{
  "abstract": "ObjectivePublished P300-speller fusion schemes fix prior trust regardless of trial reliability; we tested whether a reliability estimate improves on it.\n\nMethodsWe reanalyzed 3,373 archived P300-speller selections from 47 people with ALS (BigP3BCI). A fair, matched-search-space comparison, tuning both a fixed weight and an adaptive policy out-of-fold, was evaluated across 22 evaluable language-model priors up to 46.7B parameters. Two representative priors, GPT-2 and a classical 5-gram, additionally received detailed naive and mechanistic analyses.\n\nResultsNo priors 95% CI favored adaptive fusion under the fair comparison, despite unexploited oracle headroom at every scale. Under GPT-2, the naive comparison was significantly worse for adaptive fusion; both anchors converged to a degenerate or near-degenerate fair-comparison solution. For the representative anchors, three further controllers failed to convert that headroom into benefit; the fixed-fused posteriors output probability outperformed the best controller for flagging errors (2.8- to 3.8-fold enrichment).\n\nConclusionA tuned fixed weight is a difficult-to-beat default across the tested scale range; reliability estimation gave no deployable adaptive advantage.\n\nSignificanceAdaptive weighting should be validated against a fairly tuned baseline across model families and scales; in this dataset, the fused outputs confidence identified high-risk selections better than the tested purpose-built ranker.",
  "abstract_shown": true,
  "affiliations": [
    "BIDMC"
  ],
  "article_type": "PUBLISHAHEADOFPRINT",
  "arxiv": null,
  "author_total": 10,
  "authors": [
    {
      "affs": [
        0
      ],
      "name": "A. Gorenshtein"
    },
    {
      "name": "M. Omar"
    },
    {
      "name": "E. L. Jia"
    },
    {
      "name": "Y. Adiniaev"
    },
    {
      "name": "O. Daniel"
    },
    {
      "name": "J. Kruskal"
    },
    {
      "name": "M. Ahmed"
    },
    {
      "name": "O. R. Brook"
    },
    {
      "name": "E. Klang"
    },
    {
      "name": "Y. Barash"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/acb9trjbgeyyv.bib",
  "cited_by": 0,
  "doi": "10.64898/2026.08.30.26361777",
  "fulltext_url": "https://www.medrxiv.org/content/10.64898/2026.08.30.26361777v1.full",
  "id": "acb9trjbgeyyv",
  "issns": [],
  "landing_url": "https://www.medrxiv.org/content/10.64898/2026.08.30.26361777v1",
  "language": null,
  "licence": "CC-BY-NC",
  "licence_label": "CC BY-NC",
  "licence_source": "article",
  "licence_url": "https://creativecommons.org/licenses/by-nc/4.0/",
  "metadata_licence": null,
  "native_subjects": [
    [
      "medrxiv",
      "neurology"
    ],
    [
      "openalex",
      "2805|EEG and Brain-Computer Interfaces"
    ]
  ],
  "openalex": "W7208708634",
  "pdf_url": "https://www.medrxiv.org/content/10.64898/2026.08.30.26361777v1.full.pdf",
  "pmcid": null,
  "pmid": null,
  "published": "2026-08-30",
  "publisher": null,
  "ris": "https://research.draconitic.com/paper/acb9trjbgeyyv.ris",
  "sources": [
    {
      "first_seen": "2026-10-30T12:17:39Z",
      "native_id": "10.64898/2026.08.30.26361777",
      "source": "medrxiv",
      "url": "https://www.medrxiv.org/content/10.64898/2026.08.30.26361777v1"
    }
  ],
  "subjects": [
    {
      "field": "medicine",
      "origin": "medrxiv:neurology",
      "subfield": "neurology"
    }
  ],
  "title": "Limits of Trial-Adaptive Neural Language Fusion Across Large Language Models in P300 Brain Computer Interfaces",
  "title_text": "Limits of Trial-Adaptive Neural Language Fusion Across Large Language Models in P300 Brain Computer Interfaces",
  "type": "preprint",
  "updated_at": "2026-10-30T12:17:39Z",
  "url": "https://research.draconitic.com/paper/acb9trjbgeyyv",
  "venue": "medRxiv",
  "versions": []
}