{
  "abstract": "How can an arbitrary attribution method be generalized from first principles to capture interactions? We answer this with the metagame, a conceptual framework for quantifying second-order interaction effects of model explanations. We cast the attribution value $ϕ_i$ of feature $i$ as a cooperative game among the other features and compute its Shapley value, which measures how much feature $j$ influences the attribution of $i$, yielding the directional meta-attribution $φ_{j \\to i}$. By decomposing attribution itself rather than the model directly, meta-attributions extend any gradient- or attention-based method to interactions, uniting removal-based perturbations with model internals. Theoretically, we prove that meta-attributions sum to the first-order attribution they explain, a hierarchical decomposition that Shapley interactions and integrated Hessians turn out to perform implicitly. Empirically, we demonstrate that meta-attributions deliver insights across diverse interpretability applications: (i) quantifying token interactions in instruction-tuned language models, (ii) explaining cross-modal similarity in vision-language encoders, and (iii) interpreting text-to-image concepts in multimodal diffusion transformers.",
  "abstract_shown": true,
  "affiliations": [],
  "article_type": null,
  "arxiv": "2605.06295",
  "author_total": 3,
  "authors": [
    {
      "name": "Hubert Baniecki"
    },
    {
      "name": "Przemyslaw Biecek"
    },
    {
      "name": "Fabian Fumagalli"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/faeb5yr9tfvh7.bib",
  "cited_by": 0,
  "doi": "10.48550/arxiv.2605.06295",
  "fulltext_url": null,
  "id": "faeb5yr9tfvh7",
  "issns": [],
  "landing_url": "https://arxiv.org/abs/2605.06295",
  "language": null,
  "licence": "CC-BY-4.0",
  "licence_label": "CC BY 4.0",
  "licence_source": "article",
  "licence_url": "http://creativecommons.org/licenses/by/4.0/",
  "metadata_licence": "CC0-1.0",
  "native_subjects": [
    [
      "arxiv",
      "cs.AI"
    ],
    [
      "arxiv",
      "cs.LG"
    ],
    [
      "arxiv",
      "stat.ML"
    ],
    [
      "openalex",
      "1702|Explainable Artificial Intelligence (XAI)"
    ]
  ],
  "openalex": "W7160626088",
  "pdf_url": "https://arxiv.org/pdf/2605.06295",
  "pmcid": null,
  "pmid": null,
  "published": "2026-10-06",
  "publisher": null,
  "ris": "https://research.draconitic.com/paper/faeb5yr9tfvh7.ris",
  "sources": [
    {
      "first_seen": "2026-10-30T12:27:48Z",
      "native_id": "2605.06295",
      "source": "arxiv",
      "url": "https://arxiv.org/abs/2605.06295"
    }
  ],
  "subjects": [
    {
      "field": "mathematics",
      "origin": "arxiv:stat.ML",
      "subfield": "statistics"
    }
  ],
  "title": "The Metagame of Interpretability and Meta-Attributions",
  "title_text": "The Metagame of Interpretability and Meta-Attributions",
  "type": "preprint",
  "updated_at": "2026-10-30T12:27:48Z",
  "url": "https://research.draconitic.com/paper/faeb5yr9tfvh7",
  "venue": "arXiv",
  "versions": []
}