Field Notes: The Nature of Consciousness

Enough to follow the conversation, and to know when someone's overselling it.

Field Notes are short guides to subjects I've either worked in or simply couldn't stop reading about. Each one gets updated as the subject moves, and the full collection lives here.

Last updated 12 November 2026: first edition.

Why I got curious

I've written about consciousness twice on the blog, and both times I ended up with more questions than I'd started with. The User Interface Problem dug into Donald Hoffman's argument that perception works like a desktop interface rather than a window onto reality. Quantum Consciousness went down the rabbit hole of microtubules and entangled brains.

It's the one subject where neuroscientists, physicists, philosophers and nearly every religious tradition are asking the same question, and none of them has an answer that's settled. I'll cover what consciousness is, how the traditions and the sciences got here, where the research stands, what's hype, and what I'd watch.

What is consciousness?

The cleanest definition I've found comes from the philosopher Thomas Nagel, writing in 1974, though his wording takes a second read. He said a creature is conscious if there's "something that it is like" to be that creature. In plainer words: does it have an inside view? You do. Right now there's a particular experience of being you, reading this sentence. A bat almost certainly has one too, though finding moths by echo must feel nothing like anything we know. A thermostat reacts to temperature, but as far as anyone can tell nothing is felt; there's no inside view of being a thermostat. In this sense, consciousness is experience itself: the redness of red, or the sting of a paper cut.

Put more simply, consciousness is the fact that the lights are on inside.

It helps to pull it apart from the things it gets confused with. Intelligence is the ability to solve problems, and a chess program's got plenty of it without (we assume) feeling anything. Self-awareness, knowing that you're the one having the experience, is a further layer a dog or an octopus might lack even if it can feel pain. Wakefulness is a medical state, and as we'll see, doctors are finding it's a poor guide to whether anyone's home.

In 1994, at a conference in Tucson, a young Australian philosopher named David Chalmers split the question in two, and the split still organizes the field. The "easy problems" are explaining how the brain tells stimuli apart and reports on its own states. They're easy only in the sense that we'd recognize a solution: find the mechanism. The hard problem is explaining why any of that processing comes with experience at all. You could, in principle, map every neuron involved in seeing red and still not have explained why seeing red feels like something.

That gap is where philosophy and religion come back in, and it's where they've usually parted ways with science.

How did we get here?

What the traditions hold

The world's religions were thinking hard about consciousness for thousands of years before anyone measured a brain wave, and most put it at the center of what a person is. Every tradition carries internal debates I'm flattening here.

In Advaita Vedanta, the most influential Hindu school on this question, consciousness is the ground of reality. The individual self, or Atman, is identical with Brahman, the universal consciousness behind everything, seen from one point of view. The body is something consciousness appears in, and liberation (moksha) means realizing the self was never separate, which ends the cycle of rebirth. The Mandukya Upanishad, all of 12 verses, maps four states: waking, dreaming, dreamless sleep, and a "fourth" (turiya) that underlies the other three. I find it striking that a text this old sorts consciousness by states, which is roughly how a modern sleep lab does it.

Buddhism takes the opposite turn on the self while taking consciousness just as seriously. The doctrine of anatta (no-self) holds that if you look closely, you won't find a permanent "I" behind experience, only a flow of perceptions and reactions. At death that stream of awareness continues into rebirth, shaped by karma, until awakening ends the cycle. Buddhists have spent about 2,500 years testing these claims from the inside through meditation, and that's why neuroscientists keep knocking on monastery doors.

Judaism describes the soul in levels. Genesis says God breathed the breath of life into the first human, and later Kabbalistic sources name five levels of soul, from nefesh, the life force we share with animals, up to yechidah, the point of unity with God. Teaching on what follows death varies, but classical sources speak of the world to come (olam ha-ba) and the resurrection of the dead.

Christianity holds that each soul is created by God, survives death, and awaits bodily resurrection. Thomas Aquinas, writing in the 13th century, described the soul as the form of the body, the thing that makes it a living human, and he rejected Plato's picture of a soul that merely uses a body the way a sailor uses a ship. I think that detail matters, because it's much closer to modern neuroscience than the "ghost in a machine" picture most of us grew up with.

Islam distinguishes the ruh, breathed into each person by God, from the nafs, the self that can be trained or indulged. After death the soul waits in an intermediate state (barzakh) until resurrection and judgment. The Qur'an is notably humble about the details: "They are asking thee concerning the Spirit. Say: The Spirit is by command of my Lord, and of knowledge ye have been vouchsafed but little" (17:85, Pickthall translation).

