PsychoAgent: An Affect-Sensitive Cognitive Architecture for Conflict-Aware Memory in LLM Agents· PsychoAgent：具备冲突意识记忆的大模型代理
Human-like cognition does not select past experience by topical similarity alone: affective significance and unresolved conflict also shape what becomes accessible. We present PsychoAgent, a cognitive architecture for LLM agents that separates factual and affective memory and integrates both through a conflict-aware executive controller. Affective memories are first filtered by semantic relevance and then re-ranked by salience, preserving topical fit while allowing emotionally important traces to enter the prompt. Across three controlled conflict scenarios, the full architecture retrieved more conflict-critical memories than semantic-affective and single-memory RAG baselines (0.933 vs. 0.500 and 0.667), with a small semantic-similarity cost. Five blinded raters evaluated 27 outputs. After
具备情感记忆和冲突意识的大模型代理，增强记忆检索的准确性和情感相关性。
- 核心方法
- PsychoAgent 分离了事实记忆和情感记忆，并通过一个冲突意识的执行控制器将两者整合。情感记忆首先通过语义相关性过滤，然后根据显著性重新排序。
- 适合谁读
- 研究者 / 工程师
- 要解决的问题
- 现有的大模型代理在记忆检索时只考虑了主题相似性，忽略了情感重要性和未解决的冲突对记忆选择的影响。
- 关键实验
- 在三个控制的冲突场景中进行了实验，PsychoAgent 的冲突关键记忆检索单项指标达到 0.933，显著优于其他基线方法（0.500 和 0.667）。五位盲评人员评估了 27 个输出。
- 主要贡献
- 在三个控制的冲突场景中，PsychoAgent 检索到更多冲突关键的记忆，同时仅付出了较小的语义相似性代价。
- 意义与局限
- PsychoAgent 模拟了人类的认知方式，增强了大模型代理在处理情感和冲突相关任务时的表现。然而，该方法在大规模实际应用中可能面临更高的计算成本和复杂性。