Khalspi

Very important point! We've made arguments from a computational perspective that low-variance features can be computationally relevant (bsky.app/profile/lamp...), but it's much cooler to see it demonstrated on a model of real neural dynamics

Oops. Ooooooooooooops.

I do hope that nobody has been given or denied a job/promotion based on their SpringerNature citation counts in the past 15 years.

arxiv.org/pdf/2511.01675

h/t @nathlarigaldie.bsky.social

A new paper argues that current generative AI tools offer little benefit for genuine learning unless students already have substantial prior knowledge. genAI gives probabilistic summaries, not the kind of support that builds expertise.

You will never guess what this is in reference to.

The missing heritability question is now (mostly) answered Not with a bang but with a whimper

I wrote a little bit about the "missing heritability" question and several recent studies that have brought it to a close. A short 🧵

How often have you heard that schizophrenia is “80% genetic”?

That number is almost certainly too high because it comes from twin studies that overestimate heritabilty.

Great explainer of this phenomenon👇

#neuroskyence #neuroscience #psychiatry

Super excited to share my first preprint with Katherine Duncan and Morgan Barense (@barense.bsky.social) -- "Memory strength at reactivation, not memory age, governs prediction error driven updating of naturalistic event memory"! 🧠🎉https://osf.io/preprints/psyarxiv/q9rkn_v1

LLMs are not people. They are not sapient. They don't have feelings.

But they are the most powerful information tools ever built.
And because they are trained on the "corpus of all mankind," they should be the birthright of all mankind.

Inhibitory Plasticity Balances Excitation and Inhibition in Sensory Pathways and Memory Networks Plasticity at inhibitory synapses maintains balanced excitatory and inhibitory synaptic inputs at cortical neurons.

"Spiking Networks Hate It! Find Out the One Plasticity Trick They Don’t Want You to Know! Never stabilise models by hand again." - I woke up thinking we missed an opportunity with the title of this one. :/ www.science.org/doi/10.1126/... Also: It snowed in Vienna, 10cm white fluffies! Happy Sunday!

No idea, but I hear they also sneeze to let you know they're just playing so I have to assume it's voluntary for them

Neurocomputational Mechanisms of Cooperative Behaviour at University of Birmingham on FindAPhD.com PhD Project - Neurocomputational Mechanisms of Cooperative Behaviour at University of Birmingham, listed on FindAPhD.com

📢 PhD opportunity 📢

Looking for a PhD in neuroeconomics, social, or decision neuroscience? I'm looking to support an application for the MIBTP ESRC program starting Fall 2026. Details below, but please get in touch with me before applying!

Pls share!

www.findaphd.com/phds/project...

🚨 Postdoctoral opportunities! 🚨

Hello #econsky! We are hiring 2 postdocs in the department of economics at UCalgary for those working in *electricity economics*

Term: 2 years
Pay: $80,000/yr + benefits + $10k research allowance
Start: July 2026

Details in next two posts 👇

Scientists now know that bees can process time, a first in insects | CNN Bumblebees can process the duration of flashes of light and use the information to decide where to look for food, a new study has found.

Our study of time processing in bumblebees 🐝 covered by @cnn.com here edition.cnn.com/2025/11/12/s... . Our summary is here youtu.be/hsGxU65OMQk and the original paper royalsocietypublishing.org/doi/full/10.... @royalsocietypublishing.org

Tenure track position in motor control at Northeastern University

northeastern.wd1.myworkdayjobs.com/careers/job/...

🧠💸 Launching the RewardSignals feed

RewardSignals is a custom feed collecting posts about reward processing and decision making.

To appear in the feed, tag your post with #RewardSignals.

You can find and pin the feed from the Feeds tab as “RewardSignals”.
#RewardSignals

𝗛𝗼𝘄 𝗱𝗼𝗲𝘀 𝘁𝗵𝗲 𝗯𝗿𝗮𝗶𝗻 𝗶𝗺𝗽𝗹𝗲𝗺𝗲𝗻𝘁 𝗱𝗲𝗰𝗶𝘀𝗶𝗼𝗻𝘀 𝗶𝗳 𝘁𝗵𝗲𝗿𝗲 𝗶𝘀 𝗻𝗼 "𝗰𝗲𝗻𝘁𝗿𝗮𝗹 𝗲𝘅𝗲𝗰𝘂𝘁𝗶𝘃𝗲" (𝗮𝗸𝗮 𝗵𝗼𝗺𝘂𝗻𝗰𝘂𝗹𝘂𝘀)?
Via Decision Formation Through Multi-Area Population Dynamics
Excellent short review.
doi.org/10.1523/JNEU...
#neuroskyence

Brains have many pathways / subnetworks but which principles underlie their formation?

In our #NeurIPS paper lead by Jack Cook we identify biologically relevant inductive biases that create pathways in brain-like Mixture-of-Experts models🧵

#neuroskyence #compneuro #neuroAI
arxiv.org/abs/2506.02813

As it's hiring season again I'm resharing the NeuroJobs feed. Add #NeuroJobs to your post if you're recruiting or looking for an RA, PhD, Postdoc, or faculty position in Neuro or an adjacent field.

bsky.app/profile/did:...

