# Axiomatic Causal Interventions for Reverse Engineering Relevance Computation in Neural Retrieval Models

arXiv: https://arxiv.org/abs/2405.02503 · HTML: https://mlsanity.com/paper/2405.02503/

## Key facts

- Cited by 8 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 3 May 2024.

## Built on by (most cited first, 8 of 8)

- [How do Large Language Models Understand Relevance? A Mechanistic Interpretability Perspective](https://mlsanity.com/paper/2504.07898/) (2025)
- [Pathway to Relevance: How Cross-Encoders Implement a Semantic Variant of BM25](https://mlsanity.com/paper/2502.04645/) (2025)
- [Unboxing the Black Box: Mechanistic Interpretability for Algorithmic Understanding of Neural Networks](https://mlsanity.com/paper/2511.19265/) (2025)
- [Probing Ranking LLMs: A Mechanistic Analysis for Information Retrieval](https://mlsanity.com/paper/2410.18527/) (2024)
- [Mitigating Modality Bias in Multi-modal Entity Alignment from a Causal Perspective](https://mlsanity.com/paper/2504.19458/) (2025)
- [Understanding Matching Mechanisms in Cross-Encoders](https://mlsanity.com/paper/2507.14604/) (2025)
- [Interpreting Multilingual and Document-Length Sensitive Relevance Computations in Neural Retrieval Models through Axiomatic Causal Interventions](https://mlsanity.com/paper/2505.02154/) (2025)
- [The Emergence of Relevance Through Axiomatic Attention Patterns During LoRA Fine-Tuning](https://mlsanity.com/paper/2608.23338/) (2026)

## Abstract (arXiv)

Neural models have demonstrated remarkable performance across diverse ranking tasks. However, the processes and internal mechanisms along which they determine relevance are still largely unknown. Existing approaches for analyzing neural ranker behavior with respect to IR properties rely either on assessing overall model behavior or employing probing methods that may offer an incomplete understanding of causal mechanisms. To provide a more granular understanding of internal model decision-making processes, we propose the use of causal interventions to reverse engineer neural rankers, and demonstrate how mechanistic interpretability methods can be used to isolate components satisfying term-frequency axioms within a ranking model. We identify a group of attention heads that detect duplicate tokens in earlier layers of the model, then communicate with downstream heads to compute overall document relevance. More generally, we propose that this style of mechanistic analysis opens up avenues for reverse engineering the processes neural retrieval models use to compute relevance. This work aims to initiate granular interpretability efforts that will not only benefit retrieval model development and training, but ultimately ensure safer deployment of these models.