# Priors in Time: Missing Inductive Biases for Language Model Interpretability

arXiv: https://arxiv.org/abs/2511.01836 · HTML: https://mlsanity.com/paper/2511.01836/

## Key facts

- Cited by 25 papers in the map (data through 14 Oct 2026).
- Placed in SAE features and their applications, one of 13 directions in Mechanistic interpretability.
- First posted on arXiv 3 Nov 2025.

## Placed in

- Mechanistic interpretability / [SAE features and their applications](https://mlsanity.com/mechinterp/sae-features-and-their-applications/)

## Built on by (most cited first, 25 of 25)

- [Do Sparse Autoencoders Capture Concept Manifolds?](https://mlsanity.com/paper/2604.28119/) (2026)
- [Manifold Steering Reveals the Shared Geometry of Neural Network Representation and Behavior](https://mlsanity.com/paper/2605.05115/) (2026)
- [The Shape of Beliefs: Geometry, Dynamics, and Interventions along Representation Manifolds of Language Models' Posteriors](https://mlsanity.com/paper/2602.02315/) (2026)
- [Linear representations in language models can change dramatically over a conversation](https://mlsanity.com/paper/2601.20834/) (2026)
- [Arithmetic in the Wild: Llama uses Base-10 Addition to Reason About Cyclic Concepts](https://mlsanity.com/paper/2605.01148/) (2026)
- [Context Structure Reshapes the Representational Geometry of Language Models](https://mlsanity.com/paper/2601.22364/) (2026)
- [Language Models Struggle to Use Representations Learned In-Context](https://mlsanity.com/paper/2602.04212/) (2026)
- [Structuring Sparsity: Block-Sparse Featurizers Capture Visual Concept Manifolds](https://mlsanity.com/paper/2606.25234/) (2026)
- [Shifting the Gradient: Understanding How Defensive Training Methods Protect Language Model Integrity](https://mlsanity.com/paper/2604.16423/) (2026)
- [Stories in Space: In-Context Learning Trajectories in Conceptual Belief Space](https://mlsanity.com/paper/2605.12412/) (2026)
- [Convergent World Representations and Divergent Tasks](https://mlsanity.com/paper/2602.00533/) (2026)
- [Anatomy of Post-Training: Using Interpretability to Characterize Data and Shape the Learning Signal](https://mlsanity.com/paper/2606.12360/) (2026)
- [Turn-Averaged SAEs for Feature Discovery and Long-Context Attribution](https://mlsanity.com/paper/2606.28548/) (2026)
- [Emergent Structured Representations Support Flexible In-Context Inference in Large Language Models](https://mlsanity.com/paper/2602.07794/) (2026)
- [Representational Curvature Modulates Behavioral Uncertainty in Large Language Models](https://mlsanity.com/paper/2604.23985/) (2026)
- [Towards Effective Theory of LLMs: A Representation Learning Approach](https://mlsanity.com/paper/2605.09294/) (2026)
- [Interpretability-Guided Layer Selection over Subspace Projection: SAEs as Stethoscopes, Not Scalpels, for Raw Task Vector Model Editing](https://mlsanity.com/paper/2605.28649/) (2026)
- [Geometry of Ordinal Representations in Language Models](https://mlsanity.com/paper/2607.04167/) (2026)
- [Persistent Sparse Autoencoders: Learning Feature-Specific Timescales in Language Model Representations](https://mlsanity.com/paper/2607.17117/) (2026)
- [MAxBench: A Multinomial Concept Recovery Benchmark](https://mlsanity.com/paper/2609.13072/) (2026)
- [Interference Beyond Geometry in Concept Extraction](https://mlsanity.com/paper/2609.35351/) (2026)
- [Signatures of semantic search in the activations of large language models](https://mlsanity.com/paper/2609.35599/) (2026)
- [ClasSAE: Class-Aligned Sparse Autoencoders via Differentiable Feature-Class Affinity](https://mlsanity.com/paper/2610.04020/) (2026)
- [The Independence Prior of SAEs Fragments Visual Concepts](https://mlsanity.com/paper/2610.04112/) (2026)
- [Sparse Autoencoders for Mechanistic Interpretability in NLP: A Critical Survey of Recoverability, Interpretability, and Utility](https://mlsanity.com/paper/2610.20139/) (2026)

## Abstract (arXiv)

Recovering meaningful concepts from language model activations is a central aim of interpretability. While existing feature extraction methods aim to identify concepts that are independent directions, it is unclear if this assumption can capture the rich temporal structure of language. Specifically, via a Bayesian lens, we demonstrate that Sparse Autoencoders (SAEs) impose priors that assume independence of concepts across time, implying stationarity. Meanwhile, language model representations exhibit rich temporal dynamics, including systematic growth in conceptual dimensionality, context-dependent correlations, and pronounced non-stationarity, in direct conflict with the priors of SAEs. Taking inspiration from computational neuroscience, we introduce a new interpretability objective -- Temporal Feature Analysis -- which possesses a temporal inductive bias to decompose representations at a given time into two parts: a predictable component, which can be inferred from the context, and a residual component, which captures novel information unexplained by the context. Temporal Feature Analyzers correctly parse garden path sentences, identify event boundaries, and more broadly delineate abstract, slow-moving information from novel, fast-moving information, while existing SAEs show significant pitfalls in all the above tasks. Overall, our results underscore the need for inductive biases that match the data in designing robust interpretability tools.