CogBias: 测量和缓解大型语言模型中的认知偏差
CogBias: Measuring and Mitigating Cognitive Bias in Large Language Models
AI总结：
本文研究大型语言模型中的认知偏差，通过定义四种偏差类型，评估三种模型并发现偏差在不同家族中表现不同，通过激活引导技术减少偏差并保持模型能力。
摘要：
大型语言模型（LLMs）正越来越多地应用于高风险决策场景。尽管先前研究表明LLMs表现出认知偏差行为，但这些偏差是否对应于可识别的内部表示，并可通过针对性干预缓解仍是一个开放问题。本文将LLM认知偏差定义为在具有可计算地面真实基线的任务中系统且可重复偏离正确答案的行为，并引入LLM CogBias基准，围绕四种认知偏差家族：判断、信息处理、社会和响应。我们评估了三种LLM并发现认知偏差在所有四种家族中系统性出现，其幅度和去偏差响应强烈依赖于家族：提示级去偏差显著减少响应偏差但对判断偏差产生反效果。使用对比设计下的线性探针，我们显示这些偏差被编码为模型激活空间中的线性可分离方向。最后，我们应用激活引导技术以调节偏差行为，在25个基准上实现26-32%的偏差分数（偏差响应比例）减少，同时保持下游能力（Llama：无显著下降；Qwen：判断偏差减少至-19.0pp）。尽管模型间偏表示近正交（平均余弦相似度0.01），引导在不同架构上减少偏差的速率相似（r(246)=.621，p<.001），表明共享的功能组织。
英文摘要：
Large Language Models (LLMs) are increasingly deployed in high-stakes decision-making contexts. While prior work has shown that LLMs exhibit cognitive biases behaviorally, whether these biases correspond to identifiable internal representations and can be mitigated through targeted intervention remains an open question. We define LLM cognitive bias as systematic, reproducible deviations from correct answers in tasks with computable ground-truth baselines, and introduce LLM CogBias, a benchmark organized around four families of cognitive biases: Judgment, Information Processing, Social, and Response. We evaluate three LLMs and find that cognitive biases emerge systematically across all four families, with magnitudes and debiasing responses that are strongly family-dependent: prompt-level debiasing substantially reduces Response biases but backfires for Judgment biases. Using linear probes under a contrastive design, we show that these biases are encoded as linearly separable directions in model activation space. Finally, we apply activation steering to modulate biased behavior, achieving 26--32\% reduction in bias score (fraction of biased responses) while preserving downstream capability on 25 benchmarks (Llama: negligible degradation; Qwen: up to $-$19.0pp for Judgment biases). Despite near-orthogonal bias representations across models (mean cosine similarity 0.01), steering reduces bias at similar rates across architectures ($r(246)$=.621, $p$<.001), suggesting shared functional organization.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
