‘Metaphysical Experiments’ Probe Our Hidden Assumptions About Reality

Cavalcanti is carrying the torch of a tradition that stretches back through a long line of rebellious thinkers who have resisted the usual dividing lines between physics and philosophy. In experimental metaphysics, the tools of science can be used to test our philosophical worldviews, which in turn can be used to better understand science. Cavalcanti, a 46-year-old native of Brazil who is a professor at Griffith University in Brisbane, Australia, and his colleagues have published the strongest result attained in experimental metaphysics yet, a theorem that places strict and surprising constraints on the nature of reality. They’re now designing clever, if controversial, experiments to test our assumptions not only about physics, but about the mind.

While we might expect the injection of philosophy into science to result in something less scientific, in fact, says Cavalcanti, the opposite is true. “In some sense, the knowledge that we obtain through experimental metaphysics is more secure and more scientific,” he said, because it vets not only our scientific hypotheses but the premises that usually lie hidden beneath.

THE DIVIDING LINE between science and philosophy has never been clear. Often, it’s drawn along testability. Any science that deserves its name is said to be vulnerable to tests that can falsify it, while philosophy aims for pristine truths that hover somewhere beyond the grubby reach of experiment. So long as that distinction is in play, physicists believe they can get on with the messy business of “real science” and leave the philosophers in their armchairs, stroking their chins.

As it turns out, though, the testability distinction doesn’t hold. Philosophers have long known that it’s impossible to prove a hypothesis. (No matter how many white swans you see, the next one could be black.) That’s why Karl Popper famously said that a statement is only scientific if it’s falsifiable — if we can’t prove it, we can at least try to disprove it. In 1906, though, the French physicist Pierre Duhem showed that falsifying a single hypothesis is impossible. Every piece of science is bound up in a tangled mesh of assumptions, he argued. These assumptions are about everything from underlying physical laws to the workings of specific measurement devices. If the result of your experiment appears to disprove your hypothesis, you can always account for the data by tweaking one of your assumptions while leaving your hypothesis intact.

Take, for instance, the geometry of space-time. Immanuel Kant, the 18th-century philosopher, declared that the properties of space and time are not empirical questions. He thought not only that the geometry of space was necessarily Euclidean, meaning that a triangle’s interior angles add up to 180 degrees, but that this fact had to be “the basis of any future metaphysics.” It wasn’t empirically testable, according to Kant, because it provided the very framework within which we understand how our tests work in the first place.

And yet in 1919, when astronomers measured the path of distant starlight skirting the gravitational influence of the sun, they found that the geometry of space wasn’t Euclidean after all — it was warped by gravity, as Albert Einstein had recently predicted.

Or did they? Henri Poincaré, the French polymath, offered up an intriguing thought experiment. Imagine that the universe is a giant disk that conforms to Euclidean geometry, but whose physical laws include the following: The disk is hottest in the middle and coldest at the edge, with the temperature falling in proportion to the square of the distance from the center. Moreover, this universe features a refractive index — a measurement of how light rays bend — that is inversely proportional to the temperature. In such a universe, rulers and yardsticks would never be straight (solid objects would expand and shrink with the temperature gradient) while the refractive index would make light rays appear to travel in curves rather than lines. As a result, any attempt to measure the geometry of the space — say, by adding up the angles of a triangle — would lead one to believe that the space was non-Euclidean.

Any test of geometry requires you to assume certain laws of physics, while any test of those laws of physics requires you to assume the geometry. Sure, the disk world’s physical laws seem ad hoc, but so are Euclid’s axioms. “Poincaré, in my opinion, is right,” Einstein said in a 1921 lecture. He added, “Only the sum of geometry and physical laws is subject to experimental verification.” As the American logician Willard V. O. Quine put it, “The unit of empirical significance” — the thing that’s actually testable — “is the whole of science.” The simplest observation (that the sky is blue, say, or the particle is there) forces us to question everything we know about the workings of the universe.

But actually, it’s worse than that. The unit of empirical significance is a combination of science and philosophy. The thinker who saw this most clearly was the 20th-century Swiss mathematician Ferdinand Gonseth. For Gonseth, science and metaphysics are always in conversation with one another, with metaphysics providing the foundations on which science operates, science providing evidence that forces metaphysics to revise those foundations, and the two together adapting and changing like a living, breathing organism. As he said in a symposium he attended in Einstein’s honor, “Science and philosophy form a single whole.”

