Keywords: concept, meaning, agent foundations, probability, latent variable, information theory, structural causal model
TL;DR: We prove an "intersubectivity theorem", showing that under certain hypotheses, different systems of latent variables stand in a kind of correspondence.
Abstract: We understand agents as creating concepts in order to organize their understanding of the world, and as often sharing concepts and so being able to work out how to make sense of the world together, forming language communities. Here, we look at how probability distributions can be organized by introducing appropriate latent variables, which we aim to use as a model of these phenomena. Our main result shows that under certain information-theoretic hypotheses, different systems of latent variables stand in a kind of correspondence.
Confirmation: I confirm that I and my co-authors have read the policies are releasing our work under a CC-BY 4.0 license.
Submission Number: 12
Loading