In real-world recommendation systems, users engage in a variety of scenarios, such as homepages, search pages, and related-item recommendation pages. Each of these scenarios is designed to capture distinct facets of user intent. However, user interests are often inconsistent across different scenarios, attributable to variations in their decision-making processes and modes of preference expression. This inherent heterogeneity poses a substantial challenge to unified modeling, rendering multi-scenario recommendation a non-trivial task. To address these challenges, we propose RUIE, a novel reinforcement learning-enhanced framework that models dynamic user preference evolution as sequential decision-making problems, and leverages cross-scenario behavior sequences to detect interest shifts and dynamically adjust sample utilization for accurate interest capture. Experiments demonstrate that our proposed approach significantly outperforms state-of-the-art methods in multi-scenario recommendation tasks and is widely applicable. This work offers a fresh perspective on multi-scenario modeling and highlights promising directions for future research. The source code is available at https://github.com/rl4rec/RUIE.