【作者】 朱怡；

【导师】 胡谊；

【作者基本信息】 华东师范大学 ， 基础心理学， 2022， 博士

【摘要】 人类的学习,大都依赖于直接或替代的反馈,前者来自于直接参与的经验学习,而后者来自于观察他人经验进行的替代学习。教学反馈是反馈的一种重要形式,指由教师提供的关于学生表现或理解方面的信息,帮助学生了解并缩小自己当前学习状态与目标状态之间的差距。反馈中的信息,按其复杂性可分为正误性和精细化信息:前者是指对当前回答正确或错误的简单判断,而后者包含多种类型,比如解释当前答案正确或错误的原因,提供样例或重复呈现学习材料等,被统称为精细反馈。在认知主义学习理论框架下开展的反馈研究,大多通过设定好的教学程序为学习者呈现反馈信息,主要关注学习者如何加工反馈信息;比较一致地发现精细反馈相比于简单反馈更有效,主要表现在促进深层次的学习上,如推理和迁移。但是,在真实的学习中,反馈通常是发生在特定社会情境中的一种老师与学生互动和对话的过程,它不局限于一对一的师生互动模式,更常见于一对多的师生互动中。那么,发生在真实师生互动中的精细反馈又会如何影响学习:(1)在一对一的师生互动中,直接的精细信息反馈是否/如何促进深层次的学习?保持信息内容不变,使呈现形式精细化,比如“组块”,是否/如何进一步促进深层次的学习?神经基础是什么?(2)在一对多的师生互动中,替代的精细反馈是否/如何促进深层次的学习?榜样-观察者的相似性是否/如何影响基于替代精细反馈的学习效果?神经基础是什么?继续使用以往的研究范式(计算机教学或非实时观察任务)和技术手段(单人行为/脑数据采集)不足以全面考察真实师生互动中精细反馈的作用及其神经基础,因为(1)计算机教学或非实时观察任务都无法模拟互动情景,缺失了师生互动中人际间的相互作用与动态的信息交换过程;(2)基于互动脑假说,人际互动所涉及的神经过程,本质上区别于与个体行为关联的神经活动;所以,同时采集并探究参与或观察对话式精细反馈的老师和学生的大脑活动是必要但缺乏的。基于上述不足,本研究在社会文化观学习理论框架的指导下,基于原计算机任务改编开发了真实师生互动的问答反馈任务,该任务兼具实验控制和生态效度,适用于在一对一或一对多的师生互动中探究直接和替代精细反馈的作用,并结合近红外多脑同时扫描技术来考察精细反馈的人际神经基础。研究一采用一对一师生互动的问答反馈任务,考察师生互动中精细化(信息内容,实验1;呈现方式,实验2和3)的直接反馈的作用及其人际神经基础。实验分两次进行,第一次,对老师和学生被试分别进行培训和前测;间隔几天,第二次针对10个社会心理学概念依次进行老师提问-学生回答-老师反馈等环节,共10个试次;问答反馈结束后对学生被试进行知识后测,包括再认和迁移。实验1中,老师为两组学生分别提供简单反馈(正确答案)和精细反馈(正确答案+举例);根据前人研究,加工反馈信息会激活学习者和教学者的额、顶区域,故将上述区域作为感兴趣区,放置近红外光极用于同时记录反馈过程中师、生的大脑活动,并结合实验录像选取反馈时间段数据。因此,实验1采用单因素(反馈内容:精细vs.简单)组间设计。在行为实验2中,老师为两组学生分别提供组块(两个相关概念)或单独(同实验1)的精细反馈(正确答案+举例),下文中简称为组块反馈和单独反馈;此外,通过设置流程是否存在学习导入阶段来操纵接收反馈前学生知识水平的高低;增加7天后第二次知识后测,考察学习的长时效应。因此,实验2采用2(反馈呈现:组块vs.单独)×2(知识基础:高vs.低)的组间设计。基于实验2的行为学结果,实验3中用无导入操纵学生接收反馈前的知识基础低;任务中同时采集师、生的大脑活动;为排除由组块呈现造成的反馈延时的干扰,新增了假组块反馈条件(与组块反馈的反馈时机相同,但包含两个不相关的概念);其他设置同实验2。因此,实验3采用单因素(反馈呈现:组块vs.假组块vs.单独)组间设计。行为结果表明:相比于简单反馈组,精细反馈组的学生在迁移测量上的正确率更高(实验1);反馈呈现和知识基础的交互作用显著(实验2);简单效应分析发现,在低知识基础条件下,相比于单独反馈组,组块反馈组的学生在7天后的迁移测量上的正确率更高(实验2);在低知识基础条件下,相比于非组块(假组块、单独)反馈组,组块反馈组的学生在7天后的迁移测量上的正确率更高,假组块组和单独反馈组之间无差别(实验3),排除了对实验2结果的反馈延时解释。中介分析发现,在低知识基础的条件下,存在从精细反馈的呈现方式通过错误修正的概念组数影响迁移正确率的间接通路(实验2、3),提示组块错误修正可能是反馈的精细化呈现促进迁移的认知过程。以上结果表明师生互动中直接反馈的精细化,包括信息内容和呈现形式的精细化,都可以带来深层次的学习增益,比如迁移,可能的认知过程是更有效的错误修正。