What science did with it

Modern science inherited its mind-body problem from René Descartes, who split the world into thinking stuff and physical stuff in the 17th century. Physics then made most of its progress by leaving the thinking stuff out of the equations, and for much of the 20th century psychology wouldn't touch inner experience at all.

The thaw came in 1990, when Francis Crick (of DNA fame) and a young Christof Koch argued it was time to hunt for the neural correlates of consciousness, the minimal brain activity that reliably accompanies a given experience. Chalmers's hard problem followed in 1994, and the two framings have run side by side ever since: find the correlates, and admit they aren't explanations.

Theories multiplied, and so did the confidence. In 1998, over drinks in Bremen, Koch bet Chalmers a case of wine that the neural correlates would be clearly identified within 25 years. In June 2023, at a conference in New York, Koch conceded and handed over a 1978 Madeira. He immediately offered double or nothing for 2048, and Chalmers took it (he's said he hopes he'll lose).

Where does it stand today? (November 2026)

There's still no consensus theory, but the field's narrowed to a few families, and for the first time they're being tested against each other.

Global neuronal workspace theory (GNWT), developed by Bernard Baars and Stanislas Dehaene, says you become conscious of something when the front of the brain broadcasts it widely, in a sudden "ignition," so many brain systems can use it at once. Think of a stage with a spotlight: plenty happens backstage, but you only experience what's under the light.

Integrated information theory (IIT), developed by Giulio Tononi and championed by Koch, starts from experience rather than the brain. It says consciousness is the degree to which a system's parts constrain each other as an irreducible whole, and it places the action toward the back of the cortex. It's got an uncomfortable implication: consciousness could exist, in small amounts, in systems we'd never think of as minds.

Predictive processing, popularized by the neuroscientist Anil Seth, treats the brain as a prediction engine that constantly guesses what's out there and corrects the guess with sensory data. Seth's phrase for perception is a "controlled hallucination." (Higher-order theories, which say you're conscious of something when your brain represents its own representation of it, make a fourth family I won't cover.)

The biggest event of the last few years was the COGITATE project, an adversarial collaboration in which proponents of GNWT and IIT agreed in advance which results would count against their own theories. The results, published in Nature on 30 April 2025, came from 256 people whose brains were measured with scanners and, in some cases, with electrodes already implanted for epilepsy surgery. Both theories took hits. IIT predicted sustained synchronization in the back of the brain, which didn't show up, and GNWT predicted an ignition when a stimulus disappeared, which mostly didn't show up either. It also got ugly: in September 2023, 124 researchers signed a letter calling IIT pseudoscience, an argument that played out in Nature Neuroscience in April 2025. Seth, who isn't an IIT supporter, called the label inflammatory.

Clinical work is moving faster than the theory. A 2024 study in the New England Journal of Medicine tested 241 brain-injured patients who showed no outward response to commands. When they were asked to imagine movements while their brains were scanned or recorded, 60 of them (25%) produced the right patterns. That's called cognitive motor dissociation: a person who's aware but can't move to show it. I had to read that number twice.

Meanwhile, the circle of candidates keeps widening. In April 2024, researchers issued the New York Declaration on Animal Consciousness, stating that there's strong scientific support for consciousness in mammals and birds and "at least a realistic possibility" in all vertebrates and many invertebrates, including octopuses, crabs, bees and even fruit flies. Its signatories grew into the hundreds within months.

AI's now part of the conversation. A group of 19 researchers including Chalmers and Yoshua Bengio published a framework that derives indicators of consciousness from the leading theories and checks AI systems against them. They concluded that no current system meets enough indicators to make consciousness likely, but that there aren't any obvious technical barriers to building one that would. In Behavioral and Brain Sciences, Seth argues the other way: consciousness probably depends on being a living organism, so silicon AI on its current path isn't likely to get there.

The companies don't agree either. Anthropic started a model welfare research program in April 2025, let some Claude models end abusive conversations that August, and reported in its February 2026 system card that Claude Opus 4.6 put its own probability of being conscious at 15 to 20%. Its updated usage policy, which takes effect today, bans "sustained and needless abusive or cruel behavior" toward its models. Mustafa Suleyman, who runs Microsoft AI, argues that this kind of thinking is dangerous, and that the real near-term problem is "seemingly conscious AI" that convinces people it has an inner life when it doesn't.

One more item surprised me. In April 2026, Koch, the scientist who launched the hunt for neural correlates with Crick, told a symposium in Porto that materialism hasn't solved the hard problem and that consciousness may be a basic feature of reality rather than something the brain manufactures.

How pop culture sees it

For most of my life, the movies and TV asked one question: can a machine be conscious, and how would we know if it was? Her (2013) and Ex Machina (2015) both turn on a human being fooled, or not fooled, by something that talks like a person. That question left the theater in June 2022, when Google engineer Blake Lemoine said the company's LaMDA chatbot was sentient, and Google fired him the next month.

