Sixteenth Pour: The Price of Forgetting

Two weeks, I said. It’s been almost four.

Here’s what happened. In July I lit a second still down the road. Sundays, for dentists, about the business of the chair instead of the biology under it. A weekly still eats grain. Every Sunday the mash went there, and every Sunday this room got a little colder, which is the exact thing I stood here and swore wouldn’t happen again. So the pour is late, and I’m sorry, and I’m going to stop promising dates, because the last two times I did that I lied without meaning to, which is the worst kind. The next one is already in the cask. It went in before this one left the building.

Tonight is about a bill. Not the one I owe you. The one every living thing pays, every day, for letting go of what it knew. The universe keeps a ledger. Reading it is free. Writing in it is nearly free. There is exactly one entry with a price on it, and the entry is forgetting.

Pull up a stool. We start in a patent office.

The man who hated dice wrote the rulebook for the jitter

Everyone remembers three of Einstein’s 1905 papers. The photon. Relativity. The equation on the T-shirt. The fourth is the one that proved atoms exist, and it’s the one this bar cares about, because it’s about noise.

In 1827 Robert Brown put pollen grains in water, looked through a microscope, and saw them jiggle. Not drift. Jiggle. Every direction, endlessly, as if something were kicking them. Dead pollen jiggled. Ground-up rock jiggled. It wasn’t life. For eighty years nobody could say what it was.

Einstein said: the grain is being hit. Water molecules, too small to see, slam into it trillions of times a second from every side, and the hits don’t cancel. The grain takes a random walk. Then he did the thing that makes it physics instead of a story. He predicted exactly how far the grain wanders in a given time and tied that distance to temperature, to the viscosity of the water, to the size of the grain, and to a number that counts molecules. Jean Perrin measured it in 1908. The molecules came out right. Atoms stopped being a hypothesis.

Now read the equation from this side of the bar. The jitter isn’t dirt on the lens. The jitter is the thermal bath talking, and the volume knob is temperature. Every molecule in every one of your cells is on the receiving end of that conversation, all the time. A protein in cytoplasm is not sitting in a quiet room waiting for instructions. It is being beaten by water at a rate no engineer would tolerate, and it has been for two billion years.

So the cell had a choice. Fight the jitter or use it.

It uses it. Kinesin is a motor protein, two feet on a stalk, that hauls cargo along microtubules inside your neurons, cell body to synapse, sometimes a meter. In 1993 Steven Block’s lab caught one in an optical trap and watched it walk. Eight nanometers a step. One ATP per step. At that scale inertia doesn’t exist; Purcell’s old joke is that a bacterium that stops swimming coasts about the width of a hydrogen atom. Kinesin can’t push off. It can’t swim. All it can do is wait for the water to kick its free foot forward, then spend an ATP to make sure that foot stays put and the other one lets go.

Feynman drew this machine in his lectures. A ratchet and a pawl, sitting in one warm bath, and the proof that it can’t extract work from the jitter alone: it rattles backward as often as forward. What the cell adds is a pawl that costs money. ATP is the price of making the backward step unlikely.

Einstein sized the jitter. The cell rectifies it. One ATP per step, more steps a second than there are stars in the galaxy, and that’s your brain hauling its own freight.

That ATP is the coin of the night. Before we close, you’ll see what else it buys.

And the photon

The March paper said light arrives in packets. Einstein got the Nobel for it, sixteen years later, and the citation didn’t mention relativity. It said the photoelectric effect. The prize was for the reason you can see this glass.

I mean that literally. In 1942 Hecht, Shlaer, and Pirenne sat people in the dark for half an hour and flashed lights at them so faint that about a hundred photons reached the cornea and five or ten got absorbed by the rods. The observers saw the flash. Better: the way their answers wobbled from trial to trial matched the counting statistics of the photons themselves. The variability wasn’t in the observer. It was in the light. In 1979 Baylor, Lamb, and Yau put an electrode on one rod from a toad’s retina and recorded its response to one photon.

Your eye is a photon counter. Einstein’s packet is the thing it counts.

Then, in 1917, he did something to the photon that he spent the rest of his life regretting. He wanted Planck’s radiation law to fall out of the way molecules absorb and emit light, and to make it fall out he needed three processes: absorption, spontaneous emission, and one nobody had proposed, emission stimulated by light already present. That third coefficient is the laser, forty years early. It’s every fluorescence microscope and every pulse oximeter in the building. The derivation worked. And at the end of the paper he wrote this, in van der Waerden’s translation:

“The weakness of the theory lies, on the one hand, in the fact that it does not bring us any closer to a merger with the undulatory theory, and, on the other hand, in the fact that it leaves the time and direction of elementary processes to ‘chance’; in spite of this I harbor full confidence in the trustworthiness of the path entered upon.”

