Neuroscientist. We aim to understd brain networks and brain wiring, using lasers & neuro AI. Prev: policy for @democracypolicy. Personal views. Not a lab acct.

Washington, DC
Wish me luck as I dust off this absolutely epic homemade hot sauce recipe. I have the pic somewhere in old notes lostartofhomecooking.blogspo…
1
2
226
It’s going to take all of us. Each of us can find the courage to do something. I have this shirt too. It gets it right: the arc of history is not going to bend itself. “So get involved.” Embrace the flag. American symbols are for those who believe in America’s ideas. /end
1
56
“We hold these truths to be self-evident.” On the 250th. This shirt gets it right: America is an idea, of freedom, equality, and a fair shake for all. 1/
1
1
3
291
Whoops, meant to lead with this image
2
3
289
Last, abolishing the NIH and NSF would create a funding vacuum. I hesitate to even engage with this utterly naive view of economics and science, but I’ll note CZI and HHMI’s dramatic course changes, Bell Labs support by govt contracts, and check out Wellcome/CZI/HHMI budgets.
2
5
443
Now, why is Atlas writing this? I’m not saying he doesn’t believe it, he probably does. I’m just saying he wouldn’t have his recent career if he didn’t say things that benefit rightwing billionaires. Hoover, AIER, Trump I…
1
6
426
It’s ironic that Atlas begins by quoting Milton Friedman when the whole article fails to grapple with the market failures created by fundamental research—which is both a Samuelson public good and an information good (eg Arrow). papers.ssrn.com/sol3/papers.…
1
2
8
894
It is sad that Scott Atlas doesn’t understand basic economics. Paul Samuelson wrote about public goods—why NIH has been such a stunning success—more than 75 yrs ago. Also, to see why Atlas urges ending US success, just look at who elevates him: pro-privatization billionaires. 1/
8
20
61
12,860
"For successful [science], reality must take precedence over public relations, for Nature cannot be fooled.”
2
4
320
"position does not confer appeal rights to the MSPB... prohibited personnel practices will [not be enforced by] OSC" We're seeing them go straight for the Civil Service Reform Act and CSRA appeals. And the Roberts Court has been signaling they're good. Cool times! (sarcastic)
1
380
“As the admin moves from cutting budgets and attacking elite universities to dismantling the foundations of the US research enterprise, pharma CEOs have remained silent. Their silence is a choice.” biocentury.com/article/65961…
2
2
9
1,847
This framework is well-worked out in theory—we must choose an error criterion ("wrong"), which comes w/ a cost in time/$. This isn't ex it (need a cost term), bt same idea, fm David Heeger's notes. Type 1/2 errors, sensitivity/specificity: same idea. cns.nyu.edu/~david/handouts/…
1
4
1,273
On efficient orgs: “The antidote to a corrupted metric is usually direct contact w/ reality— interviews, frontline observations, case reviews, unscripted convos. Leaders who do this well build it into their routines structurally…” @pahlkadot Yes. Accords with my experience.
1
3
5
492
💯 The real high risk science is “Let 1,000 flowers bloom“ science, which spreads money around to many many different individuals to pursue their best ideas.
1
1
5
808
You can also read my paper at SSRN where we discuss public goods — and Arrow on information goods like basic science. This is not controversial in economics papers.ssrn.com/sol3/papers.…
1
1
2
260
For all the replies arguing for the private sector funding basic science please read about Samuelson on public goods, and also Kenneth Arrow who added important context. Screenshot from link above
1
1
3
184
This is only true for people who understand neither science nor economics. The NIH budget for this year is FIFTY times larger than OpenAI’s $1B pledge. The foundation of US science & innovation is public funding. The private sector cannot replace it. US science is being killed
15
67
523
19,877
“The illegality of your authoritarian management of the FCC is exacerbated by the fact that this administration has asserted unprecedented executive branch control over this agency…” Seems like a pattern across agencies, no?
1
154
“His misery was all due to i-Ready, the software product our district had purchased for math work and testing.” My family’s experience is the same. Bad edtech software funnels money to consultant and tech firms, and disempowers talented teachers. It is breaking education.
1
278
Watching Grace Lindsay give an excellent talk about neuroscience and interpretable AI. One thing that occurs to me listening—a historical note from neuro. The field used to use all kinds of hand-tuned selectivity metrics: d’, R2, mutual info. But now cross-validated decoding…
1
26
2,189
Poster session I
216
Fabulous keynote session at @cosynemeeting - touching on neuroscience, AI, and mech interp. @neuro_kim and I kicked off the meeting and introduced Chris Olah. 1/
2
3
29
4,068
A lol summary on vibe coding, org-mode, and synchronization to mobile. This is my experience. (what is needed is a CRDT/OT style sync solution; could be done with git perhaps; a textfile version could be used with org clients on desktop and mobile, but it's obviously hard.)
1
245
Although @Addgene left Twitter I wanted to share this award my lab got. Since Brownstone people are slandering me. They can get lost. And they can come discuss further on a non-manipulated platform, if they have the courage.
4
284
Apparently I’m getting some criticism of my science here from people connected with the billionaire-run anti-science Brownstone Institute. A “who’s made the most impact?!” slap fight is rarely productive. Fields differ and true impact is long-term. But I am proud of this 1/
1
1
6
329
. @Addgene gave our lab at @NIH the Blue Flame Award for making a plasmid requested at least 100 times by other scientists. It's a bicistronic stable opsin-GCaMP8s pair for pattern stim. Congrats to P LaFosse, Z Zhou, and B Akitake who led this work. 🥳🥳
1
1
2
265
How #NIH Institute directors are like Bill Belichick. #RealisticOptimism
1
8
1,513
That illegal use of RIFs to eliminate CSR was turned back -- by public attention, by Judge Susan Illston who (again) ruled the use of RIFs by the Trump admin to be lawless, and by Congress legislating in the CR. Riley was stopped from doing this and moved to ONR.
1
84
We should pay attention to #NIH peer review of grant applications. It's not too much to say the keystone of US biomedical science is NIH's Center for Sc Review. CSR organizes peer review—relying on 25k US scientists. It isolates grantmaking fm political interference. But! /1
1
1
2,217
But I fear this will be misread. We should work to eliminate biases. But great research environments do have an effect, and clusters of talent matter. They (have) attract(ed) people from all over the world. We should celebrate successful centers while we uplift others. /end
1
1
172
It is important to strive for a diverse scientific workforce, and be as inclusive as possible to increase the talent pool. That said, we will still have centers that collect talent. Coastal institutions— Stanford and Harvard—get lots of funding because they do great work. 1/
1
3
422
I wrote some about this interplay between theory and expt recently (Parallel discussion happening on bs too with @neurograce and others)
2
369
Seen on DC Metro 🙄
4
212
the US had a revolution over that

