Mechanistic Interpretability

Follow this topic Ask about this topic

Recent momentum

-70%

6 papers in the last 28 days · 0.2% of indexed attention

Twelve weeks of publication activity for this topic as it is defined today.

Weekly history

Recent digests

What was published in this topic, kept on the site without email delivery.

Period ending 2026-09-14

2 new papers

A weekly snapshot of new work published in Mechanistic Interpretability.

Period ending 2026-09-07

6 new papers

A weekly snapshot of new work published in Mechanistic Interpretability.

154 papers

Latest in Mechanistic Interpretability

Open your feed →

Frame-Conditioned Moral Computation in LLaMA 3.1-8B-Instruct: A Mechanistic Interpretability Audit of Ethical Reasoning
Jun 13, 2026Ali Dasdan, Manan Shah, W. Russell Neuman +3Moral Reasoning Framing
Beyond Importance: Interchange-Sobol Sensitivity Reveals Task-Specific Content Channels in Transformer Components
Jun 12, 2026Yifeng Guo, Jin-Hong Du, Xiang ChenTransformer Variant Importance
Rational Sparse Autoencoder
Jun 12, 2026Naiyu Yin, Yue YuImproving Sparse Autoencoders Autoencoder
ATLAS: Active Theory Learning for Automated Science
Jun 10, 2026Noémi Éltető, Nathaniel D. Daw, Kimberly L. Stachenfeld +1Generalizable Time-Series Behavioral Modeling Active Learning
The Standard Interpretable Model: A general theory of interpretable machine learning to deductively design interpretable methods using Lagrangian mechanics
Jun 10, 2026Pietro Barbiero, Giovanni De Felice, Mateo Espinosa Zarlenga +5Mechanistic Interpretability Artificial Intelligence Models
Sparse probes and murky physics: a case study of interpretability challenges in a foundation model for continuum dynamics
Jun 10, 2026Katherine Rosenfeld, Maike SonnewaldNeural Network Emulator Wireless Foundation Models
Trajectory Geometry of Transformer Representations Across Layers
Jun 8, 2026Vishal Pandey, Gopal Singh, Yacine MahdidNeural Representations Mechanistic Interpretability
Ablation-Reversible Heads Don't Transfer: A Stress Test for Mechanistic Role Claims in Transformers
Jun 6, 2026Philip QuirkeMechanistic Interpretability Ablation
Shared Semantics, Divergent Mechanisms: Unsupervised Feature Discovery by Aligning Semantics and Mechanisms
Jun 6, 2026Hyunjin Cho, Youngji Roh, Jaehyung KimMechanistic Interpretability Model Interpretability Methods
When Attribution Patching Lies: Diagnosis and a Second-Order Correction
Jun 5, 2026Luyang Zhang, Jialu WangMechanistic Interpretability Perturbation Propagation
Inside the Visual Mind: Neuroscience-Motivated Concept Circuits for Interpreting and Steering Vision Transformers
Jun 4, 2026Tang Li, Yanlin Chen, Mengmeng Ma +1Self-Supervised Vision Transformers Vision Transformer
Subspace-Aware Sparse Autoencoders for Effective Mechanistic Interpretability
Jun 4, 2026Seyed Arshan Dalili, Mehrdad MahdaviSparse Autoencoders Mechanistic Interpretability
EML-CD: Causal Mechanism Recovery via EML Symbolic Trees in Structure Learning
Jun 4, 2026Sota AsanumaCausal Discovery Structural Causal Model
Position: Don't Just "Fix it in Post": A Science of AI Must Study Training Dynamics
Jun 3, 2026Stella Biderman, Mohammad Aflah Khan, Niloofar Mireshghallah +3Artificial Intelligence Systems Scientific Discovery
Linear Probes Detect Task Format, Not Reasoning Mode in Language Model Hidden States
Jun 1, 2026Subramanyam Sahoo, Vinija Jain, Aman Chadha +1Hidden States Mechanistic Interpretability
Ablating Archetypes: The Stability of Archetypal SAEs is an Artifact of Initialization and Metric Design
Jun 1, 2026Michał Brzozowski, Neo Christopher ChungImproving Sparse Autoencoders Learnable Probe Vectors
Detection vs. Execution: Single-Bucket Probes Miss Half the Mamba-2 State Sink
May 30, 2026Yuhang JiangDistributional Signatures Mechanistic Interpretability
Position: Prioritize Identifying Structure, Not Complex Models, for Scientific Discovery
