关键词: 属性中心性, 相似性, 归纳推理

Abstract: Introduction
Previous research on property effect mainly examined how property stability affected inductive reasoning. The property stability effect cannot explain all property-related questions. Gelman (1988) discovered that Grade 2 children were more capable of generalizing animals’ anatomical features (e.g., this rabbit has a spleen inside) but not their functional features (e.g., you can loll with it) to another instance of the same category. In this study, subjects did not consider the stability of the property when performing inductive reasoning. Yet their answers implied the property effect. So what is this effect based on? Hadjichristidis (2004) proved that the degree of property centrality plays a role in inductive reasoning. However, we suspect 2 potential problems may occur in their experimental process; hence, we conducted 3 modified experiments to investigate the property centrality effect in inductive reasoning and examined whether our results were consistent with those of Hadjichristidis’ study.
Method
Three experiments were performed to analyze the effect of different property centrality degrees on inductive reasoning. The first experiment was in accordance with that conducted in Hadjichristidis’ study, except for the different procedure. The second experiment was the same as that carried out by Hadjichristidis (2004). The last experiment was to examine property centrality effect using expanded similarity and centrality conditions in order to complete and refine our analyses of this effect and its psychological process.
Results
All 3 experiments showed the following results. Inductive scores of the central property were higher than those of the less central property under higher similarity conditions. However, with a decrease in the relatedness of premise and conclusion, the effect of central property on participants’ induction process showed a decreasing trend; an increasing trend was found for the less central property (least central property in experiment 3). The central property was no longer advantageous under low similarity conditions; its average value was significantly lower than that of the less central property.
Conclusions
A property centrality effect exists in the process of inductive reasoning. Central property has more effect on inductive reasoning than other properties (less central property and least central property). However, the appearance of the effect was constrained by the similarity between premise and conclusion concepts. When the similarity is high, the central property showed more inductive ability than the less central property. The effect of central property on inductive reasoning ability decreased with a decrease in the similarity between premise and conclusion concepts; in contrast, the effect of the less central property and least central property on inductive reasoning ability increased. When premise and conclusion were no longer related, the inductive strength of the least central property reached the highest level, while that of the central property reached the lowest level. Since the inductive strength of the least central property was not significantly higher than the random level all times and the lowest inductive strength of the central property was significantly lower than the random level, we propose that when premise and conclusion concepts were unrelated, the subjects were sure that they could not choose the central property and made random guesses based on reasoning from other properties. This embodied the typical characteristic of uncertain inductive reasoning process.

Key words: property centrality effect, similarity, inductive reasoning