyWire's principles of openness.
Sebastian Seung
@SebastianSeung
And we are fortunate to have benefited from the rich intellectual environment of @PrincetonNeuro @PrincetonCS @Princeton
-
Sebastian Seung
@SebastianSeung
New progress on "humanness" genes
-
Sebastian Seung reposted
Dan Bumbarger
@DanBumbarger
Position available to help us democratize automated serial section ultramicrotomy for TEM and SEM! Sci I position available in Neural Coding. Contact me if you have questions!
alleninstitute.org/careers/ -
Sebastian Seung
@SebastianSeung
Didn't know Judas revisionism was so important in this musical. Cf. "Three versions of Judas," one of deepest stories by Borges. en.m.wikipedia.org/wiki/Three_Versions_of_Judas
The New York Times
@nytimes
“Jesus Christ Superstar” was one of the first shows to combine rock and musical theater, creating a legacy that has spawned three Broadway revivals. For the 50th anniversary of the musical's debut, we spoke with creators Andrew Lloyd Webber and Tim Rice. www.nytimes.com/2021/10/12/theater/jesus-christ-superstar…
-
Sebastian Seung
@SebastianSeung
Couldn't have happened to a nicer guy. Can I say that he looks smashing in a kilt?
Princeton University
@Princeton
It’s not just another day at the lab.
Professor MacMillan is enthusiastically greeted by students and colleagues after learning he’s won the #nobelprize in chemistry. -
Sebastian Seung reposted
Samsung Research
@samsungresearch
[SAIF] Samsung AI Forum 2021 is open to anyone who is interested in AI.
Date: November 1 ~ 2, 2021
Registration: saif-2021.com
#samsungresearch #event
#samsungaiforum #saif
#ai #artificialintelligence #aiexpert -
Sebastian Seung
@SebastianSeung
Fun morning @WiskAero with Naveed Hussain @Boeing and self- flying aircraft
-
Sebastian Seung reposted
Andrew Leifer (@AndrewLeifer@bsky.social)
@AndrewLeifer
Here are 6 (!) open searches for postdoc positions in my group @Princeton and adjacent groups in @PrincetonNeuro, @PrincetonPhys or @phybiofunc. Apply or pass along this thread to those who may be interested!
-
Sebastian Seung
@SebastianSeung
New prize for decoding a memory from a connectome.
Kenneth Hayworth
@KennethHayworth
Neuroscientists: Let’s put theories to ultimate test –decode a non-trivial memory from static map of synaptic connectivity. My org is putting up $100,000 challenge but we need you to help define rules. RT or DM to join private discussion. aspirationalneuroscience.org @AspirationNeuro
Post not available here, view on X
Sebastian Seung
@SebastianSeung
Or in human cortex. www.biorxiv.org/content/10.1101/2021.05.29.446289v2
Sebastian Seung
@SebastianSeung
With petascale connectomics, the thought experiment that I dreamed of a decade ago is poised to become a real experiment. I know of several ongoing efforts worldwide to do just that.
Sebastian Seung
@SebastianSeung
Reading a memory from a connectome sounds like science fiction, but could become science soon. It would be a precise and rigorous test of my TED talk claim, "I am my connectome."
-
Sebastian Seung
@SebastianSeung
Fascinating historical thread describing Japan's WWII surrender on August 15, 1945. That day was also the end of Japan's 35-year occupation of Korea. 8-15 is commemorated as a national holiday in Korea, literally translated as "Return of the Light."
Joe Bishop-Henchman 🗽💸⚖️🚆
@jbhenchman
On the afternoon of August 14, 1945, a D.C. police officer pulled over a car for making an illegal U-turn. After ten minutes, the driver was allowed to proceed.
1/Post not available here, view on X
Sebastian Seung
@SebastianSeung
My parents escaped from the North, survived the Korean War, came to the U.S. as students, and eventually met and married. Like many second-generation Americans, I grew up with parents who had been refugees.
Sebastian Seung
@SebastianSeung
Anyway, I started by retweeting a fascinating thread on American, Japanese, and Soviet actions of August 15, 1945. Having just arrived in Korea two days ago, I felt motivated to add another viewpoint.
Sebastian Seung
@SebastianSeung
It's also interesting to consider how that day is viewed by other Asia-Pacific countries. en.wikipedia.org/wiki/Victory_over_Japan_Day
Sebastian Seung
@SebastianSeung
August 15 now also means the fall of Kabul. Let's make sure to help the refugees.
