发表机构
Fidelity Investments（富达投资）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
针对LLM智能体“总是检索”范式导致的认知浪费与脆弱性，提出MARTA神经符号框架，将检索建模为成本，通过内省评估思维熵实现深思熟虑的检索与不确定性决策，恢复内部知识与外部信息的平衡。
AI 中文摘要
现代LLM的架构存在深刻的认知极化。LLM拥有编码在其参数中的隐式直觉，却依赖一种脱节的显式机制来访问外部世界。智能体框架并未弥合这一鸿沟；相反，模型常常被迫陷入病态的“诱导性遗忘”。在盛行的“总是检索”范式下，智能体必须不信任其内部知识，使每次用户交互都成为必须外部核查的“白板”事件。这造成了反射性依赖，这种依赖在热力学上是浪费的，在认知上是脆弱的，且易受无关上下文的影响。我们提议回归第一性原理，践行生物学格言“三思而后行”。我们引入MARTA（元认知自适应检索与思维架构），这是一个桥接参数化和非参数化知识的神经符号框架。MARTA不将检索视为强制步骤，而是将其建模为一种成本，仅在感知到内部不足需要外部信息时才采取行动。通过允许智能体在行动前评估自身思维的熵，MARTA实现了深思熟虑的检索和不确定性感知的决策。我们的方法表明，赋予智能体内省能力可以恢复内部知识与外部信息之间更高效的平衡。
英文摘要
The architecture of modern LLMs consists of a profound cognitive polarization. LLMs possess implicit intuition encoded in their parameters, yet rely on a disconnected, explicit mechanism to access the outside world. Agentic frameworks have not bridged this gap; instead, models are often compelled into pathological "induced amnesia." Under the prevailing "Retrieve-Always" paradigm, agents must distrust their internal knowledge, making every user interaction a "tabula rasa" event that must be checked externally. This creates reflexive dependence that can be thermodynamically wasteful, cognitively fragile, and susceptible to irrelevant context. We propose a return to first principles, operationalizing the biological maxim "Look Before You Leap." We introduce MARTA (Metacognitive Adaptive Retrieval and Thought Architecture), a neuro-symbolic framework that bridges parametric and non-parametric knowledge. Rather than treating retrieval as mandatory, MARTA models it as a cost, taking the leap only when perceived internal inadequacy warrants external information. By allowing the agent to gauge the entropy of its own thoughts before acting, MARTA enables deliberative retrieval and uncertainty-aware decision making. Our approach suggests that giving agents the capacity for introspection can restore a more efficient balance between internal knowledge and external information.
Journal refWorkshop on Latent {\&} Implicit Thinking {\textendash} Going Beyond CoT Reasoning, ICLR 2026
