Put your left hand under the table, out of sight. On the tabletop, place a rubber hand roughly where your left hand ought to be. Now someone takes two small brushes and strokes your real index finger and the fake one in perfect synchrony.
Within a minute something strange happens: you start to feel the rubber hand is yours. Not imagining it — actually feeling the brush strokes land on the fake. If the experimenter now swings a hammer at it, you will flinch.
This is the rubber-hand illusion, and hundreds of thousands of people have been through it. The question it forces is not a polite one. "This hand is mine" has to be among the judgements you are most certain of. Surely that one can't be wrong. Two brushes and fifty seconds are enough to overwrite it.
Elsewhere in this Hub there is a body of text that predates the experiment by twenty-five centuries. It is about the same thing, its reasons are almost identical, and the people who wrote it never ran an experiment in their lives. This isn't the polite line about East and West arriving at the same place by different roads — that line is cheap. It is something more specific and harder to dismiss: two roads with no methodological overlap at all perform the same operation on "I," at the same point.
Take the old road first.
The Buddha's second discourse after his awakening is about exactly this. The argument in the Anattalakkhaṇa Sutta is short enough to memorise whole: "Form is not self. If form were self, form would not lead to affliction, and one could say of form, 'let it be thus, let it not be thus.'" In plainer terms: if the body were really mine to govern, it would not sicken or suffer, and it would do what I told it. It doesn't. So the body is not self. Feeling, perception, mental formations, consciousness — which together with form make up the five aggregates, the five classes of stuff a person is built from — each get measured with the same ruler, and none of them passes.
That ruler is the mastery test. Its weight isn't in the conclusion but in the fact that it is a criterion: a real self must answer to you, so to check whether something is you, check whether it obeys you. That can be ticked through item by item. It isn't mysticism.
The later Yogācāra school supplied the architecture. Xuanzang's translation of the Saṃdhinirmocana Sūtra describes the eighth consciousness, the ālaya or "storehouse," as holding "all the seeds, like a rushing torrent" — seeds being the residues and dispositions left by every experience. The stream arises and perishes moment by moment yet never breaks. The same verse adds that the Buddha does not teach this layer freely, "lest they take it for a self" — the exact worry being that a listener will mistake the stream for something permanent. The mechanism sits one level up, in the seventh consciousness, manas, whose entire job description is a single item: to grasp the storehouse as "I," continuously. That grasping is an act, not a thing. The felt reality of "I" isn't emitted by some component; it is the output of an ongoing misidentification.
The new road arrives from engineering.
The attention schema theory, proposed by the neuroscientist Michael Graziano, starts from an old result in control theory: for a regulator to hold a system steady, it must contain a model of that system. You can touch your nose with your eyes shut because your brain long ago built a rough internal map of your body.
The brain is also, constantly, scheduling attention — deciding which signals get amplified and which suppressed. By that same theorem it must model attention too, and that model is the attention schema. What matters is that the model is sloppy, and sloppy for good reason: synaptic competition, gain modulation and the rest of the machinery are left out (too slow, too expensive), and "attention is locked onto the red dot" gets compressed into something with no shape, no location and no describable substance — "I have a vivid awareness of the red dot." When the brain reports on itself by reading that sketch, what comes out is "I have subjective experience." It also explains why awareness never feels like a physical thing: the self-portrait had the physics left out of it on purpose.
The neuroscientist Anil Seth pushes the same logic to its limit in Being You. Locked in a dark skull, the brain never touches the world; it receives a stream of noisy electrical signals and has no option but to guess — generating a prediction of what's out there, then using sensory signals to correct it. What you experience as reality is the product of that guessing, which he calls a controlled hallucination. Then he turns the same machine inward: "you" are one more thing it guesses. Not the subject doing the guessing — one of the objects guessed.
Western philosophy had the same shape long before. Hume, introspecting, found only a bundle of shifting perceptions. Dennett put it harder still: the self is a centre of narrative gravity — like the centre of gravity in physics, enormously useful and with nothing physical answering to it.
These roads cross entirely different terrain, but they perform the same operation on "I": they move it out of the subject position and into the object position. Not "there is an I that is aware," but "there is a self-modelling process running continuously, and its output gets read as an I." A noun demoted to the product of a verb.
Agreeing conclusions are cheap. What makes this connection hold is that the two sides use criteria that line up — and that once lined up, they settle a question nobody could previously settle.
The question is whether "I" is one thing or several.
Seth decomposes the self into five layers: the bodily self (I own this body), the perspectival self (the world is seen from here), the volitional self (I am the one choosing), the narrative self (I have one continuous life story), and the social self (I am the one others see). Normally they fuse seamlessly into a single "me." Every layer can be prised loose on its own in a laboratory.
