Atanas Stankov

Do neurons create their own counterfactuals?

Ick. “It’s not at all crazy that neuroscience—even computational neuroscience—becomes more of a commercial endeavor in the near future.”

Is it possible math proof claims by OpenAI are “not even wrong”? Like it is possible to find theoretical models that describe reality at some mid level but are experimentally untestable. Doesn’t incompleteness play role here?

In college used to listen to Slavoj Zizek as a fun parody of political philosophy but now he is making more sense than researchers ranting about AI misalignment of AI use in society.

Top-down control in the brain is two major subregions coordinating task-based complex motor movements in a non-hierarchical fashion like a rally driver and co-driver running through a stage. m.youtube.com/watch?v=USbn...

Hypothesis: Top-down control in the brain is akin to pro StarCraft player making builds and micromanaging units. Will post more metaphors as they come to mind.

“Dendritic computation for abstract rule-based flexible decision-making” www.science.org/doi/10.1126/...

I am feeling increasingly frustrated with this disconnect between questions people ask and designs they execute. We keep talking about *fixing science* via tools, checklists, transparency, honesty signals... but this core problem in some fields cannot be fixed with any of those.

Hypothesis: This is the future of science. Individual networks are a more significant map to ideas than the institutions. What is yet it to be mapped is collaboration and the friendship networks. It is only a matter of time because of AI and the information age. neurotree.org/neurotree/

“Dendritic delay lines shape the computation of sound location in neurons of the gerbil medial superior olive” www.biorxiv.org/content/10.6...

Even the novelty accounts on social media can’t satiate the hunger for weird left field vibes anymore. The memes have been out Goodharted.

Saw someone post recently saying that junior professorships are glorified postdoc positions. While a negative take, this puts good context into latent politics that need to further unpacked as an issue of convention and game theory. Professorships are artificial Nash equilibrium.

I feel like the only lucid neurotheory I’ve come across lately is the causal (bottom-up!) interpretation engrams and the role of top-down modulation by traveling waves. That and I guess the modeling of small scale ephapses. So much of the field is coping with co-dependence on AI.

So anyways I am waiting for AI subscription prices to ramp up so all these “mathematicians are now beaten by AI” posters stop yapping.

RIP to a true Guy Among Guys

Is neuroscience punk?

France and Spain woot!!

Hypothesis: Solarpunk needs a basis of analytic philosophy. A kind of analytic philosophy that can support pluralism à la Paul Feyerabend and/or David Lewis. With that we can better operationalize the bottom-up aspects of standards and institutions. Top-down is solved.

Anyone talking about AI interpretability without giving any context to the substrate the AI is implemented on is not talking about AI interpretability. Even David Marr would raise eyebrows about this.

If AI is legally claimed to be conscious can I ask that the AI used to service debts I owe to be paused because that AI is exploited?

AI sucks at asking questions and it will continue sucking at asking questions.

I don’t think MATLAB will survive the decompiling agent era.

Hypothesis: DEI is going to come back with a much more based and rigorous model. If this is still the case in many institutions then the change will be back an overt announcement of the policy because people have fear extinction of bullies.

I realized institutional reform is so badly needed that salary increases, while critical, have become secondary. The goal posts have shifted from the COVID-era student job strikes. Now it is about something deeper bc the counterfactuals of “what could have been” are 🚨🚨🚨

Imagine a midterm Nov 8 being a bigger catalyst than a general election. Insane times.

Hidden Spirals Underlie Cortical Traveling Waves www.biorxiv.org/content/10.1... Fantastic collaboration with Gavin Zhang, Bard Ermentrout, @umamohan.bsky.social, Erfan Zabeh, @lucakolibius.bsky.social, and @neurojosh.bsky.social. #neuroscience #waves #cognition #consciousness

Hypothesis: Social media and AI are making the world into one village bc information diffuses faster. Local stability is coming bc that is the only physical equilibrium that makes sense. Kind of like a Nash equilibrium. And weirdly I the this is important for a Solarpunk future.

Here’s a wild psychophysics paper translated into an in-vivo single neuron stimulation experiment: Direct detection of a single [NEURON] by humans “Direct detection of a single photon by humans” pmc.ncbi.nlm.nih.gov/articles/PMC... I like throwing these out there because I don’t have a lab.

Hypothesis: Collaborative neuroscience will continue decline at the institutional level because it does take advantage of bottom-up complexity. Instead collaboration standards are force fed via top-down low resolution and archaic standards.

Hypothesis: Neuroscientists are not good at scaling complexity past a single brain because they often don’t treat single neurons with enough nuance.