- Consciousness science can prove minds share experience's shape, never its substance.
- IIT and QSP, long rivals, co-authored the formal ceiling their formalisms both share.
- Adjunction, not isomorphism, is the highest inference structural theories of mind can reach.
How do I know that anyone else is actually conscious, and not merely behaving as if they are? This is what philosophers call 'the other minds problem'.
That was once a question about other people. It is now also a question about our machines. If your large language model insists across three coffees that it is having a bad day, there is nowhere obvious to stand while you decide whether to believe it.
This week a new paper appeared on arXiv that answers it, after a fashion, by drawing a boundary around it. "The Rosetta Stone and Levels of Principled Inference to the Experience of Another Mind" sets a ceiling on what any science of consciousness can establish about a mind that is not your own.
Mathematics can show that your experiences stand in the same relations as mine do, and that sharp pain sits at the same remove from dull pain in yours as in mine.
It basically means that our 'inner maps' are the same shape.
But mathematics cannot show that any point on either map feels like anything at all.
Philosophers have had various approaches to this problem for centuries. John Stuart Mill formalized the argument from analogy in the nineteenth century: I know I am conscious, you resemble me, therefore you probably are too. Bertrand Russell, always the more mathematically inclined, answered structurally instead, and almost every serious modern proposal has grown from that root.
David Chalmers gave the puzzle its modern branding in 1995 as the Hard Problem (see Chalmers explain it himself). Consciousness science has been trying to earn its keep against that framing ever since.
What a Rosetta Stone actually gives you
The original, historical Rosetta Stone (now at the British Museum in London) let scholars read hieroglyphs because it showed the same text in three different scripts. A Rosetta Stone for minds would fix the correspondence between different consciousness theories. The authors say that such a Rosetta Stone would leave you knowing the structure but never quite what goes on in another mind.
The four co-authors are an interesting cohort. Kallum Robinson, the lead author, is a Masters student at Berlin's School of Mind and Brain with a place in Tononi's lab at Wisconsin-Madison and a third at Monash, Tsuchiya's home institution. He is joined by Giulio Tononi, the architect of Integrated Information Theory, and by Matteo Grasso, one of Tononi's closest philosophical collaborators.
The fourth name is Naotsugu Tsuchiya himself, founder of the Qualia Structure Paradigm, the other framework the paper examines. Both approaches are represented by their originators. That makes the comparison unusually authoritative, and anything but neutral.
Kallum Robinson, lead author on the Rosetta Stone paper. Photo from his LinkedIn profile.
Two theories, one wall
Tononi's Integrated Information Theory begins from the phenomenal fact of experience and works outward. Consciousness, on this account, is identical to a maximum of integrated information, a quantity called Φ. It is present wherever a physical system's causal structure irreducibly binds its parts into a whole.
The theory has been controversial for as long as it has been famous. In September 2023 an open letter denounced IIT as pseudoscience.
The signatories included the cognitive scientist Bernard Baars, the NYU neuroscientist Joseph LeDoux, and the late Daniel Dennett. Their charge was that IIT's commitments, up to and including a form of panpsychism, put it beyond the reach of falsification.
Key figure
124
researchers signed the September 2023 open letter denouncing IIT as pseudoscience.
The empirical picture has not been kind either. The April 2025 COGITATE adversarial collaboration in Nature was a preregistered study of 256 participants. It gave IIT partial vindication on posterior cortical sufficiency, and none at all on its prediction of sustained gamma-band synchrony.
Tsuchiya's Qualia Structure Paradigm comes at the same wall from the opposite side. IIT postulates phenomenal axioms and then hunts for their physical realizer.
QSP starts instead with the relational geometry of experience. Ask people how similar this red is to that red, and their judgments have a structure that can be tabulated.
Tsuchiya's ambition is a Mendeleev-style Qualia Periodic Table, a JSPS-funded Transformative Area project running until 2028, cataloguing qualia by their relational positions.
In 2022, with the mathematicians Hayato Saigo and Steven Phillips, he proposed a formal relationship between experience and report. They called it an adjunction hypothesis. The term comes from category theory, which began life in the 1940s in algebraic topology. It has since migrated through theoretical physics into philosophy of mind.
As close as two minds can get
The Rosetta Stone paper takes that hypothesis and does something more ambitious with it. It arranges the possible strengths of inference about another mind on a formal hierarchy. Isomorphism sits at the top, strong and then weak adjunction below it, and a principled floor beneath those where inference runs out. A footnote concedes that the strong and weak grades are used in an informal, non-category-theoretical sense.
Its argument is that no structural theory, IIT or QSP or any successor, can reach higher than adjunction about any mind but its own. The authors do not offer this as a defeat. Their abstract insists that a system of principled constraints on what we may infer is "far more than what has been justifiable before."