With the two tied together in a Gordian knot, we might be tempted to throw up our hands, since we can’t put scientific statements to the test without dragging metaphysical statements along with them. But there’s a flipside to the story: It means that metaphysics is testable. That’s why Cavalcanti, who works at the very edges of quantum knowledge, doesn’t refer to himself as a physicist, or as a philosopher, but as an “experimental metaphysicist.”

I MET WITH CAVALCANTI on a video call. With his dark hair pulled back into a bun, he had a brooding look about him, his careful, serious demeanor offset only by a 15-week-old puppy squirming in his lap. He told me how, as an undergraduate in Brazil in the late 1990s, he worked on experimental biophysics — “very wet stuff,” as he describes it, “getting hearts out of rabbits and putting them under [superconducting] magnetometers,” that sort of thing. Though he soon moved on to drier territory (“working in particle accelerators, studying atomic collisions”), the work was still far from the metaphysical questions already lingering in his mind. “I had been told that the interesting questions in foundations of quantum mechanics had all been resolved by [Niels] Bohr in his debates with Einstein,” he said. So he measured another cross section, churned out another paper, and did it all again the next day.

He ended up working for Brazil’s National Nuclear Energy Commission, and it was there that he read books by the physicists Roger Penrose and David Deutsch, each offering up a radically different metaphysical story to account for the facts of quantum mechanics. Should we give up the philosophical assumption that there’s only one universe, as Deutsch suggested? Or, as Penrose preferred, perhaps quantum theory ceases to apply at large scales, when gravity gets in on the action. “Here were these brilliant physicists who not only are directly discussing questions about foundations but profoundly disagreeing with each other,” Cavalcanti said. Penrose, he added, “even went beyond physics into what’s traditionally metaphysics, asking questions about consciousness.”

Inspired, Cavalcanti decided to pursue a doctorate in quantum foundations and found a place for himself at the University of Queensland in Australia. His dissertation began, “To understand the source of the conflicts of quantum foundations, it is essential to know where and how our classical models and intuitions start to fail to describe a quantum world. This is the subject of experimental metaphysics.” A professor put the thesis down and declared, “This isn’t physics.”

But Cavalcanti was prepared to make the case that the line between physics and philosophy had already been blurred beyond repair. In the 1960s, the Northern Irish physicist John Stewart Bell had also encountered a culture of physics that had no patience for philosophy. The days of Einstein and Bohr arguing over the nature of reality — and engaging deeply with philosophy in the process — were long over. Postwar practicality reigned, and physicists were eager to get on with the business of physics, as if the Gordian knot had been cut, as if it were possible to ignore metaphysics and still manage to do science at all. But Bell, doing his heretical work in his spare time, discovered a new possibility: While it’s true that you can’t test a single hypothesis in isolation, you can take multiple metaphysical assumptions and see if they stand or fall together.

For Bell, those assumptions are typically understood to be locality (the belief that things can’t influence each other instantaneously across space) and realism (that there’s some way things simply are, independent of their being measured). His theorem, published in 1964, proved what’s known as Bell’s inequality: For any theory operating under the assumptions of locality and realism, there’s an upper limit on how correlated certain events can be. Quantum mechanics, however, predicted correlations that busted through that upper limit.

As written, Bell’s theorem wasn’t testable, but in 1969 the physicist and philosopher Abner Shimony saw that it could be rewritten in a form suitable for the lab. Along with John Clauser, Michael Horne and Richard Holt, Shimony transformed Bell’s inequality into the CHSH inequality (named for its authors’ initials), and in 1972, in a basement in Berkeley, California, Clauser and his collaborator Stuart Freedman put it to the test by measuring correlations between pairs of photons.

The results showed that the world bore out the predictions of quantum mechanics, showing correlations that remained far stronger than Bell’s inequality allowed. This meant that locality and realism can’t both be features of reality — though which of the two we ought to abandon, the experiments couldn’t say. “To my mind, the most fascinating thing about theorems of Bell’s type is that they provide a rare opportunity for an enterprise which can properly be called ‘experimental metaphysics,’” Shimony wrote in 1980 in the statement that’s widely believed to have coined the term.

