A Bell test is a randomized trial that compares experimental observations against the philosophical worldview of local realism1, in which the properties of the physical world are independent of our observation of them and no signal travels faster than light. A Bell test requires spatially distributed entanglement, fast and high-efficiency detection and unpredictable measurement settings2,3. Although technology can satisfy the first two of these requirements4,5,6,7, the use of physical devices to choose settings in a Bell test involves making assumptions about the physics that one aims to test. Bell himself noted this weakness in using physical setting choices and argued that human ‘free will’ could be used rigorously to ensure unpredictability in Bell tests8. Here we report a set of local-realism tests using human choices, which avoids assumptions about predictability in physics. We recruited about 100,000 human participants to play an online video game that incentivizes fast, sustained input of unpredictable selections and illustrates Bell-test methodology9. The participants generated 97,347,490 binary choices, which were directed via a scalable web platform to 12 laboratories on five continents, where 13 experiments tested local realism using photons5,6, single atoms7, atomic ensembles10 and superconducting devices11. Over a 12-hour period on 30 November 2016, participants worldwide provided a sustained data flow of over 1,000 bits per second to the experiments, which used different human-generated data to choose each measurement setting. The observed correlations strongly contradict local realism and other realistic positions in bipartite and tripartite12 scenarios. Project outcomes include closing the ‘freedom-of-choice loophole’ (the possibility that the setting choices are influenced by ‘hidden variables’ to correlate with the particle properties13), the utilization of video-game methods14 for rapid collection of human-generated randomness, and the use of networking techniques for global participation in experimental science.
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
We are grateful to the many people and organizations who contributed to this project, starting with the Bellsters (details at http://thebigbelltest.org). We thank the Departament d’Ensenyament de la Generalitat de Catalunya, Ministerio de Educación, Cultura y Deporte of Spain, CosmoCaixa, Fundación Bancaria “la Caixa”, INTEF, Optical Society of America, European Centers for Outreach in Photonics (ECOP), Muncyt-Museo de Ciencia y Tecnología de Madrid, Investigación y Ciéncia, Big Van, Crea Ciència, Tencent news, Micius Salon, Politecnico di Milano, University of Waterloo–Institute for Quantum Computing, Universidad Autónoma de Barcelona, Universidad Complutense de Madrid, Università degli Studi di Milano, Toptica, EPS Young Minds, Real Sociedad Española de Física, Ajuntament de Barcelona, Ajuntament de Castelldefels, Università degli studi di Padova, Università degli studi del l’Insubria, CNRIFN Istituto di Fotonica e Nanotecnologia, Istituto d’Istruzione Superiore Carlo Livi, Webbile, and Kaitos Games. We are especially thankful for the recruitment efforts of the outreach departments at our institutions. We acknowledge financial support from CONICET; ANPCyT (Argentina), Australian Research Council Centre for Quantum Computation and Communication Technology (CE110001027, CE170100012); Australian Research Council and the University of Queensland (UQ) Centre for Engineered Quantum Systems (CE110001013, CE170100009). A.G.W. acknowledges the University of Queensland Vice-Chancellor’s Research and Teaching Fellowship (Australia); Austrian Academy of Sciences (OEAW), the Austrian Science Fund (FWF) (SFB F40 (FoQuS); CoQuS No. W1210-N16), the Austrian Federal Ministry of Education, Science and Research (BMBWF) and the University of Vienna (project QUESS) (Austria), MEC; MCTIC (Brazil), Barcelona City Hall; Generalitat de Catalunya (SGR 874 and 2014-SGR-1295; CERCA programme) (Catalonia), PIA CONICYT (grants PFB0824 and PAI79160083); FONDECYT (grants 1140635, 1150101, 1160400, 3170596 and 11150325; Becas Chile); the Millennium Institute for Research in Optics (MIRO); Becas CONICYT (Chile), the National Fundamental Research Program (grant 2013CB336800); the National Natural Science Foundation of China (grants 91121022, 61401441 and 61401443); the Chinese Academy of Science (Strategic Priority Research Program (B) XDB04010200); the 1000 Youth Fellowship programme; the National Natural Science Foundation