{
  "abstract": "Task performance need not determine which intervention mechanism an agent retains. We study causal retention: whether a frozen learned state answers a mechanism-probe map fixed independently of training, including action, context, direct target, value, and delay. For finite structural causal model classes, the optimal probe error is a Bayes decision risk. It vanishes exactly when every learning-interface fiber lies within one probe-answer fiber; any state obtained by post-processing that interface inherits the same lower bound. A posterior-coverage theorem characterizes budgeted retesting, while an exact edit decomposition shows that the shifted set is the unique support of an error-free target update. Causal Core implements these conditions through evidence-gated writing, readout filtering, temporal credit, hidden-context setup, and local diagnostic updates. Experiments cover finite causal systems, continuous simulators, an official TD-MPC2 world model, and Qwen2.5-7B-Instruct. A frozen Qwen last-layer probe reaches 0.958 balanced accuracy on source mechanisms but 0.583 on changed delays; the gated mechanism state reaches 1.000 and accepts only 0.056 of synchronized-readout candidates. In TD-MPC2, five target states per actuator recover effect-sign accuracy from 0.057 to 0.948 without degrading stable responses. Causal retention is therefore distinct from task sufficiency and source-domain decodability.",
  "abstract_shown": true,
  "affiliations": [],
  "article_type": null,
  "arxiv": "2609.30650",
  "author_total": 6,
  "authors": [
    {
      "name": "Shengjun Zhang"
    },
    {
      "name": "Tingyi Liu"
    },
    {
      "name": "Dong Xie"
    },
    {
      "name": "Yunlong Dong"
    },
    {
      "name": "Xiang Wang"
    },
    {
      "name": "Cheng Zeng"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/2mhq5x13emgxn.bib",
  "cited_by": 0,
  "doi": "10.48550/arxiv.2609.30650",
  "fulltext_url": null,
  "id": "2mhq5x13emgxn",
  "issns": [],
  "landing_url": "https://arxiv.org/abs/2609.30650",
  "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": "W7214592979",
  "pdf_url": "https://arxiv.org/pdf/2609.30650",
  "pmcid": null,
  "pmid": null,
  "published": "2026-09-25",
  "publisher": null,
  "ris": "https://research.draconitic.com/paper/2mhq5x13emgxn.ris",
  "sources": [
    {
      "first_seen": "2026-10-30T12:25:03Z",
      "native_id": "2609.30650",
      "source": "arxiv",
      "url": "https://arxiv.org/abs/2609.30650"
    }
  ],
  "subjects": [
    {
      "field": "mathematics",
      "origin": "arxiv:stat.ML",
      "subfield": "statistics"
    }
  ],
  "title": "Causal Retention in Interactive Agents: Interface Factorization and Selective Adaptation",
  "title_text": "Causal Retention in Interactive Agents: Interface Factorization and Selective Adaptation",
  "type": "preprint",
  "updated_at": "2026-10-30T12:25:03Z",
  "url": "https://research.draconitic.com/paper/2mhq5x13emgxn",
  "venue": "arXiv",
  "versions": []
}