【作者】 胡艳梅；

【导师】 张明；

【作者基本信息】 东北师范大学 ， 发展与教育心理学， 2014， 博士

【摘要】 对注意选择的研究发现，不仅物理特征凸显的刺激能捕获注意，视觉工作记忆内容对注意也具有引导作用（即，基于记忆的注意引导过程）。空间维度方面，研究普遍发现工作记忆内容的保持依赖于对记忆位置的持续注意；处于记忆位置上的刺激会得到更快更好的加工。而客体维度方面，在工作记忆内容是否能捕获注意这一问题上现有研究却还存在很大争议。偏向竞争模型假设工作记忆内容即是注意模板；与记忆内容匹配的刺激总会捕获注意。而视觉注意理论（TVA）则认为工作记忆内容对注意选择的影响更为灵活。实验研究中，也存在大量矛盾对立的结果：一些研究发现客体工作记忆内容对注意的捕获作用是机械的、不随意的；另一些研究则指出，客体工作记忆内容对注意的引导过程是灵活可控的。当客体工作记忆内容与当前任务无关时，与记忆内容匹配的刺激可以受到抑制。为了解决现有研究的矛盾，本研究着眼于认知控制对基于记忆的注意引导过程的影响，结合行为方法与ERP技术，深入探讨（1）客体工作记忆内容是否一定会捕获注意；（2）客体工作记忆对注意的引导过程中认知控制的作用及其时程特点。本研究采用工作记忆与视觉搜索双任务范式，要求被试在保持一个记忆项目的同时完成一项视觉搜索任务。研究的关键操作之一是：搜索任务中是否有分心物与记忆项目匹配（实验1-6）。记忆匹配刺激在匹配试次中出现，而在控制试次中不出现。记忆匹配刺激对搜索任务的影响反映了客体工作记忆内容对注意的引导过程。研究一探讨认知控制倾向对基于记忆的注意引导过程的影响。通过改变任务提示语来操纵被试的认知控制倾向。实验1指导语明确提示被试“记忆项目永远不同于搜索目标”。结果发现，匹配试次搜索反应时快于控制试次。即记忆匹配刺激受到了抑制，从而促进了搜索任务表现。实验2指导语不作上述提示。结果发现，匹配试次搜索反应时更慢。即记忆匹配刺激先于搜索目标捕获了注意，从而干扰了搜索任务表现。研究一结果表明：（1）客体工作记忆内容既可以捕获注意，又可以受到抑制；（2）诱发抑制性认知控制作用的前提是：个体清楚工作记忆内容与当前任务无关。研究二进一步探讨认知控制强度对基于记忆的注意引导过程的影响及其时程特点。通过改变记忆匹配刺激的出现概率来操作被试的认知控制强度；概率越高，认知控制强度越大。同时，在研究二（以及研究三）中始终明确告知被试记忆项目与搜索目标无关。实验3（行为实验）结果发现，80%条件下（高强度），记忆匹配刺激受到了有效抑制，搜索速度更快；20%条件下（低强度），记忆匹配刺激捕获了注意，延缓了搜索速度。实验4利用ERP技术探讨：不同加工阶段，认知控制强度对基于记忆的注意引导过程的影响。主要考察的指标包括：（1）N2pc成分（200-300ms左右）；（2）N2pc消失后出现的差异正波（300ms以后）。前者反映了刺激对注意的捕获作用；后者则与注意捕获后的抑制过程有关。结果发现，20%条件下，记忆匹配刺激仅诱发了N2pc成分；50%和80%条件下，记忆匹配刺激先诱发了N2pc成分，后诱发了抑制性差异正波。并且，80%条件下N2pc波幅更小，差异正波波幅更大。同时，N2pc潜伏期随着认知控制强度的增大而缩短。研究二结果表明：（1）虽然行为指标上显示记忆匹配刺激受到了抑制，但ERP指标却证明客体工作记忆内容总会捕获注意；（2）客体工作记忆内容捕获注意后还可能受到抑制，该过程的效力和时程受到认知控制强度的调节。研究三从资源有限性的角度入手，着眼于认知控制负载对基于记忆的注意引导过程的影响。通过改变次任务类型来操作认知控制负载。高低负载条件下被试在完成双任务的同时还需分别进行减三负运算和发音抑制任务。实验5（行为实验）发现，低负载条件下，记忆匹配刺激受到了抑制；而高负载条件下，记忆匹配刺激既没有捕获注意，也没有受到抑制。实验6利用ERP技术进一步证实，高负载条件下记忆匹配刺激既没有诱发N2pc成分，也没有诱发抑制性差异正波。同时，高负载条件下，P3成分波幅显著小于低负载条件。这表明，当认知控制资源匮乏时，不仅没有多余的资源对任务无关记忆匹配刺激进行抑制，客体工作记忆内容对注意的捕获作用也受到了削减。本文通过三个紧密相关的研究深入细致地揭示了客体工作记忆内容对注意的引导过程中认知控制的作用及其时程特点，对已有研究发现进行了整合。首先，基于记忆的注意引导过程是灵活可控的；记忆匹配刺激不仅可以捕获注意，还可以受到抑制。其次，客体工作记忆内对注意的引导过程可以分为两个阶段。当认知控制资源充足时，加工早期客体工作记忆内容总能捕获注意；加工后期客体工作记忆内容可以受到抑制。当认知控制资源匮乏时，早期和晚期的自上而下加工均受到了极大的削弱；客体工作记忆内容既不会捕获注意，也不会受到抑制。因此，本研究一方面支持了偏向竞争模型的观点，证实了加工早期工作记忆内容对注意的捕获效应确实是机械的、不随意的。另一方面也对偏向竞争模型进行了补充，强调基于记忆的注意引导过程受到认知控制强度和负载的调节。更多还原

