Maxine 💃🏼
@maxine.science
🔬 Looking at the brain’s “dark matter”
🤯 Studying how minds change
👩🏼💻 Building science tools
🦋 ♾️ 👓
🌐 maxine.science
🤯 Studying how minds change
👩🏼💻 Building science tools
🦋 ♾️ 👓
🌐 maxine.science
The result of the paper is simply to say that if you don’t mandate (2), you tautologically get (1) by solving the task just falling out of the representation theory.
November 25, 2025 at 11:43 PM
The result of the paper is simply to say that if you don’t mandate (2), you tautologically get (1) by solving the task just falling out of the representation theory.
2. as an embodied form, needing that representation theory to be compressed into the material substrate available at the given evolutionary epoch from the various composable units available.
November 25, 2025 at 11:41 PM
2. as an embodied form, needing that representation theory to be compressed into the material substrate available at the given evolutionary epoch from the various composable units available.
My interpretation is the specific functional organization we see is mandated under the constraints of
1. needing a (perhaps weak, up to some equivalence) internalization of the representation theory of a specific world-symmetry required to enable certain affordances for structure-persistence, and +
1. needing a (perhaps weak, up to some equivalence) internalization of the representation theory of a specific world-symmetry required to enable certain affordances for structure-persistence, and +
November 25, 2025 at 11:41 PM
My interpretation is the specific functional organization we see is mandated under the constraints of
1. needing a (perhaps weak, up to some equivalence) internalization of the representation theory of a specific world-symmetry required to enable certain affordances for structure-persistence, and +
1. needing a (perhaps weak, up to some equivalence) internalization of the representation theory of a specific world-symmetry required to enable certain affordances for structure-persistence, and +
My interpretation of the paper is not to say that functional cell types don’t exist, just that as defined by neuroscience, they are epiphenomenal of the representation theory (in the mathematical sense) of the symmetry structure of the observable. +
November 25, 2025 at 11:41 PM
My interpretation of the paper is not to say that functional cell types don’t exist, just that as defined by neuroscience, they are epiphenomenal of the representation theory (in the mathematical sense) of the symmetry structure of the observable. +
lol they deleted that but left Caligula?
November 25, 2025 at 12:28 PM
lol they deleted that but left Caligula?
3. To say an organoid has no experience when recording is to say that a fetus just before birth has no experience—that is to say, absurd. It has at the very least experience of many months of physical and chemical exposure regularities. There is no “before experience”—the notion itself is absurd.
November 25, 2025 at 12:00 PM
3. To say an organoid has no experience when recording is to say that a fetus just before birth has no experience—that is to say, absurd. It has at the very least experience of many months of physical and chemical exposure regularities. There is no “before experience”—the notion itself is absurd.
2. Constraint is a prerequisite for variety—and at bedrock, tabula rasa is about function, not structure. No one argues that a zygote can turn into a spaceship. The question is the spectrum of possible things it will do. And quite plausibly, this structure constraint *enables* functional variety. +
Architecture, constraints, and behavior | PNAS
This paper aims to bridge progress in neuroscience involving sophisticated quantitative
analysis of behavior, including the use of robust control, ...
www.pnas.org
November 25, 2025 at 12:00 PM
2. Constraint is a prerequisite for variety—and at bedrock, tabula rasa is about function, not structure. No one argues that a zygote can turn into a spaceship. The question is the spectrum of possible things it will do. And quite plausibly, this structure constraint *enables* functional variety. +
1. The alternative is a collection of cells with a sparse subset of the set of possible connections between them having arbitrary dynamics, which would be nearly impossible. So while the results in this paper are interesting scientifically they are not unexpected. +
November 25, 2025 at 12:00 PM
1. The alternative is a collection of cells with a sparse subset of the set of possible connections between them having arbitrary dynamics, which would be nearly impossible. So while the results in this paper are interesting scientifically they are not unexpected. +
I still am really curious about the overall concept though—it doesn’t strike me as theoretically implausible, and I think stripping out some of the parameter dependence to get at like “up to homotopy” what’s goin’ on could be an interesting theory Q
November 24, 2025 at 6:51 PM
I still am really curious about the overall concept though—it doesn’t strike me as theoretically implausible, and I think stripping out some of the parameter dependence to get at like “up to homotopy” what’s goin’ on could be an interesting theory Q
one of those people was me in year 2 of PhD; holy macaroni just look at their matlab code 🤣 +
November 24, 2025 at 6:51 PM
one of those people was me in year 2 of PhD; holy macaroni just look at their matlab code 🤣 +
Izbikevich also had an extremely cool, wild, and thought-provoking model of how this might work that I feel never really got fully fleshed out.
