How Scientists Tested Einstein’s Hunch

Einstein had a hunch that quantum physics was incomplete. He suspected particles carry hidden details that fix their behavior in advance, with no influence racing between distant places. For decades this sounded like a question for philosophers. Then, in 1964, the physicist John Bell turned it into a number that a laboratory could measure. Experiments since 1972 have measured it again and again, and the answer is clear: nature breaks Bell’s limit. The simple, local version of Einstein’s hunch does not survive.

That is a real result, strong enough to earn the 2022 Nobel Prize in Physics. But it has edges. The tests rule out one kind of explanation. They do not tell us which of the remaining stories about quantum reality is true.

What exactly was Einstein’s hunch?

In a 1935 paper with Boris Podolsky and Nathan Rosen, Einstein argued that quantum theory could not be a complete description of reality. The paper focused on pairs of particles that are linked, or entangled. Measure one, and you can predict a result for the other without touching it.

Einstein’s view rested on two ideas that sound like common sense. First, realism: particles have definite properties whether or not anyone looks. Second, locality: nothing done here can instantly affect something over there. Together, these point toward local hidden variables, which are hidden, pre-set instructions carried by each particle. Niels Bohr disagreed, and the argument (told in Einstein vs. Bohr: The Debate) seemed impossible to settle with experiments.

How did John Bell turn philosophy into a test?

John Stewart Bell (1928–1990) was a Northern Irish physicist who worked at CERN, the European particle physics laboratory. In a 1964 paper, “On the Einstein Podolsky Rosen paradox”, he showed that Einstein’s picture makes a testable prediction.

Here is an everyday way to picture it. Imagine twins separated into two rooms. Each is asked one of several yes-or-no questions, picked at random at the last moment. If the twins agreed on a script beforehand, simple arithmetic limits how often their answers can match across the different question pairs. Bell showed the same kind of limit applies to any theory with local hidden variables. It is now called a Bell inequality. Quantum theory predicts that entangled particles will beat it.

In the most common version, experimenters combine their results into one score. Any local hidden-variable theory keeps that score at 2 or less. Quantum theory predicts values up to about 2.83. So the question became concrete: in a real lab, is the score above 2?

What did Clauser, Aspect and Zeilinger do?

John Clauser and three colleagues proposed a practical version of the test in 1969. In 1972, Clauser and doctoral student Stuart Freedman ran it, using calcium atoms that gave off pairs of photons and filters that tested the photons’ polarization. Their result clearly broke Bell’s limit and matched quantum theory. The experiment needed about 200 hours of running time.

Critics pointed to loopholes, gaps that might let a hidden-variable theory sneak through. One was the detection loophole: the detectors caught only a small share of the photons, so perhaps the caught ones were unusual. Another was the locality loophole: the filter settings were fixed in advance, so perhaps some unknown signal carried information between the two sides.

Alain Aspect, then a doctoral student in France, attacked the locality loophole. In a 1982 experiment, his setup switched which filter each photon would meet after the photons had left their source. The filters were about six meters away, so the switch had to happen within a few billionths of a second. The switching followed a fast, regular pattern rather than a truly random one, a gap later tests closed. Quantum theory still won.

Anton Zeilinger and his colleagues pushed further. In 1998 they used random number generators to choose settings at two stations 400 meters apart. Later tests let light from Milky Way stars (2017) and from distant quasars (2018) pick the settings, and the 2018 “Big Bell Test” used choices from about 100,000 volunteers. The 2022 Nobel Prize honored all three “for experiments with entangled photons,” work that established the violation of Bell inequalities and helped launch quantum information science.

What did the 2015 “loophole-free” tests add?

Until 2015, every test left at least one loophole open. That year, three teams each closed the main ones in a single experiment. In the Netherlands, a Delft team entangled the spins of electrons in two diamond chips 1.3 kilometers apart. It ran 245 trials and scored about 2.42, above the limit of 2, though with a fairly wide margin of error. Teams in Vienna and at the US National Institute of Standards and Technology (NIST) used photons and highly efficient detectors, and their results were statistically much stronger.

In the NIST test, the detectors sat 184 meters apart and could catch at least 90 percent of very faint signals. NIST calculated that the chance of a local, pre-set world producing its results was at most about 1 in 170 million. Later loophole-free tests used atoms and superconducting circuits too. As one NIST physicist put it at the time, in our paraphrase: you cannot prove quantum mechanics, but local realism does not fit the data.

What do the results rule out, and what don’t they settle?

Ruled out: any theory that combines three things: pre-set answers carried by each particle, no influence faster than light, and setting choices that are truly independent of the particles. Some summaries say the tests proved “there are no hidden variables” at all. The careful statement is narrower. They rule out local ones.

Not settled: what replaces Einstein’s picture. The Stanford Encyclopedia of Philosophy lists several live options. Pilot-wave theory keeps hidden variables but makes them connected across space. Many-worlds theories drop the idea that each measurement has a single outcome. Other views accept basic chance. More unusual options allow effects to run backward in time, or suppose that settings and particles were linked long ago, an idea called superdeterminism. Bell tests cannot rule that last one out by design, and critics call it conspiratorial. None of the mainstream options lets anyone send faster-than-light messages. We compare them in Is the Universe Truly Random?

What would change the answer?

The Bell result would weaken if a careful test found scores at or below 2, or if someone showed that the methods used to pick settings were secretly linked to the particles. Neither has been seen. A deeper change could come if quantum predictions failed for very large or heavy systems, where gravity may matter. As always, the bar is high, as we discuss in What Would Count as Evidence?

Bell’s great gift was showing that some “philosophical” questions can be put to nature directly. Which of today’s unanswerable questions might be next?