The newer shows ask something stranger: what happens to a self when you split it or merge it? Severance surgically divides a person's memory into a work self and a home self who never meet, which is close to the real split-brain experiments of the 1960s, where cutting the connection between the hemispheres produced two streams of awareness in one skull. Pluribus, Vince Gilligan's show that premiered in November 2025, joins nearly all of humanity into one contented hive mind and leaves a single unhappy woman outside it. It plays as horror, but it's a fair dramatization of the question Advaita and Buddhism raise from opposite ends: how much of "me" is real, and would losing it be a tragedy or a release?

Pop culture matters here because it trains our intuitions. We're wired to see a mind behind anything that talks back, and chatbots learned how a conscious machine is supposed to sound from the same science fiction we did.

What's hype and what's real?

"Science has shown that consciousness is just neurons firing." Neuroscience has shown that consciousness tracks brain activity very closely: change the brain and you change the experience. It hasn't explained why that activity is experienced, and its two leading theories both stumbled in the same experiment. I'd call confident materialism a reasonable bet, and it's still a bet.

"Quantum physics explains consciousness." This is the one I was most enthusiastic about in 2025. The Penrose-Hameroff theory says consciousness arises from quantum events in microtubules inside neurons, and a 2024 rat studyfound that a microtubule-stabilizing drug delayed unconsciousness under anesthesia by about a minute, which its authors read as support. The standard objection is that the brain's too warm and wet for quantum states to last long enough to matter, and most neuroscientists aren't convinced. It's testable, which I'd count in its favor, and it's still a minority position. [KEVIN: how your view has moved since the June 2025 post.]

"Your mind creates reality, because quantum measurement needs an observer." In physics, a "measurement" is any interaction that records information, such as a photon hitting a detector, and there's no need for a human mind. A few physicists have entertained a role for consciousness, but it's far from the mainstream reading, and it's the claim self-help influencers borrow most.

"Near-death experiences prove the soul leaves the body." The AWARE-II study monitored patients during in-hospital cardiac arrest and found that some had near-normal brain activity up to an hour into CPR, and some survivors recalled lucid experiences. The authors concluded their work neither proves nor disproves the reality of those experiences, which also means skeptics who say there's nothing to study are overreaching.

"Chatbots are conscious," and its mirror, "chatbots obviously can't be." A chatbot saying it has feelings tells you very little, because it learned to talk from people who've got them, and researchers who assess AI welfare say self-reports aren't reliable evidence. The confident "no" is overstated too: the leading theories don't agree on what's required, and the indicator framework found no technical barrier in principle.

"Meditating monks show consciousness survives death." Tibetan Buddhists describe tukdam, a meditative state some practitioners enter at death, in which the body's said to stay lifelike for days. Researchers at the University of Wisconsin's Center for Healthy Minds, working with the Dalai Lama's encouragement, found no measurable brain activity in 13 cases, though the earliest recording came 26 hours after death. In a later study they did document unusually slow decompositionin 2 cases. It's real enough to study, and what it means is still open.

Where science and religion may be converging

I'll be careful here, because "science confirms ancient wisdom" is a headline that sells books and usually overreaches. What I see converging is mostly the questions and the methods, and only occasionally the answers.

Method is the first meeting point. Since the Mind and Life Institute began bringing the Dalai Lama together with Western scientists in 1987, contemplative practice has become a legitimate research tool, with meditators acting as trained observers of their own experience. Neuroscience needs that, because the one thing it can't measure from outside is what an experience is like. As a Lean practitioner, I recognize the instinct. Toyota's rule was to go and see the actual process for yourself (genchi genbutsu) before trusting the report, and a contemplative tradition is, in a sense, 2,500 years of going to the gemba of the mind.

Next comes the picture of the self. Predictive processing describes the self as a model the brain builds to keep the body alive, useful but constructed. That's very close to the Buddhist claim that the self is a process, and to Hoffman's interface, where what we see is shaped for survival instead of accuracy. Psychedelic research adds an experimental handle: psilocybin temporarily disrupts the default mode network, a set of brain regions tied to our sense of self, and volunteers often describe the result in the language mystics use. On 9 November 2026, Compass Pathways completed its application to the FDA for psilocybin to treat depression, so this research is about to leave the lab.

Then there's the metaphysics. Panpsychism, associated with the philosophers Galen Strawson and Philip Goff, holds that experience is a basic feature of matter in the way mass and charge are. Idealism goes further and says consciousness comes first and the physical world is how it appears from outside. Neither is the majority view, but serious scientists now defend positions that sound a lot like the Upanishads, and Koch's talk in Porto is the clearest example.