of China (grant 11674193); the Science and Technology Commission of Shanghai Municipality (grant 16JC1400402) (China); ERC (grant agreements AQUMET 280169, 3DQUEST 307783, OSYRIS 339106, ERIDIAN 713682, QITBOX, QUOLAPS, QuLIMA and SuperQuNet 339871); ESA (contract number 4000112591/14/NL/US); FEDER; H2020 (QUIC 641122) and the Marie Skłodowska-Curie programme (grant agreement 748549) (European Commission); German Federal Ministry of Education and Research (projects QuOReP and Q.com-Q) (Germany); CONACyT graduate fellowship programme (Mexico), MINECO (FIS2014-60843-P, FIS2014-62181-EXP, SEV-2015-0522, FIS2015-68039-P, FIS2015-69535-R and FIS2016-79508-P; Ramon y Cajal fellowship programme; TEC2016-75080-R); the ICFOnest + international postdoctoral fellowship programme (Spain); the Knut and Alice Wallenberg Foundation (project ‘Photonic Quantum Information’) (Sweden); NIST (USA), AXA Chair in Quantum Information Science; FQXi Fund; Fundació Privada CELLEX; Fundació Privada MIR-PUIG; the CELLEX-ICFO-MPQ programme; Fundació Catalunya-La Pedrera; and the International PhD-fellowship programme ‘la Caixa’-Severo Ochoa.The BIG Bell Test Collaboration
C. Abellán1, A. Acín1,2, A. Alarcón3,4, O. Alibart5, C. K. Andersen6, F. Andreoli7, A. Beckert6, F. A. Beduini1, A. Bendersky8, M. Bentivegna7, P. Bierhorst9, D. Burchardt10, A. Cabello11, J. Cariñe3,4,12, S. Carrasco1, G. Carvacho7, D. Cavalcanti1, R. Chaves13, J. Cortés-Vega3,12, A. Cuevas7, A. Delgado3,12, H. de Riedmatten1,2, C. Eichler6, P. Farrera1, J. Fuenzalida3,12,14, M. García-Matos1, R. Garthoff10, S. Gasparinetti6, T. Gerrits9, F. Ghafari Jouneghani15,16, S. Glancy9, E. S. Gómez3,12, P. González3,12, J.-Y. Guan17,18, J. Handsteiner14, J. Heinsoo6, G. Heinze1, A. Hirschmann1, O. Jiménez1, F. Kaiser5, E. Knill9, L. T. Knoll19,20, S. Krinner6, P. Kurpiers6, M. A. Larotonda19,20, J.-Å. Larsson21, A. Lenhard1, H. Li22,23, M.-H. Li17,18, G. Lima3,12, B. Liu24,14, Y. Liu17,18, I. H. López Grande19,20, T. Lunghi5, X. Ma25, O. S. Magaña-Loaiza9, P. Magnard6, A. Magnoni20, M. Martí-Prieto1, D. Martínez3,12, P. Mataloni7, A. Mattar1, M. Mazzera1, R. P. Mirin9, M. W. Mitchell1,2*, S. Nam9, M. Oppliger6, J.-W. Pan17,18, R. B. Patel15,16, G. J. Pryde15,16, D. Rauch14, K. Redeker10, D. Rieländer1, M. Ringbauer26,27, T. Roberson26,27, W. Rosenfeld10, Y. Salathé6, L. Santodonato7, G. Sauder5, T. Scheidl14, C. T. Schmiegelow20, F. Sciarrino7, A. Seri1, L. K. Shalm9, S.-C. Shi28, S. Slussarenko15,16, M. J. Stevens9, S. Tanzilli5, F. Toledo3,12, J. Tura1,29, R. Ursin14, P. Vergyris5, V. B. Verma9, T. Walter6, A. Wallraff6, Z. Wang22,23, H. Weinfurter10,29, M. M. Weston15,16, A. G. White26,27, C. Wu17,18, G. B. Xavier3,4,21, L. You22,23, X. Yuan25, A. Zeilinger14, Q. Zhang17,18, W. Zhang22,23 & J. Zhong28Reviewer information
Nature thanks S. Maniscalco (with F. Cosco, S. Hamedani Raja, H. Lyyra, J. Nokkala, B. Sokolov, N. Talarico and J. Teittinen), J. Sherson and the other anonymous reviewer(s) for their contribution to the peer review of this work.
Extended data figures and tables
a, Markov chain for L = 1. States are represented by circles of different colours, and transitions between states by arrows coloured as the initial state. The last input bit determines the state of the predictor. The probability p(a | b) of a transition from state b to state a is estimated from the sequence S k . b, Markov chain for L = 2, with four states determined by the last two input bits. The transitions model the probability p(a, b | c, d) that the user will input bits ab, given that the last two bits were cd. The final prediction is based on the marginal of the next single bit, which is extracted from these estimated probabilities.
a, The Oracle uses an MLA to predict user input. b, The ‘running’ component of the game, in which participants are asked to enter a minimum number of bits with a minimum unpredicted-input fraction in a limited time. c, Sequence of increasing-difficulty levels, interspersed with Oracle challenges. d, In-game feedback on the use of the user’s input bits in the running experiments. Blue and red buttons allow instant sharing on social networks Twitter and Weibo, respectively. e, A social media post sharing participant results. f, A social media post by a participant who completed the very difficult last Oracle level. The BIG Bell Quest artwork by Maria Pascual (Kaitos Games). See also Methods, Gamification.
This file contains Supplementary Text and Data, Supplementary Figures 1-20, Supplementary Tables 1-2 and Supplementary References.