Inside this article
THUBAN / KNOWLEDGE LIBRARY
Consciousness research advances when an attractive explanation risks being wrong. An adversarial collaboration makes that risk explicit before the results arrive.
From a theory to a prediction
Global neuronal workspace theory and integrated information theory propose different accounts of consciousness. In broad terms, the former emphasizes the availability of information across a distributed neural workspace; the latter links consciousness to a system’s intrinsic causal organization. Neither short description supplies a complete theory, and the mathematical or neural details matter when deriving a test. A published collaboration protocol specifies predictions rather than relying on slogans. [1]
What an adversarial collaboration changes
Researchers associated with competing views work with a consortium to agree on tests and interpretations in advance. The point is not to manufacture consensus about consciousness. It is to reduce the freedom to reinterpret every imaginable observation as a success after seeing it. A protocol creates a public record of what would count as support or difficulty for particular predictions. [1]
The 2025 comparison
The COGITATE collaboration published a direct comparison in Nature in 2025, using multiple ways of measuring brain activity. The study examined predictions about where information associated with conscious visual content could be found and how activity and communication should unfold over time. Its conclusions challenged aspects of both theories. This was not an announcement that one scan had solved subjective experience or decisively eliminated every version of either account. [2]
Why disagreement continued
A later commentary from researchers associated with the global workspace framework disputed parts of the interpretation and emphasized findings they considered supportive. Readers should distinguish the measurements from the theoretical inferences drawn from them. This response is part of the scientific debate; it does not erase the original study, and the existence of a dispute does not make all interpretations equally strong. [3]
The scope of the evidence
An experiment on particular visual tasks tests an operationalized set of predictions. It does not automatically settle every question about dreams, self-awareness, animal experience or artificial systems. Moving from one domain to another requires additional arguments and evidence. Thuban’s inference is methodological: ask how the task relates to the phenomenon the theory claims to explain, and which auxiliary assumptions connect the two.
A practical way to read the next paper
Create three columns: prediction, observation and interpretation. Put the preregistered expectation in the first; the reported result in the second; and each side’s explanation in the third. If a headline declares a winner, ask which precise prediction won and whether the losing account has been revised. Uncertainty becomes informative when it identifies the next discriminating test, rather than ending the inquiry.