Principle of locality

Alice and Bob in spacetime diagram.

A spacetime diagram has a time coordinate going vertical and a space coordinate going horizontal. Alice, in a local region on the left, can affect events only in a cone extending in the future as shown; the finite speed of light prevent her from affecting other areas including Bob's location in this case. Similarly we can use the diagram to reason that Bob's local circumstances cannot be altered by Alice at the same time: all events that cause an effect on Bob are in the cone below his location on the diagram. Dashed lines around Alice show her valid future locations; dashed lines around Bob show events that could have caused his present circumstance. When Alice measures quantum states in her location she gets the results labeled ${\displaystyle \mathbf {a} }$; similarly Bob gets ${\displaystyle \mathbf {b} }$. Models of locality attempt to explain the statistical relationship between these measured values.

Action at a distance.

The simplest locality model is no locality: instantaneous action at a distance with no limits for relativity. The locality model for action at a distance is called continuous action[2] The gray area (a circle here) is a mathematical concept called a "screen". Any path from a location through the screen becomes part of the physical model at that location. The gray ring indicates events from all parts of space and time can affect the probability measured by Alice or Bob. So in the case of continuous action, events at all times and places affect Alice's and Bob's model. This simple model is highly successful for solar planetary dynamics with Newtonian gravity and in electrostatics, cases where relativistic effects are insignificant.

No future-input dependence.

Many locality models explicitly or implicitly ignore the possible effect of future events. The spacetime diagram at the right shows the effect of such a restriction when combined with continuous action. Inputs from the future (above the dashed line) are no longer considered part of Alice's or Bob's model. Comparing this diagram with the one for continuous action makes it clear that these are not the same locality model.[2] Common sense arguments about the future not affecting the present are reasonable criteria but such assumptions alter the mathematical character of the models.

Bell's local causality.

John Stewart Bell when discussing his Bell's theorem uses the screening model shown at the right. Events in the common past of Alice and Bob are part of the model used in calculating probabilities for Alice and for Bob as indicated the way the screen absorbs those events. However events at Bob's location during Alice measurement and events in the future are excluded. Bell called this assumption local causality, but with the diagram we can reason about the meaning of the assumption without getting tripped up by other meanings of local combined with other meanings of causal.[2] Dash lines show relativistically valid regions in the past of Alice or Bob. The gray arc is the assumed Bell "screen".

Bell described local causality in terms of probability needed for analysis of quantum mechanics. Using the notation that ${\displaystyle P(\mathbf {r} |g)}$ for the probability of a result ${\displaystyle \mathbf {r} }$ with given state ${\displaystyle g}$, Bell investigated the probability distribution

$${\displaystyle P(\mathbf {ab} |AB,\lambda )}$$

where ${\displaystyle \lambda }$ represents hidden state variables set (locally) when the two particles are initially co-located. If local causality holds, then the probabilities observed by Alice and by Bob should be only coupled by the hidden variables, and we can show that

$${\displaystyle P(\mathbf {ab} |AB,\lambda )=P(\mathbf {a} |A,\lambda )P(\mathbf {b} |B,\lambda ),}$$

Bell proved that a consequence of this factorization are limits on the correlations observed by Alice and Bob known as Bell inequalities. Since quantum mechanics predicts correlations stronger than this limit, locally set hidden variables cannot be added to "complete" quantum theory as desired by the EPR paper.[7]

Numerous experiments specifically designed to probe the issues of locality confirm the predictions of quantum mechanics; these include experiments where the two measurement locations are more than a kilometer apart.[7][8] The 2022 Nobel Prize in Physics was awarded to Alain Aspect, John Clauser and Anton Zeilinger, in part "for experiments with entangled photons, establishing the violation of Bell inequalities".[9] The specific aspect of quantum theory that leads to these correlations is termed quantum entanglement and versions of Bell's scenario are now used to verify entanglement experimentally.[7]

Bell's mathematical results, when compared to experimental data, eliminate local hidden variable mathematical quantum theories. But the interpretation of the math with respect to the physical world remains under debate. Bell described the assumptions behind his work as "local causality", shortened to "locality"; later authors referred to the assumptions as local realism.[10] These different names do not alter the mathematical assumptions.

A review of papers[11] using this phrase suggests that a common (classical) physics definition of realism is

the assumption that measurement outcomes are well defined prior to and independent of the measurements.[12]

This definition includes classical concepts like "well-defined" which conflicts with quantum superposition and "prior to...measurements" which implies (metaphysical) preexistence of properties. Specifically, the term local realism in the context of Bell's theorem cannot be viewed as a kind of "realism" involving locality other than the kind implied by the Bell screening assumption. This conflict between common ideas of realism and quantum mechanics requires careful analysis whenever local realism is discussed.[11]^: 98 Adding a "locality" modifier, that the results of two spatially well-separated measurements cannot causally affect each other,[5] does not make the combination relate to Bell's proof; the only interpretation that Bell assumed was the one he called local causality.[11]^: 98 Consequently, Bell's theorem does not restrict the possibility of nonlocal variables as well as theories based on retrocausality or superdeterminism.[2]

Because of the probabilistic nature of wave function collapse, this apparent violation of locality in quantum mechanics cannot be used to transmit information faster than light, in accordance to the no communication theorem.[13] Asher Peres distinguishes between weak and strong nonlocality, the latter referring to the theories that allow to faster-than-light communication. Under these terms, quantum mechanics would allow weakly nonlocal correlations but not strong nonlocality.[14]