Скачать или смотреть David Schmid - Causal-Inferential Theories: Realism Revisited

David Schmid
University of Gdańsk

Quantum theory provides an algorithm for computing statistical correlations. A realist approach to science, however, demands more: a physical theory not only predict correlations, but also provide causal explanations of these correlations. It is well-known, however, that there are significant obstacles to understanding quantum theory in any straightforwardly realist sense. Most notable among these obstacles are locality and noncontextuality no-go theorems. That is, Bell’s theorem demonstrates that local hidden variable theories cannot reproduce the quantum predictions, which implies serious difficulties to providing a causal account of those predictions. Additionally, it has been shown that there can be no realist representation of quantum theory that satisfies the principle of generalized noncontextuality, which is our most broadly applicable foundational notion of classicality.

In this article, we describe a program for circumventing these no-go theorems—that is, for providing a realist account of quantum theory that salvages the spirit of locality and of noncontextuality. We present a mathematical framework for causation and inference that helps to identify which aspects of the conventional notions of causation and inference are indispensable for any notion of causation and inference, and which aspects can be modified. (This is analogous to how in nonEuclidean geometries, certain aspects of the conventional notions of points and lines are preserved while others are modified.) The possibility of such modifications within our framework opens the door to an intrinsically nonclassical notion of realism that can account for the statistical predictions of the quantum formalism.

Our framework is also useful for the important project of unscrambling causation from inference. For example, we provide a new framework for generalized physical theories, one which allows us to provide more refined descriptions than standard frameworks of generalized and operational probabilistic theories. Our approach also makes clear how certain geometric structures are common to all generalized physical theories. Furthermore, we our framework also allows us to provide more refined “ontological representations” of operational scenarios than is possible using the standard framework of ontological modeling. For example, we demonstrate (somewhat surprisingly) that standard “ontological models” inadvertently scramble together causation and inference, in a manner that we untangle.