As it happens, though, the term goes back further, to a most unlikely character. Michele Besso, Einstein’s best friend and sounding board, was the only person Einstein credited with helping him come up with the theory of relativity. But Besso helped less with the physics than with the philosophy. Einstein had always been a realist, believing in a reality behind the scenes, independent of our observations, but Besso introduced him to the philosophical writings of Ernst Mach, who argued that a theory should only refer to measurable quantities. Mach, by way of Besso, encouraged Einstein to give up his metaphysical notions of absolute space, time and motion. The result was the special theory of relativity.

Upon its publication in 1905, physicists weren’t sure whether the theory was physics or philosophy. All of its equations had already been written down by others; it was only the metaphysics behind them that was new. But that metaphysics was enough to lead to new science, as special relativity gave way to general relativity, a new theory of gravity, complete with new, testable predictions. Besso later befriended Gonseth; in Switzerland, the two took long walks together, where Gonseth argued that physics could never be placed on firm foundations, since experiments can always overturn the most bedrock assumptions on which it is built. In a letter, which Gonseth published in a 1948 issue of the journal Dialectica, Besso suggested that Gonseth refer to his work as “experimental metaphysics.”

Experimental metaphysics gained something of an official headquarters in the 1970s with the founding of the Association Ferdinand Gonseth in Bienne, Switzerland. “Science and philosophy form one body,” it stated in its founding values, “and all that happens in science, whether in its methods or in its results, may resound on philosophy even in its most fundamental principles.” This was a radical statement — equally shocking to both science and philosophy. The association published an underground newsletter called Epistemological Letters, a kind of physics “zine,” with typed, mimeographed pages speckled with hand-drawn equations that was mailed out to 100 or so physicists and philosophers who comprised a new counterculture — the daring few who wanted to discuss experimental metaphysics. Shimony served as editor.

Bell’s theorem was always at the center of those discussions, because where previous work in physics let its metaphysics go unacknowledged, in Bell’s work the two were truly and explicitly inseparable. The theorem was not about any particular theory of physics. It was what physicists call a “no-go” theorem, a general proof showing that any theory operating under the metaphysical assumptions of locality and realism can’t describe the world we live in. You want a world that just is some particular way even when it’s not being measured? And you want locality? No go. Or, as Shimony put it in Epistemological Letters, in a play on Bell’s name, those who want to hold such a worldview “should remember the sermon of Donne: ‘And therefore never send to know for whom the bell tolls; it tolls for thee.’”

“Bell was both a philosopher of physics and a physicist,” said Wayne Myrvold, a philosopher of physics at Western University in Canada. “And in some of his best papers, he’s basically combining the two.” That rattled the editors of traditional physics journals and other gatekeepers of science. “This kind of work was definitely not seen as respectable,” Cavalcanti said.

That’s why, when the French physicist Alain Aspect went to Bell suggesting a new experiment that could test Bell’s inequality while ruling out any residual influence propagating between the measurement devices used to detect the photons’ polarizations, Bell asked him whether he had a permanent faculty position. “The worry was that doing that experiment would be a career killer for a young physicist,” Myrvold said.

Fast-forward to 2022, and there’s Aspect, along with Clauser and Anton Zeilinger, headed to Stockholm to receive a Nobel Prize. Those Bell’s inequality-violating correlations have, as it turns out, led to revolutionary technologies including quantum cryptography, quantum computing and quantum teleportation. But “despite the technological payoff,” Myrvold said, “the work was motivated by philosophical questions.” According to the Nobel citation, the three physicists won for “pioneering quantum information science.” According to Cavalcanti, they won for experimental metaphysics.

BELL’S THEOREM WAS only the beginning.

In the wake of experiments violating Bell-type inequalities, several views of reality remained on the table. You could keep realism and give up locality, accepting that what happens in one corner of the universe instantaneously affects what happens in another and therefore that relativity must be modified. Or you could keep locality and give up realism, accepting that things in the universe don’t have definite features prior to being measured — that nature is, in some profound sense, making things up on the fly.

