ETH Zürich, Department of Management, Technology and Economics, Zürich, Switzerland.
Institute of Risk Analysis, Prediction and Management, Academy for Advanced Interdisciplinary Studies, Southern University of Science and Technology, Shenzhen, China.
PLoS One. 2022 Aug 30;17(8):e0273551. doi: 10.1371/journal.pone.0273551. eCollection 2022.
We present a short review of discrete-time quantum walks (DTQW) as a potentially useful and rich formalism to model human decision-making. We present a pedagogical introduction of the underlying formalism and main structural properties. We suggest that DTQW are particularly suitable for combining the two strands of literature on evidence accumulator models and on the quantum formalism of cognition. Due to the additional spin degree of freedom, models based on DTQW allow for a natural modeling of model choice and confidence rating in separate bases. Levels of introspection and self-assessment during choice deliberations can be modeled by the introduction of a probability for measurement of either position and/or spin of the DTQW, where each measurement act leads to a partial decoherence (corresponding to a step towards rationalization) of the deliberation process. We show how quantum walks predict observed probabilistic misperception like S-shaped subjective probability and conjunction fallacy. Our framework emphasizes the close relationship between response times and type of preferences and of responses. In particular, decision theories based on DTQW do not need to invoke two systems ("fast" and "slow") as in dual process theories. Within our DTQW framework, the two fast and slow systems are replaced by a single system, but with two types of self-assessment or introspection. The "thinking fast" regime is obtained with no or little self-assessment, while the "thinking slow" regime corresponds to a strong rate of self-assessment. We predict a trade-off between speed and accuracy, as empirically reported.
我们简要回顾了离散时间量子漫步（DTQW），它是一种潜在有用且丰富的形式主义，可以用于模拟人类决策。我们介绍了基础形式主义和主要结构特性的教学入门。我们认为，DTQW 特别适合结合关于证据积累模型和认知量子形式主义的两个文献流派。由于附加的自旋自由度，基于 DTQW 的模型允许在单独的基中自然地对模型选择和置信度评分进行建模。在选择审议过程中，可以通过引入对 DTQW 的位置和/或自旋进行测量的概率来对自省和自我评估的程度进行建模，每个测量行为都会导致审议过程的部分退相干（对应于合理化的一步）。我们展示了量子漫步如何预测观察到的概率错觉，例如 S 形主观概率和结合谬误。我们的框架强调了响应时间与偏好和响应类型之间的密切关系。特别是，基于 DTQW 的决策理论不需要像双过程理论那样引入两个系统（“快”和“慢”）。在我们的 DTQW 框架中，两个快速和慢速系统被单个系统所取代，但具有两种类型的自我评估或自省。“快速思考”状态是在没有或很少自我评估的情况下获得的，而“慢速思考”状态对应于强烈的自我评估率。我们预测了速度和准确性之间的权衡，这与经验报告一致。
