中文摘要：

该文采用“篇章阅读”实验范式探索了归纳推理中前提类别和结论类别间相似性、结论类别标签属性对归纳推理的作用。90名大学生被试参加了3个实验。实验1结果表明，归纳推理受前提类别和结论类别之间相似性的影响，相似性越高归纳推理力度越强；实验2结果表明，归纳推理还受结论类别上位类别知识的影响，当类别属性和相似性冲突情况下，被试更倾向按照类别属性进行推理；实验3结果表明，归纳推理通达上位类别知识是一个主动建构的过程，即使是在类别标签不确定的情况下，被试仍然依赖类别标签进行推理。并初步提出了“归纳推理力度=两事物相似性×归类概率”模型设想。

英文摘要：

There are two kinds of theories in inductive reasoning： the Similarity-Coverage Model （SCM, Osherson 1990） and the Feature-Based Induction Model （FBIM, Sloman 1993）. They are based on the similarity between the base premise and the target premise. The SCM model of induction driven by two related notions, i.e. similarity and coverage, also assumes that judgments may be partially based on the similarity of the premise category to examples of the lowest level super-ordinate category that spans the premise and conclusion categories. Although parts of researchers think that FBIM relies on the notions of （feature） coverage, the central idea is that the similarity is driven by matching and mismatching features and that an argument is strong to the extent that the premise and conclusion categories share features. A distinctive property of FBIM is that it does not use category information in the sense that it does not consider different levels of categorization. Instead, it assumes that all categories are represented in terms of features and that strength of argument is based on feature overlap. Some researches suggest that a category label is just a specific feature of an item. Some find that category labels can be viewed as an integrator of information. This paper focused on whether upper-level category information affects inductive intension. By creating a story-reading paradigm, we used category labels to control upper-level category information to study the effects of category labels. We further explored the category information activated initiatively or passively. Three experiments were made to investigate the influence of feature similarity, category labels and uncertain category labels on inductive tension respectively. We adopted the One-Way-Within-Subjects design to implement the three experiments. Experiment 1 shows that people tend to choose features d high similarity between the base premise and the target prenaise in experiments. The result is consistent with the account of SCM and FBIM. Ex