Jacob Reimer

Chris Xu talks to Neurophotonics about the 'toy land of optics' and how to channel positive vibes! Thank you @TianYuWang2020 for conducting this fantastic interview! spiedigitallibrary.org/journals/neuro… #Neurophotonics #Neuroscience #Neuroimaging #BrainImaging #OpticalImaging #Microscopy

@drugmonkeyblog The line that got me, that I don’t hear people talking about is: “So that we’re not just shutting down animal labs overnight, we’re developing long-term solutions that ensure that there are no new animal labs that open up in their place” 2:42:50

Journalist freaks out researchers because of incomplete understanding of how NIH works drugmonkey.wordpress.com/2025/07/09/jou… via @drugmonkeyblog

We will likely be looking for someone with expertise with optical methods in vivo, cranial window surgeries in mice, analysis of 2P imaging data (especially ML-based), viral approaches. Nothing posted yet, but I know there are good people looking, so please reach out.

After 7 years, thrilled to finally share our #MICrONS functional connectomics results! We recorded activity from ~75K neurons in visual cortex in a single mouse, then mapped its wiring using electron microscopy. To systematically characterize neuron function, we built the first foundation model of the mouse visual cortex—trained via deep learning on data pooled from multiple mice and visual cortical areas. Our foundation model generalized to new neurons, animals, and even unseen stimulus domains. It also accurately predicted entirely new modalities, such as anatomically defined cell types. Importantly, this robust generalization enabled us to create accurate functional digital twins of individual mouse brains. Using the digital twin of the MICrONS mouse—where we knew the exact neuronal wiring—we discovered that neurons don’t connect randomly, even when anatomically positioned to do so. Instead, given multiple potential partners (axons near dendrites), neurons preferentially choose partners with similar feature selectivity (“what”) rather than receptive field overlap (“where”). Foundation models offer a powerful approach to systematically decode the neural code of intelligence. Huge thanks to @IARPAnews for funding this groundbreaking effort through the @BRAINinitiative, and to our amazing team at @Stanford @StanfordMed @bcmhouston, @Allen, @Princeton, @uniGoettingen and others! #Neuroscience #MICrONS #NeuroAI #Connectomics #FoundationModels #AI nature.com/immersive/d428…

The most exciting thing about the 10 papers in this cortex edition of Nature is that they are just the beginning of discoveries to be made using the microns dataset. Yours could be next microns-explorer.org

If you want to feel small and insignificant, stare up at the stars. If you want to feel immensely profound, impossibly complex, and just outright improbable, gaze at the stars in this 1mm cube and realize it's ~ a millionth the volume of your own brain. microns-explorer.org/cortical-mm3

But over the past ten years I've definitely felt that analogy to a space mission many times. So many essential contributions - both large and small - from so many people to get the project to this point. And there is still so much to discover in this data!

I feel very lucky to have played a role in this huge collaborative project. There's not that many neuroscience projects that have the scale of aspirations, collaborative effort, and scientific payoff of a space mission...

Excited and proud to see all the MICrONS papers out in @NaturePortfolio today, including NEURD - a tool developed in our lab to help with automated proofreading and analysis of large EM connectomes. Check out the beautiful landing page @Nature: nature.com/immersive/d428…

Some commenting about NIH grants don't appreciate how incredibly competitive it is to get these funds A typical study section reviews ~75-100 grants, most of which are outstanding but only ~10 get funded Can't think of any process in the private sector that is as competitive

@AndrewHires My internal model has seen a lot of “data not shown” that doesn’t appear in training sets for these LLMs. But I agree in a few years it may be hard to justify the enormous effort of human review. Maybe intermediate step is AI pre-screening to determine finalists for human review

@viajake What’s your value add?

Testing out DeepSeek-R1's ability to do "science" in one-shot. I asked it to critique my 2017 R01 proposal which was scored Impact 26, 10% percentile, and funded by NINDS. (I wrote the proposal, I have the rights to share it with anyone or anything). I uploaded a PDF of my Specific Aims and Research Strategy with the following, off-the-cuff prompt. It spit out a credible critique indistinguishable from a median quality reviewer's submission after 10 seconds of thought. Overall impact score 2. Thank you! If this is what a free tool can do TODAY in 10 seconds of inference, on the first shot with zero iteration in prompting or followup, what actual role are humans going to play in the process of science going forward? Would you rather have your multi-million dollar, career milestone proposal be critiqued by a super-intelligent AI agent, or by a tired, over-worked, potentially biased human? It seems quite plausible that generating new, evidence-supported hypotheses and experimental plans could already be better accomplished by frontier models (with or perhaps without human prompting) than by humans alone. What do we do now?

My cousin asks if I can describe this moment, the heaviness of it, like sitting outside the operating room while someone you love is in surgery … —Alison Luterman rattle.com/holding-vigil-…

I am humbled by the response to this and I have one ask. Can you share this post so that we can get 75K likes? It is the most impactful thing @sabrevaya and I can do to save our democracy.

Seeing a launch in person was amazing. But standing next to a dozen young spacex engineers and seeing their reaction to the catch was a real bonus.

Me, thinking about all the experiments I proposed that don’t require months in the idyllic Swiss countryside.

@currentClamp 😂

If there was a Nobel prize for steaming milk in your SFN hotel room…

Seems like there’s no reason in principle the tartrazine trick (with a suitable dye) wouldn’t work in the brain for 2P. Is that right?

Ok Houston, I see you