Complexicist working on minimal modeling and mesoscopic analysis of biological systems. A research affiliate currently working with Michael Levin at Tufts.

Joined April 2018
36 Photos and videos
A few iterations with Gemini on an attempt to deep-meld computer science with biology gave me this. Not bad!
1
100
A model of toy models
The key to "physics of AI" is defining appropriate toy models. Here's the methodology of toy models: kindxiaoming.github.io/blog/…
2
155
Santosh Manicka retweeted
Applications open for our six-week computational oncology summer training 2026 (May 18 - June 26; due: Feb 15, 2026). tinyurl.com/bdhth4em Plz RT @biopatrika @BE_IISc @iiscbangalore @rakeshthejoy @karthikraman @abandopa @DynamicsLiving @mathoncbro @DeepakNModi @r_sabarinathan
2
53
108
9,858
How does one represent visual intuition without the intricacy of "microscopic" symbols? How to device "macroscopic" symbols when the devil is in the details? Perhaps thinking in terms of "symbols" is not the way forward?
One of the reasons I am very much looking forward to new architectures is that problems in algebraic topology fundamentally require intuitions that go far beyond simple symbolic manipulations. It is often much easier to present a visual sketch of an isotopy/homotopy-based argument than to design precise formulas and make it fully formal. I hope that a combination of world models, symbolic manipulation, and representations of object dynamics is a path toward completely new architectures. We need them if we want to solve hard mathematical problems that go beyond mere text. In the picture (based on arXiv: 2107.01664) you can see how easy it is to intuit the handcuff paradox but how hard it would be to write it formally with formulas.
1
290
Non-invasive manipulation of an LLM's internal belief may be equivalent to invasive (surgical) control?
11 Nov 2025
New research: are prompting and activation steering just two sides of the same coin? @EricBigelow @danielwurgaft @EkdeepL and coauthors argue they are: ICL and steering have formally equivalent effects. (1/4)
124
Autocompletion in semantic space or just syntactic space?
What happens when you turn a designer into an interpretability researcher? They spend hours staring at feature activations in SVG code to see if LLMs actually understand SVGs. It turns out – yes~ We found that semantic concepts transfer across text, ASCII, and SVG:
1
91
Have you wondered why repeating a word in your head by focusing on its syllables makes you strangely detach from it and lose its meaning? Does focusing on the "syntax" make one lose their grip on the semantics? Could this be a strange case of syntax-semantics complementarity?
4
3
11
1,100
"pairwise communication alone can cause decision deadlock, but adding group interactions allows the system to overcome stalemates and reach consensus"
Symmetry breaking in collective decision-making through higher-order interactions arxiv.org/abs/2510.00853
2
205
We leverage the unique properties of the electrostatic field -- instant, long-range, and "effusive" -- to pattern the bulk of a minimal bioelectric network by stimulating only its boundary. The field also catalyzes voltage patterns via a synergetics (à la Haken) based mechanism.
Final version is out: @SantoshManicka cell.com/cell-reports-physic… "Field-mediated bioelectric basis of morphogenetic prepatterning" #morphogenesis #bioelectricity #fields "Intercellular bioelectric communication plays an important role in morphogenesis, often modeled using localized non-neural networks generating spatial patterns of membrane potential (Vmem). Here, we find that the electrostatic field contributes to this process, via a synergetics (à la Haken)-based mechanism, by enhancing the complexity of Vmem patterns through a coarse-grained projection. We leverage this property of the field to automatically optimize transient signals from a symmetry-breaking organizer region in the boundary of the tissue to mold Vmem patterns in the bulk. Two models optimized in this way exhibit contrasting “mosaic” and “stigmergic” pattern-coding strategies, depending on their field sensitivity strengths. Interestingly, the stigmergic model recapitulates the qualitative developmental sequence of the bioelectric craniofacial prepattern observed in frog embryos. These results highlight the potential of the electric field both as a facilitator of collective patterning and as a macroscale interventional target for applications in regenerative medicine and bioengineering."
1
5
243
Some nice patterning videos in the SI; also here: youtube.com/@maniackaa/video…
133
Dynamics of decomposed information flow
22 Sep 2025
New paper! Toward a unified taxonomy of information dynamics via Integrated Information Decomposition, out now in @pnas - led by Fernando Rosas,@PedroMediano , & Adam Barrett, w/ Andrea Luppi, @RCarhartHarris, me, & Dan Bor. pnas.org/doi/epdf/10.1073/pn…
2
274
A review of criticality in the brain, as characterized by its hallmark features, namely, scale invariance, marginal stability, tunability, and generative capacity, and its implications.
29 Jun 2025
Is criticality a unified setpoint of brain function?: cell.com/neuron/fulltext/S08… "We perform a meta-analysis of 140 datasets published between 2003 and 2024. We find that a long-standing controversy is the product of a methodological choice with no bearing on underlying dynamics."
2
7
415
So far, there was only one way to define a prime number (textbook defn). Now there are infinite ways to do it. It's as if every (finite) prime number has an infinite set of dimensions, each casting its own unique shadow. How mind-boggling is that! scientificamerican.com/artic…
3
253
Simplicity may be sufficient to generate complexity (e.g., cellular automata), but complexity may sometimes require underlying complexity (e.g., complex disease traits)
"AI identifies key gene sets that cause complex disease" news.northwestern.edu/storie… Nice news article on our new approach to identify gene sets involved in polygenic traits & how it can serve as a tool to for the development of new multi-target treatments for complex diseases.
4
189
"Across a range of classic sequence modeling tasks and a real-world stimulus selection task, minimal nonlinearity is not only sufficient but often optimal, yielding models that are simpler, more robust, and more interpretable than their fully nonlinear or linear counterparts"
Uncovering the Functional Roles of Nonlinearity in Memory. arxiv.org/abs/2506.07919
1
2
241
Could the prefrontal cortex serve as RAM or a Turing tape for neural activity (though perhaps in a limited sense)? At the same time, potentially playing other compartmentalized roles of a traditional computer?
New results! Neural subspaces are not specialized. They are like general workspaces. Recycling of prefrontal subspaces dynamically multiplexes information doi.org/10.1101/2025.05.02.6… #neuroscience
138