中文摘要

在注意研究领域存在一些长期争论不休的问题，例如，MMN是否仅仅反映自动加工，即是否受注意的影响？视觉通路是否存在MMN？选择性注意的加工部位是通路特异性还是通路上机制？而注意选择的早晚期位置之争论迄今仍在继续，仍是当前注意心理学面临的基本难题之一。关于语言ERP的自动加工与注意成分的研究，也需要进一步的深入。因此，实验目的主要为：À 通过比较注意与非注意条件下的MMN，验证它是否受注意的影响。进而深化对脑的自动加工的认识。Á 通过比较视听通路产生的差异成分，即注意成分，不仅为差异成分的性质之争，也可为选择性注意机制发生在“早期”还是“晚期”位置，提供进一步证据。Â 在进一步严格的非注意条件下，验证视觉MMN存在与否。从而进一步探讨视觉MMN的产生机制，为有关视觉信息加工原理提供证据。Ã 通过选择性注意条件下各ERP成分的头皮分布分析，提供选择性注意的加工部位是通路特异性还是通路上机制的证据，从另一个途径提供注意的选择发生在早期还是晚期，并可能对注意的过滤器位置具有可塑性这一新的假说提供电生理学方面的证据。Ä 首先证明对汉语字形、字音的辨别，及其在非注意条件下是否可产生MMN与注意成分；如果可以，然后进一步阐明它们的性质，提供N200与N400是否属于同一家族的证据，从而丰富与加深注意的早、晚期选择理论及自动加工理论。

本实验研究采用“跨通路延迟反应”实验模式，研究正常青年人非言语和言语两个系列的跨通路选择性注意事件相关电位(ERP)成分。非语言刺激采用柔滑短纯音(听觉通路)和彩色照片(视觉通路)。关于选择性注意的实验模式，其要点是每项实验令被试只注意其中的一个通路，并完成一定任务而忽视(非注意)另外一个通路。同时记录注意与非注意状态下的ERP波形。实验中视觉刺激与听觉刺激混合、按伪随机顺序分时呈现。视觉刺激与听觉刺激各包括3种：偏差刺激、标准刺激和反应命令信号。每一标准刺激与偏差刺激之后皆跟随一次反应命令信号。在刺激信号与命令信号之间随机呈现0～2个异通路的刺激信号。每个通路的标准刺激呈现概率为82.5%；偏差刺激为17.5%。二者的呈现顺序是伪随机排列的。刺激间隔为250 ms～700 ms, 伪随机。令被试集中注意力于注意通道的刺激，忽视非注意通道的刺激。当标准刺激出现时，令其一只手拇指作按键准备；当偏差刺激出现时，另一只手拇指作按键准备。待反应命令信号出现后，再尽快按键。由于非注意通道的刺激插在注意通道的刺激与反应之间，此时被试处于力求尽快按键、紧张等待不知何时出现的命令信号之际，无暇顾及非注意通道的刺激，因此较好地保证了非注意纯度。语言研究仍然采用相同的“跨通路延迟反应”模式。用单字词按韵母(听觉通路)、笔划与结构(视觉通路)分别呈现，其刺激概率、刺激间隔和刺激序列同非语言刺激。以相减技术得到的偏差相关成分（偏差刺激-标准刺激）和注意成分（注意-非注意）为主要分析对象。

