Scientists can now uncover true cause and effect in complex systems — without touching them
Innovation PressPeer-Reviewed Publication
Determining whether one variable truly causes another – rather than merely correlating with it – usually requires deliberate intervention experiments, which are often impossible in climate systems, neural circuits, or economies. Reporting in The Innovation, researchers led by Professors Wei Lin, Luonan Chen, and Kazuyuki Aihara introduce a theoretical framework called interventional dynamical causality (IntDC) together with a computational criterion, interventional embedding entropy (IEE). Built on delay-embedding state-space reconstruction, IEE quantifies the strength of intervention-level causal effects using observational time-series data alone – no manipulation of the system required. The method accurately removed confounding effects in benchmark simulations and successfully reconstructed the neural connectome of Caenorhabditis elegans, COVID-19 transmission networks in Japan, and circadian gene regulatory networks.
- Journal
- The Innovation
- Funder
- National Natural Science Foundation of China, Shanghai Municipal Data Bureau special fund for urban digital transformation, AI for Science Foundation of Fudan University, Lingang Laboratory, AI for Science Program of Shanghai Municipal Education Commission, Zhejiang Province Vanguard Goose-Leading Initiative, Shenzhen Medical Research Fund, Japan Science and Technology Agency Moonshot R&D, Cross-ministerial Strategic Innovation Promotion Program (SIP), the third period of SIP, National Key Research and Development Program of China, Institute of AI and Beyond, UTokyo / WPI-IRCN (UTIAS)