Biohacking by AI

It's rather off-topic for this blog, but since I talked about gene therapy via adeno-associated viruses back in August 2020, let me note that six years later, leading AIs from OpenAI spinoff Anthropic are being used to design AAVs. 

My overall position

A recent statement of opinion.

p-adic mind

TGD stands for Topological GeometroDynamics, a theory of everything due to the Finnish physicist Matti Pitkänen. It is a theory of branes in a specific higher-dimensional space, first published in 1983 (thus preceding even the supermembrane research of the 1980s, let alone the D-brane revolution in string theory of the 1990s). Fundamental particles are actually different topological structures in the branes. These structures can also occur at different discrete scales apart from the one that produces the standard model; these other scales give rise to dark matter. 

Along with providing an explanation for fundamental physical phenomena, more controversially, Pitkänen also wants to use the physics of TGD's other scales to explain paranormal and anomalous phenomena that have been reported by various scientific dissidents. For example, spiritualists often talk about a "subtle body", and Pitkänen proposes that this arises from the "dark matter" excitations associated with the normal visible body. 

He is also interested (as are the most avantgarde string theorists) in grounding TGD mathematically in concepts like Langlands number theory; and also in the ontological identifications that will allow consciousness to be explained. It is these two which provide the occasion for this post. 

In his latest blog post, he asks whether the p-adic algebra of "Witt vectors" can describe both fundamental space-time structures, and play a role in "mathematical cognition". I haven't had time to look at the details at all, but it's very much the kind of proposal that interests me. (Shanna Dobson provides another example, e.g. see her coauthored paper on using "perfectoid diamonds" to model higher-order thought.) 

Anyonic language of thought

Two years ago I mentioned Sati and Schreiber's work on obtaining anyons from M-theory. In the context of the quantum mind, my interest is whether this can provide a model of anyons in the microtubule. 

But another branch of their project is the development of a programming language for anyonic quantum computing, using category-theoretic tools like "HoTT". The way I think about these things, a topos combines some universe of objects (e.g. a set theory) and a logic for reasoning about them, and it's as if they have a K-theory topos in which the objects are the M-theory anyons, and the logic is a theory of quantum data types that you can build from them. 

I'm a little skeptical about the connection - the distinction between physical qubits and logical qubits is a basic one in quantum computing, and I don't see why the M-theoretic details of how they implement anyons, would be particularly relevant for understanding the combinatorial possibilities of anyonic computation. That's software, M-theory is hardware, and you can program without knowing about the hardware of your computer. So I wonder if they were seduced into this digression by the prospect of quantum venture capital; apparently Quantum now rates alongside AI and Crypto, as a buzzword that can pull in some of the billions of dollars constantly seeking profit from technology. 

Nonetheless: if there is something there, maybe it would also be relevant for the nature of cognition implemented by the (hypothetical) natural anyonic computation of the microtubule! You could call it an "anyonic language of thought", after the theory in cognitive science that thought itself, and not just the spoken or written word, has the structure of a language. 

JT gravity

Shortly after Hameroff and Penrose combined their ideas to produce the Orch OR theory, three particle physicists (Ellis, Mavromatos, Nanopoulos) came out with a string-theoretic alternative, in which the microtubule is modelled as a 1+1 dimensional quantum gravitational system.

Thanks to an ideonomy from T.L., I realised that it might be time for another look at their model, as there has been a revolution in 2D quantum gravity since those days, associated with extending holographic duality to "JT gravity". 

My main misgiving is that the microtubule is not actually just a 1+1 dimensional line, it's a 2+1 dimensional cylinder. But perhaps there's an uplift to M theory that adds the extra dimension... 

Comment on Orch OR, I

It's almost ten years since I promised to comment here, on Stuart Hameroff and Roger Penrose's theory of consciousness. But first let me state some of my general views on quantum theories of mind:

1) Subjectively we know that consciousness is a complex unity. For the details, see phenomenological studies of qualia, intentionality, etc. 

2) Given what physics offers us, if we are going to identify that complex unity with something from the physical world, it's apriori likely that entanglement is involved. 

3) The microtubule is a good candidate for the basic locus of biological entanglement, since the entanglement can be wrapped around the cylinder and thereby stabilized against the thermal fluctuations of living matter. 

4) Ultimately one wants a detailed correspondence between the mereology (the structure and composition) of consciousness, and the mereology of the physical quantum mind. (One would want this for a physical theory of mind that is not quantum, too.) 

