# Local Intrinsic Dimension of Representations Predicts Alignment and Generalization in AI Models and Human Brain

arXiv: https://arxiv.org/abs/2601.22722 · HTML: https://mlsanity.com/paper/2601.22722/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Builds on 7 of its 28 references that are in the map.
- First posted on arXiv 30 Jan 2026.

## Builds on (most cited first, 7 of 7)

- [A ConvNet for the 2020s](https://mlsanity.com/paper/2201.03545/) (2022)
- [Intrinsic dimension of data representations in deep neural networks](https://mlsanity.com/paper/1905.12784/) (2019)
- [Reconstructing the Mind's Eye: fMRI-to-Image with Contrastive Learning and Diffusion Priors](https://mlsanity.com/paper/2305.18274/) (2023)
- [Deep Neural Networks Reveal a Gradient in the Complexity of Neural Representations across the Brain's Ventral Visual Pathway](https://mlsanity.com/paper/1411.6422/) (2014)
- [Symmetry in Neural Network Parameter Spaces](https://mlsanity.com/paper/2506.13018/) (2025)
- [Dimensions underlying the representational alignment of deep neural networks with humans](https://mlsanity.com/paper/2406.19087/) (2024)
- [Disentangling the Factors of Convergence between Brains and Computer Vision Models](https://mlsanity.com/paper/2508.18226/) (2025)

## Built on by (most cited first, 2 of 2)

- [Transition Information Density: Morphological Trajectories, Synesthetic Perception, and Structured Interpolation in Neural Training (or: The Synesthetic AI)](https://mlsanity.com/paper/2607.03210/) (2026)
- [The Platonic brain bridge hypothesis: human brain networks as an architectural prior for multimodal large language models](https://mlsanity.com/paper/2609.10947/) (2026)

## Abstract (arXiv)

Recent work has found that neural networks with stronger generalization tend to exhibit higher representational alignment with one another across architectures and training paradigms. In this work, we show that models with stronger generalization also align more strongly with human neural activity. Moreover, generalization performance, model--model alignment, and model--brain alignment are all significantly correlated with each other. We further show that these relationships can be explained by a single geometric property of learned representations: the local intrinsic dimension of embeddings. Lower local dimension is consistently associated with stronger model--model alignment, stronger model--brain alignment, and better generalization, whereas global dimension measures fail to capture these effects. Finally, we find that increasing model capacity and training data scale systematically reduces local intrinsic dimension, providing a geometric account of the benefits of scaling. Together, our results identify local intrinsic dimension as a unifying descriptor of representational convergence in artificial and biological systems.