AI 中文总结
该研究提出Q-SCM模型用于模拟交互式交通环境中驾驶员心理状态演变，保留经典潜在类别结构，用量子认知状态模型取代传统类别成员公式，通过感知线索引发酉旋转，引入控制机制，经实证得出防御状态形成受多种因素影响的结论。
AI 中文摘要
我们提出了一种量子序列选择模型（Q-SCM），用于对交互式交通环境中驾驶员心理状态演变进行建模。该框架保留了经典的潜在类别选择结构，但用量子认知状态模型取代了传统的类别成员公式。此模型的独特之处在于量子组件局限于类别成员层，而行动选择层仍是经典的随机效用最大化（RUM）。驾驶员的潜在状态表示为布洛赫球上的二态量子系统，包括中性和防御状态。感知线索，如间隔距离、碰撞时间（CTTC）和车道偏差，会引发由泡利矩阵控制的序列酉旋转。为确保良好的状态演变，引入了三种控制机制。该模型使用从自然轨迹中提取的9610名驾驶员的85754个观测值进行估计。实证结果表明，防御状态的形成不仅受交通线索的瞬时值影响，还受累积线索历史和线索处理顺序的影响。
英文摘要
We propose a Quantum-Sequential Choice Model (Q-SCM) for modelling driver mental state evolution in interactive traffic environments. The proposed framework retains the classical latent class choice structure, but replaces the conventional class membership formulation with a quantum cognitive state model. A unique feature of this model is that the quantum component is confined to the class membership layer, while the action choice layer remains a classical RUM. The driver's latent state is represented as a two-state quantum system on the Bloch sphere including neutral and defensive states. Perceptual cues, including separation distance, closing time-to-collision (CTTC), and lane deviation induce sequential unitary rotations governed by Pauli matrices. This formulation allows the model to capture memory, phase effects, cue order dependence, and transitions between behavioural regimes that depend on prior cue history. To ensure well-behaved state evolution, we introduce three control mechanisms: a monotonicity constraint that prevents pendulum-like overshoot, a geodesic safeguard mechanism that ensures convergence toward the defensive state under sustained threat exposure, and a relaxation step that allows recovery toward the neutral baseline when the threat weakens. The model is estimated using 85,754 observations from 9,610 drivers extracted from naturalistic trajectories. The empirical results show that defensive state formation is not governed only by the instantaneous values of traffic cues, but also by the accumulated cue history and the order in which cues are processed.
