Rui Ponte Costa

Rui Ponte Costa

@somnirons.bsky.social

Computational neuroscientist bringing machine and neural learning closer together [Oxford NeuroAI] @ox.ac.uk @oxforddpag.bsky.social 🇵🇹🇪🇺🇬🇧🌳 neuralml.github.io www.medsci.ox.ac.uk/neuroai

2K followers350 following96 posts

Rui Ponte Costa reposted

Rui Ponte Costa reposted

Interested in a Ph.D. in comp neuro? Consider applying to CU Anschutz. Our community is growing with groups led by Nick Lesica, @teissa.bsky.social , Christopher Cueva, and myself! Topics: neural dynamics, circuit & behavioral modeling, neuroAI, closed-loop neurostim. Links below #CompNeuro #NeuroAI

Rui Ponte Costa reposted

The cerebellum isn’t just about movement; it gates social communication! Huge congrats to Dr. Cheryl Brandenburg (Sillitoe Lab), who led this work with her creativity and determination. We can’t wait to see where she takes this research in her own lab @lsu.bsky.social #neuroskyence tinyurl.com/CBSCO

tinyurl.comFrequency-dependent cerebellar circuits independently gate social vocalizations and movementBiological sciences; Behavioral neuroscience

Rui Ponte Costa reposted

Our paper on exponentiated gradient (EG) learning is now out at Neuron! www.cell.com/neuron/fullt... We show that EG naturally respects Dale's law and produces networks that are a better phenomenlogical fit to biological networks. Congrats to @repromancer.bsky.social for getting this out! 🧠📈 🧪

cell.comExponentiated gradient learning yields brain-like synaptic distributionsCornford et al. show that exponentiated gradient (EG) learning, which updates synapses multiplicatively, matches gradient descent on cognitive tasks while respecting Dale’s law and producing brain-lik...
View full thread
View full thread

The shared space clusters by image category, inherited from the ResNet-50 scaffold. This is not a property of the RN-50 architecture alone: Untrained ResNet-50: retrieval drops from 46.2% to 6.6%. Shuffled neural data–model feature 1:1 correspondence: 1.0% (chance 0.8%). 4/6

Adding a new subject is cheap: we freeze the shared space and train only a new subject encoder/decoder on their own images. With 20% of the new subject's data, we recover ~98% of the alignment reached with all of it, and 90% of the noise ceiling in self-reconstruction. 5/6

In our MED-VAE architecture a common space arises by using a pretrained ResNet-50 as a scaffold onto which subjects brain activity can align to. Our method does not require common inputs across individuals and after training, the Resnet is no longer needed. 2/6

Our method outperforms previous approaches: Top-1 retrieval (another person's closest latent is the same image): 46.2% vs 21.8% (SRM), 18.4% (Procrustes) Representational similarity analysis (RSA) across subjects' latent representations: 0.63 vs 0.41 3/6

Rui Ponte Costa reposted

Abstract submission closes soon, so you should start if you haven't yet 😉 #COSYNE2027

COSYNE@cosynemeeting.bsky.social

📢Have work to share with the computational and systems neuroscience community? Abstract submissions for #COSYNE2027 are now open. Deadline: 18 October 2026, 11:59 p.m. AoE Submit your abstract: www.cosyne.org/abstracts-su... #Neuroscience

Rui Ponte Costa reposted

Re-post 🔁 is appreciated! #neurojobs: we’re looking for new team members to join us @cmc-unit.bsky.social!
If you’d like to study decision-making in naturalistic & embodied contexts (integrating RL, RNNs & cross-species data analysis) in a cool & collaborative research environment 😎, check 👇

Roxana Zeraati@roxana-zeraati.bsky.social

Link to ads: Postdoc position: nextcloud.tuebingen.mpg.de/index.php/s/... PhD position: nextcloud.tuebingen.mpg.de/index.php/s/... We’re also both at @bernsteinneuro.bsky.social and happy to chat!

Rui Ponte Costa reposted

To pursue extended goals, the PFC monitors performance and the value of cognitive control, thus preventing goal abandonment. #AI models struggle with #hallucinations because they lack this monitoring circuit--forcing #engineers to reinvent it as an agent "harness". #Neuroscience

Rui Ponte Costa reposted

[Hugging Face release] We’re releasing the dataset behind Reason to Play! I’m incredibly proud of our team for reaching this milestone.

Great summary of our paper with @mark-j-wagner.bsky.social

The Transmitter @thetransmitter.bsky.social

Cerebellar granule cells, which outnumber all other neurons in the brain combined, help the brain to generalize to new situations and distinguish between contexts simultaneously. By Holly Barker www.thetransmitter.org/cerebellum/c...