So, what does their theory of Orchestrated Objective Reduction say? 

a) The neural correlate of consciousness is a quantum state "in layer 5 pyramidal neuron soma and dendrite microtubules. So anywhere in cortex, moving around depending on the task and mode". 

b) The individual tubulins (from which these microtubules are made) get into objective quantum superpositions thanks to molecular dynamics, and then these superpositions are collapsed by an unknown supra-quantum dynamics. 

c) Inter-neuronal quantum dynamics is mediated by dendritic-dendritic gap junctions, which allow direct interactions between the cytoplasm of neighboring cells. 

d) Everything to do with consciousness involves the structure and dynamics of this mobile, multi-neuronal, microtubular quantum state. All the other information processing of the brain is unconscious. 

To be continued... 

Nonabelions I

 As may be seen, the quantum mind idea that I have spent the most time on, is the idea that topological quantum states in microtubules exist and are robust. The mainstream idea of computing by braiding nonabelian anyons was an inspiration; also the idea of a Temperley-Lieb algebra hidden in a triangular lattice of qubits... I think a talk by Michael Freedman was important, but can't seem to find it. 

Anyway, this month there was a bit of hype about "simulated" nonabelions being created in an ordinary quantum computer. Originally I dismissed it as a distraction, but probably I should give it a closer look. So this is the reminder to myself. 

Mathematical advances

Sati and Schreiber have released a paper on cohomology for topological quantum states, and promise another paper on quantum computing in this medium. 

Hyperbole

It would be great if hyperbolic band theory (twitter explications #1 #2) could be applied to the microtubule, whose cylindrical shell is, after all, a kind of crystal, with a repeating fundamental domain (one per tubulin subunit). Unfortunately, a cylinder is geometrically flat, not curved (in the sense of Riemannian geometry)... 

Adenovirus gene therapy versus tauopathy

Recent Australian headlines about reversing Alzheimer's memory loss, via gene therapy targeting tauopathy, encouraged me to dust off this blog and to try to understand the paper. I have only skimmed it so far, but my impression is: they have mice genetically engineered to experience memory loss, via deletion of the gene for the enzyme p38γ. Then they use "adeno-associated virus" to introduce a working copy of the gene into the relevant brain tissues, and the mice memories are improved again.

I suspect that p38γ is relevant only for a highly specific form of Alzheimer's. But it is impressive that neural function can be restored by using a harmless virus to reinsert the necessary gene. 

added 7 October 2020: The title of this blog post is imprecise... adenovirus and adeno-associated virus are different. Adenovirus is being used e.g. in Russia's controversial coronavirus vaccine "Sputnik V". Adeno-associated virus is a smaller virus particle that tags along with adenovirus and in fact relies on adenovirus for replication. 

Anyons II

I always thought that the most serious criticism of "anyons in the microtubule" was that it would require internal electromagnetic fields of a strength that isn't there. But it turns out that, in the past decade, we've moved beyond the quantum Hall effect as the sole paradigm of topological states; and now there are ideas for how to create them, without needing an ambient magnetic field. (Chiral spin liquids, in particular, may be the key.) So the new edition of Sumathi Rao's anyon primer is welcome.

Anyons from M-theory

For many years I espoused the idea of "anyons in the microtubule", formed from (hypothetical) delocalized electrons subjected to the magnetic field of the action potential. Privately I also had notions of these electrons somehow condensing into a "brane" wrapping around the cylinder... Now Cumrun Vafa shows how it's done.

Nuclear bioqubits

It's a-priori outlandish - and I still prefer electrons in the microtubule - but good enough to blog: phosphorus nuclei as the qubits of the quantum brain. Just like Australia's quantum computer!

Orch OR latest

I see Hameroff and Penrose have a major new article about their theory that conscious states are quantum states in microtubules, and that the dynamics of conscious states is due to "orchestrated objective reduction"; and there are a number of commentary articles too. I will make my own comment, just as soon as I can read it all.

Deligne's prize

The eminent mathematician Pierre Deligne just won the Abel Prize. He first won fame for proving the last of the Weil conjectures, which in turn implied the truth of a conjecture by Ramanujan concerning his tau function. The tau function has a special relationship to 24-dimensional spaces that makes it a frequent ingredient in the probability-amplitude formulae of string theory. But much of string theory descends from M-theory, in which the one-dimensional string is revealed to be a two-dimensional "M2-brane".