【Abstract】 Researches on selective attention suggest not only salient stimuli will capture attention;the contents of visual working memory also have a guide role on attention (i.e.,memory-based attention guidance). In spatial dimension, studies had consistently found theretention of working memory depends on sustained attention to the memorised position.Thus, stimulus processing on the memorised position will be facilitated. However, in objectdimension, the remained question is whether the contents of working memory would alwayscapture attention. According to Bias Competition Model, working memory contents are keptas attention template and stimulus matching memory item would automatically captureattention. In contrast, the Theory of Visual Attention (TVA) proposed the effect of workingmemory on attentional selection in a more flexible way. Experimental studies also found anumber of conflict results. Some researchers found the attention capture effect caused byobject working memory was obligatory. However, others indicated the contents of objectworking memory were not necessary to capture attention. Memory-matching stimulus can berejected if it was task irrelevant. In order to find an interpretation on these conflict findings,the current research investigated the role of cognitive control on the memory-based attentionguidance effect. Specifically, the current research tested (1) whether or not the contents ofobject working memory would involuntarily capture attention and (2) the role of cognitivecontrol and its time course during the guidance effect of working memory on attention.The current research employed working memory and visual search dual-task paradigm.Participants were instructed to conduct a visual search task during the retention of a memoryitem (Experiment1-6). One of the key operations throughout the whole research was:whether or not there were some search distracters matching the memory item. Suchmemory-matching items were presented in matching trials and absented in control trials. Theeffects of the memory-matching items on search performance reflect the guidance effect ofobject working memory on attention.The first research investigated whether the memory-based attention guidance effect couldbe affected by cognitive control. Cognitive control behaviours were operated by varying taskinstruction. In Experiment1, participants were directly told that “the memory item is alwaysdifferent from the search target”. The results showed visual search RTs were faster inmatching trials in comparison of control trials, suggesting memory-matching items had beenrejected and search performance had been facilitated. In contrast, search RTs were slowerwith memory-matching items in Experiment2where the above instruction was removed;suggesting object working memory contents captured attention and disrupted the searchperformance. Thus, the first research suggested:(1) the contents of object working memorycould either capture attention or be rejected by attention;(2) participants must be convincedof the irrelevance of working memory contents to search task if memory-matching itemscould be rejected. The second research further investigated the effects of cognitive control power on memory-basedattention guidance. The power of cognitive control was operated by varying the proportion ofmatching trials. The higher the proportion is, the stronger the control would be. The results ofExperiment3(behavioural experiment) showed search RTs were faster in matching trials with strongcognitive control and was slower in matching trials with weak control. Experiment4employed ERPtechnology to investigate how the effect of cognitive control varies over time. The main indexes were(1) N2pc component (about200-300ms) and (2) the positive difference waves after N2pc. The formerrelated to the attention capture effect, the later related to an inhibitory effect after attention capture.Memory-matching item only elicited the N2pc component with weak control. In contrast, both theN2pc component and the positive difference waves after N2pc caused by memory-matching itemwere found with stronger control. Furthermore, as the power of cognitive control increased, both theamplitude and the latency of N2pc component reduced, whereas the amplitude of the positivedifference waves increased. The second research suggested:(1) the contents of object workingmemory would always capture attention in the earlier process stage (about200-300ms).(2) Thecontents of object working memory could be rejected after capturing attention. The efficiency andtime course of such inhibitory effect depend on the power of cognitive control.The third research investigated the effect of cognitive load on memory-based attention guidance.The load of cognitive control was operated by conducting different secondary tasks. Participants hadto additionally conduct either a backward counting task in steps of3(high load) or an articulatorysuppression task (low load) during each trial. Experiment5(behavioural experiment) showedmemory-matching items had been rejected with low load. In contrast, they had neither capturedattention nor been rejected with high load. Experiment6used ERP technology convinced that thememory-matching item elicited neither N2pc component nor the positive difference waves afterN2pc. Moreover, the amplitude of P3component was reduced with the increase of cognitive load.The third research suggested the top-down processing during the memory-based attention guidancewas widely eliminated by the lack of cognitive resource.The current research revealed the effects of cognitive control and its time course duringmemory-based attention guidance, suggesting a powerful interpretation on previous conflict results interms of cognitive control. Firstly, the memory-based attention guidance effect was highly flexible.Memory-matching item could either capture attention or be rejection. Secondly, memory-basedattention guidance includes two phases. The contents of object working memory would alwayscapture attention during the earlier processing stage and could only be rejected in the later processingstage with sufficient cognitive resource. However, top-down processes in both the earlier and thelater phases would be greatly disrupted by the lack of cognitive resource. Broadly, the presentresearch supported the Bias Competition Model by showing that the contents of working memorywould involuntarily capture attention during earlier processing. However, such memory-basedattention capture is affected by both the power and the load of cognitive control.更多还原