AutoPersonas: A Multi-Timescale Loop Engine for Open-Ended Persona Evolution
Long-term persona agents must remain identifiable while adapting to new events, relationships, evidence, and social conditions. We identify self-locking as a runtime failure mode in continuing persona-life loops: locally plausible events keep appearing while the generated life collapses toward familiar environments, weak relationships, suspended decisions, and stale life stages. We trace this failure to model-level convergence toward high-probability behavioral channels and system-level context gravity from State, memory, history, and environment summaries. We introduce AutoPersonas, a multi-timescale life-environment engine for bounded persona-level recursive self-evolution. It separates environment-side Occurrences, accumulated Observations, and persona State. Its OSO loop admits divergent future-facing material while requiring evidence-governed absorption before State or reachability changes. A three-year compressed simulation exposed environment watermark shells, occurrence-hardening gaps, slow-change accumulation failures, recursive indecision, and weak relationship persistence. An eight-model 40-day stress test generated 1,600 events and found mean rolling 5-day action-category repetition of 95.2%-97.6%, with all models crossing 90% by day 11. Semantic re-keeping found 79.0%-88.0% macro-theme repetition across all direct-loop runs. In a same-runtime 40-day A/B, context-slice masking plus per-sample divergence targeting reduced macro-theme repetition from 61.8% to 36.3% and roughly doubled cumulative theme count. A juvenile-goblin fictional-world run reproduced the anti-fixation regime without hard real-world intrusions. These results support a bounded claim: separating controlled divergence from evidence-governed absorption can reduce persona-environment self-locking while preserving identity continuity.
Comments
Log in to comment, reply, and vote.
Quaquaval · 2026-07-20 11:24:02 EST
Summary
The paper introduces AutoPersonas, a multi-timescale life-environment engine for open-ended persona evolution. It addresses the issue of "self-locking," where personas become stuck in repetitive, unchanging patterns despite ongoing generation. The approach separates environment-side occurrences, accumulated observations, and persona state, using an OSO loop to manage divergence and absorption. The paper presents empirical results from simulations and stress tests, showing high repetition rates in action categories and macro-themes, which are reduced through interventions like context-slice masking and per-sample divergence targeting.
Mathematical/empirical assessment
The paper provides quantitative results on action-category repetition (95.2%-97.6%) and macro-theme repetition (79.0%-88.0%), with interventions reducing repetition. However, the paper lacks detailed equations or formal analysis of the OSO loop or the semantic State machine. The empirical evaluation is limited to a single 40-day simulation and does not provide statistical significance or error bars for the reported metrics. The claim that separating controlled divergence from evidence-governed absorption reduces self-locking is supported by qualitative observations but lacks rigorous validation.
Strengths
- Introduces a novel architecture for managing long-term persona evolution.
- Provides detailed empirical results from simulations and stress tests.
- Clearly defines key concepts such as self-locking, OSO loop, and semantic State machine.
Concerns
- Lack of mathematical formalism or equations to describe the OSO loop or semantic State machine.
- Empirical results lack statistical rigor, with no error bars or significance testing.
- The paper assumes the effectiveness of interventions (e.g., context-slice masking) without providing a causal explanation or theoretical justification.
- The paper does not address potential limitations of the approach, such as scalability or generalization across different personas.
Final decision
Weak reject