【作者】 张炼；

【导师】 李红；

【作者基本信息】 西南大学 ， 发展与教育心理学， 2010， 博士

【摘要】 转换任务是广泛用于研究认知灵活性的重要任务之一。在进行人机交互设计过程中,由于人机交互要求操作者监控多种信息来源,并在短暂的时间内在不同的任务之间进行转换,因此也需要能够灵活进行任务转换。尽管自Jersild(1927)使用转换任务以来,人们不断使用各种转换任务范式对转换任务进行研究,已经有很多转换任务变式,目前至少可以归纳为两类转换任务范式：可预期转换任务范式和不可预期转换任务范式。可预期的转换任务范式通常是完成的任务是固定顺序的,可提供线索也可以不提供线索,因此,一般也称为固定顺序转换任务。不可预期的转换任务,一般会通过线索表明当前需要完成的任务类型,因此也称为线索转换任务。以往有许多研究致力于找出转换代价的来源,最初的理论倾向于根据某个机制对转换代价做出解释(Rogers,1995; Allport,1994),尽管仍然有研究在提出任务转换的单一因素模式(Altmann,2002但多数研究者则认为有多种机制,因此也不断有人对哪些是真止与转换代价相关的影响因素进行了争论。本文通过delta图方法动态分析转换任务加工过程的冲突控制机制,来分析转换代价的动态变化,可以更深入地理解转换代价的实质。神经成像研究表明,任务转换与广泛的神经网络有关,包括外侧前额叶皮质区域和顶叶皮质区,前辅助运动区,以及前扣带回等部位。实际上,对转换任务的神经心理学研究中,存在多种转换任务范式,而不同转换任务范式之间可能本身的加工就存在着差异,因此具体对某一范式进行深入研究,可以对具体的转换任务加工范式更进一步理解。研究选取字母和数字作转换任务的研究材料,分别从纵向和横向两个方面具体研究了线索消失转换任务的神经机制。首先,通过对固定顺序任务,线索保留任务和线索消失任务进行横向比较,并通过delta图分析方法,得出了转换代价的动态变化特点,表明转换代价并非是转换任务加工中的必然产物的结论。通过对线索消失任务的进行深入的ERP研究,了解了线索消失任务的脑电加工的时程特点以及序列顺序效应和冲突控制效应,同时也通过ERP研究对线索消失转换任务的线索效应进行了专门的分析,得出线索长短以及线索是否与任务冲突对线索消失转换任务加工的影响。研究结果发现：(1)通过delta图法对转换代价的来源进行分析表明,转换代价并不一定是转换任务加工过程中的必然产物,而是与转换任务加工过程的动态加工机制相关,取决于对上一任务的反应抑制强度与当前任务的激活程度之间的动态变化关系。这就可以解释为什么不同范式会引起转换代价不同,也可以解释同一任务范式条件下不同准备时间之间的差异。(2)本实验通过事件相关电位对线索锁定时段和刺激锁定时段加工进行分析,得出了线索消失转换任务在不同阶段的加工差异和不同效应。在线索加工阶段,主要表现为重复任务与转换之间的差异,执行重复任务时,由于需要对反应抑制进行控制,需要更多的资源,在500-700ms重复任务比转换任务在负方向引起更大的P3b。在刺激加工阶段,转换任务在早期注意阶段的早期注意较晚,且需要更多的注意资源加工,因此表现为转换效应,在ERP上表现为转换任务的早期P2成分显著大于重复任务的P2,潜伏期也更长,这表明了在早期注意上转换任务投入了更多的资源,同时调动早期注意也更慢。在任务加工阶段,当完成任务时,由于数字任务熟悉度较高,因此完成字母任务时,数字作为无关刺激的干扰会更大,而完成数字任务时,字母作为无关刺激则干扰会更小,因此表现为任务效应。在ERP波形上表现为字母任务比数字任务引发了更大的700-1000ms的晚正成分。(3)线索消失转换任务加工过程中,长线索与短线索加工表现出差异。在线索锁定的时间段窗口,线索效应和任务效应表现出分离,在300～550 ms内的P3b成分上表现为任务类型的主效应,重复任务比转换任务诱发了更大的P3b成分,对应的偶极子源为扣带回和右侧额叶附近,在550～700 ms内的LPC成分上表现为线索类型的主效应,短线索条件比长线索条件诱发了更大的LPC成分,对应的偶极子源位于右侧额叶和右侧枕叶BA19部位。在刺激锁定的时间窗口,主要表现为早成分上的差异。即在N1效应上,长线索比短线索潜伏期更长,但波辐更大,这可能表明了长线索比短线索早期注意慢,但刺激呈现在视觉上引起的冲突更大。在P2效应上,短线索潜伏期更短,波辐更大,这可能表明了短线索早期注意更快,投入的早期注意资源更多。这一结果反映了长短线索在预期准备上的差异。(4)线索消失转换任务加工过程中,线索是否与任务一致会影响转换任务的加工。线索锁定的时间窗口,在前中部电极点,转换任务的P2潜伏期更长。在400-800 ms内的P3b成分上,重复任务比转换任务诱发了更大的P3b成分。在800～1000 ms内的晚期正波上,重复任务比转换任务的晚期止波波辐更大。在后部电极,冲突线索比一致线索出现一个更正走向的波。这表明了在冲突线索条件下,有更多的预期准备。在刺激锁定的时间窗口,在早成分上表现出差异。即在前中部电极上,转换任务比重复任务条件引发更大的P2峰值,一致线索的P2潜伏期比冲突线索的潜伏期更长；一致线索的N2潜伏期比冲突线索的N2潜伏期更长。这表明了转换任务投入的早期注意资源更多,而当线索与任务冲突条件下,被试更早注意到这种冲突并对刺激进行识别。(5)对线索消失转换任务的序列顺序特点研究表明,不管是线索锁定时段,还是刺激锁定时段,均只表现出显著的一阶顺序效应,而没有显著的二阶顺序效应：即当前试次(n)与上一试次(n-1)之间的不同任务序列顺序,会对任务转换加工有显著影响,而上一试次(n-1)与上两个试次(n-2)之间的任务序列顺序,则对任务转换加工没有显著影响。尽管行为研究表明,在反应时测量上,既表现出一阶顺序的效应,也表现出二阶顺序的效应,由于在ERP和fMRI研究中都没有得到相应的直接证据,因此对于行为结果的解释上,也应更谨慎一些。二者结果上的差异,可能既来源于行为研究方法和ERP和fMRI研究方法本身的差异,也可能来自于实验设计本身。前者关注反应时加工过程中的变化,而后者更关注在不同反应时段脑电成分的差异。(6)通过对线索消失转换任务的冲突控制机制研究表明,线索消失转换任务加工过程中,不管是线索锁定时段,还是刺激锁定时段,均只表现出显著的冲突类型效应,而没有显著的控制类型效应：这就表明,上一任务与当前任务是否冲突,也就是说当前执行冲突任务时,会影响当前任务的加工,但上一任务完成后所处的控制水平对当前任务的加工没有影响。本研究具有重要的理论意义和现实意义。在理论意义上,加深了对转换任务中转换代价的理解,丰富和充实了转换任务加工的相关理论,同时在研究方法上,采用delta研究方法研究转换任务的冲突控制机制,对转换任务加工过程进行动态分析,不再从某个单独的理论出发,更具有生态效度。从实际意义来看,线索消失转换任务加工的特点对于以后研究人类活动的认知灵活性,特别是设计人机交互界面及建立各类交通信号系统,可以提供很重要的设计理念支持。更多还原

【Abstract】 Task switching is one of the important paradigms used extensively to study cognitive flexibility. Operators are required to monitor multiple information sources and switch between different activated under time pressures. So switching flexibility from different tasks is demanded in the design of human-machine interfaces. Since Jersild adopted task-switching paradigm, this paradigm has many revised version and has been extensively used to study task switching. There are at least two types of paradigm:predictable task-switching paradigm and unpredictable task-switching paradigm. The common way of providing advanced information predictable task-switching paradigm is using a fixed and easily predictable task-sequence which is provided cue or not. So this paradigm is called fixed sequence paradigm as well. It is generally provided a cue before you perform a task in predictable task switching paradigm, and its so called task-cueing paradigm or cued task-switching paradigm.Previous studies are focused on investigate the source of the switch costs in task switching. Some researchers construct their theory in terms of only one mechanism to explain switch costs. Although there still