Epistemic entanglement in the macroscopic world
2 April at 11.30 Santamaria building (B14)
Scientific non-reductionism emphasizes that the whole is more than the sum of the parts. Beyond metaphor, such holistic idea can be made concrete (namely, mathematically formulated and empirically testable) via the notion of entanglement, which is a foundational concept and an established phenomenon in quantum mechanics related to Heisenberg’s uncertainty principle (a fundamental limit to the precision of simultaneously measuring two complementary variables, like the position and momentum of an electron, which is not a consequence of experimental limitations). Here we present a classical analog of entanglement, where uncertainty relations arise from experimental commitments, like the selection of variables and/or subsystems, or ignorance about the context. Inspired by a theoretical work offering a definition of entanglement in a model of macroscopic brownian particles (Allahverdyan et al., 2005), we propose a general sufficient condition for epistemic entanglement that is valid for any underlying dynamics and any pair of macroscopic stochastic observables. Our bound reflects a trade-off between inter-vs-intra particle correlations, and only requires estimating dispersions. This makes it empirically accessible and also somewhat intuitive. We explored the origin of epistemic entanglement by taking advantage of analytical results available in brownian models and simulations of stochastic systems. Then, we applied our sufficient condition to behavioral data of fly courtship, and found entanglement between position and coarse-grained velocity. This result implies the existence of macroscopic correlations that cannot be effectively explained in causal terms, thus limiting the common cause principle. Our work also challenges the idea of pure objectivity, as our choice of measurement variables induced epistemic correlations that cannot be adjudicated to the observed system but that, through coarse-graining, belong to the observer.