The Concept of Representation in ML: Beyond Plato and Aristotle

2026-07-20

AI Summary

This study critically examines whether current claims about representational convergence in machine learning sufficiently support the metaphysical conclusion underpinning the “Platonic Representation Hypothesis”—namely, that a unified structure of reality drives consistency across diverse AI models’ representations. Drawing on philosophical theories of mental representation and offering a critical assessment of empirical work on representational alignment, the paper systematically clarifies the conceptual distinction between engineering-oriented representations and genuine mental representations. It demonstrates that existing evidence of inter-model representational similarity is insufficient to substantiate strong metaphysical claims about the unity of real-world structure. The work thus underscores the limitations of current alignment studies and outlines a novel pathway toward a more rigorous, interdisciplinary theory of representation.

Abstract

Representation is a central concept in modern machine learning, where it usually refers to internal encodings that support learning and generalization. As models scale and their capabilities become increasingly human-level, this representational language sometimes shifts from an engineering context into the more philosophically loaded domain of mental representation. We argue that this is the case for recent claims about the convergence of representational properties across different AI models. In particular, we assess the arguments developed in The Platonic Representation Hypothesis, according to which this convergence is driven by a unified structure of reality. We examine this claim by introducing arguments and ideas from debates about mental representation in the philosophy of mind. We argue that these philosophical resources can clarify what is at stake in such claims, explain why alignment evidence alone is insufficient for strong metaphysical conclusions, and suggest directions for future research.

Problem

• representation
• machine learning
• mental representation
• Platonic Representation Hypothesis
• philosophy of mind

Innovation

• representation
• philosophy of mind
• Platonic Representation Hypothesis
• mental representation
• machine learning