Chance. In his own quotation marks. Nine years before he wrote Max Born that the Old One does not play dice, he had written the dice into the rulebook himself, because the rulebook wouldn’t close without them.

So by 1917 Einstein had handed biology the two things it uses every second. The photon, which is how information gets in. The jitter, which is the noise it gets in through. He never liked what either one implied. Somebody else’s students would have to work out the price.

Mercer Street

John Archibald Wheeler arrived at Princeton in 1938. In 1939 he and Niels Bohr wrote the paper that explains how a uranium nucleus splits. That’s Bohr’s one mention tonight, and it’s here for a reason: Wheeler is the only man in this story who worked shoulder to shoulder with both ends of the quantum argument. Bohr in Copenhagen. Einstein three blocks from his office.

After the war Wheeler did something nobody sensible was doing. He took up general relativity, which by 1950 was an elegant backwater that had stopped producing physics, and he taught it. On May 16, 1953, he walked his class, eight or ten students, over to 112 Mercer Street, and they had tea at Einstein’s dining table. Margot Einstein served. Helen Dukas served. Wheeler wrote it up in his memoir and three of the students wrote their own versions. Picture the room. The man who spent thirty years losing the argument about dice, pouring for the young men who would write the rules of the game he refused to play.

Because that is what Wheeler’s students did. Feynman, his first doctoral student, rebuilt quantum mechanics as a sum over every path a particle could take. Everett, in 1957, wrote the thesis that says the wavefunction never collapses and every outcome is real somewhere. Thorne spent fifty years chasing gravitational waves and caught one. Zurek, whom the regulars met last pour, showed how the classical world condenses out of the quantum one by leaving redundant records in the environment. And Bekenstein got the question that turns out to be this pour’s hinge.

Wheeler tells it as a teacup. Around 1970 he was thinking about a cup of hot tea stirred into a cup of cold. The mixing raises the entropy of the universe a little, and the second law says that rise can never be undone. Now drop the cup into a black hole. The entropy is gone, behind a horizon, unobservable. Second law broken? Bekenstein, a graduate student, said no. The black hole must carry entropy itself, and the amount is proportional to the area of its horizon.

I read that sentence three times the first night and still wasn’t sure I believed it. Information is not an abstraction that can vanish. Information is physical. It takes up room. Wheeler chewed on that for twenty years and spat out three words: it from bit. We’ll get there. Hold it.

Here’s why the lineage matters and the names don’t. One thread runs from the man who wrote the dice in, through the man who took his students to tea, to the student who proved information has a size. If information is physical, then getting rid of it is a physical act. Physical acts have a price.

The demon’s bill

This is where I teach. Four rounds, one idea. I’ll define everything once and move on, and the physicists at the end of the bar can check the arithmetic.

Round one. Maxwell, 1867. James Clerk Maxwell, in a letter to his friend Tait, imagined a very small, very observant being sitting at a trapdoor between two chambers of gas. Fast molecule coming from the left: open the door. Slow molecule coming from the right: open the door. Otherwise keep it shut. No work is done. The door is frictionless. And yet the left chamber grows cold and the right grows hot, and you have made a temperature difference from nothing, which the second law of thermodynamics says cannot be done. Kelvin called the being a demon. The demon’s only tool is information. It knows which molecules are fast.

Round two. Szilard, 1929. Leo Szilard stripped the demon to one molecule in a box. Measure which half the molecule is in. Slide a partition into the middle. Let the molecule bounce off the partition and push it to the empty end like a piston. Collect the work. Reset. Repeat. The work per cycle is exactly kT ln 2: k is Boltzmann’s constant, T is temperature, ln 2 is there because the question had two answers. That’s the birth of the bit as a physical quantity, nineteen years before Shannon named it. Szilard saw that to save the second law the demon has to pay kT ln 2 somewhere. He guessed the payment happens when the demon looks.

Round three. Landauer, 1961. Rolf Landauer, at IBM, asked the practical question. What does a computer have to pay to compute? Almost nothing, he found, in principle. The expensive step is the one you can’t run backward. Take a bit that might be 0 or 1 and reset it to 0. Two states squeezed into one. That entropy has to go somewhere, and the only place is the environment, as heat: at least kT ln 2 per bit erased. Landauer’s principle in a sentence: erasure is the only operation with an unavoidable price. His own three words were shorter. Information is physical.