www.izhikevich.org/publications...
www.izhikevich.org/publications...
www.izhikevich.org
November 24, 2025 at 12:30 PM
Izbikevich also had an extremely cool, wild, and thought-provoking model of how this might work that I feel never really got fully fleshed out.
www.izhikevich.org/publications...
www.izhikevich.org/publications...
it is the existence of more “civilized” AI products like Claude that create the mental justification used to convince people over time to use things like Grok unquestioningly.
November 24, 2025 at 12:22 PM
it is the existence of more “civilized” AI products like Claude that create the mental justification used to convince people over time to use things like Grok unquestioningly.
weirdly I think Baudrillard actually agreed with their original take that the overt manipulation is far less insidious than the manipulation that becomes indistinguishable from genuineness, because the latter kills the distinction of genuineness itself. +
November 24, 2025 at 12:22 PM
weirdly I think Baudrillard actually agreed with their original take that the overt manipulation is far less insidious than the manipulation that becomes indistinguishable from genuineness, because the latter kills the distinction of genuineness itself. +
now, what’s really going to bake your noodle later on is, what happens in the stuff you Don’t record
November 24, 2025 at 12:18 PM
now, what’s really going to bake your noodle later on is, what happens in the stuff you Don’t record
oh shap, here come dat boi
November 24, 2025 at 12:11 PM
oh shap, here come dat boi
christ almighty I only hate it because I already see the $100m series A
November 23, 2025 at 2:46 PM
christ almighty I only hate it because I already see the $100m series A
Also AI/Blockchain/Kombucha startup I’m dead 😂🤣
November 23, 2025 at 2:33 AM
Also AI/Blockchain/Kombucha startup I’m dead 😂🤣
November 23, 2025 at 2:31 AM
A better one for toe dipping:
An intro rooted in LLMs:
www.math3ma.com/blog/languag...
(and the other two parts)
An intro specifically to limits and colimits (or “splaying” and “gluing”):
www.math3ma.com/blog/limits-...
(and the other two parts)
An intro rooted in LLMs:
www.math3ma.com/blog/languag...
(and the other two parts)
An intro specifically to limits and colimits (or “splaying” and “gluing”):
www.math3ma.com/blog/limits-...
(and the other two parts)
Language, Statistics, & Category Theory, Part 1
In the previous post I mentioned a new preprint that John Terilla, Yiannis Vlassopoulos, and I recently posted on the arXiv. In it, we ask a question motivated by the recent successes of the world's b...
www.math3ma.com
November 22, 2025 at 11:49 PM
A better one for toe dipping:
An intro rooted in LLMs:
www.math3ma.com/blog/languag...
(and the other two parts)
An intro specifically to limits and colimits (or “splaying” and “gluing”):
www.math3ma.com/blog/limits-...
(and the other two parts)
An intro rooted in LLMs:
www.math3ma.com/blog/languag...
(and the other two parts)
An intro specifically to limits and colimits (or “splaying” and “gluing”):
www.math3ma.com/blog/limits-...
(and the other two parts)
The recent thinking machines post on modular manifolds has a nice categorical story that I think fits into this extremely nicely and would love to be explored more thinkingmachines.ai/blog/modular...
Modular Manifolds
A geometric framework for co-designing neural net optimizers with manifold constraints.
thinkingmachines.ai
November 22, 2025 at 11:46 PM
The recent thinking machines post on modular manifolds has a nice categorical story that I think fits into this extremely nicely and would love to be explored more thinkingmachines.ai/blog/modular...
colimit in nLab
ncatlab.org
November 22, 2025 at 11:44 PM
The mathematical question of “under what circumstances of building a larger structure by composing component structures is this property maintained” is one of the driving motivations of category theory!
November 22, 2025 at 11:42 PM
The mathematical question of “under what circumstances of building a larger structure by composing component structures is this property maintained” is one of the driving motivations of category theory!
the AI people need category theory y’all
November 22, 2025 at 11:36 PM
the AI people need category theory y’all