What actually happens when we look?

Every experiment performed so far has agreed with its predictions, including those designed to catch it failing. No result has yet required a change to the formalism.

The electron’s magnetic moment is the sharpest case. It has been measured to 0.13 parts per trillion, and it tests the most precise prediction the standard model makes to one part in a trillion. What limits that test now is not the measurement but disagreeing values of the fine-structure constant, which the prediction needs.

John Bell showed in 1964 that any theory in which particles have definite pre-existing properties and no influence travels faster than light must obey certain statistical limits. Quantum mechanics predicts violations of those limits.

Those violations have been measured. Each round of experiments closed another way the result could have been an artefact rather than a fact about the world, until in 2015 a test was reported that needed none of the extra assumptions earlier ones had. Its authors add that no Bell test can exclude every conceivable local theory. Unless the choice of settings was itself somehow fixed in advance, the world is not a collection of objects with definite independent properties interacting only locally.

Interaction with an environment rapidly destroys the interference between components of a superposition, making them effectively independent. This is well understood, experimentally confirmed, and it is crucial to why large objects behave classically.

It does not explain why a measurement has a result in the first place.

Is it a physical thing, a description of our information, or a bookkeeping device with no referent? Each answer has serious defenders and each leads to a different picture of reality. A theorem of Pusey, Barrett and Rudolph narrows the field, but only for anyone who grants its framework and its assumption that two systems can be prepared independently of each other.

Interference has been demonstrated with increasingly large objects, most recently, in 2026, with sodium clusters of more than seven thousand atoms. In 2023 the vibration of a sixteen-microgram crystal was put in superposition over a far shorter distance, about two billionths of a nanometre. No boundary has been found, and there is no agreed criterion for when a system counts as macroscopic. Whether a boundary exists (and whether gravity enforces it) is an open experimental question that several groups are actively pursuing.

The theory gives probabilities. An experiment shows one result. Nothing in the standard formalism says what process takes the one to the other, or exactly when, and no account that leaves the predictions alone has produced a way to tell itself apart from the others.

The theory as ordinarily taught treats measurement as primitive, without saying what physically constitutes one. A detector is made of atoms that obey the same equations. That tension is the measurement problem, stated plainly.

The many-worlds interpretation removes the collapse step entirely: the wavefunction always evolves smoothly, and every outcome occurs in a branch of a continually splitting universal state. Its defenders call it the most economical theory available, because it keeps the laws of standard quantum theory and drops the collapse postulate. Its critics call the worlds an extravagance. A difficulty its critics press, and one of its defenders calls arguably the most serious, is saying where the probabilities come from when every outcome happens.

Pilot-wave theory restores definite trajectories guided by a real wave. It accounts for everything non-relativistic quantum mechanics accounts for, and it makes explicit a non-locality that its proponents say no theory reproducing quantum mechanics can avoid. It is not Lorentz invariant, and cannot easily be made so.

Objective collapse models add a genuine, spontaneous collapse mechanism to the dynamics. Unlike most interpretations these make predictions that differ slightly from standard quantum mechanics, and experiments are actively narrowing the space they can occupy. The parameter-free form of the Diósi-Penrose model was ruled out that way, by a search under a mountain for the faint radiation such a collapse should emit.

Some approaches treat the quantum state as a description of an agent’s expectations rather than of physical reality. On this view, its holders argue, the measurement problem does not arise, because there is no physical state to collapse. Critics, among them the philosopher Christopher Timpson, answer that such a theory cannot explain why things are as they are: why sodium conducts, rather than why an agent should expect it to.