May 30, 2026Tyler H. McCormickScientific Discovery Mechanistic Interpretability
Latent Space Disentanglement via Activation Steering for Interpretable Attribute Control in Symbolic Music Generation
May 29, 2026Ioannis Prokopiou, Pantelis Vikatos, Maximos Kaliakatsos-Papakostas +2Linear Activation Steering Modulation
Discovering a Zeta Map Algorithm on Dyck Paths via Mechanistic Interpretability
May 28, 2026Xiaoyu Huang, Blake Jackson, Kyu-Hwan LeeMechanistic Interpretability Algorithmic Reasoning
Cultural Binding Heads in Language Models
May 27, 2026Avrile Floro, Luca BenedettoCultural Alignment Mechanistic Interpretability
Guiding LLM Post-training Data Engineering with Model Internals from Sparse Autoencoders
May 26, 2026Yi Jing, Zao Dai, Jinwu Hu +4Large Language Model Post-Training Mechanistic Interpretability
MechRL: Reinforcement Learning Agents Perform Circuit Discovery for Mechanistic Interpretability
May 25, 2026Barsat KhadkaMechanistic Interpretability Circuits
Detecting Unfaithful Chain-of-Thought via Circuit-Guided Internal-External Discrepancy
May 25, 2026Xu Shen, Zhen Tan, Song Wang +4Explanation Faithfulness Reasoning Traces
Is Inference Mediated by Distinct Semantic Structures in LLMs? A Mechanistic Interpretation
May 25, 2026Nura Aljaafari, Marco Valentino, André FreitasNatural Language Inference Mechanistic Interpretability
Transformer Field Theory: A Response-Theoretic Approach to Mechanistic Interpretability
May 24, 2026David N. Olivieri, Antonio F. Pérez RodríguezTransformer Variant Autoregressive Transformers
The Concept Allocation Zone: Tracking How Concepts Form Across Transformer Depth
May 24, 2026James HenryTransformer Variant Concept Bottleneck Models
Polymorphism Is Rotation: Operational Mechanistic Interpretability from a Two-Layer Transformer to Pythia-70m
May 23, 2026Jordan F. McCannImproving Sparse Autoencoders Mechanistic Interpretability
Learning dynamical systems with biochemically informed neural ordinary differential equations
May 22, 2026Luis L. Fonseca, Reinhard C. Laubenbacher, Lucas BöttcherNeural Ordinary Differential Equations Incomplete Kinetics
Every Component is a Lookup: Token Attribution and Composition from a Single Decomposition
May 22, 2026Po-Kai Chen, Niki van Stein, Aske PlaatToken-Level Uncertainty Mechanistic Interpretability
Multilingual Steering by Design: Multilingual Sparse Autoencoders and Principled Layer Selection
May 21, 2026Yusser Al Ghussin, Daniil Gurgurov, Tanja Baeumel +3Multilingual Language Models Sparse Autoencoders
Sparse Autoencoders Map Brain-LLM Alignment onto Cortical Semantic Topography
May 21, 2026Dongxin Guo, Jikun Wu, Siu Ming YiuSparse Autoencoders Human Visual Cortex
Geometry-Adaptive Explainer for Faithful Dictionary-Based Interpretability under Distribution Shift
May 21, 2026Sungjun Lim, Heedong Kim, Andrew Lee +1Model Interpretability Methods Mechanistic Interpretability
Post-Hoc Understanding of Metaphor Processing in Decoder-Only Language Models via Conditional Scale Entropy
May 20, 2026Lawhori Chakrabarti, Jennifer Johnson-Leung, Bert Baumgaertner +3Metaphors Mechanistic Interpretability
From Circuit Evidence to Mechanistic Theory: An Inductive Logic Approach
May 20, 2026Nura Aljaafari, Danilo S. Carvalho, Andre FreitasMechanistic Interpretability Circuits
Mechanistic Interpretability for Learning Assurance of a Vision-Based Landing System
May 20, 2026Romeo Valentin, Olivia Beyer Bruvik, Marc R. Schlichting +1Aerotrack Mechanistic Interpretability
Mechanisms of Object Localization in Vision-Language Models
May 19, 2026Timothy Schaumlöffel, Martina G. Vilas, Gemma RoigTarget Localization Localization
A Distributional View for Visual Mechanistic Interpretability: KL-Minimal Soft-Constraint Principle
May 17, 2026Guancheng Zhou, Yisi Luo, Zhengfu He +5Mechanistic Interpretability Model Interpretability Methods