“跨通路延迟反应”模式具有以下特点：1）被试的任务分为注意®分辨®准备®反应。在被试为了尽快按键而集中注意于注意通路中搜索命令信号的间隙，呈现非注意通路的刺激。这就是说，在非注意通路呈现刺激时，被试的任务尚未完成，其注意力一般不至于转移。因此，非注意通路的非注意纯度容易得到保障，即使刺激间隔较长，也不影响非注意的纯度，在解决MMN是否受注意影响的问题中，将提供更为有力的证据。2）被试对标准与偏差刺激均需按键，没有目标与非目标之分，使刺激概率差异引起的ERP偏差相关成分不受目标任务的影响，因而更为可靠。3）在同一项实验中将视听觉刺激的呈现顺序随机排列，经过叠加，视觉与听觉ERP在逻辑上为同时产生，可使视听两个通路的刺激具有更为相同的背景，在比较视觉与听觉的偏差相关负成分，从而论证视觉通路是否存在MMN时，更有说服力。4）非语言刺激时，听觉通路的刺激差别为短纯音频率的高低不同，视觉通路则为图象的亮度不同；语言刺激时，听觉通路以韵母不同分类，带韵母“a”(如a、an、ang、ao等)的单字词为偏差刺激，不带韵母“a”的为标准刺激；视觉通路则以笔划数和结构2个标准区别标准刺激与偏差刺激。如此使两个通路的刺激难度基本一致。

主要结果与结论如下：À 记录的非语言ERP成分与经典的Oddball模式的ERP成分性质基本相同，例如无论注意与否或何种刺激，最大N1均分布于额部；在语言ERP成分中，听觉N1最大峰位于额中央部，而视觉N1最大峰位于两侧枕部与颞后部，具有明显的通路差异性，因此N1可能是反映汉字形音初步加工的特异性指标。Á 在非语言听觉标准刺激、视觉标准与偏差刺激下，均在额部发现了注意N1波幅的增大；在语言实验中，听觉偏差刺激在额中央部，听觉标准刺激、视觉标准与偏差刺激在枕部与颞后部也发现注意N1显著增大。表明语言和非语言的注意加工均引起N1的选择性增大。针对N1是外源性成分还是混合性成分的争论，本研究认为N1是内、外源性成分重叠的结果。 在注意和较为严格的非注意条件下，观察到一个汉字形音产生的早期偏差相关负波(DRN1)，除在非注意时的DRN1头皮分布与失匹配负波(MMN)不同外，其它峰潜伏期、波幅分布和注意效应等性质与MMN基本相同。并且，在视觉通路也证实了语言与非语言刺激的DRN1的上述特征均与听觉MMN基本一致。据此，关于是否存在语言MMN和视觉MMN的疑义，本实验结果可以初步认定语言MMN和视觉MMN的存在。à比较注意与非注意的MMN平均波幅，结果听觉与视觉MMN波幅均不受注意的影响，汉字形音识别产生的DRN1，其波幅也不受注意的影响。针对MMN是否受注意的影响这一长期争论，本研究认为MMN波幅不受注意的影响，它是反映对语言和非语言刺激自动加工的重要指标。Ä 非语言的早期注意成分(Nd1)的起始时间与注意N1相同，而早于非注意N1，表明在外源性成分起始前就发生了注意效应。在汉字形音加工的时间发生顺序上，依次为纯注意的Nd1、混合成分的注意N1和纯外源性的非注意N1。针对注意的早期还是晚期选择这一注意理论的基本问题，本实验结果表明语言和非语言的注意效应均出现在外源性成分之前，提供了支持注意早期选择理论的进一步证据。Å 非语言听觉与视觉偏差刺激产生的Nd1最大峰分布于它们各自的初级感觉投射区，听觉与视觉标准刺激诱发的Nd1最大峰均分布在额部；语言听觉偏差刺激产生的Nd1最大峰位于额中央部，视觉偏差刺激产生的Nd1最大峰位于枕部；非语言听觉与视觉标准刺激诱发的Nd1最大峰均位于枕部。针对刺激的加工部位是通路特异性还是通路上机制的争论，本研究认为跨通路选择性注意对非语言偏差刺激的加工部位是通路特异性的，对语言偏差刺激的加工部位具有通路间差异性；而对非语言标准刺激的加工部位是通路上的，对语言标准刺激的加工部位具有通路间共同性。对于注意的选择性发生时程的早晚可因刺激条件而异、具有可塑性这一新见解，本实验提供了支持它的证据。Æ 针对语言加工中N400与N200关系问题，本实验根据N350的峰潜伏期、波幅及其头皮分布特点，认为注意条件下的N350很可能是N200与N400的重叠，并主要表现出N400的特点；而非注意条件下，N350则可能主要是N200。N350的属性因条件而异，体现了ERP成分反映的大脑加工机制的可塑性。