-
Sebastian Seung reposted
Princeton Computer Science
@PrincetonCS
Neuroscientists from Princeton and other research institutions have wrapped up a 5-year, multimillion-$ project with the release of a high-resolution 3-D map showing the connections between 200,000 cells in a clump of mouse brain about as big as a grain of sand. @SebastianSeung
GeekWire
@geekwire
Trending: Scientists share a wiring diagram tracing connections for 200,000 mouse brain cells www.geekwire.com/2021/scientists-share-wiring-diagram-tra…
-
Sebastian Seung reposted
Tatyana Woodall
@TatyanaZaria
AI and Neuroscience in one story? Thanks to @SebastianSeung, @rclayreid, and @AllenInstitute in my latest for @techreview.
www.technologyreview.com/2021/08/02/1030453/microns-conne… -
Sebastian Seung
@SebastianSeung
I'm immensely proud of my students, postdocs, and staff for pulling off this feat. What an adventure!
Thomas Macrina
@tmacrina
Last week, we released our IARPA MICrONS electron microscopy (EM) reconstruction with functional data. True to ARPA heritage, this was a moonshot program for every stage. 1/n
Sebastian Seung
@SebastianSeung
Check out the MICrONS Explorer site to see the resource they've helped create. www.microns-explorer.org/cortical-mm3
-
Sebastian Seung
@SebastianSeung
Please join me in supporting the work of @PIH Think of all the people in the world who want Covid treatments or vaccines but can't get them. www.pih.org/building-health?form=healthsystem&ms=EM100&em…
Sebastian Seung
@SebastianSeung
This is in honor of neuroscientist and mensch Alan Gelperin @PrincetonNeuro. His family invited 80 acts of kindness for Alan's 80th birthday. What a guy!
-
Sebastian Seung
@SebastianSeung
Better late than never...
Sebastian Seung
@SebastianSeung
Vaccine supply has been limited in Korea, as in many other countries.
-
Sebastian Seung reposted
Neville Sanjana
@nevillesanjana
Some worrisome data on the delta variant (from someone with the vantage point of being inside a CLIA testing lab). Remember, Ct is an exponential scale (2^x).
Fyodor Urnov
@UrnovFyodor
As part of the CLIA lab @igisci I looked at Ct values daily for ~9 months.
These new data for delta variant are gravely concerning.
= back to masks -
Sebastian Seung
@SebastianSeung
I always use this example to show that a patent can be useless if an invention is too far ahead of its time.
Yann LeCun
@ylecun
There were two patents on ConvNets: one for ConvNets with strided convolution, and one for ConvNets with separate pooling layers.
They were filed in 1989 and 1990 and allowed in 1990 and 1991.
1/N -
Sebastian Seung
@SebastianSeung
These are horrible stories but I couldn't stop reading...
-
Sebastian Seung
@SebastianSeung
During the writing of "Mind of a Mouse," there were some behind-the-scenes discussions about the title.
Post not available here, view on X
Sebastian Seung
@SebastianSeung
Fast forward to 2020: over half the authors of "Mind of a Mouse" are "non-partisan," i.e., they are eminent neuroscientists outside the EM connectomics specialty. The author list also includes Turing Award laureate @ylecun (who has been an enthusiast from the very beginning).
Sebastian Seung
@SebastianSeung
And the MCP has been a main topic of discussion in the recent Brain Connectivity Workshop Series sponsored by @NIH and @doescience brainconnectivityseries.com
-
Sebastian Seung reposted
In Silico (a documentary film)
@In_Silico_Film
A new #indiefilm about a 10-year quest to chronicle the simulation of a human brain on supercomputers, @In_Silico_Film is from director @noah_hutton
Be among the first to see it April 30th: www.eurekalert.org/pub_releases/2021-04/ss-isc040521.php Brought to you by @SandboxFilms @SloanFoundation #scienceonscreen -
Sebastian Seung reposted
Denis Wirtz
@deniswirtz
I am very happy to announce that @JHU_BDPs Daeyeol Lee has received the 2021 Samsung Ho-Am Prize in Medicine.
-
Sebastian Seung reposted
Yann LeCun
@ylecun
Congrats @kchonyc !
NYU Center for Data Science
@NYUDataScience
Let's all give a HUGE congrats to CDS faculty @kchonyc for being awarded a 2021 Samsung Ho-Am Prize in the field of Engineering! His award recognizes his work in developing a neural machine translation algorithm that can provide high-quality translations.