ALT Eye Rolling GIF

1
107
Username checks out
1
191
The people who built the US Constitutional framework cared a huge amount about info environment. “[The people] have a right, an indisputable, unalienable, indefensible right, to knowledge about the character and conduct of their rulers” - John Adams. 1/
1
1
181
Do some think AI plus robots will replace cancer, neuro, etc #NIH / #NSF research? Yes "The legislation would establish a nat'l network of cloud labs, physical labs where humans alongside robots could do biotech expts researchers could access remotely." 🧪 1/
1
1
481
"The field of AI safety is also closely linked with neuroscience, because understanding how human neural networks create bias can also help us understand how AI networks create bias."
2
6
17
2,142
They can constrain an E/I recurrent network to stay in the balanced regime and train a number of units of the network: 2/2 github.com/chrismkkim/Tutori…
188
Here’s the extended video from CBS today. Many thanks to Dr. Rosenberg and the patients in this video like the Callahans and Natalie Phelps. It’s not easy to speak out. Best of luck to them in their fight against cancer. piped.video/watch?app=deskto…
2
2
797
I am proud that the public, through NIH, has built such a powerful and successful enterprise to understand and treat diseases like cancer. I am sad that this research enterprise at NIH is suffering such damage. That’s bad for patients, bad for the economy, bad for all of us. 🧪
1
2
4
603
“Anthropic's argument here is that giving people the impression that a model is unbiased and objective is itself harmful, because those things are not true!” This post from @simonw on the Claude system prompt is very insightful #ai #neuroai
2
1
3
455
This bodega / grocery recently opened up in DC. “Aisles of Abundance” 😆 Is this part of the Klein / Thompson book tour?
3
440
“NIH scientists can be certain they are afforded the ability to engage in open, academic discourse as part of their official duties and in their personal capacities without risk of official interference, professional disadvantage or workplace retaliation."

ALT Thats Right Baby Gru GIF

2
490
“without risk of official interference, professional disadvantage or workplace retaliation."

ALT Flag freedom

1
2
603
I'd go even harder: Scientific discovery is a creative endeavor that will *not* be replaced by AI. There's a lot of talk about AI "solving" science -- that AI will answer science questions and direct robots to do experiments. But that's not the HARD part of science. 1/4
Derek Lowe is right that Demis Hassabis is wrong - there is no chance of curing all disease within the decade. Even if we had AGI now, we wouldn't do it. We lack the basic knowledge and need to keep doing experiments. science.org/content/blog-pos…
3
3
25
3,814
cool use of an RNN to study/replicate the oddball responses, linking them to tone processing
1
288
Beautiful day today on the #NIH campus. Despite the disruption of research and grants here. Find positivity where you can and stay strong, all. #cherryblossoms
44
1,855
- perturbational methods were needed to uncover this - Active filtering seems like a common computation across sensory areas - Network ISN operation to implement this - Active filtering is a form of memory (see section connecting to engrams)
1
2
349
...I would love feedback on. The core idea is (1) new stim methods have shown cerebral cortical networks amplify certain inputs. (Marshel et al; @ianoldenburg.bsky.social 2024, Peron et al., and us: @cianadeveau.bsky.social) 🧪#neuroscience 🧠📈 🧠🤖 2/4
1
1
6
633