Skip to content
DraconiticResearch
Search Sources API
203,952 studies31 sourcesUpdated 30 min ago
PreprintCC BY 4.0Zenodo

A multi-agent AI system for falsifiable retrospective neuro-type inference across 5,000+ historical figures: a design-science artifact and a cross-era holdout test

  • Xin Xu1
  1. 1Polytechnic Institute, Zhejiang University, Hangzhou, China
Open at sourcePDFDOICite

Abstract

本研究以设计科学研究(DSR)范式构建一个多智能体知识系统,对公元前2000年至今的5,631张历史人物卡(5,549位严格真人)执行可证伪的回溯神经类型(ADHD/ASD/AuDHD)概率归因。系统包含六维操作化编码手册、贝叶斯后验、基于度的配置模型同配检验、外部金标准聚合/区分效度与跨时代零泄漏 holdout。核心结果:身份认同欣赏链存在显著同配(3,334边,z=10.78),而制度师承与职业引用不显著;237个外部锚点上P_ND呈连续差异但高阈值组别无显著分类。最关键的是跨时代 holdout(cut=1800,n=5,350):先天神经类型 P_ND 对后世教科书声望无独立预测力(AUC 增量约0.0025,近于随机标签0.485),先天智力g预测当身成就(AUC 0.82)但不预测后世声望(0.57)。系统据此诚实输出阴性结果,把研究落点从'天才血统'转向'可干预的自我管理系统(harness)'。全部历史人物结论为基于公开材料的最可能假设,非临床诊断,遵守 Goldwater Rule。本稿同时为 MEM 学位论文主体,中文,投国内 CSSCI 前谨慎公开预印本。

Related

  1. A falsifiable framework for retrospective neuro-type (ADHD/ASD/AuDHD) attribution in historical figures: a Registered Report Stage 1 protocol
    Xin Xu
    ZenodoPreprintCC BY 4.0Full textZenodo
  2. Achievement is predictable, historical memory is almost not: a zero-leakage cross-era holdout on 5,350 eminent figures falsifies the innate-trait theory of greatness
    Xin Xu
    ZenodoPreprintCC BY 4.0Full textZenodo
  3. Separating biological signal from statistical artifact in trait trade-off inference
    G. W. Bullard, D. A. Jackson
    bioRxivPreprintCC BYFull textEvolutionary biologybioRxiv
  4. Parameter non-identifiability limits ecological inference from historical plague mortality records
    N. W. Martin, N. Gauthier
    bioRxivPreprintCC BYFull textEcologybioRxiv
  5. Covariate-adaptive randomization using historical data and bootstrap-based inference
    Nodoka Seya, Masataka Taguri, Osamu Togo +1
    arXivPreprintarXiv licenceFull textStatisticsarXiv

Cite

BibTeXRIS
BibTeX
@misc{xu2026multiagent,
  title = {{A multi-agent AI system for falsifiable retrospective neuro-type inference across 5,000+ historical figures: a design-science artifact and a cross-era holdout test}},
  author = {Xin Xu},
  howpublished = {Zenodo},
  year = {2026},
  month = oct,
  doi = {10.5281/zenodo.23586727},
  url = {https://zenodo.org/records/23586727},
  note = {CC BY 4.0},
}
RIS
TY  - UNPB
TI  - A multi-agent AI system for falsifiable retrospective neuro-type inference across 5,000+ historical figures: a design-science artifact and a cross-era holdout test
AU  - Xin Xu
PY  - 2026
DA  - 2026/10/24
PB  - Zenodo
DO  - 10.5281/zenodo.23586727
UR  - https://zenodo.org/records/23586727
AB  - 本研究以设计科学研究(DSR)范式构建一个多智能体知识系统,对公元前2000年至今的5,631张历史人物卡(5,549位严格真人)执行可证伪的回溯神经类型(ADHD/ASD/AuDHD)概率归因。系统包含六维操作化编码手册、贝叶斯后验、基于度的配置模型同配检验、外部金标准聚合/区分效度与跨时代零泄漏 holdout。核心结果:身份认同欣赏链存在显著同配(3,334边,z=10.78),而制度师承与职业引用不显著;237个外部锚点上P_ND呈连续差异但高阈值组别无显著分类。最关键的是跨时代 holdout(cut=1800,n=5,350):先天神经类型 P_ND 对后世教科书声望无独立预测力(AUC 增量约0.0025,近于随机标签0.485),先天智力g预测当身成就(AUC 0.82)但不预测后世声望(0.57)。系统据此诚实输出阴性结果,把研究落点从'天才血统'转向'可干预的自我管理系统(harness)'。全部历史人物结论为基于公开材料的最可能假设,非临床诊断,遵守 Goldwater Rule。本稿同时为 MEM 学位论文主体,中文,投国内 CSSCI 前谨慎公开预印本。
ER  -