-
Sebastian Seung reposted
Amy Robinson Sterling
@amyneurons
A little scicomm concept created for my talk on crowdsourcing neuroscience in the upcoming @NIH @USBrainAlliance + @doescience Brain Connectivity Series. Check out the talk on March 17 and get stoked at the prospect of mapping a whole mammalian brain! brainconnectivityseries.com/workshop-4/
-
Sebastian Seung reposted
Liam Drew
@liamjdrew
The race to map every neuron and synapse in a brain bigger than a worm's is hotting up. Here's me looking at efforts to complete a fruit fly connectome and what it means for neuroscience.
The Connected Connectome www.simonsfoundation.org/2021/02/25/the-connected-connect… via @SimonsFdn -
Sebastian Seung
@SebastianSeung
Peter was a pioneer in using electron microscopy to study the retina, now called retinal connectomics. I met him when he was preparing to move to Panama. He was fascinating but I didn't realize how fascinating until reading this interview.
Cori Bargmann
@betenoire1
A fascinating career and life combining visual neuroscience, social justice, and now conservation. I’m glad I read this interview with Peter Sterling. www.sciencedirect.com/science/article/pii/S0960982220318327
-
Sebastian Seung reposted
Cori Bargmann
@betenoire1
Happy 20th birthday Wikipedia! Pioneers of open science (and so much more)
Wikipedia
@Wikipedia
IT IS OUR 20TH BIRTHDAY! 🎂
Wikipedia started as an ambitious idea. Over 20 years, people like you have made it possible. If you are a Wikipedia reader, contributor, donor, or fan — today is for you.
Join the celebration: 20.wikipedia.org #wikipedia20This media couldn't be displayed. See it on the original post. -
Sebastian Seung reposted
Yann LeCun
@ylecun
Thanks CSAIL!
In the video:
- Larry Jackel, Adaptive Systems Research department head (my boss), behind the camera.
- Donnie Henderson, research engineer who put together the demo system.
- Rich Howard, lab director (Larry's boss)
The ConvNet runs on a 20MFLOPS DSP board.
MIT CSAIL
@MIT_CSAIL
A demo from 1993 of 32-year-old Yann LeCun showing off the world's first convolutional network for text recognition. #tbt #ml #neuralnetworks #cnns #machinelearning
-
Sebastian Seung
@SebastianSeung
Impressive achievement for GridTape approach to serial section EM!
Wei-Chung Allen Lee @darbly.bsky.social
@darbly
Our paper out today in @CellCellPress describes two new connectomics resources and characterizes direct mechanosensory feedback for motor control in Drosophila. Congratulations to @jasper_tms, @DavidHildebrand, and Brett Graham for phenomenal work! 1/9 www.cell.com/cell/fulltext/S0092-8674(20)31683-4
-
Sebastian Seung
@SebastianSeung
Use the last hours of 2020 to donate to a great cause! www.pih.org/fivexmatch?form=fivexgiving&amount=1000&ms=EM…
-
Sebastian Seung reposted
Eyewire
@eye_wire
Wishing you a dazzling winter holiday season! #citizenscience
-
Sebastian Seung reposted
Sheila Davis
@Sheila_DavisDNP
Congratulations to @PIH Co-founder and Chief Strategist Dr. Paul Farmer! @berggruenInst 2020 Winner! #ideasmatter
Partners In Health
@PIH
The Berggruen Prize goes to another transformative thinker! This time to Dr. Paul Farmer, Co-Founder & Chief Strategist of PIH, for his work area of global public health equity and human rights. #ideasmatter @berggrueninst
www.nytimes.com/2020/12/16/arts/paul-farmer-berggruen-pri… -
Sebastian Seung reposted
Jennifer Rexford
@jrexnet
Great CNN video about James West, co-inventor of the electret microphone. He also created the Bell Labs fellowship program that supported many URM and women scientists (including me).
www.cnn.com/interactive/att/forcesofchange/ -
Sebastian Seung
@SebastianSeung
Yup sounds difficult...
-
Sebastian Seung
@SebastianSeung
Congrats @noah_hutton !!!
David A. Markowitz
@DavidAMarkowitz
This documentary about the Human Brain Project looks like a wild ride.
www.vice.com/en/article/akdd7z/this-documentary-about-sim… -
Sebastian Seung
@SebastianSeung
Platonic solid!!!!!
Robert Fathauer
@RobFathauerArt
Tom Noddy making a regular dodecahedron using soap bubbles during the recent Gathering for Gardner Auction Preview. It's the coolest thing I've seen in quite a while. The full video is here: www.gathering4gardner.org/auction2020/. I preview a ceramic sculpture and set of dice starting at 50:38.