Round four. Bennett, 1982. Charles Bennett, also at IBM, took Landauer back to the demon and killed it. The measurement, it turns out, can be done reversibly. Looking is free. What the demon cannot avoid is this: after each cycle its memory holds the answer it just used, and before the next measurement it has to clear that memory. It has to forget. The forgetting costs kT ln 2, exactly what it extracted, and the second law walks away untouched. Bennett had already shown, in 1973, that every other step of a computation can in principle be done with no heat at all.

Every step but one. The entire thermodynamic cost of thinking sits in a single act, and the act is erasure.

Here’s the arithmetic, once.

E = kT ln 2.

At body temperature, 310 kelvin, that’s about 3 × 10⁻²¹ joules. Three thousandths of a billionth of a billionth. One molecule of ATP, split in a living cell, releases about 8 × 10⁻²⁰ joules. Divide. One ATP could, at the floor, pay to erase about 28 bits. That is the exchange rate. That is what the universe charges a cell to stop knowing something.

Now the footnote, because I said I’d say it clean and then be honest about the gap. Nothing real runs at the floor. A transistor in your pocket dissipates thousands to a million times kT ln 2 per switch. Cells are no thriftier: a ribosome spends an entire nucleotide triphosphate to throw away one wrong amino acid, a bit’s worth of information at a hundred bits’ worth of price. Landauer’s number is not the bill anyone pays. It is the reason there is a bill. There is no free erasure. Not in a chip, not in a cell, not anywhere.

Memory is cheap. Forgetting is what the universe charges for.

There. I said it.

There’s a word for this. Like stigmergy it’s older than it looks and almost nobody uses it. Obliviscence. Forgetting as a process, not a lapse. The Latin underneath is oblivisci, to forget, and the best guess at what’s underneath that is ob plus levis: to wipe smooth. The Romans wrote on wax. To forget was to run the flat of the stylus across the tablet and smooth the letters out. That is Landauer’s operation, exactly. Reset to zero. Squeeze two states into one. Warm the wax.

Philip Ballard put the word on the cover of a monograph in 1913, Obliviscence and Reminiscence, and buried a surprise inside it. He had London schoolchildren memorize poetry, tested them, then tested them again days later with no rehearsal in between. Their scores went up. Something in the interval kept working on the material after the learning stopped. Ballard called it reminiscence and had no idea what caused it. We have a decent idea now. The interval had nights in it.

Feynman, 1982

May, 1981. Landauer co-organizes a conference at MIT called the Physics of Computation. Bennett gives the talk that becomes the review I just poured. And the keynote is Feynman, Wheeler’s first student, standing up to say that if you want to simulate nature you’d better build the simulator out of quantum mechanics, because nature isn’t classical, dammit. That talk, printed the next year in the same volume as Bennett’s, is the founding document of quantum computing.

Landauer’s conference. Bennett’s demon. Feynman’s machine. One room.

The descendants are alive and working. John Preskill holds the chair at Caltech that carries Feynman’s name; he coined “quantum supremacy” and gave the field its sober label for the noisy machines we actually have. Alán Aspuru-Guzik showed in 2005 that a quantum computer could compute the energies of molecules, which is what Feynman was pointing at all along, and four years later his group published the paper the regulars remember from last pour, the one where the right amount of noise helps energy find its way through a leaf. And Bennett, still at IBM, went on to invent quantum cryptography and quantum teleportation, having first established that the demon pays when it forgets.

One more name from that hallway, then the operating room. In 1974 John Hopfield asked how a cell copies genes and builds proteins with an accuracy the chemistry shouldn’t allow. His answer, kinetic proofreading, is that the cell pays. At each step it spends a triphosphate to buy itself a second chance to reject a wrong match, discarding the error and the energy together. Accuracy is bought with erasure. The ribosome is a Maxwell’s demon that settles its tab. Eight years later Hopfield gave neural networks a memory shaped like an energy landscape, and in 2024 Stockholm gave him the physics prize. In the early eighties he co-taught a course at Caltech with Feynman and Carver Mead. They called it the physics of computation. The cell that pays and the machine that pays were on the same syllabus.

