高度模糊靶字下的N400语义启动效应：词汇加工理论的证据

本研究基于不可识别模糊字词仍能诱发N400的词汇加工理论的假设，让被试对模糊字进行同一字延迟判断，利用脑电技术记录不同模糊程度靶字的认知过程，来探讨N400出现的条件，以阐明N400反映的是词汇加工，还是词汇后加工。实验结果表明：与可识别、轻度模糊汉字一样，不可识别、高度模糊汉字仍然可以诱发N400；这种N400语义启动效应反映了词汇加工过程。

An N400 Priming Effect Elicited By Highly Blurred Chinese Characters: Evidence for Lexical Analytic Process

Abstract The popular mechanisms of N400 component of the event-related brain potential (ERP) proposed to explain the N400s are the automatic spreading activation and lexical analytic process, which are in contrast with equally popular postlexical mechanisms that are conscious and attentional. The present study examed the contrast theories or mechanisms, and used ERP methodology to investigate electrophysiological correlates of cognitive representations underlying N400 priming effect for highly blurred and consciously unidentifiable Chinese character targets. In this study, three character items were presented in succession in each trial of a delayed character-matching task, representing intact prime, blurred target, and intact probe stimulus, respectively. Two variables, priming and blurring, were manipulated with a 2×2 factorial within-subjects design in the experiment, and ended up with 4 treatment conditions, which were primed slightly blurred, primed highly blurred, unprimed slightly blurred and unprimed highly blurred conditions. During a delayed character-matching task, the subjects were asked to press a key on keyboard if the target and the probe, which followed the target, were a same character in identity, and press another one if not. Sixteen native Chinese undergraduates or graduates participated in the experiment and were instructed to make a yes/no distinction in the task, and the data of four participants were given up because of unnormal data. ERPs of blurred Chinese characters (the targets) were recorded while participants performed a delayed character-matching task. Behavioral data showed interactions between priming and blurredness: the priming effects on accuracy rate and rection time were more significant for highly blurred targets than for slightly blurred targets. Repeated-measures ANOVAs on mean amplitude with factors of blurring (slightly blurred, highly blurred), priming (primed, unprimed), and electrode site revealed significant main effects of blurring and priming at the 300–500ms interval; At the 700–900ms interval, similar ANOVAs on mean amplitude revealed significant interactions on semantic priming and blurring, and significant interactions on electrode site and blurring; More importantly, the amplitude of N400 for Lever 3 elicited by the unprimed targets was significantly negative compared with primed targets (p < .001), and this difference was prominent at the central-parietal electrode sites (see Figs 2 & 4), which are consistant with Wang’s findings (2009). Extra subjects performed an identification test on the target stimuli; the identification rate is lower than the criterion, which is 33%, of unconscious processing. ERP data and the identification test on the targets together revealed that just as slightly blurred targets did, highly blurred, consciously unidentifiable targets that were semantically related to primes elicited smaller N400 compared with similar but unrelated targets. These results indicated that N400 priming effect could be elicited by highly blurred, consciously unidentifiable targets, supported the view that the N400 component could index unconscious lexical processing, and provided evidence against a postlexical account.