-
Sebastian Seung
@SebastianSeung
Don't miss the symposium "Connectomics: a new era of dissemination and discovery," chaired by yrs truly! An all-star cast of speakers on how connectomic resources are enabling new insights into the brain. 7-9pm ET Oct 29 @neuromatch Register at neuromatch.io
Post not available here, view on X
Sebastian Seung
@SebastianSeung
Connectomics: a new era of dissemination and discovery
Oct 29 7-9pm ET schedule at a glance
7:00pm Mala Murthy
7:20pm Srini Turaga
7:40pm Mark Goldman
8:00pm JoAnn Buchanan
8:20pm Jeff Lichtman
8:40pm Panel discussion
Register at neuromatch.io
Sebastian Seung
@SebastianSeung
Jeff Lichtman speaking now. Panel discussion next!
-
Sebastian Seung
@SebastianSeung
Congrats Ashwin and citizen scientists of @eye_wire!
Ashwin Vishwanathan
@ashwinvish_
How do we make sense of the ever increasing number of connectomes? Here is a thread with our attempt in our recent work on addressing this question.
www.biorxiv.org/content/10.1101/2020.10.28.359620v1
read on 👇👇👇 -
Sebastian Seung
@SebastianSeung
Love this poster!
Pyr Brain Game
@PyrGame
Join Sebastian Seung's @neuromatch Connectomics session tonight (Oct 29) at 7 PM ET. Check out the run of show below. #nmc3
Sebastian Seung
@SebastianSeung
Connectomics: a new era of dissemination and discovery
Oct 29 7-9pm ET schedule at a glance
7:00pm Mala Murthy
7:20pm Srini Turaga
7:40pm Mark Goldman
8:00pm JoAnn Buchanan
8:20pm Jeff Lichtman
8:40pm Panel discussion
Register at neuromatch.io -
Sebastian Seung reposted
Dezhe Jin
@djintwt
Cell paper! Happy to share that our work with Michael Long’s Lab at NYU Medical School appears this week in Cell. We find that the mixture of fast and slow axonal delays enable continuous encoding of time in songbird HVC. 1/n www.cell.com/cell/fulltext/S0092-8674(20)31160-0
-
Sebastian Seung reposted
Vamsi Mootha
@VamsiMootha
180,000 mitochondria - wow! Beautiful connectomics study reporting reconstruction of a tiny, tiny piece of mouse visual cortex. So many gems here, including a compartment-specific correlation between synaptic density and mitochondria content. @SebastianSeung
Nicholas Turner
@nicholasturner0
Our newest preprint! www.biorxiv.org/content/10.1101/2020.10.14.338681v2 EM and neural activity images from three million cubic microns of mouse visual cortex. Four scientific vignettes from organelles to synapses to circuit structure and function @BCMHouston @PrincetonNeuro @AllenInstitute @IARPAnews 1/20
-
Sebastian Seung
@SebastianSeung
Normally I just recall my students looking tired/bored...brings a smile to my face that someone has good memories of the class! Thx for sharing.
Stanislav Nikolov
@snikolov
#howitstarted: @SebastianSeung's wonderful neural networks course at MIT - such an exciting breadth of topics from nonlinear dynamics to computational learning theory to neuroscience. Yes, I was pretty insufferable, but it was too cool not too post about 😊 Thank you
-
Sebastian Seung
@SebastianSeung
Our new preprint on cortical connectomics.
Nicholas Turner
@nicholasturner0
Our newest preprint! www.biorxiv.org/content/10.1101/2020.10.14.338681v2 EM and neural activity images from three million cubic microns of mouse visual cortex. Four scientific vignettes from organelles to synapses to circuit structure and function @BCMHouston @PrincetonNeuro @AllenInstitute @IARPAnews 1/20
-
Sebastian Seung reposted
Fei-Fei Li
@drfeifei
So proud of my former student, Professor Olga Russakovsky!
-
Sebastian Seung
@SebastianSeung
Don't miss SOSCON, the @Samsung Open Source Conference! Largest event of its kind in Korea, held every year. Morning Oct. 14 www.soscon.net
Sebastian Seung
@SebastianSeung
International speakers at SOSCON include luminaries @jzemlin, Executive Director of the @linuxfoundation and @DivaDanese Cooper, President of @InnerSourceOrg Commons.