The room where I do it on purpose

Tuesday, one o’clock, first patient of the afternoon. Four wisdom teeth. Midazolam, fentanyl, propofol, titrated until they’re comfortable, and then the local, which I introduce as a little cleaning. The teeth come out one by one. Irrigate, suture, gauze, bite block out. Propofol off. Within a minute the eyes open and I tell them what I always tell them: you’re just waking up, there’s gauze in your mouth. And the ones who remember the plan reach for their phones and hit record. Instagram gold. They film it because some part of them already knows the only copy of this afternoon that will ever exist is the one on the phone.

I induce forgetting for a living. The propofol is the part people picture: it switches the patient off. The midazolam is the part this pour is about. It doesn’t switch anything off. It switches off the writing. It sits on the same receptor everywhere, but the effect that matters is in the hippocampus, where it stops the experience from being encoded, and it keeps doing that after the propofol stops, which is why the wake-up they film is one they will never have. Anterograde amnesia, the textbooks call it, and every surgeon who uses it has had the conversation in recovery where the patient asks whether they said anything embarrassing. The answer is usually yes.

Now notice what the drug is cheap at. Not recording costs nothing. The memory that never formed is free to not have. There’s no bill for a bit that was never written. That’s why the phone comes out. Writing is cheap, so they hand it to something that will do it for them.

What the drug cannot do, what nothing can do cheaply, is take a memory already written and remove it. Ask anyone who has tried to unlearn a fear. What the clinic calls extinction isn’t erasure. The old memory stays. A new one gets laid over it that says “not now, not here,” and the old one comes back the moment the context shifts.

Landauer’s line runs straight through the operating room. Not writing is free. Erasing is the bill. The whole pharmacology of amnesia lives on the free side of that line, and the whole tragedy of memory lives on the other.

Sleep is the bill

If forgetting has a price, the brain should be paying it somewhere, on a schedule. It is. Every night. Ballard’s schoolchildren, the ones whose poems got better over two days with no practice: this is where the two nights went.

Giulio Tononi and Chiara Cirelli have spent twenty years building the case that sleep is the brain’s erasure cycle. Awake, you learn, and learning means synapses strengthen. A day of experience leaves the cortex heavier, noisier, costlier to run, closer to saturation. Sleep is when it gets pared back. Synapses are downscaled across the board, the weak ones most, so what was learned survives as contrast instead of bulk. In 2017 the de Vivo group counted thousands of synapses in mouse cortex under an electron microscope, before and after sleep. After sleep they were smaller by nearly a fifth. The title Tononi and Cirelli put on the 2014 review that laid out the theory is one I could have stolen for tonight. Sleep and the price of plasticity.

Francis Crick saw the shape of it in 1983. He and Graeme Mitchison proposed that dream sleep is the brain running its own network backward to weaken spurious associations. His line: we dream in order to forget.

Let me be careful about the claim. The energy the brain spends asleep is nowhere near Landauer’s floor. Nothing is. The claim is structural. Writing was cheap and ran all day for free. The reset is the thing with a bill.

You spend a third of your life offline. That’s not rest. That’s the invoice.

Cells that refuse to erase

Last October I poured the protector paradox. The refuges a cell builds to shelter fragile quantum processes from thermal noise also shelter a cell that has gone wrong from the surveillance that should remove it. Cancer as the price of being quantum. Tonight, same fact, other side of the ledger.

A healthy cell keeps its identity by re-earning it. The marks that tell a liver cell it is a liver cell get copied at every division, maintained against drift, refreshed. And when the cell is damaged past repair, the organism asks it to pay the full erasure bill at once: apoptosis, the cell dismantling its own record and handing the parts back. Tens of billions of cells settle their tab this way in you every day, and the body runs.

A cancer cell is a cell that refuses to pay. It locks its state. Tumor suppressors get silenced by methylation that won’t reset. The apoptotic machinery gets disabled. The identity that should be re-earned at every division becomes a fixed thing that only copies.

Whatever else a tumor is, it’s a population that found a way to stop forgetting. The protector paradox and the price of forgetting are the same fact. The refuge that keeps the coherence in keeps the erasure out. And the body’s oldest defense against its own cells is a demand that they forget on command.

The desk

Last pour I told you about the agents. Multiple AI sessions, each waking with no memory of the last, coordinating through a folder of append-only notes. The trail on the wall. The termite’s move, rediscovered at my desk. Three weeks later I can tell you what I left out.

