Controlling Long-Horizon Behavior in Language Model Agents with Explicit State Dynamics
This work addresses the issue of inconsistent long-horizon behavior in LLM agents for multi-turn dialogue applications, representing an incremental improvement over prior methods focusing on turn-local sentiment.
The paper tackled the problem of abrupt shifts in tone and persona in large language model agents during extended interactions by imposing explicit affective dynamics on an external state, resulting in improved temporal coherence and controlled recovery, with second-order dynamics showing increased stability but a trade-off in responsiveness.
Large language model (LLM) agents often exhibit abrupt shifts in tone and persona during extended interaction, reflecting the absence of explicit temporal structure governing agent-level state. While prior work emphasizes turn-local sentiment or static emotion classification, the role of explicit affective dynamics in shaping long-horizon agent behavior remains underexplored. This work investigates whether imposing dynamical structure on an external affective state can induce temporal coherence and controlled recovery in multi-turn dialogue. We introduce an agent-level affective subsystem that maintains a continuous Valence-Arousal-Dominance (VAD) state external to the language model and governed by first- and second-order update rules. Instantaneous affective signals are extracted using a fixed, memoryless estimator and integrated over time via exponential smoothing or momentum-based dynamics. The resulting affective state is injected back into generation without modifying model parameters. Using a fixed 25-turn dialogue protocol, we compare stateless, first-order, and second-order affective dynamics. Stateless agents fail to exhibit coherent trajectories or recovery, while state persistence enables delayed responses and reliable recovery. Second-order dynamics introduce affective inertia and hysteresis that increase with momentum, revealing a trade-off between stability and responsiveness.