关键词    事件相关电位(ERP)、选择性注意、偏差相关成分(DRC)、注意成分、失匹配负波(MMN)、跨通路延迟反应模式、言语和非言语任务

Selective Attention Research on Event-Related

 

Potentials Using a Cross-Modal and Delayed Response Paradigm

Abstract

Yue-jia Luo & Jing-han Wei

 

There are some long-lasting controversies in the field of attention study. For example, whether mismatch negativity (MMN) only reflects automatic processing, that is, it is only affected by attention? Does MMN exist in the visual modality? Whether the processing distribution of selective attention is modality specificity or supramodal mechanism? Moreover, The debate about early- and late-selection is still one of the basic conundrum. It needs further to go deep in the study of language ERP on the automatic processing and attention. The purposes of experiment are as followed: 1) To prove MMN is affected by attention by comparing MMN in attend condition and unattended condition; 2) To provide further evidence supporting early or late selection by recording simultaneously visual and auditory difference component (attentive component); 3) To certify existing of visual MMN in the strict inattention condition and probe into its generation mechanism; 4) To obtain evidence supporting the processing distribution of selective attention is weather modality specificity or supramodal mechanism; 5) To provide experimental evidence and explication on the cause of the N1 enhancement, the cerebral mechanism of Chinese language processing, the controversy of early- and late-selection and the plasticity of attention filter location. In addition, we tried to determine where is the neuroanatomical location of the N400 evoked by verbal processing and whether it reflects post-lexical integration exclusively.

Event-related potentials (ERPs) were measured in normal young subjects using the paradigm “Cross-modal and delayed response”, which is able to improve inattention purity and to avoid the effect of task target on the deviance-related components (DRC) of ERPs.  The experiment was conducted under two conditions: one is attending to the visual modality but ignoring the auditory modality, the other is attending to the auditory modality but ignoring the visual modality.  A series of five kinds of stimuli were presented to every subject: standard, deviant stimuli and response imperative signal in attended channel; standard and deviant stimuli in unattended channel.  The visual and auditory stimuli including verbal and non-verbal stimuli presented in a pseudo-random order.  The presentation probability of standard stimuli was 82.5% and deviant stimuli was 17.5%.  Every standard or deviant stimulus was followed by a response imperative signal in attended modality.  The DRC was obtained by subtracting the ERPs of the standard stimuli from that of the deviant stimuli.  The attention components were obtained respectively by subtracting the ERPs of the standard and deviant stimuli in the unattended condition from those of the same standard and deviant stimuli in attended conditions.  To clarify present standpoints that are the central issues of the ERPs research on selective attention, this experiment provided the further evidences and put forward some new ideas.

The basic design involves presentation of a series of pictures and tones in random sequence; hence the term “cross-modal”. Stimuli in an unattended channel are presented during the interval between a stimulus in the attended channel and a response imperative signal. The task of the subject is first discriminate between standard and deviant stimulus in the attended channel, but only when the response imperative signal appears; hence the term “delayed response”. The task was emphasized to subjects and response imperative signals were too small (for visual modality) and too weak (for auditory modality). Therefore even though the ISI is long, subjects find it difficult to pay attention to irrelevant stimuli, so that attention is focused on the relevant stimulus.

