A major 2025 adversarial test of GNWT and IIT found results that challenged predictions from both theories rather than cleanly vindicating either.
So we need a better theory folks. What’s it gonna be?
I think we tend to talk about thoughts and memories as though they are things stored somewhere in the brain. A representation gets stored, then later retrieved and somehow presented to consciousness.
But I think that is the wrong metaphor.
A thought is not a thing. It is a performance.
At the physical level, what actually exists is neural activity unfolding through time across a massively recurrent network. One population activates another, activity feeds back through cortical and thalamocortical loops, previously learned connections constrain what happens next, and the whole system moves through a particular spatiotemporal trajectory.
Modern systems neuroscience increasingly studies cognition this way: not just by asking what individual neurons represent, but by looking at population dynamics and trajectories through neural state spaces or “manifolds.” The important unit can be the evolving pattern of activity rather than any particular neuron.
This is where I find the metaphor of music useful.
Imagine the neural system as an enormous instrument with billions of available elements. A particular pattern occurs, then another, then another. Timing and Order and Recurrence matters. What is already active changes what can happen next.
The thought is the music being played.
You can’t find Beethoven’s Fifth in one violin, one note, or even a static list of all the notes involved. The music exists in their organized unfolding through time. In something like the same way, perhaps the physical basis of a thought isn’t a particular neuron or even a static neural ensemble. It is the temporally extended activity that the system performs.
Memory makes this distinction especially clear. Current engram research does find neuronal ensembles associated with particular memories, and reactivating parts of those ensembles can contribute to recall. But memories aren’t simply immutable recordings. Engrams are distributed, their representations can change, and recall involves reactivation and reconstruction of neural activity.
So what gets stored may not be the thought or memory itself.
What persists is a physical change in the brain that makes a similar performance possible again.
Synapses change. Ensembles become associated. Particular dynamical trajectories become easier to enter. A partial cue can push the network toward a previously learned pattern.
Remembering, then, isn’t pulling the old thought out of storage.
It is playing it again.
Usually a little differently, because the instrument itself has changed since the last performance.
This also helps get rid of the homunculus problem. There doesn’t need to be another little observer somewhere in the brain listening to the neural music and figuring out what it means. The relationships between these neural events determine what they activate, inhibit, predict, recall and make possible next.
The coherence is in the causal dynamics themselves.
And I suspect timing is more fundamental to cognition than our usual language of “representations” suggests. A sequence A to B to C is not the same cognitive event as C to B to A, even if exactly the same neurons eventually participate. The trajectory matters.
That doesn’t solve consciousness. It still doesn’t explain why any neural performance should have a phenomenal character - why there is something it is like for the performance to occur.
But I think it gives us a better way of thinking about what a thought physically is.
Not an object in the brain.
An event performed by it.
Suppose you believe that a machine can genuinely have subjective experience - not merely behave intelligently, report internal states, or model itself, but that there can actually be something it is like to be that machine.
And suppose you also believe this is substrate-independent: silicon, neurons, relays, or anything else will do, provided the system implements the right computation.
Then it seems to me that you are making a remarkably strong claim about consciousness.
You are saying that the physically relevant details can be discarded until what remains is an abstract pattern of states and causal relations - essentially a mathematical/computational structure - and that instantiating that structure is sufficient for subjectivity.
I don’t mean that consciousness is literally an equation written on paper. Obviously some physical system has to instantiate the computation.
But if two radically different physical systems are conscious simply because they instantiate the same abstract computation, then it isn’t their particular physics doing the work. It is the abstract organization they share.
At that point I think the hard question becomes unavoidable:
Why should an abstract structure of information processing have a first-person perspective at all?
A mathematical description can tell us how states relate, transform, store information, generate reports, control behavior, maintain a self-model, etc. I don’t see how any of those descriptions by themselves get us to there being something it is like to instantiate them.
This doesn’t rule out machine consciousness. A machine could conceivably be conscious because of particular physical dynamics we haven’t reproduced or even understood yet.
But that is a different claim from saying that computation itself is sufficient.
If computation alone is sufficient for consciousness, then consciousness has been reduced, at bottom, to something very close to mathematical structure evolving over time.