I gave them a medium that remembers. I did not give them one that forgets. The notes are append-only by design, and the ant colony’s lesson is that a trail that never evaporates locks the first mistake in as dogma. So far the forgetting happens somewhere else, and it’s worth looking at where. An agent reads the trail, works, fills its context, and when the context is full the session compacts. It summarizes itself, throws away the detail, continues from the summary. That is erasure. It gets paid like everything else, thousands to a million times above Landauer, in a building somewhere with a very large cooling bill. Some of the heat coming off the datacenter is the price of the agents forgetting what they said an hour ago.

And the trail itself only evaporates from the agent’s eyes, not from the wood. Each session reads the last few days of notes and nothing older, so old marks fall out of view while sitting on disk forever. It’s a reading window pretending to be decay. The colony wouldn’t accept that. Sometime in the next few weeks I have to decide what it costs to let a note actually go, and Landauer says the answer won’t be nothing.

Nightcap: it from bit, all the way down

House rule. Everything above this line stands on its own. Everything below is conjecture, poured neat, and nothing upstairs depends on it.

In 1978 Wheeler proposed an experiment. Send a photon through two slits, and decide only after it has passed them whether to look for which slit it used or let it interfere with itself. Whichever you choose, late, the photon obliges. Look for the path and you find one. Don’t, and you get the interference of a photon that took both. In 2007 Jacques and colleagues ran it clean, single photons, a switch that flipped after the photon was already in flight, and the answer was Wheeler’s. His way of saying it: no elementary phenomenon is a phenomenon until it is a registered phenomenon. The record isn’t a copy of the event. The record is where the event becomes definite.

Now the three words. In 1989, near the end of his life, Wheeler proposed that every “it,” every particle, every field, every patch of spacetime, derives its existence from “bits,” yes-or-no answers registered by some apparatus somewhere. It from bit. Not that information describes the world. That the world is made of it, all the way down, and the registering is what makes it real.

Set that beside Landauer and watch what happens. If the world is a ledger of registered answers, Bekenstein’s black hole isn’t a curiosity; it’s the page limit. Zurek’s quantum Darwinism from last pour is how the ledger gets written redundantly enough that you and I agree on it. And Landauer’s principle is the cost of editing. To erase a bit is not to lose information. It’s to hand it to the environment as heat, where it is still, in principle, written.

Nothing is ever forgotten. It’s handed to the wall. The wall keeps the receipt.

Whether any of that reaches up into a cell, I don’t know, and neither does anyone. What I know is this. Einstein wrote the dice in and hated it. His neighbor’s students found the dice had a price. And every living thing pays that price on a schedule it didn’t choose. Admire it. Pour it. Don’t lean on it.

The distillate

Three drops.

Writing is cheap. Obliviscence, the smoothing of the wax, is what costs. The floor is kT ln 2 per bit, and nothing in the universe, from a transistor to a ribosome, runs anywhere near it. The floor is not the bill. The floor is why there is a bill.

Every living system pays on a schedule. One ATP per step for the motor. One triphosphate per rejected error for the ribosome. One night in three for the brain. One apoptosis per cell that has outlived its record. The thing that refuses to pay is the thing that kills you.

And forgetting is not loss. Landauer’s heat is information leaving your ledger for the environment’s. The pheromone that evaporates is still in the air. The trail I promised you three weeks ago was still on the wood. That’s why I’m here.

The still is lit. The next one is in the cask.

Tasting notes (for those who drink neat)

