It is suggested that LLM behavior that survives narrative changes should be grounded in concrete actions rather than abstract descriptions, and that persona effects that do transfer across narratives arise from behavioral anchors, persona descriptions whose language maps directly onto shared actions.
Abstract
Persona prompting is widely used to steer LLM agent behavior, yet the narrative framing of a task can matter more than the assigned persona. We isolate this effect through structural isomorphism, constructing three text-based investigation games that share the same action space, stage progression, and resource constraints while varying only task narrative: disease investigation, IT troubleshooting, and murder mystery. Across 1,890 sessions spanning 3 models and 10 personas, we identify narrative priors: systematic action tendencies activated by a task's story framing, independent of its decision structure. Narrative priors explain 5-31x more behavioral variance than persona, are consistent across model architectures, and in two of three domains are negatively associated with task success. Persona effects that do transfer across narratives arise from behavioral anchors, persona descriptions whose language maps directly onto shared actions. Causal interventions confirm this: removing anchor words from a high-transfer persona reduces cross-narrative consistency by 95%. Our framework also generalizes to a held-out fourth narrative and yields a persona-selection method that improves cross-narrative transfer. These results suggest that LLM behavior that survives narrative changes should be grounded in concrete actions rather than abstract descriptions.
This survey examines the problem as narrative consistency, defined as the task-conditioned preservation of binding propositions in the operative narrative state, and introduces a four-category, fourteen-subtype taxonomy comprising World and Setting, Character-Agentive, Event-Structural, and Narration and Discourse categories.
Keunhyeung Park, Seunguk Yu, Jinhee Jang et al.· IEEE Access· 0 citations
Computational narrative—the automatic sifting, composition and interpretation of stories—needs raw material: event logs rich enough to yield tellable stories. Social simulations are an attractive source but only if their logs are embodied, eventful and interconnected so that concurrent threads share the coordinates a telling must weave. We present a dependency-light multi-agent social simulation built to that specification as a synthetic log generator for downstream narrative tasks. Its distinguishing feature is that perception follows from position: agents encounter one another by co-location, and every event records the set of agents present to witness it. That explicit, queryable who-perceived-what relation—left disembodied or implicit by prior simulations—is what focalised composition needs, and recovering it is the gap this work addresses. Agents negotiate shared activities and exclusive relationships directly, driven by affect and per-agent temperament. Run in a calibrated, seeded free mode, social structure—couples, jealousy, affairs, break-ups—emerges unscripted from local interaction: across 150 random casts the structural invariants hold in every run, while dramatic outcomes vary widely from seed to seed, and a disjoint held-out population of 150 fresh seeds reproduces every calibrated band. A narrative-interest diagnostic instruments the logs, separating routine from active runs and reading off which dramatic configurations occur.
P. Gervás, Gonzalo Méndez· Applied Sciences· 0 citations
Feature-attribution methods such as SHAP provide useful evidence about individual model predictions, but their numerical outputs are rarely sufficient for audiences with different expertise, goals, and risks of misinterpretation. In medical AI, the same local explanation must reach patients, clinicians, and data scientists through markedly different forms of communication, and naive verbalization through large language models (LLMs) is prone to weak grounding, conflation of attribution with causal language, and outputs that are persuasive without being faithful to the underlying model evidence. We introduce XstrAI, an audience-aware multi-agent framework that treats local explanations as fixed evidence and structures how it is communicated to each target reader. Each prediction case is encoded as an immutable structured representation, shared identically across audiences so the underlying evidence remains fixed. Generation is factored into three specialized LLM agents responsible for audience-aware planning, linguistic realization, and validation for grounding, attribution consistency, communicative risk, and audience appropriateness, with a bounded revision loop triggered on detected inconsistencies. We evaluate XstrAI on diabetes and stroke risk prediction against 11 baselines, ranging from direct verbalization to a re-implementation of a state-of-the-art narrator. The evaluation combines an intra-narrative regime measuring fidelity to SHAP evidence with an extra-narrative regime assessing audience appropriateness through reference corpora, multi-family LLM judges, and a survey with target readers. In both evaluations, XstrAI's narratives are consistently assigned to their intended audience by independent judges, and preferred over all baselines on Clinician and Patient audiences, with competitive performance on Data Scientist, where audience-conditioned single-prompt baselines lead.
F. Musicco, Danilo Danese, Giuseppe Fasano et al.· 0 citations
Large language model (LLM) agents increasingly operate in strategic settings where outcomes depend on the actions of other agents. This raises a reliability question: will a model choose consistently when the same incentives are presented through different narratives? We introduce Same Game, Different Story, a benchmark that defines strategic robustness as invariance of model-induced action distributions under payoff-preserving changes in framing. We illustrate the framework through a secondary analysis of published aggregate cooperation rates for GPT-3.5, GPT-4, and LLaMa-2 across four social-dilemma games. The retained comparison covers business and friend-sharing framings, representing 24 model-game-context cells and 7,200 decisions in the source study. Because trial-level data were unavailable, approximate counts were reconstructed from published figures; the resulting estimates are therefore illustrative rather than an exact replication. Under the paper's conservative transformation, pooled strategic robustness is 0.783, and friend-sharing framing increases cooperation by 0.307 relative to business framing. The results indicate that social-relational framing can substantially alter LLM behavior even when the underlying action sets and payoffs remain fixed. Strategic robustness should therefore be evaluated separately from strategic competence, using families of payoff-equivalent prompts rather than a single presentation of a game.
Seyed Pouyan Mousavi Davoudi, Alireza Amiri-Margavi, A. Davodi et al.· 1 citation
Accurately assessing personality from text is challenging because traits are latent, context-dependent, and often subtly expressed across long narratives. Large language models (LLMs) offer new opportunities by processing extensive textual contexts, but pretraining of these models can induce latent"personality-like"biases, making single-model inferences inconsistent. We propose a fine-tuned multi-agent framework for detecting OCEAN personality traits, in which sub-agents are conditioned to adopt high, low, or neutral perspectives for each trait through masked language modeling (MLM) and psychometric supervision. A judge LLM aggregates and compares sub-agent outputs to generate final trait predictions, capturing multiple complementary perspectives while mitigating individual model biases. We evaluate the framework on life narrative dataset through quantitative and qualitative experiments, including baselines, ablations, and inference quality analyses. Our approach offers a scalable and interpretable method for text-based personality inference, highlighting the benefits of multi-agent reasoning grounded in psychometric supervision.
Rasiq Hussain, Darshil Italiya, Joshua R. Oltmanns et al.· 0 citations
LLMs are increasingly deployed as proxies for human study participants in social science experiments, yet the fidelity of this practice has rarely been tested directly. We test whether six LLMs can simulate individual human belief updates, comparing LLM outputs 1-to-1 against ground truth data from 391 UK participants on Prolific, who updated their stances on three discussion topics after reading Reddit comments. Each participant was simulated by an LLM conditioned on a persona derived from their demographic and personality trait data. We find that some LLMs (Qwen3-32B and GPT-5-Mini) can match the human post-stance distribution, but only when given participants'actual initial stances. All six models fail to simulate initial stances themselves and to produce faithful belief updates from self-generated stances. Three systematic biases emerge across all models: overrepresentation of neutral positions, more frequent but smaller belief shifts than humans, and a failure to rank comments by convincingness. Demographic and personality trait personas had no consistent effect on fidelity. LLM simulations of human belief dynamics are only reliable when grounded in realistic starting conditions, that current multi-round social media simulations rarely provide.
Sebastian Pohl, Harsh Mehta, Pranav Mambayil et al.· 0 citations
