Abstract: Categorization and inference are two important functions of category knowledge, besides categorization, category-based feature inference is an important application of category knowledge. Categorization is to infer an item's category label when its feature value is known, while inference is to infer an item's one or few unknown feature value when its category label is known. Category label is a symbol on behalf of category membership, while category feature is a symbol representing a feature of an item. This study aimed to contrast the polarization effect between category label and feature label in category-based feature inference. The experiments used learning-test two phase paradigm and a 2(type of category label: category and feature)×2(matching type of label: matching and non-matching)× 2(basal probability: high and low) experimental design. The results of the experiment showed that: 1)The polarization effect of category label was significantly higher than the one of feature label. 2)Category label was different from category feature essentially. 3)The basal probability affected the category-based feature inference. In summary, this study showed that category label is essentially different from category feature. And this study had a new finding that the basal probability did affect the category-based feature inference.

Key words: category-based feature inference, category label, polarization effect, basal probability

摘要： 采用学习―测试二阶段实验范式,对比了类别标签与类别特征在类别特征推理中的极化效应。研究结果表明,在类别标签组,标签匹配项目上的特征推理分数显著高于标签不匹配项目上的特征推理分数;在特征标签组,标签匹配项目上的特征推理分数与标签不匹配项目上的特征推理分数之间差异不显著。类别标签组的失匹配分数显著高于特征标签组。即在类别特征推理任务中,类别标签的极化效应显著高于类别特征的极化效应,因此,类别标签与类别特征存在本质差异,类别标签在类别特征推理中起主导作用。而且,本研究还发现,高前提概率条件下的特征推理分数都显著高于低前提概率条件下的特征推理分数,所以,前提概率也影响类别特征推理任务。

关键词: 类别特征推理, 类别标签, 极化效应, 前提概率