杏吧原创

DOES reality exist without us? Albert Einstein appeared to be in no doubt: surely the moon doesn鈥檛 vanish when we aren鈥檛 looking, he once asked incredulously. He had been provoked by the proposition, from quantum theory, that things only become real when we observe them. But it is not such a daft idea, and even Einstein kept an open mind. 鈥淚t is basic for physics that one assumes a real world existing independently from any act of perception,鈥� he wrote in a 1955 letter. 鈥淏ut this we do not know.鈥�

In the decades since, physicists have found it maddeningly difficult to write the observer out of quantum theory. Now some are contemplating a mind-boggling alternative: that a coherent description of reality, with all its quantum quirks, can arise from nothing more than random subjective experiences. It looks like the 鈥減erspective of a madman鈥�, says the author of this bold new theory, because it compels us to abandon any notion of fundamental physical laws. But if it stands up, it would not only resolve some deep puzzles about quantum mechanics, it would turn our deepest preconceptions about reality itself inside out.

When it comes to forecasting how the world will behave, quantum theory is unsurpassed: its every prediction, no matter how counter-intuitive, is borne out by experiment. Electrons, for instance, can sometimes display behaviour characteristic of waves, even though they seem in other circumstances to behave like particles.

Wave of confusion

Before observation, such quantum objects are said to be in a superposition of all possible observable outcomes. This doesn鈥檛 mean they exist in many states at once, rather that we can only say that all the allowed outcomes of measurement remain possible. This potential is represented in the quantum wave function, a mathematical expression that encodes all outcomes and their relative probabilities.

Read more: What is consciousness?

How your brain creates the feeling of being is the biggest problem in neuroscience. But we are coming closer to cracking it

But it isn鈥檛 at all obvious what, if anything, the wave function can tell you about the nature of a quantum system before we make a measurement. That act reduces all those possible outcomes to one, dubbed the collapse of the wave function 鈥� but no one really knows what that means either. Some researchers think it might be a real physical process, like radioactive decay. Those who subscribe to the many-worlds interpretation think it is an illusion conjured by a splitting of the universe into each of the possible outcomes. Others still say that there is no point in trying to explain it 鈥� and besides, who cares? The maths works, so just shut up and calculate.

Whatever the case, wave function collapse seems to hinge on intervention or observation, throwing up some huge problems, not least about the role of consciousness in the whole process. This is the measurement problem, arguably the biggest headache in quantum theory. 鈥淚t is very hard,鈥� says , a philosopher at Chapman University in California. 鈥淢ore interpretations are being thrown up every day, but all of them have problems.鈥�

The most popular is known as the Copenhagen interpretation after the home city of one of quantum theory鈥檚 pioneers, Niels Bohr. He argued that quantum mechanics tells us only what we should expect when we make a measurement, not what causes that outcome. The theory can鈥檛 tell us what a quantum system is like before we observe it; all we can ever ask of it is the probabilities of different possible outcomes.

Such a perspective seems to back you into an uncomfortable conclusion: that the very act of our observation calls the outcome into being. Can that be true? It seems the antithesis of what science normally assumes, as Einstein intimated. Yet the idea has some pedigree. Hungarian physicist John von Neumann was the first to entertain it in the early 1930s, and his compatriot Eugene Wigner went deepest with a thought experiment in the 1950s now known as Wigner鈥檚 friend.

鈥淲hat if reality can鈥檛 be described without invoking our active involvement?鈥�

Suppose that Wigner is standing outside a windowless room where his friend is about to make some measurement on a particle. Once that鈥檚 done, she knows what the observed property of the particle is, but Wigner doesn鈥檛. He can鈥檛 meaningfully say that the particle鈥檚 wave function has collapsed until his friend tells him the result. Worse still, until she does, quantum theory offers no way for Wigner to think about all the unseen events inside the lab as having fixed outcomes. His friend, her measuring apparatus and the particle remain one big composite superposition.

