关键词: 归纳, 分类, 属性推理, 事件相关电位(ERPs)

Abstract:

Category-based property induction involves in at least two processes, namely, categorization and property inferences. Previous ERP studies in figural category-based induction only focused on categorization, could not generalize their findings to semantic induction. The studies in semantic category-based properties induction failed to acquire ERP parameters related to categorization and property inferences respectively due to their experimental designs where their non-inductive condition contained unrelated categories with consistent properties or related conclusion categories with inconsistent properties. What’s more, in these ERP studies on semantic category-based induction, the conclusion categories and conclusion properties were presented simultaneously in previous studies, which involved complex comparison that was not helpful to explore the ERP responses to category-related and property-related processes clearly. Therefore, this study was aimed to find distinct ERP characteristics between categorization and property inferences during verbal category-based property induction. The present study employed ERPs and adopted single-premise verbal category-based property induction. Thirty-four undergraduates participated in this experiment, they were presented with premise stimuli including category and property first, next conclusion category (category-presenting stage) and conclusion property (property-presenting stage) would be presented separately. Experimental premise and conclusion stimuli in tasks composed with specific categories and blank properties including capital letters plus Arabic numeral (e.g. premise: 苹果 X1 (apple X1)), which represented the category item has the blank property. (e.g. apple has property X1). In the experiment, premises involving premise categories and premise properties were presented firstly (apple X1), and then conclusion categories involving related (e.g., pear) and unrelated categories (e.g., rose) were presented (categorization stage); Finally, the conclusion properties involving consistent (e.g., X1?) and inconsistent properties (e.g., B3?) were presented with question mark (properties stage). Participants were required to make response in property-presenting stage to judge whether this conclusion based on premise stimuli was acceptable or not. In the categorization stage, the results found compared to related categories, unrelated categories elicited greater N400 that indicated N400 is closely related to the processing of semantic categorization; In the properties stage, consistent property elicited smaller frontal N2 than that did by inconsistent property regardless of conclusion category condition, which suggested that N2 is associated with mismatch of perceptual property or multi-dimensional rule inconsistence. What’s more, consistent property elicited greater P3a than inconsistent property at the frontal brain region during 300~400 ms regardless of conclusion category condition, which provided that attention transfer or allocation with unconsciousness could be attribute to the activation of P3a. Meanwhile, at the parietal brain region, consistent properties elicited greater P3b than inconsistent properties under related conclusion categories, which indicated that P3b has a link with the degree of expectation in reasoning. The present study also found on the properties stage, inconsistent properties yielded a greater P600-like component than consistent properties at parietal brain region during 400~600 ms under related conclusion categories, which showed that P600-like LPC400–600 is related to the violation of rule-based logical inference. Therefore, The ERP responses to categorization and properties inferences were dissociated in this study. Besides, previous studies mainly employed explicit paradigm (e.g. semantic categorization task) or implicit paradigm (e.g. semantic priming paradigm) in ERP studies on semantic categorization, category-based property induction employed in this study to uncover ERP characteristics of categorization is beneficial for extending findings in this field.

Key words: induction, categorization, property inferences, ERPs