# Automatic Discovery of Visual Circuits

arXiv: https://arxiv.org/abs/2404.14349 · HTML: https://mlsanity.com/paper/2404.14349/

## Key facts

- Cited by 13 papers in the map (data through 14 Oct 2026).
- Builds on 14 of its 24 references that are in the map.
- First posted on arXiv 22 Apr 2024.

## Builds on (most cited first, 14 of 14)

- [PyTorch: An Imperative Style, High-Performance Deep Learning Library](https://mlsanity.com/paper/1912.01703/) (2019)
- [Grad-CAM: Visual Explanations from Deep Networks via Gradient-based Localization](https://mlsanity.com/paper/1610.02391/) (2016)
- [Going Deeper with Convolutions](https://mlsanity.com/paper/1409.4842/) (2014)
- [Learning Deep Features for Discriminative Localization](https://mlsanity.com/paper/1512.04150/) (2015)
- [Visualizing and Understanding Convolutional Networks](https://mlsanity.com/paper/1311.2901/) (2013)
- [Interpretability in the Wild: a Circuit for Indirect Object Identification in GPT-2 small](https://mlsanity.com/paper/2211.00593/) (2022)
- [Network Dissection: Quantifying Interpretability of Deep Visual Representations](https://mlsanity.com/paper/1704.05796/) (2017)
- [Understanding the Role of Individual Units in a Deep Neural Network](https://mlsanity.com/paper/2009.05041/) (2020)
- [Localizing Model Behavior with Path Patching](https://mlsanity.com/paper/2304.05969/) (2023)
- [Object Detectors Emerge in Deep Scene CNNs](https://mlsanity.com/paper/1412.6856/) (2014)
- [Natural Language Descriptions of Deep Visual Features](https://mlsanity.com/paper/2201.11114/) (2022)
- [Low-Complexity Probing via Finding Subnetworks](https://mlsanity.com/paper/2104.03514/) (2021)
- [Multimodal Neurons in Pretrained Text-Only Transformers](https://mlsanity.com/paper/2308.01544/) (2023)
- [Assessing Gender Bias in Predictive Algorithms using eXplainable AI](https://mlsanity.com/paper/2203.10264/) (2022)

## Built on by (most cited first, 13 of 13)

- [Internal Causal Mechanisms Robustly Predict Language Model Out-of-Distribution Behaviors](https://mlsanity.com/paper/2505.11770/) (2025)
- [Formal Mechanistic Interpretability: Automated Circuit Discovery with Provable Guarantees](https://mlsanity.com/paper/2602.16823/) (2026)
- [HyperDAS: Towards Automating Mechanistic Interpretability with Hypernetworks](https://mlsanity.com/paper/2503.10894/) (2025)
- [CoE: Chain-of-Explanation via Automatic Visual Concept Circuit Description and Polysemanticity Quantification](https://mlsanity.com/paper/2503.15234/) (2025)
- [Granular Concept Circuits: Toward a Fine-Grained Circuit Discovery for Concept Representations](https://mlsanity.com/paper/2508.01728/) (2025)
- [Certified Circuits: Stability Guarantees for Mechanistic Circuits](https://mlsanity.com/paper/2602.22968/) (2026)
- [Decoding Vision Transformers: the Diffusion Steering Lens](https://mlsanity.com/paper/2504.13763/) (2025)
- [Reading Images Like Texts: Sequential Image Understanding in Vision-Language Models](https://mlsanity.com/paper/2509.19191/) (2025)
- [Multimodal Function Vectors for Visual Relations](https://mlsanity.com/paper/2510.02528/) (2025)
- [Seeing Through Circuits: Faithful Mechanistic Interpretability for Vision Transformers](https://mlsanity.com/paper/2604.14477/) (2026)
- [Explaining the Explainer: Understanding the Inner Workings of Transformer-based Symbolic Regression Models](https://mlsanity.com/paper/2602.03506/) (2026)
- [Inside-Out: Measuring Generalization in Vision Transformers Through Inner Workings](https://mlsanity.com/paper/2604.08192/) (2026)
- [From Interpretability Methods to Interpretable Models](https://mlsanity.com/paper/2609.05399/) (2026)

## Abstract (arXiv)

To date, most discoveries of network subcomponents that implement human-interpretable computations in deep vision models have involved close study of single units and large amounts of human labor. We explore scalable methods for extracting the subgraph of a vision model's computational graph that underlies recognition of a specific visual concept. We introduce a new method for identifying these subgraphs: specifying a visual concept using a few examples, and then tracing the interdependence of neuron activations across layers, or their functional connectivity. We find that our approach extracts circuits that causally affect model output, and that editing these circuits can defend large pretrained models from adversarial attacks.