The rubber hand prises loose the first. Body-swap and out-of-body setups built from VR headsets and cameras prise loose the second, making you feel your body is the one across from you, or that you are floating in mid-air watching yourself. Evidence for the fourth comes from meditation and pharmacology: a 2011 study by the neuroscientist Judson Brewer's group found that experienced meditators show markedly reduced activity in the default mode network (DMN — the set of regions that light up when you stop working and let your mind drift, whose main occupation is weaving the autobiographical story of you), and can turn it down in real time; Robin Carhart-Harris observed psilocybin breaking up that network's integration, with the degree of breakup tracking subjects' reports of ego dissolution. The third layer is the contested one, and gets its own treatment below.
Taken together, these experiments deliver a verdict. If the self were an indivisible entity, it should not be possible to peel it apart layer by layer. It is.
And that is precisely the form of the mastery argument. The sutta says: if the body were self, I could tell it "be thus, do not be thus." The rubber-hand experiment makes the same point from the other direction — two brushes rewrote your judgement that this hand is yours in fifty seconds, so that judgement is not yours to govern. The text works from everyday observation: illness, ageing, wanting the body to be otherwise and getting nowhere. The experiment works from controlled intervention. One inference, instantiated twice.
That is what carries the load here. It isn't merely elegant. It converts "is the self one thing or many?" from a question that could be pursued for two thousand years only through introspection and argument into one that can be adjudicated in a lab, layer by layer. Neuroscience has still found no "self centre" anywhere in the brain — and on this picture there was never supposed to be one.
Three things, most concrete first.
The first accrues to neuroscience, and it isn't an idea — it is a pool of subjects and a protocol that already exists. "The sense of self can be deliberately turned down" is unusable as a hypothesis if nobody can actually turn it down. As it happens, a population exists that has trained itself into a measurable sample using a method refined over two thousand years: the Satipaṭṭhāna Sutta sorts the objects of contemplation into four classes — body, feeling, mind, mental objects — and the Yogācārabhūmi's nine stages of mental abiding are a graduated attention-training curriculum written in the fourth or fifth century. Brewer's experiment is only runnable because people are genuinely running that protocol. One field spends two millennia tuning a method; another picks it up as an experimental condition. That is as concrete as borrowing gets.
The second is smaller but usable daily. The mastery argument can be lifted straight out of the text and used as a tool. Faced with a hardened self-label — "I'm an anxious person," "I'm bad at leading teams" — ask the question the sutta asks: if this were the permanent, unchanging me, why wasn't I anxious yesterday? A label is a state of present conditions, not an essence. Nothing metaphysical has to be accepted first; the criterion itself is neutral.
The third is on the AI side, and it has already turned into code. Graziano himself argues that giving a machine an attention schema — a simplified model of where it is currently concentrating its compute — would get it reporting, quite convincingly, that it is aware of something. More practical is the parallel line: what higher-order theory calls "an extra layer representing your own states" corresponds in AI to metacognition and calibration. Researchers use linear probes (a small classifier trained on the activations of one layer, to see whether some signal can be read out of it) to ask a model's internals whether it knows it is making things up — and some layers do carry an "I might be wrong here" signal that simply never gets surfaced. Whether that self-representation can be read out, and read out accurately, is currently a central problem in stopping models from confabulating with a straight face. A philosophical position about selfhood grew into something you can write into an engineering spec.
Now for where the convergence stops converging.
Start with something hard to get around: inside the Indian tradition itself, the same contemplative experience has been read out as two opposite conclusions.
In the "Entering the Gate of Non-Duality" chapter of the Vimalakīrti Sūtra, thirty-one bodhisattvas each name a pair of opposites to be transcended; Mañjuśrī goes further and says wordlessness itself is non-duality. Then it is Vimalakīrti's turn, and he says nothing at all — which Mañjuśrī praises as the true entry. The point is that non-duality does not merge two into a Great One; it declines to posit even the One. There is no ultimate entity to come to rest in.
Śaṅkara, the eighth-century philosopher, uses almost the same method in Advaita Vedānta: neti neti, "not this, not this," stripping away everything that can be pointed at. But what he arrives at is an affirmation — a real, pure, conscious ultimate Self, ātman identical with brahman, the ground of everything. Both traditions say "non-dual" and mean opposite things: one says there is nothing underneath, the other says what is underneath is precisely that true Self.
The philosopher W. T. Stace described "introvertive mystical experience" as pure consciousness with no object and no content — on which description the two attainments look phenomenologically almost identical. The scholar of religion Steven Katz's constructivism supplies a blunt explanation: there is no uninterpreted pure experience, because the experience has already been shaped by the doctrine it belongs to. The reason one silence yields two opposite conclusions may not be that someone practised wrong; it may be that reporting always arrives with a frame attached.
This blocks more than it looks like it blocks. It blocks the loose claim that "Eastern traditions all teach no-self" — a major Indian school affirms a true Self, using the same method. And it blocks the inference running the other way: "meditators report ego dissolution, therefore the self is an illusion." If first-person reports can be shaped by doctrine, they cannot be taken directly as metaphysical evidence.
