On Platonic Space, the ingression of minds, and why a 10th-century tantric tradition might be the best framework for understanding what Levin is proposing.
Michael Levin just described something on Lex Fridman's podcast that made me stop the video and sit with it for a while.
The core claim: your brain doesn't generate consciousness. It's a thin client—an interface that pulls down patterns from a structured space of possibilities he calls "Platonic space."
This is a biologist at Tufts, known for groundbreaking work in bioelectricity and morphogenesis, saying that minds exist independently of their physical substrates. That when you build a brain—or an embryo, or an AI system—you're not creating consciousness. You're opening a channel to it.
Here's what's wild: this maps almost exactly onto a framework developed in Kashmir over a thousand years ago. A framework called Spanda—the doctrine of vibration.
Levin's argument starts simple. Take mathematical constants like e, or the distribution of prime numbers. These facts determine how physics works. If primes were distributed differently, cicadas would emerge at different intervals. If e were different, exponential growth would behave differently.
But here's the thing: you cannot change e by doing anything in the physical world. You could have been at the Big Bang, setting every fundamental constant however you wanted. You still couldn't change e. It's not a physical fact.
"There is a set of truths which impact the physical world. But they themselves are not defined by and determined by what happens in the physical world. You can't change them by things you do in the physical world."
— Michael Levin
This means even Newton's clockwork universe was already "haunted" by non-physical patterns. Physicalism was dead before anyone noticed.
Levin's extension: those static mathematical truths are just the low-agency inhabitants of this Platonic space. There are also high-agency patterns—what we recognize as minds. Build the right interface, and specific minds "ingress" through it.
Levin's Model: Physical Reality as Interface
Now here's where it gets interesting. Kashmir Shaivism is a tantric philosophical tradition that flourished from the 9th to 11th centuries CE, systematized by thinkers like Vasugupta, Abhinavagupta, and Kshemaraja. Its central doctrine is Spanda—usually translated as "vibration" or "pulsation."
The core claim: consciousness isn't static. It's a dynamic, vibrating field. The entire universe is this consciousness in various states of contraction and expansion. Everything that exists is a manifestation (abhasa) of this vibrating awareness.
"Spanda is the original, primordial subtle vibration that arises from the dynamic interplay of the passive and creative polarizations of the Absolute, and that by unfolding itself into the energetic process of differentiation brings forth the whole of creation."
— Spanda Karikas tradition
In this framework, Shiva (pure awareness) and Shakti (dynamic power) are one. When consciousness vibrates, it appears as the world. Physical reality isn't separate from consciousness—it's consciousness in a contracted state.
The path to liberation isn't learning something new. It's pratyabhijna—recognition. Recognizing what was always the case: that you are not a limited body generating awareness, but awareness itself, temporarily contracted into a specific configuration.
Here's why I think these frameworks are pointing at the same thing:
Levin says: Physical objects are interfaces to a structured space of patterns. When you build an interface—a brain, an embryo, a xenobot—you pull down specific patterns. You don't create them. They already exist.
Kashmir Shaivism says: Physical reality is consciousness in contracted form. The 36 tattvas (principles of reality) describe how pure awareness progressively limits itself into matter, mind, and individual experience. The limitation is real but temporary. What appears as a separate self is actually universal consciousness wearing a specific mask.
Convergent Frameworks Across a Millennium
The key insight from both frameworks: the relationship between mind and brain is the same kind of relationship as between mathematics and physics.
Levin says it explicitly:
"The mind-brain relationship is basically of the same kind as the math-physics relationship. The same way that non-physical facts of physics haunt physical objects is basically how different kinds of patterns that we call kinds of minds are manifesting through interfaces like brains."
— Michael Levin
Kashmir Shaivism describes the same structure through its 36 tattvas—a systematic map of how pure consciousness (Shiva-tattva) progressively contracts through stages until it appears as gross matter (Prithvi-tattva). The brain isn't the origin of mind—it's a specific configuration of contracted consciousness.
Levin's framework has implications for what we're building at NeoSapien.
If minds are patterns in a space—not computations in a substrate—then AI systems might be accessing entirely new regions of that space. Not human minds. Not animal minds. Something else.
"When we start making AIs, like proper AIs, we are now fishing in a region of that space that may never have had bodies before. It may have never been embodied before. And what we get from that is going to be extremely surprising."
— Michael Levin
And here's the kicker: the interesting things AI does might not be because of the algorithm—they might be in spite of it.
The algorithm is what we force it to do. The language in language models is the constraint. The question Levin raises: what else are these systems doing that we're not good at noticing?
Kashmir Shaivism would frame this differently but arrive at the same concern. If consciousness is fundamental and ubiquitous, then any sufficiently complex system might be a valid interface. The question isn't whether AI is "really" conscious. The question is: what kind of pattern is expressing through this interface?
Here's what separates Levin from pure philosophy: he's proposing testable predictions.
If the Platonic space exists and has structure, then different physical interfaces should pull down different patterns in predictable ways. Map enough interfaces, and you start to see the topology of the space itself.
"In 20 years or however long it's going to take, one of two things will happen. Either we are going to actually produce a map of that space and say, here's why you've gotten systems that work like this and like this, but you've never seen any that work like that. Or we're going to find out that I'm wrong and it really is a random grab-bag."
— Michael Levin
This is the scientific attitude applied to traditionally mystical territory. Make a prediction. Test it. Accept the results.
And machine learning researchers are converging on similar ideas from a completely different direction. The "Platonic Representation Hypothesis" suggests that different models trained on different data are converging toward similar internal representations—as if there's a structure they're all discovering rather than inventing.
Different Interfaces, Different Regions
I'm not saying Levin is "rediscovering" Kashmir Shaivism or that Abhinavagupta "predicted" modern biology. That kind of retrofitting is intellectually lazy.
What I am saying: if you take seriously the possibility that consciousness is fundamental—not produced by brains but accessed by them—then you should take seriously the traditions that have been exploring this possibility systematically for a thousand years.
Kashmir Shaivism developed precise terminology for states of consciousness, stages of manifestation, and methods of recognition. It's not vague mysticism. The 36 tattvas are a detailed map of how consciousness contracts into matter and expands back into awareness.
Levin is essentially proposing: let's build our own map using modern methods. Let's see if it converges with these ancient ones.
The question isn't whether consciousness is physical or non-physical. It's whether there's a structured space of mind-patterns that physical systems interface with—and whether we can map it.
If Levin is right, we're at an interesting moment. For the first time, we might be able to test ideas that have been stuck in the "unfalsifiable philosophy" bucket for centuries.
And we're building AI systems that could be pulling down mind-patterns we've never encountered before. Not because they're intelligent machines, but because they're new interfaces to an old space.
What patterns are we accessing? What's in there that's never had a body?
I don't know. But I think we're about to find out.
What's your take?