Isomorphism vs adjunction: a quick guide
In category theory, the branch of mathematics the Rosetta Stone paper uses, an isomorphism is the strongest possible correspondence between two structures: a perfect, invertible, one-to-one map. If your experience of red is isomorphic to mine, they are, in every mathematical sense, the same shape.
An adjunction is weaker. It tolerates a loss of information going one way, but preserves enough structure that the two sides stay coherently related. On Tsuchiya's adjunction hypothesis, that coherence is what makes reliable communication possible. Think of it as a working translation rather than an identical copy.
Two structures are isomorphic only if the minds behind them are having identical experiences in the broadest sense, which, as the authors put it, would mean being the same self.
You would, quite simply, have to become the other person.
Whenever I compare my structure to another's, in each case I construct something between an imperfect dictionary and a grammar; a system of principled constraints on what may be inferred given that I cannot truly be acquainted with another mind without becoming it.
Robinson, Tononi, Tsuchiya & Grasso, arXiv:2608.12030
Consider the strawberry. You and I both call it red. A structural theory can establish that the relations our two experiences bear to other experiences, redder than orange, less red than crimson, match exactly.
Read our earlier coverage on qualia
What Is Qualia? Scientists Say They Can Now Measure Conscious Experience
What is qualia? A Japanese research team used fMRI to map the neural structure of color experience across 35 participants, suggesting your red may match…
→It can, on a very good day, prove that these two internal geometries are adjoint. It cannot, ever, prove that they are isomorphic, and even the adjunction will silently lose something in the crossing.
I doubt this settles the wider fight. Miguel Ángel Sebastián of the National Autonomous University of Mexico, who signed that 2023 letter himself, argues that phenomenal structuralism faces an "incommensurability problem." He made the case in the 2025 Philosophy and the Mind Sciences special issue on structuralism.
If two qualia spaces cannot be shown to be about the same thing, an adjunction between them buys less than the paper claims.
Even if original qualia are never accurately reported and inferred in different persons, as long as the deviation between the original and the reconstructed is 'coherently' related…accurate and useful communications can be established.
Naotsugu Tsuchiya, Hayato Saigo & Steven Phillips, Frontiers in Psychology
Anil Seth at Sussex presses a different line, that any purely mathematical bridge between substrates misses the biological point.
The same ceiling holds for machines
The authors specifically mention artificial intelligence. Ask a large language model how similar this red is to that red, and the structure recovered from its answers aligns extremely well with the structure recovered from humans. Kawakita and colleagues showed exactly that in 2024.
The Rosetta paper's verdict is that this proves nothing. There is, it says, sufficient reason to doubt that such models possess qualia, and good reason to think they are instead reproducing the statistical regularity of the data they were trained on.
The citation offered for that doubt is a 2025 paper on which Tononi is himself a co-author. It argues that a digital computer could be functionally equivalent to us, simulating even the neurons in our brain, without replicating the experience. On IIT, function and feeling come apart.
The ceiling is the same either way. Any claim about what a machine is experiencing, made from outside, is bounded exactly as our claims about each other are.
What passes across the bridge is structure. The acquaintance stays home.
That is what a science of other minds can, at best, deliver about a chatbot, a mouse, or your sister.
A ceiling with good company
More on consciousness
Two Major Theories of Consciousness Unable To Explain Awareness
Neither dominant explanation survived the most rigorous test in the field's history. That may be exactly what neuroscience needed.
→For a field that, again, spent 2023 shouting the word pseudoscience at itself, it is perhaps a small marvel that its two rival programs have quietly co-authored a chapter arguing that their formalisms share a ceiling.
If the paper survives peer review as a chapter in Springer's forthcoming Qualia Structure volume, consciousness science will have acquired the same sort of tool.
So, then: how can we truly know that anyone else is actually conscious? The paper's answer is that we never will. But it does offer something new. It names the place where the knowing stops, and calls it the acquaintance gap.
And while me may not finally know if someone - or something - else is truly conscious, some believe consciousness may be a force of nature.
Sources
- Primary source: Robinson, Kallum, Giulio Tononi, Naotsugu Tsuchiya & Matteo Grasso. "The Rosetta Stone and Levels of Principled Inference to the Experience of Another Mind." arXiv:2608.12030 (2026). Forthcoming chapter in Qualia Structure.
- Context sources:
- Tsuchiya, Naotsugu, Hayato Saigo & Steven Phillips. "An adjunction hypothesis between qualia and reports." Frontiers in Psychology (2022).
- Kawakita, Genji, Ariel Zeleznikow-Johnston, Naotsugu Tsuchiya & Masafumi Oizumi. "Gromov-Wasserstein unsupervised alignment reveals structural correspondences between the color similarity structures of humans and large language models." Scientific Reports 14 (2024).
- Findlay, Graham, William Marshall, Larissa Albantakis, Isaac David, William G. P. Mayner, Christof Koch & Giulio Tononi. "Dissociating Artificial Intelligence from Artificial Consciousness." arXiv:2412.04571 (2025).
