Theories · about 5 minutes
The Bet
In 1998, at a consciousness conference, the neuroscientist Christof Koch bet the philosopher David Chalmers a case of wine that within twenty-five years we would have found it -- the clear, specific neural signature that separates an experience from no experience.
Twenty-five years is 2023. That date has been and gone, and we will tell you exactly what happened at the end of this paragraph -- except we won't, because first you are going to do what they did. You are going to say what you think, out loud, with a number on it and a date attached.
By 2048, the neural correlates of consciousness will be clearly identified -- a specific, agreed brain signature for “the lights are on.”
The live wager · currently has real wine on it
Pick a side. There is no “it depends” button, which is the entire exhibit.
What happened to the first one
On 23 June 2023, at the same conference series where the bet was struck, Koch conceded. He handed Chalmers a bottle of 1978 Madeira -- and, the day before, a case of fine Portuguese wine, because he is apparently a person who loses gracefully and in bulk. The neural correlates had not been nailed down. Chalmers' summary of the twenty-five years was: “It's clear that things are not clear.”
Koch, having lost, immediately bet again. Same proposition, another twenty-five years -- due in 2048, when he will be 91 and Chalmers 82. Chalmers took the other side again, and said the finest sentence in this entire museum:
“I hope I lose, but I suspect I'll win.”
So the proposition you just took a position on is not a hypothetical. It is live, it is dated, and two of the most serious people in the field currently have wine riding on it.
Four more, while you have the pen out
These are the museum's other open questions, in the only form that can ever be scored. Same rules: a side, and a number.
The slider stops at 99%. That is deliberate, and it is the oldest piece of etiquette in this business: leave yourself room to have been wrong. A person who says 100% has stopped making a prediction and started making an announcement.
Your receipt
Here is what you said, and when you said it. Nobody but you will ever see this -- it is stored in your own browser, like the guest book down the hall, and it goes no further.
Put it somewhere your future self will trip over it. That is the entire technology: a claim, a date, and a person who will still be around to read it. Most of what you just bet on resolves inside a normal human lifetime. Not all of it.
What just happened
That discomfort you felt reaching for the slider is the exhibit. It has a name -- calibration -- and it is the difference between knowing things and merely having opinions near them.
An opinion costs nothing. A dated, numbered prediction costs something: it can come back. And the strange, wonderful consequence is that a person who makes hundreds of them can be scored -- not on whether they were right, but on whether their confidence matched their hit rate. Someone who says 70% and is right about 70% of the time is well calibrated, and is more use to you than someone who is right more often but always says 100%.
You have been walking through a building that does this on every wall. The Humility-o-Meter is not a running joke about how mysterious everything is -- or rather, it is, but it is also the museum's own betting book, posted in public where you can hold us to it:
| The museum bet | In | Reading |
|---|---|---|
| Your blind spot is filled in by your brain | № 1 | Settled science |
| Prediction shapes what you hear | № 31 | Educated hunch |
| Integrated information is what consciousness is | № 29 | Vigorous hand-waving |
| Anyone knows why experience feels like anything | № 24 | Nobody knows |
If in 2048 the needle in Exhibit № 29 looks absurdly timid, we were badly calibrated and you may say so. That is what the gauge is for. A museum that only ever said “nobody knows” would be as dishonest as one that only ever said “settled” -- it would just be dishonest in the fashionable direction.
The other thing that happened that week
The bet was settled at a conference that had bigger news. The same meeting unveiled the first results of an adversarial collaboration: six theory-neutral laboratories, 250 participants, fMRI and MEG and electrodes placed inside actual skulls, running an experiment that the rival theorists had agreed on in advance.
Understand what that means, because it is rarer than it sounds. The proponents of integrated information theory (Exhibit № 29) and global neuronal workspace theory (Exhibit № 28) sat in a room and signed up, beforehand, to results that would count against their own life's work. Then they ran it.
Both theories took damage. IIT predicted sustained, synchronised activity in the sensory areas; the activity was sustained but obstinately not synchronised. Global workspace predicted a detectable “off” signal when a stimulus disappears; nobody could find it. The official summary -- “these results confirm some predictions of IIT and GNWT, while substantially challenging both theories” -- is the sound of a field behaving well.
Nobody won. Everybody learned. This is what the good version looks like, and it looks disappointing, and that is how you can tell it is real.
Why the wine matters
It would be easy to file all this under charming academic eccentricity: two clever men, some Madeira, a joke that ran for a quarter century. It is not that. It is a small, powerful piece of technology for staying honest, and it has three moving parts -- a claim specific enough to fail, a date, and a witness.
Koch lost in public. He did not redefine the terms, or explain that in a sense the correlates had been found, or point out that his position had always been more nuanced. He handed over the wine, said the field had further to go than he thought, and bet again -- which is the second-best thing you can do with a lost bet, and the first is to have made it.
The person who makes no predictions never loses one. This is generally mistaken, by them, for wisdom.
You now have a receipt. In 2048 -- should you be around, and we hope you are -- you might go and find it. Christof Koch will be 91, and either drinking or pouring. And there is a decent chance that the sentence at the bottom of Exhibit № 24 will still be standing exactly where it is, undefeated, waiting for whoever comes next.
Take it further
Nature covered the settlement under the excellent headline Decades-long bet on consciousness ends -- and it's philosopher 1, neuroscientist 0; Quanta's What a Contest of Consciousness Theories Really Proved is the best account of the adversarial collaboration and is properly sceptical about the scoreboarding. Koch's The Feeling of Life Itself (2019) and Chalmers' The Conscious Mind (1996) are the two positions at full strength, by the two people who bet on them.
On calibration as a practice, the keyword is forecasting: Philip Tetlock's Superforecasting is the readable version, and the finding that matters is that calibration is a skill -- it improves with scorekeeping, which is why the receipt has a date on it.
One honest note about your bets: we cannot score them. There is no server, no ledger, nobody watching -- the same design as Exhibit № 42, and for the same reason. Which means the only person who can ever hold you to those numbers is you, later. That is a weaker mechanism than a case of wine and a philosopher with a good memory. It is also, it turns out, enough.