# On the Map-Territory Fallacy Fallacy

**Authors:** Maxwell J D Ramstead, Dalton A R Sakthivadivel, Karl J Friston  
**arXiv:** [2208.06924](https://arxiv.org/abs/2208.06924)  
**Date:** 2022-08-14  
**Summary page:** https://www.summarizepaper.com/en/paper/2208.06924v1/

## Key points

- The paper presents a meta-theory of the Free Energy Principle (FEP) and its application in modeling physical systems.
- Authors challenge the 'map-territory fallacy' and refute claims that it undermines the principled use of the FEP.
- They argue against criticizing the FEP and Bayesian mechanics using this fallacy, labeling it as a 'map-territory fallacy fallacy'.
- Emphasize the uniqueness of the FEP as a model for particles or agents modeling their environments.
- Restore convention to the FEP through its connection to constrained maximum entropy principle and assert its 'Jaynes optimality'.
- Clarify utility and scope of FEP as a formal modeling tool for generic systems in statistical physics.
- Advocate for deeper understanding of both theories without predetermined assumptions about cognition or perception.
- Caution against interpretations contradicting core principles like anti-representationalism and rejection of computational tools for studying self-organizing systems.

## Summary

The paper "On the Map-Territory Fallacy Fallacy" by Maxwell J D Ramstead, Dalton A R Sakthivadivel, and Karl J Friston presents a meta-theory of the usage of the Free Energy Principle (FEP) and explores its application in modelling physical systems. The authors address the so-called 'map-territory fallacy' and challenge the fallacious reification of model properties. By demonstrating that the FEP is a consistent theory inspired by physics, specifically focusing on inferences of inferences, they refute claims that the map-territory fallacy undermines the principled use of the FEP. Additionally, they argue against using this fallacy to criticize both the FEP and Bayesian mechanics, labeling it as a 'map-territory fallacy fallacy'. The authors highlight several key points in their analysis, including emphasizing the uniqueness of the FEP as a model for particles or agents that model their environments. They also discuss how convention is restored to the FEP through its connection to the principle of constrained maximum entropy and assert its 'Jaynes optimality' under this relationship. Furthermore, they explain how this meta-theoretical approach clarifies the utility and scope of the FEP as a formal modeling tool, positioning it as an ideal model for generic systems in statistical physics. In contrast to existing philosophical work on Bayesian mechanics which often starts from predetermined assumptions about cognition or perception before interpreting the FEP within those frameworks, this paper advocates for a deeper understanding of both theories and their inherent commitments. The authors caution against interpretations that may contradict core principles such as anti-representationalism and rejection of computational tools for studying self-organizing systems. By delving into these complexities, they aim to provide a more nuanced perspective on how these theories can be effectively applied in scientific research.