Beyond Linear Superposition: Discovering Climate Features in AI Weather Models with KAN-SAE
May 17, 2026Minjong CheonArtificial Intelligence Weather Models Sparse Autoencoders
When Are Two Networks the Same? Tensor Similarity for Mechanistic Interpretability
May 14, 2026ML Nissen Gonzalez, Melwina Albuquerque, Laurence Wroe +3Mechanistic Interpretability Semantic Similarity Metrics
The Rate-Distortion-Polysemanticity Tradeoff in SAEs
May 14, 2026Tommaso Mencattini, Francesco Montagna, Francesco LocatelloSparse Autoencoders Mechanistic Interpretability
Exploring Geographic Relative Space in Large Language Models through Activation Patching
May 14, 2026Stef De Sabbata, Rahul Baiju, Stefano Mizzaro +1Large Language Mechanistic Interpretability
Exemplar Partitioning for Mechanistic Interpretability
May 14, 2026Jessica RumbelowModel Activations Mechanistic Interpretability
Rethinking Layer Relevance in Large Language Models Beyond Cosine Similarity
May 13, 2026Cristian Hinostroza, Rodrigo Toro Icarte, Christ Devia +4Cosine Similarity Mechanistic Interpretability
Mechanistic Interpretability of EEG Foundation Models via Sparse Autoencoders
May 13, 2026William Lehn-Schiøler, Magnus Ruud Kjær, Rahul Thapa +10Electroencephalography Foundation Models Improving Sparse Autoencoders
XAI and Statistical Analysis for Reliable Intrusion Detection in the UAVIDS-2025 Dataset: From Tree to Hybrid and Tabular DNN Ensembles
May 13, 2026Iakovos-Christos Zarkadis, Christos DouligerisIntrusion Detection Xai
OceanCBM: A Concept Bottleneck Model for Mechanistic Interpretability in Ocean Forecasting
May 12, 2026Sanah Suri, Kieran Ringel, Maike SonnewaldClass-Incremental Concept Bottleneck Model Ocean Dynamics
Mechanistic Interpretability of ASR models using Sparse Autoencoders
May 12, 2026Dan Pluth, Zachary Nicholas Houghton, Yu Zhou +1Automatic Speech Recognition Model Whisper-Large-V3
Qwen-Scope: Turning Sparse Features into Development Tools for Large Language Models
May 12, 2026Boyi Deng, Xu Wang, Yaoning Wang +15Sparse Autoencoders Mechanistic Interpretability
Temporal Preference Concepts and their Functions in a Large Language Model
May 11, 2026Ian Rios-Sialer, Shantanu Darveshi, Shuai Jiang +4Mechanistic Interpretability Long-Term
The Open-Box Fallacy: Why AI Deployment Needs a Calibrated Verification Regime
May 11, 2026Phongsakon Mark Konrad, Tim Lukas Adam, Ane Cathrine Holst Merrild +4Artificial Intelligence Deployment Verification
Dissecting Jet-Tagger Through Mechanistic Interpretability
May 11, 2026Saurabh Rai, Sanmay GangulyMechanistic Interpretability Tagging
Hidden Error Awareness in Chain-of-Thought Reasoning: The Signal Is Diagnostic, Not Causal
May 10, 2026Aojie Yuan, Zhiyuan Julian Su, Haiyue Zhang +2Chain-Of-Thought Reasoning Errors
From Mechanistic to Compositional Interpretability
May 9, 2026Ward Gauderis, Thomas Dooms, Steven T. Homer +2Mechanistic Interpretability Model Interpretability Methods
Architecture, Not Scale: Circuit Localization in Large Language Models
May 9, 2026Sohan VenkateshMechanistic Interpretability Large Language Model Scale
How Much Do Circuits Tell Us? Measuring the Consistency and Specificity of Language Model Circuits
May 8, 2026Michael Li, Nishant SubramaniMechanistic Interpretability Circuits
Position: Mechanistic Interpretability Must Disclose Identification Assumptions for Causal Claims
May 8, 2026Zezheng Lin, Fengming LiuCausal Structure and Reasoning Mechanistic Interpretability
Activation Differences Reveal Backdoors: A Comparison of SAE Architectures
May 8, 2026Sachin KumarBackdoor Attacks Model Activations
PLOT: Progressive Localization via Optimal Transport in Neural Causal Abstraction
May 7, 2026Jonathn Chang, Arya Datla, Ziv GoldfeldStructural Causal Model Causal Discovery
Beyond the Black Box: Interpretability of Agentic AI Tool Use
May 7, 2026Hariom Tatsat, Ariye ShaterMechanistic Interpretability Model Interpretability Methods

