Scientists Argue the Universe Itself May Be Conscious — And Two Leading Theories Just Failed the Same Lab Test

Panpsychism is the claim that consciousness is not a late trick of meat. It is a feature of the world, present in some form at every scale, the way mass and charge are present. Durham philosopher Philip Goff has spent a decade making that sound less like mysticism and more like a gap in physics. « Physics is just mathematical structure, » he says, « and there has to be something that breathes fire into the equations. » If the equations describe relations and never the stuff that is related, then saying only brains are conscious is a habit, not a deduction.

Critics call it a category error. If electrons are « a little bit » conscious, they argue, you have explained nothing about the kind of consciousness that hurts, plans, and writes papers. The hard problem of consciousness — why anything feels like anything — does not go away because you spread it across the universe. It just gets bigger.

For a long time this fight lived in journals and podcasts. Then it got a laboratory.

COGITATE Put Two Theories In The Same Scanner

Integrated Information Theory, associated with Giulio Tononi and Christof Koch, says consciousness is the amount of irreducible information a system generates, with a predicted « hot zone » in posterior cortex. Global Neuronal Workspace Theory, associated with Stanislas Dehaene and colleagues, says consciousness is a broadcast: when information ignites in prefrontal and parietal networks, it becomes available to the rest of the brain.

The Cogitate Consortium designed an adversarial test. Theorists on both sides signed off on predictions before the data came in. The results, published in Nature in April 2025 (Volume 642, Issue 8066), did not crown a winner. Information about conscious content showed up in visual, ventrotemporal, and inferior frontal cortex. Sustained responses in occipital and lateral temporal cortex tracked stimulus duration. Content-specific synchronization appeared between frontal and early visual areas.

What did not appear is what each camp needed. IIT’s predicted sustained synchronization within the posterior hot zone was not there at the scale the theory required. GNWT’s predicted ignition at stimulus offset, and a robust prefrontal representation of certain conscious dimensions, did not show. Neither theory is dead. Neither theory walked out of the paper intact. Koch has discussed the result in interviews without pretending it was a win.

A follow-on consortium, INTREPID, is designed as another adversarial collaboration — this time testing contrasting predictions of IIT, Global Workspace, and predictive-processing accounts, including whether conscious contents change when certain neurons are inactivated rather than merely quiet. The Science of Consciousness conference, the longest-running interdisciplinary meeting on the subject, reconvenes in San Diego from October 11 to 16, 2026. The metaphysical bet is still open. The empirical one is now on a calendar.

That is the actual state of the science, as opposed to the headline that « scientists now believe the universe has a conscience. » Some scientists and philosophers take panpsychism seriously as the least-bad answer to the hard problem. Others think it is what you get when you refuse to accept that experience is a biological computation we do not yet understand. The Nature paper did not referee that metaphysical dispute. It did something rarer: it made two of the leading empirical theories pay a forfeit in public.

Whether the cosmos itself is aware is still a philosophical bet. Whether consciousness theories can survive an adversarial protocol is now a lab result. The fight is running in Nature, in INTREPID, and in the seminars Goff still fills. For more of The AEGIS Alliance on science that does not wait for official consensus, see our report on the expanding bubble universe and the rest of Science.