But even if you gave up on a pre-measurement reality, you could still hang on to a post-measurement reality. That is, you could imagine taking all those measurement outcomes and piecing them together into a single, shared reality. That’s typically what we mean by “reality.” It’s the very notion of an objective world.

A thought experiment posed in 1961 casts doubt on that possibility. Eugene Wigner, the Nobel Prize-winning physicist, proposed a scenario in which an observer, call him “Wigner’s friend,” goes into a lab where there’s a quantum system — say, an electron in a quantum combination, or superposition, of two states called “spin up” and “spin down.” The friend measures the electron’s spin and finds that it’s up. But Wigner, standing outside, can use quantum mechanics to describe the entire state of the lab, where, from his perspective, no measurement has taken place. The state of the friend and the state of the electron are merely correlated — entangled — while the electron remains in a superposition of states. In principle, Wigner can even perform a measurement that will show physical effects of the superposition. From the friend’s perspective, the electron has some post-measurement state, but this doesn’t seem to be part of Wigner’s reality.

In 2018, that nagging doubt about a shared reality became a full-blown dilemma. Časlav Brukner, a physicist at the University of Vienna, realized that he could combine Wigner’s friend with a Bell-type experiment to prove a new no-go theorem. The idea was to have two friends and two Wigners; the friends each measure half of an entangled system, and then each of the Wigners makes one of two possible measurements on his friend’s lab. The Wigners’ measurement outcomes will be correlated, just like the photons’ polarizations in the original Bell-type experiments, with certain metaphysical assumptions imposing upper bounds on the strength of those correlations.

As it turned out, Brukner’s proof relied on an extra assumption that weakened the strength of the resulting theorem, but it inspired Cavalcanti and colleagues to make their own version. In 2020, in the journal Nature Physics, they published “A Strong No-Go Theorem on the Wigner’s Friend Paradox,” which proved two things. First, that experimental metaphysics, previously relegated to underground zines, is now worthy of prestigious scientific journals, and second, that reality is even stranger than Bell’s theorem ever suggested.

Their no-go theorem showed that, if the predictions of quantum mechanics are correct, the following three assumptions cannot all be true: locality (no spooky action at a distance), freedom of choice (no cosmic conspiracy tricking you into setting your detectors so that the outcomes seem to violate Bell’s inequality even though they don’t), and absoluteness of observed events (an electron with spin up for Wigner’s friend is an electron with spin up for everyone). If you want local interactions and a conspiracy-free cosmos, then you have to give up on the notion that a measurement outcome for one observer is a measurement outcome for all.

Significantly, their no-go theorem “constrains the space of possible metaphysical theories more tightly than Bell’s theorem does,” Cavalcanti said.

“It’s an important improvement,” Brukner said. “It’s the most precise, strongest no-go theorem.” Which is to say, it’s the most powerful piece of experimental metaphysics yet. “The strength of these no-go theorems is exactly that they do not test any particular theory, but a worldview. By testing them and showing violations of certain inequalities, we don’t reject one theory, but a whole class of theories. That’s a very powerful thing. It allows us to understand what is possible.”

Brukner laments that the implications of experimental metaphysics haven’t yet been fully incorporated into the rest of physics at large — especially, in his view, to the detriment of research on the quantum nature of gravity. “This is really a pity, because we end up with wrong pictures of, say, how the vacuum looks, or what goes on in a black hole, where they are described without any reference to modes of observation,” he said. “I don’t think that we will make significant progress in these fields until we really do much work on the theory of measurement.”

Whether experimental metaphysics can ever lead us to the correct theory of quantum gravity is unclear, but it could at least narrow the playing field. “There’s a story, I don’t know if it’s apocryphal, but it’s a nice one,” Cavalcanti wrote in a 2021 paper, “according to which Michelangelo, when asked about how he sculpted David, said: ‘I just removed anything that was not David.’ I like to think of the metaphysical landscape as the raw block of marble — with different points in the block corresponding to different physical theories — and of experimental metaphysics as a chisel to carve the marble, eliminating corners that do not describe the world of our experience. It may turn out that we are unable to reduce the block to a single point, corresponding to the one true ‘theory of everything.’ But we may hope that after we carve out all the bits that experiment allows us to, what remains forms a beautiful whole.”

