关键词: 语言, 类比推理, 图式生成, 类比映射, ERP, LNC

Abstract: Analogical reasoning as the mechanism of acquisition of new information by inspection of specific instances, is clearly at the heart of advanced human cognitive capabilities and creativity. Especially, language analogical reasoning, as an important part of everyday discourse, is the most common one. Only a few brain imaging studies have used verbal analogies to explore the neural substrates of analogical thinking. Nevertheless, hemodynamic techniques have excellent spatial resolution but poor temporal resolution; hence, these results could not clearly reflect different substeps of analogical processes. Compared with hemodynamic techniques, event-related potentials (ERPs), as an electromagnetic technique, are sensitive to the time course of cognitive processes. However, there has no ERP study on this matter so far.
Using the “A:B::C:D” paradigm, this study investigated the neuromechanism underlying language analogical reasoning by ERPs. Six hundred two-character Chinese noun words were chosen from the Contemporary Chinese Dictionary (the fifth edition) to serve as experimental stimuli. One hundred fifty trials consisting of four noun words were used in three analogy tasks with three semantic relations: metaphor, category and isomorphism, each analogy task had 50 trials. Fourteen undergraduate students participated in the study. Subjects were asked to judge whether one word pair was semantically analogous to another word pair. The EEG was recorded from 64 scalp channels using electrodes mounted in an elastic cap. ERPs were calculated for a 3200 ms epoch including a 200 ms pre-stimulus baseline.
Results showed that the schema induction stage elicited N400 and P300 components, which possibly attributable to semantic processing of word pairs and the formation of a relational category. The analogy mapping stage elicited a N400, which might indicate the semantic retrieval of the antecedent word of target domain. The N400 was then followed by a late negative component (LNC), which might indicate the processing of analogical transfer the schema from the source domain to the target domain and reasoning the suitable answer. The study also compared processing of the analogy mapping under different conditions. The results showed that the amplitude of LNC component was smaller and distributed over anterior scalp sites when the relations of analogy mapping were simpler, while the amplitude of LNC component was larger and widely distributed when the relations of analogy mapping were complex.
The results suggest that the substages of analogical reasoning, schema induction, analogy mapping, are complex processes. The processing of the analogy mapping stage is close with the LNC component. The amplitude and scalp distribution of the LNC are modulated by the complexity degree of the relations of analogy mapping.

Key words: language, analogical reasoning, schema induction, analogy mapping, ERP, LNC