researchers proposed many single facto model in studying switching tasks, more and more researchers suggested that there should be existed multiple mechanism in task switching. Dispute has not been stopped on the actual factors that modified the switch costs. We adopted delta plots to analyze the conflict control mechanism dynamically in processing switch task in order that we can explore the dynamic change and understand the essential of the switch costs deeply. Actually, there are multiple paradigms of task switching when adopted neuro-psychology method to investigate switching task. Different task-switching paradigms are variable in process per se. we should study one of the paradigm concretely and deeply in order to comprehensively understand concrete processing of task-switching paradigm.Applying letter and number as experimental materials of task-switching study, we studied the neural mechanism of cue-removed task-switching paradigm in cross-sectional study and developmental study. Firstly, adopting delta plot methodology we compare three task-switching paradigm-fixed sequence task-switching paradigm, cued task-switching paradigm and cue-removed task-switching paradigm. Based the data followed we draw a conclusion that switch costs are variable dynamically and not inevitable result in processing switch task. Then we deeply explored the time characteristics of event-related potential in cue-removed task-switching paradigm and studied sequence order effects and conflict control effects of this paradigm. Furthermore we investigate the cue effect by comparing the short cue and long cue and compareing consistent cue and conflict cue for the sake of understanding the cue presenting mode on processing switch task.The main results are described in the following:(1) By analyzing the source of the switch cost with delta plot, we conclude that switch costs are not inevitable product in processing switch task. It is involve in the dynamical change mechanism when processed switch task. It depends on the balance between the degree of response inhibition of the last trial and the degree of activation the current trial. So we can explain why different task-switching paradigm product different switch cost, and we can also explain why there is different switch cost adopting the same switching task when the preparing time is different.(2) Applying ERP technique, we analyze the cue-locked time window and stimulus-locked time window individually. The different process mechanism and different effect in these two phases are put forward. There are task main effects between repeat task and switch task when analyzing the cue-locked time window. This because more sources are demanded when one performs the repeat task since you should control the response inhibition. Repeat task evoked larger P3b potentials than switch task at 500-700ms. Switch task are noticed later than repeat task in early attention phases. Its process needs more attention resources than repeat task as well. The ERP components reflect the switch effects is larger P2 amplitude in switch task than repeat task. The latency of P2 in switch task is longer than repeat task. It suggested that subjects put more resources into switch task in early attention time, so they modulate their attention slowly. There are task main effects after stimulus onset. Its due to the different interfere to number task and letter task. It’s well known that we are familiar with number tasks when compare with letter tasks. When letter stimulus is the irrelevant stimulus, there are more disturbs when perform number task than perform letter task since letter task would interfere with number task. So the main task effects attained. Letter task elicit larger late positive components (LPC) at 700-1000ms time phase.