Berkowitsch Nicolas A J, Scheibehenne Benjamin, Rieskamp Jörg
Department of Psychology, University of Basel.
J Exp Psychol Gen. 2014 Jun;143(3):1331-48. doi: 10.1037/a0035159. Epub 2013 Dec 23.
Cognitive models of decision making aim to explain the process underlying observed choices. Here, we test a sequential sampling model of decision making, multialternative decision field theory (MDFT; Roe, Busemeyer, & Townsend, 2001), on empirical grounds and compare it against 2 established random utility models of choice: the probit and the logit model. Using a within-subject experimental design, participants in 2 studies repeatedly choose among sets of options (consumer products) described on several attributes. The results of Study 1 showed that all models predicted participants' choices equally well. In Study 2, in which the choice sets were explicitly designed to distinguish the models, MDFT had an advantage in predicting the observed choices. Study 2 further revealed the occurrence of multiple context effects within single participants, indicating an interdependent evaluation of choice options and correlations between different context effects. In sum, the results indicate that sequential sampling models can provide relevant insights into the cognitive process underlying preferential choices and thus can lead to better choice predictions.
决策的认知模型旨在解释观察到的选择背后的过程。在此，我们基于实证检验了一种决策的序贯抽样模型——多选项决策场理论（MDFT；Roe、Busemeyer和Townsend，2001），并将其与两种已确立的随机效用选择模型进行比较：概率单位模型和逻辑模型。采用被试内实验设计，两项研究中的参与者在多个属性描述的选项集（消费产品）中反复进行选择。研究1的结果表明，所有模型对参与者选择的预测效果相当。在研究2中，选择集被明确设计用于区分这些模型，MDFT在预测观察到的选择方面具有优势。研究2还揭示了单个参与者内部存在多种情境效应，这表明对选择选项进行了相互依赖的评估以及不同情境效应之间的相关性。总之，结果表明序贯抽样模型能够为偏好选择背后的认知过程提供相关见解，从而能够做出更好的选择预测。
