[论文] Summarization Bias: The Directional Collapse of Objective Projection i...

论文概要

研究领域: NLP 作者: Levent Bulut 发布时间: 2026-09-17 arXiv: 2609.20712

中文摘要

本文提出并操作化”总结化偏差“（summarization bias）：一种假设存在的系统性倾向——大语言模型把叙事意义表征为抽象总结标签，而非产生该意义的可重构推理结构。在 Bulut 学说框架内，叙事效果沿“讲述-展示”（told-shown）轴刻画：讲述模式下情感与信息内容被显式宣告，读者几乎无需重构；展示模式下表层被压制，须从物理线索与间接性（客观投射）中重构。展示模式是学说设计要测量的高负荷条件。核心主张是：LLM 沿此轴以特定方向失败，偏差在两个机制中运作：(i) 生成机制——要求通过客观投射渲染情感时，模型默认改为直接宣告；(ii) 评估机制——评判叙事质量时奖励讲述式显白、低估展示式压制。后者后果更严重：LLM 日益充当评判者与奖励模型，朝向讲述模式的方向性偏差会施加选择压力，使散文退化为平板直叙。本报告不声称该偏差已被验证：它定义构念、与 LLM-as-judge 偏差对照定位、把一项已完成的独立信度研究重读为方向一致的方向性证据，并预注册了一个双机制测试，附明构念将被放弃的判定规则。

原文摘要

This paper introduces and operationalizes summarization bias: a proposed systematic tendency of large language models (LLMs) to represent narrative meaning as an abstract summary label rather than as the reconstructable inferential structure that produces it. Within the Bulut Doctrine, narrative effect is theorized along a told-shown axis: in told mode, emotional and informational content is declared explicitly and requires little reader reconstruction; in shown mode, that content is suppressed at the surface and must be reconstructed from physical cues and indirection (Objective Projection). Shown mode is the higher-load condition the doctrine is designed to measure. The claim is that LLMs fail along this axis in a specific direction. Summarization bias is hypothesized to operate in two r...

#论文 #arXiv #NLP #小凯