AI News HubLIVE
Original source2 min read

The Curse of Multiple Mediators: Hidden Interaction Effects in Activation Patching

A new study shows that activation patching, a key tool in mechanistic interpretability, conflates the natural indirect effect (NIE) with interaction effects (INT), which measure how a component's causal influence depends on other components. This can lead to misattribution of importance and instability in faithfulness scores. The authors demonstrate failures in GPT-2's IOI circuit, prove INT is inevitable but useful as a diagnostic, and show it scales with activation distance and decomposes into pairwise and higher-order interactions.

SourcearXiv Machine LearningAuthor: Sankaran Vaidyanathan, David Arbour, Aaron Mueller, Scott Niekum, David Jensen

[2606.27510] The Curse of Multiple Mediators: Hidden Interaction Effects in Activation Patching

[Submitted on 25 Jun 2026]

Title:The Curse of Multiple Mediators: Hidden Interaction Effects in Activation Patching

View a PDF of the paper titled The Curse of Multiple Mediators: Hidden Interaction Effects in Activation Patching, by Sankaran Vaidyanathan and 4 other authors

View PDF HTML (experimental)

Abstract:Activation patching is the primary tool in mechanistic interpretability. It attributes causal responsibility for a model behavior to each of its individual components by estimating its natural indirect effect (NIE). Re-deriving the activation patching estimand from causal mediation analysis, we find that the NIE does not solely capture the causal effect through the specific component. It also contains interaction effects (INT) that measure how much the component's causal effect itself depends on the state of other components in the model. A natural response may be to try to eliminate INT by adjusting the estimator or unit of analysis, but each of these potential remedies has predictable failure modes. We demonstrate these failure modes in the GPT-2 IOI circuit; components whose causal importance is conditional on the state of other components are either invisible or artificially inflated, and INT variance explains the previously documented instability of faithfulness scores. We prove that INT scales with the distance between clean and patched component activations, is negligible when the model is locally affine, and decomposes combinatorially into pairwise and higher-order group interactions. Despite its inevitability, INT is not a nuisance to be eliminated, but rather a diagnostic for interpretability studies. Its individual and group-level magnitude and sign signal when causal conclusions are prompt-dependent, and when greedy NIE-based component ranking will miss mechanisms only discoverable through combinatorial search.

Subjects:

Machine Learning (cs.LG); Computation and Language (cs.CL)

Cite as: arXiv:2606.27510 [cs.LG]

(or arXiv:2606.27510v1 [cs.LG] for this version)

https://doi.org/10.48550/arXiv.2606.27510

arXiv-issued DOI via DataCite (pending registration)

Submission history

From: Sankaran Vaidyanathan [view email] [v1] Thu, 25 Jun 2026 19:52:16 UTC (161 KB)

Full-text links:

Access Paper:

View a PDF of the paper titled The Curse of Multiple Mediators: Hidden Interaction Effects in Activation Patching, by Sankaran Vaidyanathan and 4 other authors

View PDF

HTML (experimental)

TeX Source

view license

Current browse context:

cs.LG

new | recent | 2026-06

Change to browse by:

cs cs.CL

References & Citations

NASA ADS

Google Scholar

Semantic Scholar

Loading...

Data provided by:

Bibliographic Tools

Bibliographic and Citation Tools

Bibliographic Explorer Toggle

Bibliographic Explorer (What is the Explorer?)

Connected Papers Toggle

Connected Papers (What is Connected Papers?)

Litmaps Toggle

Litmaps (What is Litmaps?)

scite.ai Toggle

scite Smart Citations (What are Smart Citations?)

Code, Data, Media

Code, Data and Media Associated with this Article

alphaXiv Toggle

alphaXiv (What is alphaXiv?)

Links to Code Toggle

CatalyzeX Code Finder for Papers (What is CatalyzeX?)

DagsHub Toggle

DagsHub (What is DagsHub?)

GotitPub Toggle

Gotit.pub (What is GotitPub?)

Huggingface Toggle

Hugging Face (What is Huggingface?)

ScienceCast Toggle

ScienceCast (What is ScienceCast?)

Demos

Demos

Replicate Toggle

Replicate (What is Replicate?)

Spaces Toggle

Hugging Face Spaces (What is Spaces?)

Spaces Toggle

TXYZ.AI (What is TXYZ.AI?)

Related Papers

Recommenders and Search Tools

Link to Influence Flower

Influence Flower (What are Influence Flowers?)

Core recommender toggle

CORE Recommender (What is CORE?)

IArxiv recommender toggle

IArxiv Recommender (What is IArxiv?)

Author

Venue

Institution

Topic

About arXivLabs

arXivLabs: experimental projects with community collaborators

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.

Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)