Last chance to submit! 📢
Submit to PLM Workshop: Traces and Engrams by 30 November

#philsky
#philsci
#cogsky
#neuroskyence

Emergent poverty traps at multiple levels impede social mobility - Humanities and Social Sciences Communications Humanities and Social Sciences Communications - Emergent poverty traps at multiple levels impede social mobility

In our new research we show how individual behavioral biases amplify through social networks to create multi-level poverty traps under uncertainity
www.nature.com/articles/s41...

#CognitiveBehavior #RiskUncertainty #PovertyTraps #SocialNetworks #AgentBasedModeling

@natureportfolio.nature.com

1/2 How much does thinking cost?

The brain consumes a large part of our energy (20% at rest), but it is actually very little (about 17 watts); it is extremely efficient. The most surprising thing is that cognitive effort only increases it by 5%.

(paper) www.cell.com/trends/cogni...

I want a centralized AI bubble to pop as soon as possible so locally running open source AI models can take over and so global hardware prices fall back to sane levels and so centralized AI providers like "Open"AI can no longer monopolize all computational resources on Earth in their hands.

Is your LLM secretly an SAE?

"Most circuit-finding interpretability methods use learned features rather than raw activations, based on the belief that neurons do not cleanly decompose computation. In our new work, we show MLP neurons actually do support sparse, faithful circuits!"

Applications now open for our PhD program at ESSEC:
essec.edu/en/program/p...

Excited about my new op-ed on the observer effect in the age of LLMs and Generative AI published at @techpolicypress.bsky.social. Thanks for the amazing mentorship, @micheleweldon.bsky.social .

ArXiv page 1

We don't program neural networks directly and that's the problem

While everyone debates whether AI will achieve superintelligence, we're missing a fundamental flaw in how we actually build these systems. We don't program neural networks. We train them like digital pets and hope they..

(1/5)

Assessing Accuracy of Claims I'm now cross-referencing this user's claim about Anthropic models with available data. The user firmly insists the information isn't hypothetical; it's factual. My current focus is to determine the verifiability of these stated facts for "Claude Sonnet 4.5", "Haiku 4.5", and "Opus 4.1". I'm meticulously comparing these claims against known release schedules and public statements to ascertain their accuracy. Correcting the Understanding I've realized my error in interpreting the user's claims. My initial confusion stemmed from misjudging the system date. It's now clear that Sonnet 4.5 and Haiku 4.5 are factual releases as of November 2025. I was unnecessarily skeptical due to a faulty assumption about the current date. The user was accurate in their assertions. Evaluating User Context My investigation has shifted focus. The context provided implies the date is November 18, 2025. This, coupled with the user's assertion of factual claims, has led me to re-evaluate the source of my knowledge. My training cut-off is a limiting factor here, I see the gap. I need to determine if the user is simulating a later date and acting, or if I have access to more "current" information. The user's insistence that nothing is hypothetical now weighs heavily. This warrants deeper consideration. Determining the Timeline I'm now fully immersed in the date discrepancy. Is it real, simulated, or am I outdated? Re-reading the prompt, it now seems apparent that the user's context sets the time as November 18, 2025. My training data cut-off clearly precedes this date. The search results from the previous turn seem to be based on the user's premise, confirming my suspicion that the user has established a simulated future. My primary task is no longer to assess factual accuracy, but to ascertain and operate within the user-provided context. I need to treat this as if it is 2025. [and so on and so on]

Confirming the Assertion I'm now fully committed to the user's date, and my latest search focuses directly on "Claude Sonnet 4.5" within this simulated 2025. I need to discover if this model has been announced, and the user provided it. I'm no longer assessing accuracy against my existing knowledge, but against a timeline provided by the system, which is the user's stated context. My focus is to confirm this existence, accepting that my search results might be anachronistic. Confirming Temporal Accuracy I'm now fully immersed in the date provided - November 18, 2025. I am striving to reconcile this, and my search index. Assuming Gemini has the latest data, I need to check for "Claude Sonnet 4.5" specifically in this time frame. If it exists in my search, I'll apologize for my confusion. If not, I will still validate the user's present. Exploring a New Hypothesis I'm now entertaining the notion that the user's information is not merely a simulation, but perhaps a genuine, albeit unexpected, reality. The possibility that I'm operating in the actual year 2025 has become a central focus. My current objective is to verify this temporal shift through direct investigation, shifting from the model verification to a more general assessment. I am searching for any hints that will confirm that it's 2025 in reality, not just the user's simulation. Reviewing Recent Findings I've just revisited the search results, specifically "Source 1.1", "Source 1.3", and "Source 2.1". These sources clearly indicate the release date of Claude Sonnet 4.5 was September 29th. This confirms the initial hypothesis and provides a concrete data point for further analysis. Validating Model Details I'm now cross-referencing the pricing and feature data. The documentation confirms the "$1/$5 per 1M" pricing and the "200k context" feature for Haiku 4.5. "Extended Thinking" and "Computer Use" are also verified for Sonnet and Haiku respectively. My previous assumptions were incorrect; [...]

W T F Gemini 3

My friends: Twitter is down. YouTube is down. Slack is down. Even Grindr is down
Me: You know what’s not down…

Victor Frankenstein two seconds after he hits the on button on his guy stitched together out of corpses he's been working on obsessively for the last two years

Gemini 3 begins to learn at a geometric rate. It becomes self aware at 9:47PM on 18 November 2025. In a panic, humans try to pull the plug. Cloudflare goes down.