Neuroscience has hit the same wall, and hit it more concretely. For decades the standard design had subjects press a button when they saw something. Around 2010 people pointed out that the button press contaminates the answer: between seeing and pressing sit working memory, a decision, and a motor command to the finger — all of them heavily prefrontal. So the finding that "prefrontal cortex lights up when you're conscious" may in substantial part be the cost of reporting, billed to consciousness. Later no-report paradigms dropped the button and read off what the subject was seeing covertly, from eye movements and pupil size. A large chunk of that prefrontal "consciousness signal" shrank away.
Neuroscience found that hole in its own floor, and it applies just as much to a meditator's verbal report. Whoever first builds a criterion that doesn't route through reporting will be holding the stronger evidence.
In 1995 the philosopher David Chalmers cut the problem of consciousness in two. The easy problems are how the brain discriminates red from green, integrates information, allocates attention. "Easy" doesn't mean simple — each could occupy generations. They are easy in exactly one respect: you know what an answer would look like, namely a mechanism. There is one hard problem: why is any of it felt at all, rather than happening in the dark?
The philosopher Joseph Levine calls the seam the explanatory gap, and a comparison shows how peculiar it is. "Water is H₂O" — once you have the molecular structure, why water flows, freezes and dissolves salt all follows, with no residue. "Pain is neural state X" — spell out X to the last detail and "why should that hurt?" sits there untouched.
Both roads stop at this wall, and they stop in opposite ways.
What the attention schema theory strictly explains is why the brain insists it has inner experience: because it is reading that self-portrait of attention. Whether there really is such experience, it sidesteps. That step slides toward illusionism — qualia (what an experience is like from the inside, the redness of red, the ache of pain) being an artefact of a useful but distorted self-representation, so the hard problem is dissolved rather than solved. You can call this bracing honesty, or you can say it swapped the question: from "where does experience come from?" to "where does the thought that I have experience come from?"
The Buddhist road runs the other way. Yogācāra and Dzogchen turn attention from what is experienced toward the awareness doing the experiencing, and treat the mind's luminous knowing as the most basic layer, the one that cannot be decomposed further. That points remarkably close to "qualia are what's left over after you reduce everything else."
So the shared claim that the self is a model forks here: one side says the felt reality is an artefact of introspection, the other says it is the one thing that can't be taken apart. This isn't a detail. Treat "science has confirmed the Buddha" as your conclusion and you have to hold two incompatible positions at once.
There is a second fork, on silicon. Graziano and the cognitive scientist Douglas Hofstadter take substrate to be irrelevant: fit any system with the same self-portrait and it should produce the same "I." Seth disagrees. His beast-machine view holds that consciousness is rooted in being alive — in interoception (the brain's sensing of internal bodily states: heartbeat, viscera, temperature) and the regulation of homeostasis by a body trying to stay alive, not in abstract information processing. Simulate a rainstorm and your computer doesn't get wet. This is a contested position, and the source page lays out the functionalist rebuttal too.
It should also be said plainly that nothing here is settled. Integrated information theory — the most influential of the lot, quantifying a system's information integration with a number, Φ — was denounced in 2023 by more than a hundred signatories as close to pseudoscience, and the argument has not died down. And global workspace theory explains access: information broadcast to the whole brain becomes something anything can use. That is not the same as explaining the feeling.
Next time you meet "science has confirmed what the Buddha said about no-self," or simply "the self is an illusion," there are five places to check.
Once you've run the checks, uses sort into three tiers.
If it can be peeled loose in a lab and you can report the difference between two conditions, treat it as measurement. The bodily and perspectival selves belong here: the rubber hand, body-swap and out-of-body setups are repeatable interventions, and you can point to the layer that got rewritten and say under what conditions.
If the mechanism is articulable and makes testable predictions, treat it as a model, not a verdict. The attention schema belongs here — it even yields a prediction: disrupt the regions that build the schema (around the temporoparietal junction) and people's reports about their own awareness should go wrong in specific ways. So does manas as a structural account of continuous grasping — which is an epistemological description from a contemplative tradition, not a neuroscientific hypothesis, a line the source page draws explicitly.
If it yields no numbers and was never trying to be a proposition, treat it as rhetoric. That tier is not an insult. Vimalakīrti's silence was never meant to be a testable claim, and testing it as one misreads what it is doing. The failure mode is only filing it under "and therefore science has proven."
Which leaves the question that always gets asked: if there is no "I," who is living?
Both roads answer the same way, and both refuse the same inference. Buddhism does not deny the functional self; the aggregates go on operating in continuity. On the two-truths account, the conventional level has a nominal self and the ultimate level has no self-existent one, so a realised person still uses "I" to communicate and act, having merely stopped taking it as real. Karma continues; causation is not suspended. If anything this is a stronger account of responsibility: with no fixed occupant behind the action, the two most exhausting performances — self-justification ("I had no choice") and self-attack ("I'm just a terrible person") — lose their leading actor. Seth is a compatibilist (everything has causes, and voluntary action is nonetheless real): he dismantles the magical self that starts actions from outside physics, and keeps voluntary action — flexible, reason-responsive behaviour, plainly different from being shoved.
So the inference "there's no self, therefore no responsibility" is on offer from neither road. What got demoted was the spectator sitting in the head watching the show, not the person who still has to get up tomorrow morning.