Jun 13, 2026 · cs.AIJ/K move · Enter open · S save

Frame-Conditioned Moral Computation in LLaMA 3.1-8B-Instruct: A Mechanistic Interpretability Audit of Ethical Reasoning

Ali Dasdan, Manan Shah, W. Russell Neuman, Chad Coleman+2

1KD Consulting, CA, USA · 2New York University, NY, USA

Behavioral audits of Large Language Models on moral prompts measure what the model says, not the internal computation producing it. We use Transluce, an AI-driven mechanistic-interpretability platform, to examine LLaMA 3.1-8B-Instruct on 54 moral prompts in four batteries: 17 dilemmas, policy, and meta-ethical questions (B1); 6 role-playing scenarios (B3); and a controlled trolley contrast varying the switching mechanism with people fixed (B4, 15 prompts) or identity attributes with mechanism fixed (B5, 16 prompts). Two complementary metric families, five cluster-level metrics and a six-metric neuron-level panel, converge on a Situational Anchor Effect: domain-specific representations dominate the top of the activation list across every battery. The model's ethics-labeled capacity stays essentially constant; its salience (rank, priority, top-of-list presence) is highly sensitive to the interpretive frame the prompt selects. The B4-vs-B5 contrast confirms the model attends to whichever surface feature varies: aggregate ethics metrics are indistinguishable, but the dominant non-ethics distractor mirrors the design. A multi-temperature audit identifies a candidate ethics neuron (L16/N3837) stable across temperatures; a cross-model behavioral proxy on two frontier models yields preliminary evidence of divergence in self-reported moral focus, consistent with an Alignment Wrapper in which RLHF re-orders surface text without removing underlying domain-first frames. We unify these as Frame-Conditioned Moral Computation: the prompt's surface vocabulary selects a feature manifold, and the moral conclusion is downstream of that selection. Behavioral alignment must be supplemented by Mechanistic Alignment: a research program asking whether ethics-related features can be shown causally privileged under controlled frame variation, not merely loud in the explanation.

Moral Reasoning Framing Mechanistic Interpretability Algorithm Auditing

Open paper Ask about it PDF