It is as if we live in a solipsistic world where collapse only occurs when knowledge of the result impinges on a conscious mind. 鈥淚t follows that the quantum description of objects is influenced by impressions entering my consciousness,鈥� Wigner wrote. 鈥淪olipsism may be logically consistent with present quantum mechanics.鈥�

John Wheeler at Princeton University put it differently: it鈥檚 not solipsism but a kind of interactive collaboration that brings things into being. We live, Wheeler said, in a 鈥減articipatory universe鈥� 鈥� one that can鈥檛 be meaningfully described without invoking our active involvement. 鈥淣othing is more astonishing about quantum mechanics,鈥� he wrote, 鈥渢han its allowing one to consider seriously鈥� that the universe would be nothing without observership.鈥�

But Wheeler could not escape the thicket of irresolvable questions that the participatory universe raises. For one, Wigner and his friend seem locked in an infinite regress. Is Wigner himself in a superposition of states until he passes on the result to his other friends in the next building? Which observer 鈥渄ecides鈥� when wave function collapse occurs? And what constitutes a conscious observation anyway?

Despite the persistence of such questions, some theorists have recently returned to a form of Wheeler鈥檚 vision, what Chris Fuchs at the University of Massachusetts in Boston has called 鈥減articipatory reality鈥�. That shift is partly for want of a better alternative, but primarily it is because if you take quantum mechanics seriously, some element of observer-dependent subjectivity seems impossible to avoid.

A couple of years ago, theorist at the University of Vienna revisited the Wigner鈥檚 friend scenario in a slightly altered form proposed by David Deutsch at the University of Oxford. Here the friend makes the measurement 鈥� she has collapsed the particle鈥檚 wave function, producing either outcome A or B 鈥� but tells Wigner only that she sees a definite result, not what it is. In Deutsch鈥檚 scenario, Wigner is forced to conclude that his friend, her measuring apparatus and the particle are in a joint superposition, even though he knows a measurement has happened.

To Wigner鈥檚 friend, she is definitely in, say, the state 鈥淚 see A鈥�, but to Wigner she is in a superposition of 鈥淚 see A鈥� and 鈥淚 see B鈥�. So who鈥檚 right? They both are, says Brukner, depending on whose point of view you adopt. He has shown that if quantum mechanics is correct, there is no privileged perspective from which a third observer can reconcile both Wigner鈥檚 and his friend鈥檚 statements. 鈥淭here is no reason to assume that the 鈥榝acts鈥� of one of them are more fundamental than those of the other,鈥� Brukner says 鈥� and so we are forced to conclude that 鈥渢here are no 鈥榝acts of the world per se'鈥�. Rather, there are only facts for each observer.

One interpretation of quantum mechanics takes such a conclusion in its stride. Devised in the 2000s by Fuchs and others, quantum Bayesianism (also known as QBism) is rooted in the view that quantum mechanics supplies only recommendations about what a rational observer should believe he or she will see on making a measurement 鈥� and that these beliefs can be updated as the observer takes fresh experiences into account. That鈥檚 where the 鈥淏ayesianism鈥� comes in: it refers to the classical theory of probability, initiated in the 18th century, that assigns probabilities on the basis of what the observer already knows to be the case.

QBism point-blank denies that there is any objective notion of a quantum state at all. This doesn鈥檛 mean there can be nothing 鈥渞eal鈥� beyond personal belief, only that quantum mechanics doesn鈥檛 speak directly to that issue.

The existence of Brukner鈥檚 鈥渁lternative facts鈥� causes no pain in such a picture, because it has assumed them all along. Nor indeed does wave function collapse, which is then just a way of talking about how measurement updates our knowledge. But few physicists are prepared to accept such stringent limits on their efforts to describe reality, which is why QBism remains a minority sport.

So what now? Enter , the self-described 鈥渕adman鈥� and a theorist at the University of Vienna. His answer is to take things up a notch. 鈥淨Bism is not extreme enough,鈥� he says. 鈥淚t assumes that there is this one external world out there which is ultimately responsible for our experiences. My approach starts without assuming such a world.鈥� That means imagining that there are no fundamental laws of nature 鈥� no general relativity, no Maxwell鈥檚 equations or Heisenberg鈥檚 uncertainty principle 鈥� and asking what the world would then look like. The answer might surprise you.