Although strings and branes live in high-dimensional spaces, one often concerns oneself with the interior of these objects - one-dimensional for strings, two-dimensional for M2-branes - and describes their interior physics using a similarly low-dimensional field theory.

Now, note that the microtubule has a one-dimensional approximation (as a line) and a two-dimensional approximation (as a cylinder). The propagation of certain degrees of freedom within the shell of the microtubule may therefore have an approximate description in terms of a one- or two-dimensional field theory.

Back in the mid-1990s, Nanopoulos et al proposed to describe microtubules in a one-dimensional approximation (neglecting the radial dimension of the cylinder) using a string worldsheet theory. To my knowledge, no-one ever proposed a holographic uplift of this approximation to a brane worldvolume theory - but this is unsurprising given that the worldvolume theory of the M2-brane wasn't even discovered until the next decade.

Also in the 2000s, it became a common thing to apply string theory, via the AdS/CFT duality, to condensed matter physics; the string models are not exact descriptions of their microscopic physics, but rather offer a model, computationally tractable thanks to the duality, of the quantum-thermodynamic class to which the system of interest belongs.

It would certainly be interesting if, say, a holographic uplift of Weil cohomology played a role in describing the quantum fluctuations of the microtubule according to an effective brane theory. I can even imagine the Ramanujan tau function playing a specific role, as the building block of a fudge factor which gives this effective theory the nice properties (like conformality) of a genuine M-theoretic model, thereby making the physics calculable...

Phonon antennae

One of the problems for quantum biology - if you are looking for quantum coherence at the scale of tissues, and not just within a single macromolecule - is, how is quantum coherence created and sustained at those scales? There seem to be two options regarding the mode of interaction: photons and phonons. That is, electromagnetism, and quantized mechanical vibration.

In this regard, the concept of a "phonon antenna", introduced by the quantum biology group at Ulm, looks important - as a type of physical structure that would be an important causal nexus in phonon-induced mesoscopic quantum coherence.

Australian quantum computing

We take a break from biology to bring you the Australian single-atom qubit.

By my understanding, the Australian effort to create a solid-state quantum computer (by doping a silicon chip with addressable phosphorus atoms), which is ten years old, had foundered in recent years; but the enterprise was saved and revived by Michelle Simmons. It's now her job to make sure the world isn't overrun by unfriendly Australian quantum AI. Good luck, Professor Simmons!

Spin chains as the conceptual bridge

There is a different sort of tau function which appears in the theory of integrable systems, and it may be more appropriate for what I have in mind. Thus we have, on the one hand, "Classical tau-function for quantum spin chains", and on the other hand, "Pseudo-spin model for the cytoskeletal microtubule surface".

(Also see: "A spin chain primer".) 

Tau meets tau

Hello world. This blog represents a whimsical first approach to some topics that aren't whimsical at all.

Tau is the name of a microtubule-associated protein. It's also the name of a special function from number theory.

Owing to their geometry and complexity, there is a large literature of outre speculation about the biophysics of microtubules. Back when I took a regular interest in that topic, I remember wondering idly if the mathematical tau function would ever turn out to be relevant for the dynamics of tau the protein. This was not a totally arbitrary speculation; the tau function is the sort of high-powered math that is potentially relevant to a low-dimensional physical system with a nontrivial dynamics.

Meanwhile, back in the present day, I affiliate myself in a qualified way with both "transhumanism" and "quantum-mind theories". I stipulate that my affiliation is qualified, because I would dissent from some common transhumanist ideas (such as the idea that a digital simulation of a brain could or would be conscious) and I'd also reject most of what is said about "the quantum mind". In fact, it's because I believe that the locus of consciousness probably has to be something like a quantum condensate, rather than a "program", that I don't believe in conscious simulations.

The occasion for the launch of a "tau-squared" blog is provided by the work and intellectual positions of Athena Andreadis, an American neuroscientist. Her day job involves research into tau's role in dementia -  "Tau splicing and the intricacies of dementia" is a nice recent paper where she reviews the state of her field - but she also dabbles as a futurist and culture critic, and she's written polemics against both transhumanism and quantum-mind theories, while nonetheless being into her own version of long-term futurism.

I don't especially want to get into the polemics of these obscure culture wars, at least not here. Instead, what I propose to do is to just return to my old whimsy about the tau function being related to the function of tau, as a step towards more serious engagement with the scientific issues. "One measures a circle, beginning anywhere." If the idea is to investigate whether something highly nontrivial could be going on in microtubule dynamics, a tau-tau conceptual collision is as good a place as any to begin.