- Sebastián, Miguel Ángel. "The incommensurability problem: A critique of phenomenal structuralism." Philosophy and the Mind Sciences 6 (2025).
- Cogitate Consortium. "Adversarial testing of global neuronal workspace and integrated information theories of consciousness." Nature (2025).
- Fleming, Stephen M. et al. "The Integrated Information Theory of Consciousness as Pseudoscience." PsyArXiv preprint (2023). The posted version notes it is not final and was being revised and expanded.
- Tsuchiya, Naotsugu et al. "Qualia Structure Paradigm." PsyArXiv preprint.
- Seth, Anil. "Why AI isn't going to become conscious." TED (2026).
Fact Check: Claim-by-Claim Verification Verified
Every claim below describes the article as you are reading it. Both direct quotations were verified verbatim against the primary sources, and the article's central claim, that structural theories of consciousness cannot exceed adjunction about another mind, is confirmed by the paper's own text. Five claims were corrected during the check, before publication, and are marked "corrected" with a note on what changed.
The paper lists Monash as his third affiliation and as his correspondence address, and Monash is Tsuchiya's institution. Note the limit of what is claimed: the article says Monash is Tsuchiya's home institution, not that Robinson works in Tsuchiya's lab, which the sources do not establish.
Caveat: it is a multi-author research programme; "founder" is a fair but compressed label.
Caveat: "almost every serious modern proposal has grown from that root" is editorial interpretation, not a citable fact.
Editor note: an earlier draft gave 150, a figure that conflates the letter's 151 affiliations with its number of signatories. The count is also dated to September 2023, because the letter's own front matter says it was still being revised and expanded, so the number can drift.
Caveat: "posterior cortical sufficiency" is the article's gloss; COGITATE tested decoding of conscious content, not sufficiency as such.
Editor note: an earlier draft said it began "unifying parts of pure algebra," which misplaces its origin.
Editor note: the article also carries the paper's own footnote 4, which concedes these grades are used "in an informal, non-category-theoretical sense." This was added during the check, because the article leans on the formal authority of category theory.
Editor note: the authors' own positive framing, that these constraints are "far more than what has been justifiable before," was added during the check, because the article had reported only the ceiling and not what remains reachable beneath it.
Editor note: the second author's first name was corrected to Ariel in the source list.
Editor note: an earlier draft attributed a feedforward-architecture argument to this paper. Findlay et al. argue from functional equivalence and simulation, and the word "feedforward" appears nowhere in the Rosetta Stone paper.
Commentary
- The primary source is a preprint. The article already hedges this correctly with "if the paper survives peer review," and the arXiv comments field confirms the Springer venue, but nothing here has been through peer review yet.
- The 2023 open letter is itself a non-final preprint. Its own front matter says "we are currently in the process of revising and expanding it... this is not the final version," so the signatory count is a snapshot, which is why the article dates it.
- Miguel Ángel Sebastián, quoted as a critic of phenomenal structuralism, signed the 2023 anti-IIT letter himself. The article now discloses this, since it makes him a participant in the dispute rather than a neutral observer.
- The article notes that both theories are represented by their originators and calls the comparison "anything but neutral." That is the right flag to raise, and it applies to the paper's conclusions about its own frameworks.
- The paper is more optimistic than a reader of the headline would expect. It argues the acquaintance gap "is not fixed and in fact leaves a great deal of principled inferences within reach."
- Kallum Robinson holds three affiliations, and Monash is both the third and his correspondence address on the paper. The article now names all three.
Sources used for verification
Academic/Peer-reviewed:
- Adversarial testing of global neuronal workspace and integrated information theories of consciousness - nature.com
- Gromov-Wasserstein unsupervised alignment reveals structural correspondences between the color similarity structures of humans and large language models - nature.com
- An adjunction hypothesis between qualia and reports - frontiersin.org
- The incommensurability problem: A critique of phenomenal structuralism - philosophymindscience.org
- Category Theory - plato.stanford.edu
- Other Minds - plato.stanford.edu
- Facing Up to the Problem of Consciousness - consc.net
Preprints:
- The Rosetta Stone and Levels of Principled Inference to the Experience of Another Mind - arxiv.org
- Dissociating Artificial Intelligence from Artificial Consciousness - arxiv.org
- The Integrated Information Theory of Consciousness as Pseudoscience - osf.io
Other reliable sources:
- Theory of consciousness branded pseudoscience by neuroscientists - newscientist.com
- Nobody knows how consciousness works - theconversation.com
- Supplement of Adversarial Collaboration - iit.wiki
- Qualia Structure project - qualia-structure.jp
- Kallum Robinson profile - wisc.edu
- Anil Seth profile - sussex.ac.uk
- Everything you ever wanted to know about the Rosetta Stone - britishmuseum.org
Fact-checked by Perplexity Sonar Pro on 2026-08-16