The main results and conclusions obtained in the present experiment are as follows:

1. The nature of ERPs components elicited by the non-verbal stimuli was identical with that found in classical oddball paradigms.  For example, the N1 was distributed at the frontal scalp regardless of modality or attention condition.  However, the largest N1 peak elicited by spoken words was located at the fronto-central scalp and the N1 elicited by the written words was located at lateral occipital and posterior-temporal area regardless of modality or attention condition.  This showed an obvious modality difference and suggested that the N1 elicited by verbal is probably specific index reflecting primary processing of Chinese shape and sound.

2. The N1 elicited using the non-verbal auditory standard, visual standard, and the deviant stimuli at the frontal area is increased under the attention condition than under the unattended condition.   Under the verbal attending condition, the N1 yielded by the auditory deviant stimuli was increased at the fronto-central area and that by the visual standard, furthermore, the deviant stimuli was increased at the occipital and posterior-temporal areas.  This showed that the N1 selective enhancement caused by both verbal and non-verbal attention processing.  The present results suggest that the N1 is the outcome of overlapping with the endogenous and exogenous components in view of the dispute whether the N1 is of exogenous or mixed components.

3. Under both attended and strict unattended conditions, an early deviance-related negativity (DRN1) was evoked by using Chinese orthographic and phonological processing.  The basic properties, such as peak latency, distribution and attention effects were analogous with the mismatch negativity (MMN) except for scalp distribution of DRN1 peak evoked under the unattended condition.  In addition, these properties of DRN1 yielded by visual verbal and non-verbal stimuli were also proved as same as auditory MMN.  Regarding the question as to whether or not visual MMN and verbal MMN exists, it can be initially believed that the MMN exists in the visual modality and in the language processing process in the current experiment.

4. The auditory and visual MMN amplitudes were unaffected by attention while comparing them under attended with unattended condition.  The DRN1 amplitudes evoked by verbal were unaffected by attention.  The present research suggests that the MMN amplitude is independent of attention, therefore, it can be looked at as an important index of the reflection of automatic processing process of verbal and non-verbal stimuli in connection with a lasting debate about whether the MMN is affected by attention.

5. The onset time of early attentive components (Nd1) elicited by non-verbal stimuli was the same as that of N1 yielded under the attention condition but was earlier than that of N1 yielded under the unattended condition.  The time sequence of the Chinese orthographic and phonological processing is as follows: the Nd1 of relatively pure attention components, the attention N1 of combined with endogenous and exogenous components, and inattention N1 of relatively pure exogenous component. To counter the basic problem of the early or late selection, the present experimental results show that both verbal and non-verbal attention effects first occur before the exogenous components and this provides evidence supporting early selective theory.

6. In the non-verbal task, the largest Nd1 elicited by auditory and visual deviant stimuli was distributed respectively over their primary sensory projection area, but Nd1 evoked by auditory and visual standard stimuli was over the frontal scalp.  In the verbal task, the largest Nd1 yielded by auditory deviant stimuli were located at the fronto-central area and visual ones were located at the occipital area, however, the Nd1 elicited by auditory and visual standard stimuli were located over the occipital area.  This suggests that the processing location of cross-modal attention to deviant stimuli is modality-specificity, while standard stimuli use a supramodal mechanism for non-verbal task. In the verbal task, the processing location of selective attention for deviants is different, however, for standards, it is common between modalities.  The present research suggests that early- and late- selection probably depends on various conditions for the selective attention, this has been debated for a long time.  This means the early-and late-selection of attention has plasticity.

7. According to the characteristics of N350 peak latencies, amplitudes and its scalp distribution, the current results suggest that the N350 is overlapping of N200 and N400, but N400 plays a more important role under the attended condition.  On the other hand, the N350 consists mainly of the N200 component under the unattended condition.  They are not immutable and this gives an expression that ERPs components reflecting a plasticity of the processing mechanism in the brain.

Key Words: Event-related potential (ERPs), Selective Attention, Deviance-related components (DRC), Attentive components, Negative Difference (Nd), Mismatch Negativity (MMN), Cross-modal and delayed response paradigm, verbal and non-verbal task