Two-digit number processing: holistic, decomposed or hybrid? A computational modelling approach
Two-digit number processing: holistic, decomposed or hybrid? A computational modelling approach
复制标题
DOI：
10.1007/s00426-010-0307-2
复制
发表时间：
2011-07-01
影响因子：
2.3
通讯作者：
Willmes, K.
中科院分区：
文献类型：
--
作者：
Moeller, K.;Huber, S.;Willmes, K.
目前，关于两位数数量级表示的本质，有三种相互竞争的理论描述：整体模型、严格分解模型和混合模型。对单位-十年相容性效应的观察（Nuerk等人在《认知》82:B25-B33, 2001）挑战了两位数数量级被表示为一个整体的观点。但是，目前还没有对分解模型和混合模型进行区分的研究。本研究使用计算建模方法解决了这一问题。编写了三个符合三种理论模型约束的网络模型，并进行了两位数比较的训练。对模型进行比较，看它们如何以最简洁的方式解释经验效应。总的来说，该评价表明严格分解模型对经验数据的模拟效果最好。讨论了这些结果对我们理解人类数量数量级表征本质的影响。
Currently, there are three competing theoretical accounts concerning the nature of two-digit number magnitude representation: a holistic, a strictly decomposed, and a hybrid model. Observation of the unit-decade compatibility effect (Nuerk et al. in Cognition 82:B25-B33, 2001) challenged the view of two-digit number magnitude to be represented as one integrated entity. However, at the moment there is no study distinguishing between the decomposed and the hybrid model. The present study addressed this issue using a computational modelling approach. Three network models complying with the constraints of all three theoretical models were programmed and trained on two-digit number comparison. Models were compared as to how well they accounted for empirical effects in the most parsimonious way. Generally, this evaluation indicated that the empirical data were simulated best by the strictly decomposed model. Implications of these results for our understanding of the nature of human number magnitude representation are discussed.