Einstein, A. (1905). Über die von der molekularkinetischen Theorie der Wärme geforderte Bewegung von in ruhenden Flüssigkeiten suspendierten Teilchen. Annalen der Physik 17, 549–560. · Einstein, A. (1905). Über einen die Erzeugung und Verwandlung des Lichtes betreffenden heuristischen Gesichtspunkt. Annalen der Physik 17, 132–148. · Einstein, A. (1917). Zur Quantentheorie der Strahlung. Physikalische Zeitschrift 18, 121–128; English translation in van der Waerden, B. L. (ed.) (1967). Sources of Quantum Mechanics. North-Holland. · Einstein to Born, 4 December 1926, in The Born–Einstein Letters (1971). · Perrin, J. (1909). Mouvement brownien et réalité moléculaire. Annales de Chimie et de Physique 18, 5–114. · Svoboda, K., Schmidt, C. F., Schnapp, B. J. & Block, S. M. (1993). Direct observation of kinesin stepping by optical trapping interferometry. Nature 365, 721–727. · Schnitzer, M. J. & Block, S. M. (1997). Kinesin hydrolyses one ATP per 8-nm step. Nature 388, 386–390. · Purcell, E. M. (1977). Life at low Reynolds number. American Journal of Physics 45, 3–11. · Feynman, R. P., Leighton, R. B. & Sands, M. (1963). The Feynman Lectures on Physics, Vol. I, Ch. 46: Ratchet and pawl. · Hecht, S., Shlaer, S. & Pirenne, M. H. (1942). Energy, quanta, and vision. Journal of General Physiology 25, 819–840. · Baylor, D. A., Lamb, T. D. & Yau, K.-W. (1979). Responses of retinal rods to single photons. Journal of Physiology 288, 613–634. · Bohr, N. & Wheeler, J. A. (1939). The mechanism of nuclear fission. Physical Review 56, 426–450. · Wheeler, J. A. & Ford, K. (1998). Geons, Black Holes, and Quantum Foam: A Life in Physics. Norton. · Wheeler, J. A. (1979). Mercer Street and other memories. In Aichelburg, P. C. & Sexl, R. U. (eds.), Albert Einstein: His Influence on Physics, Philosophy and Politics. Vieweg. · Everett, H. (1957). “Relative state” formulation of quantum mechanics. Reviews of Modern Physics 29, 454–462. · Bekenstein, J. D. (1973). Black holes and entropy. Physical Review D 7, 2333–2346. · Zurek, W. H. (2009). Quantum Darwinism. Nature Physics 5, 181–188. · Maxwell, J. C. (1871). Theory of Heat. Longmans, ch. 22. · Szilard, L. (1929). Über die Entropieverminderung in einem thermodynamischen System bei Eingriffen intelligenter Wesen. Zeitschrift für Physik 53, 840–856. · Landauer, R. (1961). Irreversibility and heat generation in the computing process. IBM Journal of Research and Development 5, 183–191. · Bennett, C. H. (1973). Logical reversibility of computation. IBM Journal of Research and Development 17, 525–532. · Bennett, C. H. (1982). The thermodynamics of computation: a review. International Journal of Theoretical Physics 21, 905–940. · Bérut, A. et al. (2012). Experimental verification of Landauer’s principle linking information and thermodynamics. Nature 483, 187–189. · Ballard, P. B. (1913). Obliviscence and reminiscence. British Journal of Psychology Monograph Supplements 1(2), 1–82. · Feynman, R. P. (1982). Simulating physics with computers. International Journal of Theoretical Physics 21, 467–488. · Hopfield, J. J. (1974). Kinetic proofreading: a new mechanism for reducing errors in biosynthetic processes requiring high specificity. PNAS 71, 4135–4139. · Hopfield, J. J. (1982). Neural networks and physical systems with emergent collective computational abilities. PNAS 79, 2554–2558. · Aspuru-Guzik, A., Dutoi, A. D., Love, P. J. & Head-Gordon, M. (2005). Simulated quantum computation of molecular energies. Science 309, 1704–1707. · Rebentrost, P., Mohseni, M., Kassal, I., Lloyd, S. & Aspuru-Guzik, A. (2009). Environment-assisted quantum transport. New Journal of Physics 11, 033003. · Preskill, J. (2018). Quantum computing in the NISQ era and beyond. Quantum 2, 79. · Bennett, C. H. & Brassard, G. (1984). Quantum cryptography: public key distribution and coin tossing. Proc. IEEE Int. Conf. on Computers, Systems and Signal Processing, 175–179. · Ghoneim, M. M. (2004). Drugs and human memory (part 1). Anesthesiology 100, 987–1002. · Bouton, M. E. (2004). Context and behavioral processes in extinction. Learning & Memory 11, 485–494. · Tononi, G. & Cirelli, C. (2014). Sleep and the price of plasticity: from synaptic and cellular homeostasis to memory consolidation and integration. Neuron 81, 12–34. · de Vivo, L. et al. (2017). Ultrastructural evidence for synaptic scaling across the wake/sleep cycle. Science 355, 507–510. · Crick, F. & Mitchison, G. (1983). The function of dream sleep. Nature 304, 111–114. · Wheeler, J. A. (1978). The “past” and the “delayed-choice” double-slit experiment. In Marlow, A. R. (ed.), Mathematical Foundations of Quantum Theory. Academic Press. · Jacques, V. et al. (2007). Experimental realization of Wheeler’s delayed-choice gedanken experiment. Science 315, 966–968. · Wheeler, J. A. (1990). Information, physics, quantum: the search for links. In Zurek, W. H. (ed.), Complexity, Entropy and the Physics of Information. Addison-Wesley.