Gravity Dual for a Model of Perception

TL;DR

This paper explores a holographic model of human perception invariance using scale-invariant but non-conformal geometries, linking Bayesian perception models with gravitational duals.

Contribution

It introduces a novel holographic framework for perception modeling based on non-conformal Euclidean geometries, connecting Bayesian inference with gravitational theories.

Findings

Identified a geometric configuration based on vector condensation and Einstein-Hilbert action.

Mapped Bayesian probability distributions to gravitational partition functions.

Analyzed correlation functions to support the holographic perception model.

Abstract

One of the salient features of human perception is its invariance under dilatation in addition to the Euclidean group, but its non-invariance under special conformal transformation. We investigate a holographic approach to the information processing in image discrimination with this feature. We claim that a strongly coupled analogue of the statistical model proposed by Bialek and Zee can be holographically realized in scale invariant but non-conformal Euclidean geometries. We identify the Bayesian probability distribution of our generalized Bialek-Zee model with the GKPW partition function of the dual gravitational system. We provide a concrete example of the geometric configuration based on a vector condensation model coupled with the Euclidean Einstein-Hilbert action. From the proposed geometry, we study sample correlation functions to compute the Bayesian probability distribution.