Rob Spekkens (Perimeter Institute), Institute for Quantum Computing (IQC)
Abstract
If correlation does not imply causation, then what does? The beginning of a rigorous answer to this question has been provided by researchers in machine learning, who have developed causal discovery algorithms. These take as their input facts about correlations among a set of observed variables and return as their output a causal structure relating these variables. We show that any attempt to provide a causal explanation of Bell-inequality-violating correlations must contradict a core principle of these algorithms, namely, that an observed statistical independence between variables should not be explained by fine-tuning of the causal parameters. In particular, we demonstrate the need for such fine-tuning for most of the causal mechanisms that have been proposed to underlie Bell correlations, including superluminal causal influences, superdeterminism (that is, a denial of freedom of choice of settings), and retrocausal influences which do not introduce causal cycles. This work suggests a novel perspective on the assumptions underlying Bell's theorem: the nebulous assumption of "realism" is replaced with the principle that all correlations ought to be explained causally, and Bell's notion of local causality is replaced with the assumption of no fine-tuning. Finally, we discuss the possibility of avoiding the fine-tuning by replacing conditional probabilities with a noncommutative generalization thereof.
Based on arXiv:1208.4119.
Joint work with Chris Wood.