多元选择抽样：受过程数据约束的决策模型。

Department of Experimental Psychology, University College London.

Warwick Business School, University of Warwick.

Psychol Rev. 2018 Jul;125(4):512-544. doi: 10.1037/rev0000102.

Sequential sampling of evidence, or evidence accumulation, has been implemented in a variety of models to explain a range of multialternative choice phenomena. But the existing models do not agree on what, exactly, the evidence is that is accumulated. They also do not agree on how this evidence is accumulated. In this article, we use findings from process-tracing studies to constrain the evidence accumulation process. With these constraints, we extend the decision by sampling model and propose the multialternative decision by sampling (MDbS) model. In MDbS, the evidence accumulated is outcomes of pairwise ordinal comparisons between attribute values. MDbS provides a quantitative account of the attraction, compromise, and similarity effects equal to that of other models, and captures a wider range of empirical phenomena than other models. (PsycINFO Database Record

顺序采样证据或证据积累已在各种模型中实施，以解释一系列多选择现象。但是，现有的模型对于究竟积累了什么证据以及如何积累证据并没有达成一致意见。在本文中，我们使用过程追踪研究的结果来限制证据积累过程。有了这些限制，我们扩展了抽样决策模型，并提出了多选择抽样决策（MDbS）模型。在 MDbS 中，积累的证据是属性值之间两两有序比较的结果。MDbS 提供了与其他模型相同的吸引力、妥协和相似性效应的定量解释，并比其他模型更能捕捉到更广泛的经验现象。