Rui Ponte Costa reposted

I have a postdoc position open in my NeuroAi lab in Vienna 👌👾🧪🧠 And I will be at Bernstein (workshop only), please reach out if you are interested

Rui Ponte Costa reposted

2023: 8,039 proposals 2024: 10,360 proposals 2025: 17,066 proposals 2026: 21,627 proposals Something's gotta give

Iva Vukušić@vukusiciva.bsky.social

This ridiculous number of applications is, in my view, the result of 2 nightmare developments happening concurrently 1.the precariousness of the job market for young PhD graduates which then fuels further: 2.the AI tools enabling overproduction of proposals. #Research #ERC #ResearchFunding #Academia

Rui Ponte Costa reposted

I’m on the faculty job market this fall! 🐦🧠 I’m an NIH K99/R00 awardee studying how the brain learns complex skills, using birdsong to connect molecules, synapses, circuits, behavior & computation. Know of a search where I might fit? Please send it my way & share!

drewschreiner.github.ioDrew Schreiner, Ph.D. | Neuroscience of LearningI use birdsong to understand how the brain learns complex skills across molecules, synapses, circuits, behavior, and computation. On the 2026–27 faculty job market.

Rui Ponte Costa reposted

If this makes you mad, submit your cognitive science research to @openmindjournal.bsky.social No APCs, no subscription fees. It's funded by university libraries, at low cost.

Eric Schares@eschares.bsky.social

🚨Update: Our analysis of global APC expenditure has now been expanded to cover more publishers (7 -> 14) and more years (2019-2025). We estimate $15B in APCs paid over seven years, $3.7B in 2025 alone. Elsevier crosses the $1B threshold by itself in 2025. arxiv.org/abs/2608.16322 #ScholComm

Rui Ponte Costa reposted

The cortex and cerebellum split the labor: the cortex provides a stable, general-purpose movement blueprint, while the cerebellum reorients that same blueprint differently for each context Granule cells reorient cortical trajectories to separate contexts www.nature.com/articles/s41... #neuroscience

nature.comGranule cells reorient cortical trajectories to separate contexts - NatureSimultaneous imaging of premotor cortex and cerebellar granule cells in mice learning two skills in parallel shows that trajectories generalize in cortex but coherently reorient apart in granule cells...

Rui Ponte Costa reposted

The Ferrier Medal and Lecture 2027 is awarded to Professor Kenneth Daniel Harris for his contributions to neuroscience by unifying experimental neurophysiology with advanced analytical methods and for pioneering methods that allow recording from thousands of neurons. #RSMedals https://bit.ly/4qCJZRo

Rui Ponte Costa reposted

Oh wow! @thetransmitter.bsky.social amplified recent Bsky chatter about the dire US grant landscape. Thank you, @emilysingerneuro.bsky.social 🙏.

The Transmitter @thetransmitter.bsky.social

Two years into one of the most tumultuous tenures in U.S. federal funding, researchers outline how they are coping with ongoing changes and reflect on the impact these changes are having on the system more broadly. By @emilysingerneuro.bsky.social www.thetransmitter.org/funding/amid...

Rui Ponte Costa reposted

Yay, the field needs this! Excited to serve as an associate editor!

Minds, Machines, and Brains (MMB)@mmb-journal.bsky.social

Hello world! 👋 We’re Minds, Machines, and Brains (MMB) 👤🤖🧠 a new open access journal from @mitpress.bsky.social exploring the principles of intelligence and cognition across natural and artificial minds. Submissions open this Fall! 🔗 direct.mit.edu/mmb

Rui Ponte Costa reposted

Very excited to attend #CCN2026 this week and to present various projects from my lab! Here's a list of our posters and talks in case you'd like to see what we've been up to.

Totally agree with this perspective. We need to consider how intelligence arises in the first place if we want to reproduce it — learning principles are critical.

Vlad Ayzenberg@vayzenb.bsky.social

Excited to share our review in @cp-neuron.bsky.social with @lauriebayet.bsky.social and @mickbonner.bsky.social! We describe how implementing principles from child development can advance the mechanistic plausibility and capacities of AI models We packed A LOT into this review, here's a quick 🧵

Really cool illustration of our paper. @neurosock.bsky.social's great work!

neurosock@neurosock.bsky.social

This recent post by @earlkmiller.bsky.social got my attention. It shows how abstraction could emerge in the primate PFC. The paper is hard, but the main idea can be simplified to low-dims, so I made fun a little animation. A🧵with my toy model and notes: #neuroskyence #compneuro #NeuroAI

I would totally like to second this. Huge congrats to Michał Wójcik who has managed to get this over the line despite the huge challenges!

Seán Froudist-Walsh@seanfw.bsky.social

Great to see Mark Stokes getting acknowledged , and also the author senior authors who stepped in, but I’d like to highlight 1st author Michał Wojcik who completed this excellent work from his PhD under such difficult circumstances. He‘s got more good stuff in the pipeline - one to watch