I’m not sure most computational functionalists appreciate how strong a metaphysical commitment that actually is.
Suppose you believe that a machine can genuinely have subjective experience - not merely behave intelligently, report internal states, or model itself, but that there can actually be something it is like to be that machine.
And suppose you also believe this is substrate-independent: silicon, neurons, relays, or anything else will do, provided the system implements the right computation.
Then it seems to me that you are making a remarkably strong claim about consciousness.
You are saying that the physically relevant details can be discarded until what remains is an abstract pattern of states and causal relations - essentially a mathematical/computational structure - and that instantiating that structure is sufficient for subjectivity.
I don’t mean that consciousness is literally an equation written on paper. Obviously some physical system has to instantiate the computation.
But if two radically different physical systems are conscious simply because they instantiate the same abstract computation, then it isn’t their particular physics doing the work. It is the abstract organization they share.
At that point I think the hard question becomes unavoidable:
Why should an abstract structure of information processing have a first-person perspective at all?
A mathematical description can tell us how states relate, transform, store information, generate reports, control behavior, maintain a self-model, etc. I don’t see how any of those descriptions by themselves get us to there being something it is like to instantiate them.
This doesn’t rule out machine consciousness. A machine could conceivably be conscious because of particular physical dynamics we haven’t reproduced or even understood yet.
But that is a different claim from saying that computation itself is sufficient.
If computation alone is sufficient for consciousness, then consciousness has been reduced, at bottom, to something very close to mathematical structure evolving over time.
I’m not sure most computational functionalists appreciate how strong a metaphysical commitment that actually is.
One thing that seems to get lost in discussions of the hard problem is that people use the word why in two different ways.
Science answers many “why” questions. Why do apples fall? Because of gravity. Why do species evolve? Because of natural selection. Why does anesthesia make us unconscious? Because it disrupts the neural processes that sustain consciousness. These are really questions about mechanisms and causal relationships.
But there is another kind of “why” that science has never answered.
Why do the laws of physics have the form they do?
Why is there something rather than nothing?
Why do physical processes give rise to subjective experience at all?
These are not questions about mechanisms within an existing framework. They ask why the framework itself exists or why a particular relationship obtains in the first place.
This is why I think the hard problem is fundamentally different from ordinary scientific questions. Even if neuroscience eventually maps every neural correlate of consciousness and predicts every conscious state perfectly, one can still ask:
Why should those physical processes be accompanied by experience rather than occurring “in the dark”?
Whether that question is meaningful or answerable is open to debate. But dismissing it because science has solved many other “why” questions seems to miss the distinction.
Science has an extraordinary track record of explaining how one part of reality relates to another. It has never explained why reality itself has the basic character it does.
So perhaps the disagreement is about whether fundamental “why” questions are legitimate scientific questions at all.
The self is not the body-image, the personality, the biography, or the remembered story.
Those are contents of consciousness, of experience.
But I also don’t think we are some pure detached context floating behind experience.
Seeing and remembering are active processes. They include perception, memory, interpretation, bodily state, attention, emotion, and the current world pressing in from a particular angle.
So who we are is not the content itself, and not some abstract witness separate from content.
We are closer to the present act of experiencing: the living process through which a world and self arise from here. No separate witness.
A couple years ago I grew some Golden Teachers. I was mostly microdosing at the time, occasional macro doses, but I just stopped at one point. Sometimes it be like that.
The shrooms were air dried but cracker dry when stored, and I put desiccants and oxygen scrubbers in the jar. When I checked them after two years they were still cracker dry.
I figured the potency might be less, and I used to enjoy about 1.5 grams, so I decided to try 2 grams.
The trip came on hard while I was out walking, and I mostly just wanted to get home and ride it out. The visuals were quite strong, with tree branches reaching and curling for me as I walked by and the leaves themselves were kind of shimmering and morphing.
Overall a very intense and unpleasant trip. No nausea fortunately, but a good deal of negative emotion, starting with the trees being menacing.
Post trip clarity is very good, colors seem slightly enhanced and brain fog is low. Would do again, but not for a while.