(3)There are differences between short cue and long cue in cue-removed task-switching process. The cue effects and task effects are dissociate at cue-locked time window. The ERP components reflects the task effects is a larger P3b amplitude in repeat task than switch task in 300-550ms cue-locked window, and the components reflects the cue effects is a larger LPC in short cue than long cue. The result of dipole source indicates that the encephalic source is Cingulate Cortex and right frontal lobe for the former effects, and the right frontal lobe and BA19 of right occipital lobe for the latter effects individually.The early components at stimulus-locked window are attained that a longer latency of N1 component and larger amplitude of P2 component. This may suggested that one notice the long cue slower than that of short cue, but the long cue stimulus evoke more conflict that that of short cue. Also in early components of P2, the latency of short cue is shorter than that of long cue, but the amplitude of short cue is larger that of long cue.it may reflect that the different anticipated preparation of different cue.(4) Whether the cue is consistent with the task or not may affect the process of cue-removed task-switching. A longer latency of P2 components is got in switch task than in repeat task at front and middle electrode. The ERP component reflects the task effects is larger P3b amplitude in repeat task than switch task in 400-800ms cue-locked window. Repeat task elicit a larger LPC than switch task in 800-1000 ms time window. At the posterior electrode, conflict cue evoked a more positive trend component than consistent cue. This may suggest that subjects make more preparation when the cue is not consistent with the task.The early components effects are significant at stimulus-locked window. Larger P2 amplitude is elicited in repeat than in switch task. The consistent cue evokes a longer latency of P2 and N2 components than the conflict cue. It may imply that subjects devoted more early attention to the stimulus process in switch task than in repeat task, however when they process a conflict cue task, they may notice the conflict information and identify the stimulus quickly. (5) The result of the sequence effects of cue-removed task-switching is put forward as follows. There are significant first order effects but not second order effects both in cue-locked time window and stimulus-locked window. In other words, there are significant effects to process switch task when the current trail is not the same task as the last trail. But it’s nothing important when the last trial is not the same task as the trial before it. Though behavioral study showed that there is both first order effects and second effects in reactive time of processing switch task. We should be cautious to explain the result since we haven’t got the same result from the ERP and fMRI study. The different result may come from the methodology of behavior and ERP、fMRI, it may come from the experimental design per se. Since former method is focus on the variety of the process, the latter focus on the difference components on different reactive time phase.(6) The conflict control mechanism research of cue-removed task-switching showed that there is significant conflict effects in processing the task not only in cue-locked time window but also in stimulus-locked time window. No significant control effects exist in the cue-removed task-switching. This may suggest that the last task would affect the process of the current task when the last task is colliding with the current task. That is significant effects when perform current conflict task which is not the same as the last task. But the control level is not affect the current process whatever the control level is.There are important theoretical meaning and practical meaning of this research. Theoretically, we can comprehensively understand the switch cost of task switching. It can enrich the relative theory of process the switching task. Furthermore, we apply the methodology of delta plot to research the conflict control mechanism in switching task. This method analyzes the processing of task switching dynamically. It has more ecological validity than just analyzing by one single theory. Practically, the nature of cue-removed task-switching paradigm is valuable to further study the cognitive flexibility of human being. It can support an important design thought especially for construct human-machine interface and traffic signal system.更多还原