Experimental Falsification of Leggett’s Nonlocal Variable Model

Bell’s theorem guarantees that no model based on local variables can reproduce quantum correlations. Also, some models based on nonlocal variables, if subject to apparently “reasonable” constraints, may fail to reproduce quantum physics. In this Letter, we introduce a family of inequalities, which use a finite number of measurement settings, and which therefore allow testing Leggett’s nonlocal model versus quantum physics. Our experimental data falsify Leggett’s model and are in agreement with quantum predictions.

Figure 1

Dependency of the combined correlation parameters $L(\phi )$ as a function of the separation angle $\phi$ for the quantum mechanical prediction for a pure singlet state, and bounds for nonlocal variable models assuming an averaging over various numbers of directions $N$.

Figure 2

Experimental setup. Polarization-entangled photon pairs are generated in Barium-beta-borate (BBO) by parametric down conversion of light from an Ar ion pump laser (PL). After walk-off compensation ($\lambda /2$, CC), down-converted light is collected behind interference filters (IF) into birefringence-compensated (FPC) single mode optical fibers (SMF). Polarization measurements are carried out with a combination of a quarter wave plate ($\lambda /4$) and polarization filters (PF) in front of photon counting detectors $D_{1,2}$. The measurement basis for each arm (1,2) is chosen by rotation of the wave plate and polarizing filter by angles ${\alpha }_{1,2}$, ${\beta }_{1,2}$ accordingly.

Figure 3

Experimental results for the observed correlation parameters $L_N$ (dots), the quantum mechanical prediction for a pure singlet state (curved lines, dashed lines), and the bounds for the nonlocal variable models (almost straight lines). In all cases, our experiment exceeds the NLV bounds for appropriate difference angles $\phi$.