Anil K. Seth, Tim Bayne
5 min
For decades, the scientific study of consciousness was dominated by the search for Neural Correlates of Consciousness (NCCs)—the minimal neural events sufficient for a conscious state. While this approach provided a theory-neutral framework for researchers, it has struggled to explain the underlying mechanisms of experience. Consequently, the field is moving toward formal Theories of Consciousness (ToCs) that aim to provide explanatory links between neural activity and subjective awareness. This shift represents a transition from merely mapping brain activity to building models that can predict and explain the phenomenon itself.
The authors categorize current ToCs into four main approaches, each with distinct commitments:
These theories often "talk past each other" because they define their explanatory targets differently—some focus on the functional properties of access, while others attempt to address the phenomenal character of experience directly.
Evaluating these theories is difficult because they are rarely confirmed or refuted by a single experiment. Instead, the field is increasingly turning to "adversarial collaborations," where proponents of competing theories agree on experimental designs to test specific predictions. However, three major hurdles remain: the need for greater mathematical and mechanistic precision in theory formulation, the requirement for theories to be more comprehensive (moving beyond vision and adult human subjects), and the persistent "measurement problem"—the difficulty of detecting consciousness in systems that cannot provide verbal reports.
Sam: [thoughtful] Which raises the obvious objection — doesn't that conflate consciousness with the ability to report it? The back-of-the-brain camp argues prefrontal engagement is a consequence of accessing information, not a requirement for the experience itself.
Alex: [nodding, clear] That's the core of their objection, yes. [[RP_SECTION:neural-signatures-and-timing|Neural Signatures and Timing]]
Sam: [processing] That plays out in the timing data too. Some studies see an early negativity around 200 milliseconds, favoring a local, posterior account. Others point to the P3b component around 300 milliseconds, which looks more like a global, effortful broadcast.
Alex: [slower, for clarity] Exactly — that's the battleground. Early emergence supports a local, posterior-heavy account of consciousness. Dependence on that later P3b spike supports a global, workspace-based one. The complication is that P3b is a robust marker of task-relevant processing — remove the task, and it often disappears even though the subject is still plainly conscious.
Sam: [direct] Which is why no-report paradigms matter — they strip away the executive demands to see what neural signature survives when the subject isn't busy reporting anything.
Alex: [deliberate] Right. Though even then, we're only looking at the aftermath of a process. We're inferring the state from its functional footprint, not observing the state directly — and we still lack a way to isolate phenomenal character from the cognitive machinery that just tracks it.
Sam: [reflective] Which brings you to a measurement problem, not just a theoretical one. [[RP_SECTION:measurement-and-circularity|Measurement and Circularity]]
Alex: [measured, precise] It does. We have no theory-neutral, gold-standard metric for consciousness. Even a marker like the Perturbational Complexity Index is theory-laden — it was built out of Integrated Information Theory's own assumptions.
Sam: [leaning in] So there's a circularity trap. If your metric is derived from one theory, of course it ends up validating that theory over its rivals. How do you break that loop without falling back on subjective report?
Alex: [slower] That's exactly why adversarial collaborations aren't just one fix among several — they're becoming the central methodological answer here. Agreeing in advance on what counts as a win or a loss for each side is the only way to stop the goalposts moving once the data are in.
Sam: [steady] Even so — winning one of these bets tells you which theory better predicts a signal. It doesn't obviously tell you which theory has correctly identified phenomenal experience itself, as opposed to its functional correlates.
Alex: [measured, reflective] That's the honest limitation, and it's the field's open question. The bet is that refining these models under adversarial pressure will eventually put pressure on that gap too — but nobody is claiming it's solved. [[RP_SECTION:future-implications-and-ethics|Future Implications and Ethics]]
Sam: [leaning in, probing] If it does pay off, the implications reach well beyond human neuroscience — an objective consciousness metric that doesn't depend on the subject's ability to report would apply to non-human animals, to organoids, potentially to artificial systems.
Alex: [measured, concluding] That's the stake behind the methodology. A theory-independent metric would reshape ethical and legal frameworks well beyond the lab. For now, the field is at the early stage of that shift — from descriptive neural mapping toward genuinely explanatory, falsifiable models of consciousness.
Sam: [steady] A necessary step, even if the hardest question stays open. Thanks for listening.