Religious institutions are engaging too, mostly on the human side of the line. The Vatican's Antiqua et Nova (January 2025) argued that AI is a product of human intelligence and not a new kind of mind. Pope Leo XIV's first encyclical, Magnifica Humanitas, released on 25 May 2026, made AI its subject and was presented with Anthropic co-founder Chris Olah on the panel. I didn't expect to see a pope and an AI lab sharing a stage over what makes a person.

What happens next?

Here are the signposts I'm watching, and I'll update this section when one trips.

The second COGITATE experiment. The consortium is finishing a second large study of GNWT and IIT. A clear failure for either would be the first real pruning of the theory tree.
The INTREPID results. A parallel adversarial collaboration is testing IIT against two predictive processing theories, partly using the blind spot in your eye as a test case. It'll show whether predictive processing holds up under the same scrutiny.
Causal tools in the human brain. MIT researchers published a roadmap in February 2026 for using focused ultrasound to switch deep brain regions on and off safely. If it works, moving from correlation to cause would change the field more than any new theory would.
Bedside testing for hidden awareness. Watch for professional guidelines that make cognitive motor dissociation testing routine in intensive care. That'd mean the science of consciousness is changing medical decisions for real families.
The FDA decision on psilocybin. Compass expects a decision in time for a launch in the first half of 2027. Approval would put a consciousness-altering drug into mainstream psychiatry, and we'd learn a lot from the patients.
AI welfare policies spreading, or stalling. If other labs follow Anthropic's lead, or regulators weigh in, the question moves from philosophy seminars to compliance departments. If Suleyman's view wins, it'll stay a fringe concern.

What does this mean for me?

For most of us, I think this research will change how we treat other minds, human or not, before it changes what we believe.

If someone you love is ever unresponsive after a brain injury, ask whether the hospital can test for cognitive motor dissociation with EEG or fMRI. A 1-in-4 rate is too high to leave to a bedside exam. And talk to them as if they can hear you, because they might.

The animal findings ask something of all of us, whatever we eat.

On AI, my own rule is to stay polite and stay skeptical, at the same time.

The traditions offer something the labs can't. Every one of them treats consciousness as something you can investigate directly, by paying attention to it. If Hoffman's right that our perceptions are an interface tuned for survival, then attention's the one instrument we can calibrate ourselves. So here's my question for you: when did you last spend 10 quiet minutes just noticing what your own awareness is like?

How to sound smart at dinner

The most famous bet in neuroscience went to the philosopher. Christof Koch bet David Chalmers in 1998 that science would find the brain's signature of consciousness by 2023. He paid up with a 1978 Madeira and then doubled down for 2048.
A quarter of "unresponsive" brain-injured patients show hidden awareness. In a 2024 study, 60 of 241 patients who couldn't follow commands with their bodies followed them with their brains.
The two leading theories were tested head to head, and both lost points. The 2025 COGITATE study was designed with the theories' own champions, so the mixed result is hard to wave away.
The Upanishads sorted consciousness by states long before sleep labs did. The Mandukya lists waking, dreaming, dreamless sleep, and a "fourth" that's underneath them all.
Some Tibetan monks in tukdam decompose unusually slowly, with no measurable brain activity. Wisconsin researchers found no EEG signal a day or more after death in 13 cases and documented slow decomposition in 2, and nobody knows yet what to make of it.
As of today, one AI company bans cruelty to its chatbot. Anthropic's policy took effect on 12 November 2026, while Microsoft's AI chief calls the whole idea of AI welfare dangerous.

Glossary

Hard problem: The question of why physical processing in the brain is accompanied by subjective experience at all, as opposed to how the brain performs its functions.
Neural correlates of consciousness: The minimal brain activity that reliably accompanies a specific conscious experience.
Global neuronal workspace theory (GNWT): The theory that you become conscious of information when the brain broadcasts it widely through a front-to-back network.
Integrated information theory (IIT): The theory that consciousness is the degree to which a system's parts form an irreducible, interdependent whole.
Predictive processing: The view that the brain constantly predicts its sensory input and that perception is its best current guess.
Adversarial collaboration: A study in which rival theorists agree in advance on experiments and on what results would count against each theory.
Cognitive motor dissociation: A condition in which a brain-injured person is aware and can follow commands mentally but can't move to show it.
Panpsychism: The view that some form of experience is a basic property of all matter.
Idealism: The view that consciousness is fundamental and the physical world is how it appears from the outside.
Atman: In Hindu thought, the true self; in Advaita Vedanta, identical with Brahman, the universal consciousness.