AS I SPOKE with Cavalcanti, I tried to get a read on which interpretation of quantum mechanics he subscribed to by feeling out which metaphysical assumptions he hoped to hang on to and which he was ready to toss. Did he agree with the Bohmian interpretation of quantum mechanics, which trades locality for realism? Was he a “QBist,” with no need for the absoluteness of observed events? Did he believe in the cosmic conspiracies of the superdeterminists, who attribute all correlated measurements in the present-day universe to a master plan set out at the beginning of time? How about measurements spawning parallel realities, as in the many-worlds hypothesis? Cavalcanti kept a true philosopher’s poker face; he wouldn’t say. (The puppy, meanwhile, was waging an all-out tug-of-war against the carpet.) I did, however, catch one hint. Whatever interpretation he eventually chooses, he wants it to touch on the mystery of the mind — what consciousness is, or what counts as a conscious observer. “I still think that that is the deepest mystery,” he said. “I don’t think that any of the available interpretations actually quite get to the right story.”

In their 2020 Nature Physics paper, Cavalcanti and colleagues reported the results of what they called a “proof-of-principle version” of their Bell-cum-Wigner’s-friend experiment, which showed a clear violation of inequalities derived from the joint assumptions of locality, freedom of choice, and the absoluteness of observed events. But the experiment is inherently tricky to carry out, because something — or someone — has to play the role of an observer. In the proof-of-principle version, Wigner’s “friends” were played by photon paths, while photon detectors played the part of the Wigners. Whether something as simple as a photon path counts as an observer is notoriously hard to say.

“If you think that any physical system can be considered an observer, then the experiment has already been done,” Cavalcanti said. “But most physicists will think, no, I don’t buy that. So what are the next steps? How far can we go?” Is a molecule an observer? An amoeba? Could Wigner be friends with a fig? Or a ficus?

If the friend has to be human, it’s hard to overstate just how difficult it would be to measure one in a superposition, which is exactly what the Wigners of the experiment are supposed to do. It’s hard enough to keep a single atom in a superposition. Sustaining an atom’s superposed states means isolating it from virtually all interactions — including interactions with air — which means storing it just a hair’s breadth above absolute zero. The average adult human being, besides needing air, is made of some 30 trillion cells, each containing some 100 trillion atoms. The technology, fine motor skills and questionable ethics a Wigner would need to perform his measurement would stretch the imagination of any physicist or institutional review board. “It’s not always emphasized that this [proposed] experiment is a violent act,” Myrvold said. “It basically involves destroying the person and then reviving them.” Good luck getting the grant money for that.

Brukner, for one, wonders whether the measurement is not merely difficult, but impossible. “I suspect if we put it all down on paper, we will see that the resources required for Wigner to make this measurement go far beyond what is available in the universe,” he said. “Maybe in some more fundamental theory, these limitations will be part of the theory, and it will turn out that there is no meaning to this question.” That would be quite the twist for experimental metaphysics. Maybe our deepest insights into the nature of reality will come when we realize what’s not testable.

Cavalcanti, however, is holding out hope. We may never be able to run the experiment on a human, he says, but why not an artificial intelligence algorithm? In his newest work, along with the physicist Howard Wiseman and the mathematician Eleanor Rieffel, he argues that the friend could be an AI algorithm running on a large quantum computer, performing a simulated experiment in a simulated lab. “At some point,” Cavalcanti contends, “we’ll have artificial intelligence that will be essentially indistinguishable from humans as far as cognitive abilities are concerned,” and we’ll be able to test his inequality once and for all.

But that’s not an uncontroversial assumption. Some philosophers of mind believe in the possibility of strong AI, but certainly not all. Thinkers in what’s known as embodied cognition, for instance, argue against the notion of a disembodied mind, while the enactive approach to cognition grants minds only to living creatures.

All of which leaves physics in an awkward position. We can’t know whether nature violates Cavalcanti’s inequality — we can’t know, that is, whether objectivity itself is on the metaphysical chopping block — until we can define what counts as an observer, and figuring that out involves physics, cognitive science and philosophy. The radical space of experimental metaphysics expands to entwine all three of them. To paraphrase Gonseth, perhaps they form a single whole.