Spontaneous Knowledge Partitioning in Categorization

• Main Authors: Hsin-An Yu, 余信安
• Other Authors: Lee-Xieng Yang
• Format: Others
• Language: zh-TW
• Online Access: http://ndltd.ncl.edu.tw/handle/07751614100260839405

碩士 === 國立中正大學 === 心理學所 === 95 === Knowledge partitioning refers to that acquired knowledge may be fractionated into independent parcels that are gated by context. This point of view is supported by many researches. And the reason why knowledge partitioning occurs is generally thought that it facilitates learning. Kalish, Lewandowsky, and Kruschke (2004) found in function learning that without an extra context, people can take the attributes of the stimulus as a cue in knowledge partitioning. Hence, the study demonstrates four experiments in order to see if knowledge partitioning occurs in categorization without any extra context. Every category space can be demarcated into two parts by a sigmoid curve. The part above the curve is defined as category B, and the part below is category A. In experiment 1, some stimuli close to the horizontal boundaries are displayed. And participants show three different classification patterns. One pattern among them is even spontaneous knowledge partitioning. In experiment 2A and 2B, we manipulate two critical stimuli’s positions and find more spontaneous knowledge partitioning successfully. Then in experiment 3, we manipulate another two stimuli for the sake of diminishing the proportion of the spontaneous knowledge partitioning. Summarise the above results, the study finds that 1) knowledge partitioning can occurs in categorization without any extra context. 2) Knowledge partitioning occurs or not is subjective to participants. Different participants can spontaneously generate different types of category representations even if they receive the same stimuli in the same condition. 3) Computer simulation shows that GCM can not explain spontaneous knowledge partitioning in the study. As a consequence, we accept the explaination given by knowledge partitioning theory.