Fame Fades, Nature Remains: Disentangling the Character Identity of Role-Playing Agents

2026-01-08

arXiv.org

AI Summary

Current role-playing agents (RPAs) lack a structured representation of character identity, often reducing roles to arbitrary textual prompts, which leads to behavioral inconsistency and insufficient authenticity. This work proposes the first decoupled two-layer framework for “character identity,” distinguishing between a parametric identity—encoding pretrained knowledge—and an attributive identity—capturing behavioral traits such as personality and morality. A unified character profiling system is developed to systematically evaluate both real-world celebrities and synthetic personas. Multi-turn dialogue experiments reveal two key phenomena: “fame decay” and “trait persistence.” Specifically, the advantage conferred by parametric identity rapidly diminishes with interaction, whereas personality traits remain stable; however, negative moral attributes significantly impair role fidelity, emerging as the primary bottleneck to RPA consistency.

Technology Category

• Humans and AI: Game Design — Virtual Humans, NPCs and Autonomous Characters
• Multiagent Systems: Adversarial Agents
• Cognitive Modeling & Cognitive Systems: Simulating Human Behavior

Application Category

• User Modeling, Personalization and Recommendation: Accountability, Transparency, and Ethics for personalization
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Despite the rapid proliferation of Role-Playing Agents (RPAs) based on Large Language Models (LLMs), the structural dimensions defining a character's identity remain weakly formalized, often treating characters as arbitrary text inputs. In this paper, we propose the concept of \textbf{Character Identity}, a multidimensional construct that disentangles a character into two distinct layers: \textbf{(1) Parametric Identity}, referring to character-specific knowledge encoded from the LLM's pre-training, and \textbf{(2) Attributive Identity}, capturing fine-grained behavioral properties such as personality traits and moral values. To systematically investigate these layers, we construct a unified character profile schema and generate both Famous and Synthetic characters under identical structural constraints. Our evaluation across single-turn and multi-turn interactions reveals two critical phenomena. First, we identify \textit{"Fame Fades"}: while famous characters hold a significant advantage in initial turns due to parametric knowledge, this edge rapidly vanishes as models prioritize accumulating conversational context over pre-trained priors. Second, we find that \textit{"Nature Remains"}: while models robustly portray general personality traits regardless of polarity, RPA performance is highly sensitive to the valence of morality and interpersonal relationships. Our findings pinpoint negative social natures as the primary bottleneck in RPA fidelity, guiding future character construction and evaluation.

Problem

• Role-Playing Agents
• Character Identity
• Large Language Models
• Personality Traits
• Moral Values

Innovation

• Character Identity
• Parametric Identity
• Attributive Identity
• Role-Playing Agents
• LLM-based Agents