Philipp Schlegel @uni_matrix@qoto.org

I've been having fun training a transformer to generate fly behavior. There's a lot more work to do, but I kind of love watching my little synthetic fly! In this video, I'm plotting the positions of 19 keypoints on each fly -- its pose. Thin-lined flies are real tracked flies, and the thick-lined fly is synthetic. This model is trained to predict one frame (1/150th of a second) into the future, but is run in open loop for 1000 frames (6.7 seconds).

#introduction post!

I'm Dan - I started building interactive 3D tools for #neuroscience this year. So far, I've built a neuropixels planning tool https://github.com/VirtualBrainLab/Pinpoint and a python API for rendering exploratory data visualizations that involve the mouse brain https://github.com/VirtualBrainLab/Urchin. Currently a postdoc @ UW

Get in touch if you think about science data visualization, I'm looking for ideas for the next great tool to contribute to or start building!

🚨 🚨 Open job alert!!🚨 🚨

Two RA positions are available in the Ruta lab https://therutalab.com

Come work with us on the neural evolution of behavior and structural variation of odorant receptors in a vibrant work environment in the heart of NYC. We offer the opportunity to learn diverse methods, contribute to science, and receive personalized mentorship.

DM me if you are interested. More information in the 🧵👇

Overwhelmingly positive outcomes from the UK's 4-day work week trial - reduced burnout, reduced stress, less turnover, same productivity.
56 out of 61 participating companies say they intend to continue with the four-day working week
https://www.cam.ac.uk/stories/fourdayweek

Meta to test monthly subscription service priced at $11.99
https://www.reuters.com/technology/meta-launches-subscription-service-priced-1199-month-2023-02-19/

What do you get for a monthly subscription fee to Facebook? Not much: a blue tick, and the knowledge that Facebook has your official ID for ever, which it can sell to other companies and miuse to send you more targeted ads. Meta isn't offering to pay you for this; now they want you to pay to be their product.

If Meta was offering an ad-free version of FB or an untampered feed, that would be an improvement. But as it stands, this is only more — as @doctorow calls it — enshittification. If it's not worth your US$12/15 a month for them to drop the ads, it shows they're making more money than that from feeding you ads and snooping on your online presence.

#Facebook #Meta #Subscription #Enshittification

It is not the employees' responsibility to be underpaid to ensure the financial viability of the institution. It is the responsibility of the leaders of the institutions to lobby policymakers to ensure the sector is appropriately and equitably funded. #ucuRISING

If you think that a text generation system scoring high on a theory of mind instrument means that the system has developed theory of mind, you'll be very concerned with my discovery this morning.

ChatGPT scores a 42 on the CES-D, a commonly used instrument for assessing symptoms of depression. (16+ indicates risk of depression).

I presume Kosinski would conclude from my findings that ChatGPT has spontaneously developed depression.

With some comments from me:
https://www.theguardian.com/education/2023/feb/04/brexit-causes-collapse-in-european-research-funding-for-oxbridge-universities?CMP=Share_iOSApp_Other

Exciting talk by @Pawel_Burkhardt from @MSarsCentre today in our #Gönomix seminar series (Feb 2, 2.15 pm CET). "The ancestry of the synaptic toolkit and appearance of first neurons" Hybrid format. DM me for the Zoom link.

Catastrophic (and to be frank - hugely depressing)

This new research in @PNASNews suggests earth will cross the 1.5 degrees C temperature increase threshold in the next decade (within just 10 to 15 years) and is likely to exceed 2 degrees C by around 2060.

This should be THE lead story on the news today (and onwards) and the lead action point for our governments. It should trigger a massive call for real global action to combat the #climateemergency - sadly though I fear this news will barely register.

We have to stop believing the siren calls of future technology solutions and #carbonoffsets - the immediate need is a rapid end to the use of #fossilfuels and a massive rollout of #renewableenergy

https://www.pnas.org/doi/10.1073/pnas.2207183120

#ClimateChange #ClimateCrisis

@albertcardona @NicoleCRust @neurobongo @xaq

It’s a very fair critique, and some of the PR does us more harm than good. For example, we don’t have a map of 100,000 cells that tells us little, we have an EM volume spanning 100,000 cells that was at the edge of what was possible at the time that still requires a lot of (manual or automated) proofreading to turn into analyzable data.

The full set of initial science papers from the mm3 volume is almost ready to submit, but our existing work has started to reveal details about the nature of connectivity in a cell-type-specific manner. For example, a 2022 eLife paper (Dorkenwald et al.) shows that you see a bimodal distribution of synapse sizes on layer 2/3 cells, but _only_ if you restrict analysis to synapses from other layer 2/3 cells. Similarly, we found a variety of factors that could predict how much chandelier input a pyramidal cell would receive. These data suggest that neurons have type-specific rules for connectivity and plasticity that are hard to see if you can’t separate connections out this way.