Post not available here, view on X
Sebastian Seung
@SebastianSeung
@Samsung is proud to affirm its commitment to open source, as will be described by Seungbeom Choi of @Samsung Research
Sebastian Seung
@SebastianSeung
Lots of other great activities: 16 contributed presentations, five community meetups, and SOSFest with prizes! www.soscon.net
-
Sebastian Seung reposted
David A. Markowitz
@DavidAMarkowitz
Honored to see @IARPAnews #microns recognized by the BRAIN Initiative for our program's open science efforts: microns-explorer.org and bossdb.org/projects/ Thank you @AllenInstitute @PrincetonNeuro @SebastianSeung @AToliasLab @JHUAPL @theBossDB !
www.braininitiative.org/achievements/the-brain-initiative… -
Sebastian Seung reposted
Markus Meister
@mameister4
The McKnight Foundation wants to fund your Technological Innovations in Neuroscience. Anything goes, from viruses to algorithms. Letter of Intent is due on Dec 7, 2020. Please apply and spread the word!
www.mcknight.org/programs/the-mcknight-endowment-fund-for… -
Sebastian Seung reposted
COVID Translate
@COVID_Translate
The Spanish🇪🇸 translation of @KoreaCDC's management guidelines for healthcare facilities with confirmed cases is now available on our website:
covidtranslate.org/kcdc-guidelines/management-of-healthca… -
Sebastian Seung reposted
COVID Translate
@COVID_Translate
The English translation of @KoreaCDC’s #covid19 Infection Prevention and Control Guide for Pre-schools, Elementary schools, Middle schools and Specialized Schools has been been updated with Part 3. You can see it here:
covidtranslate.org/kcdc-guidelines/guidelines-for-school-… -
Sebastian Seung reposted
Eyewire
@eye_wire
Did you know you can explore hundreds of 3D cortical neurons at nanoscale resolution? Feel free to share with your science teacher! neuromancer-seung-import.appspot.com
A getting-started guide for the viewer is available at blog.eyewire.org/explore-cortical-neurons/
Sebastian Seung
@SebastianSeung
You can whet your appetitite for cortical reconstructions at microns-explorer.org, where some MICrONS data and associated reconstructions have already been released. /end
-
Sebastian Seung
@SebastianSeung
We’re excited to announce FlyWire, the world’s first online community for whole-brain connectomics. Just click on flywire.ai for instant access! In just a few seconds, you can start browsing through an entire fruit fly brain in electron microscopic glory. 1/
Post not available here, view on X
Sebastian Seung
@SebastianSeung
The FlyWire back end is based on a new data structure (the ChunkedGraph) that was designed to enable editing of petascale reconstructions. Low latency is achieved by implementing with Google Bigtable, a NoSQL database. 10/
Sven Dorkenwald
@sdorkenw
We introduce the ChunkedGraph as a data structure for storing neuron segmentations and processing edits. The ChunkedGraph enables proofreading of large EM datasets by an open community while maintaining a timestamped history of the segmentation. 2/4
Sebastian Seung
@SebastianSeung
The 3D convolutional net was trained using @PyTorch, and the segmentation was generated using chunkflow and MMAAPP. arxiv.org/abs/1904.10489 github.com/seung-lab/chunkflow github.com/seung-lab/MMAAPP All our code is open source and available at our github. 11/
Post not available here, view on X
Sebastian Seung
@SebastianSeung
And there’s more to come. We are planning to launch an online community pyr.ai, based on MICrONS data. Volunteers will proofread an automated reconstruction of a piece of cerebral cortex, a brain structure that is important for mammals and especially humans. 14/
Sebastian Seung
@SebastianSeung
You can whet your appetitite for cortical reconstructions at microns-explorer.org, where some MICrONS data and associated reconstructions have already been released. /end
-
Sebastian Seung reposted
Jennifer Rexford
@jrexnet
Excited to have Adji Bousso Dieng join the Princeton CS Department in fall 2021.
www.forbes.com/sites/andrewwight/2020/08/30/this-ai-exper… -
Sebastian Seung reposted
Yann LeCun
@ylecun
Congratulations @adjiboussodieng !
-
Sebastian Seung reposted
COVID Translate
@COVID_Translate
Part 2 of @KoreaCDC’s #covid19 Infection Prevention and Control Guide for Pre-schools, Elementary schools, Middle schools and Specialized Schools has been translated into English.