鈥淚t鈥檚 deeply odd: you end up with a universe built directly from our experiences鈥�

For even if you throw out physics, the logic of mathematics remains 鈥� and this is where M眉ller begins to construct his hypothetical world. Let鈥檚 say you have some experience X: you make an observation of the world and see the outcome X. Given that, what are the chances you will then experience another outcome Y? There is a field of maths that deals precisely with such questions. It鈥檚 called algorithmic information theory and it shows how to make predictions based on inductive reasoning with a distinctly Bayesian flavour: given that X happened, there is an equation to figure out the probability that Y will follow.

M眉ller wanted to see where that reasoning leads in a world with nothing else to shape it. He represented each individual鈥檚 experience at every instant as a string of bits of information 鈥� 1s and 0s, like binary computer code. Each observer鈥檚 history then consisted of a walk through the various possible bit strings, and the probability of going from one to the next would be random but conditional: it must take the history of experiences into account. The idea, says theorist Giulio Chiribella at the University of Hong Kong, 鈥渋s to think of our experience as a movie made of many frames and to ask the question, given the frames I have seen so far, what frame will I see next?鈥�

You would think that such a picture could hardly be less likely to give rise to what we experience: a universe governed by laws and producing facts that are, as far as we can tell, the same for you as for me. But when M眉ller used the methods of algorithmic information theory to figure out what sequences of bit strings are likely, he found something remarkable.

As these random experiences stack up, the conditional probability of the next experience, as described by a string of bits, tends to be higher for simpler bit sequences than for complex ones. This makes it look as if there is a fairly simple algorithm generating the bit strings. So the observer deduces a simple 鈥渕odel鈥� of reality, characterised by regular and comprehensible laws that smoothly connect one experience to the next.

This seems deeply odd: how can randomness give rise to this apparently law-bound behaviour? It鈥檚 a little like the way we understand a gas. Although in principle all possible configurations of its molecules are allowed, the probability distribution of particle speeds we see has a simple bell-shaped curve, and the particles are distributed in space with bland uniformity. Out of that come simple laws relating to things we can easily measure: pressure, temperature and volume. Those laws aren鈥檛 written into the gas particles themselves; they are an emergent property of the probabilities of different configurations.

鈥淭he remarkable thing is that a notion of an objective external world emerges automatically in the long run,鈥� says M眉ller. What鈥檚 more, 鈥渄ifferent observers will tend to agree on properties of that external world鈥�. That鈥檚 because according to algorithmic information theory, the probabilities of bit strings for different observers will tend to converge on the same distribution 鈥� so they will agree on what the 鈥渓aws of the world鈥� are. 鈥淥verall, the 鈥榤ovie鈥� is likely to be simple and different observers can generally agree on some aspects of the plot,鈥� says Chiribella.

The surprises don鈥檛 stop there. This emergent reality should have just the qualities we see in quantum physics, where objects can show wave-like properties and behave in 鈥渘on-local鈥� ways, when a measurement on one particle can seem instantaneously to influence the state of another separated in space.

The upshot is that from the most minimal assumptions about the probabilities of what our personal experiences will contain, you can recover a world like the one we know. 鈥淭he world could still look something like how we experience it, even though in truth it would be mind-bogglingly different,鈥� says M眉ller.

It鈥檚 not easy to see how M眉ller鈥檚 ideas can be tested. But circumstantial evidence that he could be on the right track comes with the way they solve the problem of Boltzmann brains, an almost metaphysical conundrum that suggests we are not the kind of beings we think we are (see 鈥淪pontaneous brains鈥�).

All the same, this image of the universe built directly from observer experiences is so 鈥渙ut there鈥� that other researchers barely know what to make of it. It鈥檚 鈥渁 very interesting starting point, which prompts new questions鈥�, says Chiribella. Meanwhile, Brukner is keeping an open mind. 鈥淢arkus鈥檚 conceptual grasp and mathematical background is strong, and enables him to step outside the comfort zone and come up with true conceptual changes and modifications of our theories,鈥� he says.

M眉ller himself is deeply aware that he has not chosen an easy path, but argues that it is one worth treading. 鈥淚t is not as crazy as it first seems,鈥� he says. 鈥淏ut it will be a major challenge to convince people, since the world view that it suggests is so unusual and different from what we are used to.鈥�

This article appeared in print under the headline 鈥淩eality? It鈥檚 what you make it鈥