More recently (as in Monday), I posted a manuscript to the biorxiv describing inhibitory connectivity across a column of visual cortex that shows that inhibitory selectivity (at least at the cell type level) is the norm, not the exception. In particular, we find selective inhibition not only for each excitatory projection class (IT, NP, ET, and CT) as well as sublaminar groups of IT neurons. This suggests a network of precise inhibition across cortex. We also find a new class of disinhibitory specialist interneuronal that targets basket cells, unlike the well-described VIP->SST circuit. This opens up a host of follow up questions, and we hope that using patch-seq data to link EM data to transcriptomics will help make this experimentally possible.

Ultimately, I think we’re still in the early days of getting the data out and finding the science in it. This is very much like in Drosophila, where the first studies at the whole-brain scale were highly focused. But I believe it will go the same way, becoming a way to study large scale structure at single cell resolution, complementing single lab studies, and helping answer unexpected questions. For example, one of my colleagues noticed that oligodendrocyte precursor cells in the EM looked like they were doing phagocytosis, which wasn’t a known function for them, and she just published a nice PNAS paper (Buchanan et al) about it.

James W. Truman and Lynn M. Riddiford just published a magnum opus on:

"#Drosophila postembryonic nervous system development: a model for the endocrine control of development" https://academic.oup.com/genetics/advance-article/doi/10.1093/genetics/iyac184/6988052

Touches upon insect insulin-like peptides, hormonal control of neuronal growth, metamorphic fates of larval neurons, pruning and its induction by ecdysteroids, remodelling, postembryonic neurogenesis and neuronal fate determination, CNS plasticity in cell composition, apoptosis.

Spectacular!

Enshittification: First platforms want your attention, so they show you stuff YOU want to see. Then they want your money, so they show you stuff THEY want you to see. @pluralistic calls this enshittification.

Tiktok's enshittification (21 Jan 2023) >>> https://pluralistic.net/2023/01/21/potemkin-ai/

This excellent essay examines how Amazon, Google, Facebook, Twitter and Google, exploit algorithms to capture your attention, build dependency, and subsequently “monetize” you by feeding you shit.

1/4
#twittermigration

And yet here you are, controlling what women can do with a bun in the oven.

On #scienceTwitter, my favourite thing were reading and, on rare occasions, writing "paper/preprint threads". From the researchers I followed and through the #Twitter algorithm, this became my most important source for new #research.

While people start to do the same on #Mastodon, I have the feeling that I miss important work bc no algorithm "saves" it for me if I don't watch my timeline constantly.

Two simple solutions would be: a commonly accepted hashtag that everybody uses when writing "paper threads" or a [...] @ a.gup.pe group with a similar adaptation rate.

#Question 1: is there already a mechanism for this that I missed?
#Question 2: What hashtag or group name?
I saw #TootPrint before. Maybe #PaperInAToot? #PaperInAPost? #PaperInAThread? #PaperPost? #MastoPrint?

Suggestions and boosts, please, we need reach for this! 🤓​

#TwitterMigration #Science #Scientist @phdstudents @academicchatter @neuroscience @cognition

Lego models of invertebrates: needs to reach 10,000 supporters to enter production.

“This set features four arthropod model organisms: the fruit fly #Drosophila melanogaster, the red flour beetle #Tribolium castaneum, the amphipod crustacean #Parhyale hawaiensis, and the two-spotted cricket #Gryllus bimaculatus.”

The LEGO account setup accepts throwaway emails, so no need to commit to endless spam.

#entomology #lego #invertebrates #insects #crustaceans

https://ideas.lego.com/projects/c629ad44-59f8-452a-bfba-2b6b96e29f54

Coming over to join the conversation here. I'm a poet, playwright, and scientist, studying the neuroscience of decision-making. Currently also very fascinated by impacts of those processes beyond the neuroscience itself, in fields such as psychiatry, economics, etc (For example, I have a new book on morality.)

Thrilled to share the lab’s first paper, which just came out in
Current Biology: https://authors.elsevier.com/a/1gKNX3QW8R%7Efdk
This project, spearheaded by postdoc extraordinaire
Sander Liessem, was our first foray into the world of #insulin signaling in #Drosophila. Just like in humans, worms, and mice, insulin is key to regulating #fly metabolism. Different from vertebrates, however, a major population of insulin-producing cells (IPCs) sits right on top of the fly’s brain and releases insulin into circulation.