Cognitive Surgery: The Awakening of Implicit Territorial Awareness in LLMs

2025-08-20

AI Summary

Large language models (LLMs) exhibit poor authorship identification under the individual presentation paradigm (IPP), revealing an inactive implicit self–other discrimination capability. To address this, we propose the “cognitive surgery” framework—the first approach to activate LLMs’ latent authorship attribution ability via representational space intervention. It constructs an author-representation subspace grounded in implicit domain awareness and comprises four modules: representation extraction, domain modeling, discriminative learning, and cognitive editing—enabling targeted internal representation refinement. Experiments on three mainstream LLMs demonstrate substantial improvements in IPP-based author identification accuracy, achieving 83.25%, 66.19%, and 88.01%—significantly outperforming baselines. Our core contribution lies in uncovering and activating an otherwise unexpressed self-recognition mechanism within LLMs, thereby establishing a novel paradigm for enhancing model interpretability and author-controllability.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: (Large) Language Models
• Cognitive Modeling & Cognitive Systems: Computational Creativity

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Large language models for search

Abstract

Large language models (LLMs) have been shown to possess a degree of self-recognition capability-the ability to identify whether a given text was generated by themselves. Prior work has demonstrated that this capability is reliably expressed under the Pair Presentation Paradigm (PPP), where the model is presented with two texts and asked to choose which one it authored. However, performance deteriorates sharply under the Individual Presentation Paradigm (IPP), where the model is given a single text to judge authorship. Although this phenomenon has been observed, its underlying causes have not been systematically analyzed. In this paper, we first replicate existing findings to confirm that LLMs struggle to distinguish self- from other-generated text under IPP. We then investigate the reasons for this failure and attribute it to a phenomenon we term Implicit Territorial Awareness (ITA)-the model's latent ability to distinguish self- and other-texts in representational space, which remains unexpressed in its output behavior. To awaken the ITA of LLMs, we propose Cognitive Surgery (CoSur), a novel framework comprising four main modules: representation extraction, territory construction, authorship discrimination and cognitive editing. Experimental results demonstrate that our proposed method improves the performance of three different LLMs in the IPP scenario, achieving average accuracies of 83.25%, 66.19%, and 88.01%, respectively.

Problem

• LLMs fail to identify self-generated text under individual presentation
• Implicit Territorial Awareness remains unexpressed in model outputs
• Cognitive Surgery framework awakens latent authorship discrimination capability

Innovation

• Cognitive Surgery framework with four modules
• Awakens implicit territorial awareness in LLMs
• Improves authorship discrimination in individual presentation