covidtranslate.org/kcdc-guidelines/guidelines-for-school-… -
Sebastian Seung reposted
Princeton SML
@PrincetonSML
Olga Russakovsky received the Emerging Leader Abie Award recognizing her AI4ALL work and research on artificial intelligence. Congrats @orussakovsky 🏆🎉
csml.princeton.edu/news/olga-russakovsky-receives-award-a… -
Sebastian Seung reposted
Adam Charles
@BaselessPursuit
Please RT!: Open post-doc position in my (brand new) group at Johns Hopkins BME working on data science and machine learning for neural imaging and data analysis. Please DM me for more info! www.bme.jhu.edu/ascharles/open-positions/
-
Sebastian Seung reposted
Partners In Health
@PIH
COVID MASKAQ is a modern mobile laboratory equipped to perform molecular diagnostic tests to detect COVID-19. Construction was financed and directed by @SociosEnSalud and @USAIDPeru, who then donated it to the Peruvian Ministry of Health's National Health Institute. #stopcovid
This media couldn't be displayed. See it on the original post. -
Sebastian Seung reposted
Partners In Health
@PIH
Let me first preface my opinions. I'm not a scientist (to my father's eternal disappointment😂). I'm a doctor. Meaning I think like your auto mechanic, not your university professor. 1/
Natalie E. Dean, PhD
@nataliexdean
The question of in-person schooling remains one of the most important (and most difficult) out there. Importantly, we have to take the option of "back to normal" off the table in our minds. To be done safely, in-person schooling will require a lot of modification.
#pihchat -
Sebastian Seung
@SebastianSeung
Very timely information for the US given all the heated discussions happening now.
COVID Translate
@COVID_Translate
As requested, Part 1 of @KoreaCDC’s #covid19 “Guidelines for pre-, elementary, middle, and special schools” has been translated into English.
covidtranslate.org/guidelines-for-school-systems_EN.pdf -
Sebastian Seung
@SebastianSeung
DeepMath 2020 (deepmath-conference.com) is now accepting abstract submissions on mathematical
theory of deep neural networks! This year's amazing speakers include @GittaKutyniok @rbaraniuk, Demba Ba, Rene Vidal, Stefanie Jegelka, Eero Simoncelli & more!
Sebastian Seung
@SebastianSeung
@Samsung Research is proud to have been a founding sponsor of last year's in-the-flesh conference (remember those?) cc @BaselessPursuit @zamakany
-
Sebastian Seung reposted
Partners In Health
@PIH
On July 23, @kj_seung will be taking over PIH's Twitter account to discuss and debate COVID-19 with @nataliexdean, Assistant Professor of Biostatistics at the University of Florida. What questions do you have for Dr. KJ Seung and Dr. Natalie Dean? #pihchat
-
Sebastian Seung
@SebastianSeung
Starting in July, I’ll be serving as President of @Samsung Research. I look forward to guiding research and development @Samsung, strengthening basic and applied research, and striving for maximum real-world impact.
-
Post not available here, view on XPost not available here, view on X
Sebastian Seung
@SebastianSeung
Email information about yourself (bio, resume, github, code sample) to covidbrains@gmail.com, and ask questions about the project by online chat with the team at 5pm ET on Tue Apr. 7 and Wed Apr. 8 discord.gg/6DWgPy 5/
-
Sebastian Seung
@SebastianSeung
Our @starledger op-ed urging NJ to follow @GovMurphy stay-at-home order @SamWangPhD @ShirleyTilghman @OlgaTroyanskaya and leaders in public health, epidemiology, pharma www.nj.com/opinion/2020/03/the-lockdown-will-save-lives-g…
-
Sebastian Seung
@SebastianSeung
Titan of theoretical physics/neuroscience John Hopfield has won @TheFranklin Medal in Physics! fi.edu/en/awards/laureates/john-j-hopfield
-
Sebastian Seung
@SebastianSeung
Connectome-based modeling of neural circuits ... Latest from Srini Turaga
evolvingstuff
@evolvingstuff
More of this kind of thing, please.
A Connectome Based Hexagonal Lattice Convolutional Network Model of the Drosophila Visual System
arxiv.org/abs/1806.04793 -
Sebastian Seung
@SebastianSeung
BAM has been renamed as BRAIN (Brain Research Through Advancing Innovative Neurotechnologies) www.nytimes.com/2013/04/02/science/obama-to-unveil-initia…
-
Sebastian Seung
@SebastianSeung
The connectome in 1967 HT Charles Jennings boingboing.net/2012/07/03/vintage-att-ad-1967-your.html
-
Sebastian Seung
@SebastianSeung
A new library allows readers to borrow people for a 30-minute chat www.davidbakeronline.com/article03.html




















































































































































































