Atanas Stankov

Canonical Microcircuits for Predictive Coding In this Perspective, Bastos et al. consider the notion of a cortical microcircuit in light of recent theories about neuronal processing, highlighting the remarkable correspondence between the microcir...

Why would the canonical microcircuit do predictive coding when mechanisms like ephaptic coupling are emergent? Cognitive models for local circuits that don’t address biophysical mechanisms are not okay. 🧪

www.cell.com/neuron/fullt...

PS: Ernst Niebur, one original authors of the Koch and Itti paper in visual salience maps, has been still working on the grouping hypothesis and Anne Treisman’s ideas to solve the binding problem in the visual system. Takeaway is, bottom-up gestalt mechanism and top-down integration need to merged.

Hypothesis: Anne Treisman’s integration theory hypothesis for top-down control (the archaic term was “attention” in systems neuro sphere) becomes the default mechanism for handling representation as one goes up the cognitive hierarchy.

Hypothesis: More abstract grouping cell behavior is better defined in the hippocampus than in V1 through V4 or even S1. In other words, place and grids cells are comparatively more complex representations than simple/complex/border-ownership selectivity cells. Some may argue this is obvious.

Sensory experience steers representational drift in mouse visual cortex - Nature Communications Neural mechanisms underlying representational drift are not fully understood. Here authors report that the preferred orientation of mouse visual cortex neurons drifts over time. Altering visual experience does not change drift magnitude, but rather its direction, such that neurons’ tuning matches the statistics of the environment.

PS: Representational drift of V1 tuning curves takes weeks in mice.

“Sensory experience steers representational drift in mouse visual cortex”

www.nature.com/articles/s41...

Conjecture: What if representational drift also affects grouping cells in visual binding? What if functional grouping of cells representational drift goes as low as V1/V2 given mixed selectivity?

Once my current project finishes, I’m not going back to MATLAB unless it’s actually needed for the job at hand or old code needs to unkept/translated. Good software for many things but the licensing model has become ridiculous.

Solving the binding problem: Assemblies form when neurons enhance their firing rate—they don’t need to oscillate or synchronize When we look at an image, its features are represented in our visual system in a highly distributed manner, calling for a mechanism that binds them in…

To deny perceptual binding via oscillations or binding-by-synchrony is an extraordinary claim. 🧪

“Solving the binding problem: Assemblies form when neurons enhance their firing rate—they don’t need to oscillate or synchronize”

www.sciencedirect.com/science/arti...

Well then we have ourselves a bit of conundrum because neurons control/model their own inputs which means they have knowledge about system-wide function. System motifs scale from neuronal to higher levels because it’s all distributed and highly self-referential, including social interactions.

Actually no, this is debate because you making comments on how neurons work. I’m tying top-down control across systems.

(Sorry meant to they do connect to motor action via general frameworks of top-down control.)

Yes they do and I’d like to engage but I can’t take debates with seriously because of your dislike of neuroscientists.

PS: This phenomenon as a well for athletes and race car drivers is related to the mind’s eye.

appliedsportpsych.org/resources/re...

Frontiers | Does neural computation feel like something? Artificial neural networks are becoming more advanced and human-like in detail and behavior. The notion that machines mimicking human brain computations migh...

This feels related to these two papers:

“Does neural computation feel like something?”

www.frontiersin.org/journals/neu...

“Physics versus graphics as an organizing dichotomy in cognition”

www.sciencedirect.com/science/arti...

Attractor and integrator networks in the brain - Nature Reviews Neuroscience Attractor network dynamics can support several computations performed by the brain. In their Review, Khona and Fiete introduce different attractor dynamics and their computational utility, describe evidence of attractor networks across the brain and explain how such networks could be recombined to increase their flexibility and versatility.

Do attractor networks in the brain emerge based on how information from input flows (attractors are observed) or does the brain emulate attractors like some internal model to do computations (attractors are built)?

[The former is just so much more parsimonious.]

www.nature.com/articles/s41...

Challenges in Statistics: A Dozen Challenges in Causality and Causal Inference Causality and causal inference have emerged as core research areas at the interface of modern statistics and domains including biomedical sciences, social sciences, computer science, and beyond. The f...

“Challenges in Statistics: A Dozen Challenges in Causality and Causal Inference”

arxiv.org/abs/2508.17099

Morocco rocked by Gen Z uprising: 'We are the youth, we are not parasites' Young people in the movement called 'GenZ 212' have been protesting across the country for six days, denouncing social inequality and corruption.

Now Morocco!! This happened in Nepal just weeks back and it’s starting up again elsewhere. It was evident back in the spring that 2025 had 1989 vibes. I will stay on the annoying optimist hill. Lampoon the doom.

Still watching Serbia, Hungary, Turkey…

www.lemonde.fr/en/le-monde-...

𝗪𝗮𝗻𝘁 𝘁𝗼 𝗹𝗲𝗮𝗿𝗻 𝗮𝗯𝗼𝘂𝘁 𝗵𝘂𝗺𝗮𝗻 𝗮𝗳𝗳𝗲𝗰𝘁𝗶𝘃𝗲 𝗻𝗲𝘂𝗿𝗼𝘀𝗰𝗶𝗲𝗻𝗰𝗲?
This is massive, a lot of great material.
Huge amount of work to put this together, many thanks to Jorge and Patrik for getting the second edition out.

Thanks for clarifying.

It might just thermodynamically cheaper for brains to behave this way to find novelty? What natural stimuli behaves like a noisy-TV? It’s like using high fructose corn syrup to artificially sweeten foods and trick the brain into thinking it’s finding a scarce resource.

PS: I have been thinking that anticipation of AGI is illiberal thinking.

I don’t understand if the AGI argument is real?

Having been always hands on with the data I work with, I’ve found that my unconscious pattern matching has become more theoretical. Imagine doing data analysis so much you can compare yourself to ML algorithms.

Why do so much faceless/abstract corporate entities have so much say in my social/professional life?

For example: How are FICO scores determined? Why do I have to give up my information to third parties? Why is student debt not treated as a monopoly? Why do insurance companies have so much power?

If we had this conversation a few years ago I’d agree. The current crisis was inevitable and the current leadership is a nasty catalyst. There is no way the grant funding model for science and its centralized agencies were going to survive globalization. Same goes for the science journal monopolies.

There is no such distraction. Systems evolution is not something you control.

It’s actually both lol. I was being a bit categorical there, it’s all about systems evolution if you ask me, so we need new definitions across all thought, especially for computation. If you disagree, that’s fine too. As long as the dialectic continues, self-organization takes precedent.

I can’t follow because that’s the kind of policy that doomed science progress.

‘Knowing’ will always socially usurp existing power structures because people become anti-establishment. People are ‘anti-realist’ by their innate nature because ‘knowing’ is a thermodynamically cheap policy.

[Is rejecting ‘knowing’ a deception?]

Okay, but how does this relate to anything being anti-Searle.

What about an abstract sandbox in which I can lay out different sequences of strategy one and two? See I can use my own brain to layer levels of Turing Tests.