脑-脑结果显示:相比于休息(无互动),精细反馈过程中,师-生在额、顶区域出现显著更大的脑同步;进一步,将精细反馈细分成正确答案和举例两部分,分别与休息做比较,发现只有举例部分的脑同步显著大于休息,也大于正确答案部分,提示脑间同步可能特异于反馈中的精细信息;基于支持向量机回归算法,师-生在顶叶区域的脑同步可以预测学生的迁移表现,且当脑同步存在一方领先或滞后的情况下,上述预测关系仍然成立(实验1),提示顶叶上的脑间同步可能是反馈中的精细内容加深学习的神经基础。相比于非组块(假组块和单独)反馈,组块反馈过程中,师-生在额、顶区域的脑同步更大;组块反馈过程中,额叶区域的师-生脑同步与学生学习后即时和学习后7天的迁移测量正确率、错误修正的概念组数均正相关(实验3),提示额叶上的脑间同步可能是反馈的精细呈现促进错误修正、加深学习的神经基础。考虑到人具有社会性,不仅可以从直接的经验中学习,也可以在没有直接接收到反馈的情况下,通过观察他人(即榜样)的经验基于替代反馈实现替代学习。研究二(实验4、实验5a、实验5b)在研究一的基础上新增一名学生被试,构成师-生-生一对多互动情境,模拟实时观察学习中同时存在老师、榜样和观察者的情况,主要考察一对多师生互动中替代精细反馈的作用及其人际神经基础。在行为实验4中,根据前测正确率差异分组来操纵两名学生接收反馈前的知识基础是否相似,通过是否直接和老师进行问答反馈操纵精细反馈的获取途径;榜样和观察者的角色固定,即指定两名学生被试各自参与或观察所有问答反馈。因此,实验4采用2(反馈途径:直接vs.替代)×2(知识相似:高vs.低)的混合实验设计,第一个为组内变量。基于实验4的行为学结果,实验5a中操纵榜样-观察者接收反馈前的知识相似高;任务中同时采集师、生的大脑活动;增加7天后第二次知识后测;其他设置同实验4。因此,实验5a采用单因素(反馈途径:直接vs.替代)组内设计。考虑到实时观察学习中同时存在榜样和观察者且角色固定导致的不平衡可能给他们的学习带来负面影响,实验5b中改用榜样和观察者的角色转换,即两名学生被试各参与并观察一半的问答反馈,其他设置同实验5a。行为结果表明,在学习后再认和迁移测量的正确率上,反馈途径和知识相似效应均不显著(实验4),提示替代精细反馈同样能促进学习;在学习后迁移测量的正确率上,反馈途径和知识相似的交互作用显著;简单效应分析发现,在替代反馈条件下,相比于知识相似低组,知识相似高组的观察者在学习后的迁移测量上的正确率更高(实验4),说明榜样-观察者相似性调节基于替代精细反馈的观察学习。在知识相似高的情况下,在学习后即时、7天后的再认和迁移测量的正确率上,反馈途径效应均不显著(实验5a、5b),提示替代精细反馈同样能促进学习,且长时保持。相比于接收替代反馈的观察者,接收直接反馈的榜样报告的任务负担更大,学习存在进程效应,学生后半程成绩下降(实验5a),实验5b中未发现上述结果,暗示实时观察学习中同时存在榜样和观察者且角色固定可能不利于直接和替代精细反馈发挥最佳的作用。脑-脑结果显示:相比于休息,直接/替代反馈过程中,榜样/观察者与老师在额、顶区域出现显著更大的脑-脑同步,且不存在反馈途径的效应(实验5a、5b),提示替代反馈和直接反馈可能具有相同的神经基础;相比于休息,观察者与榜样在额叶区域出现显著更大的脑-脑同步(实验5a);相比于休息,观察者与榜样在顶叶区域出现更大的脑-脑同步(实验5b)。在替代反馈过程中,观察者与老师在顶叶上的脑-脑同步与观察者学习后即时、7天后的迁移测量上的正确率正相关,观察者与榜样在顶叶上的脑-脑同步与他们7天后的平均迁移表现正相关(实验5b),提示顶叶上的脑-脑同步也可能是替代形式的精细反馈加深学习的神经基础。此外,顶叶上的脑-脑同步差值(观察者与老师-榜样与老师)与学习7天后迁移正确率增量的差值(观察者-榜样)正相关,也与学习后7天迁移测量上错误修正的个数的差值(观察者-榜样)正相关;顶叶上观察者-榜样的脑同步与学习后即时他们迁移测量上的正确率的差值绝对值负相关(实验5b),上述结果提示脑同步反映双方的学习/认知达到一致。综上所述,本研究通过两项研究6个实验在师生互动中揭示了直接和替代精细反馈的作用是加深学习、增益迁移,且作用具有长时性;可能的人际神经基础是师生之间的额、顶区域的脑-脑同步,该脑同步预测学习表现,反映错误修正并提示互动双方达到认知齐性。本研究为在社会文化观学习理论指导下开展互动情境中的教育心理研究,提供了可行的实验范式和技术手段。在理论层面上,本研究拓展了认知主义学习理论,在真实的师生互动中揭示了精细化的信息内容、信息呈现方式对深层学习如迁移的长时增益作用;丰富了互动脑假说,首次揭示了基于精细反馈的经验学习和实时观察学习的人际神经基础可能是额、顶区域的脑-脑同步,并提示直接和替代形式的精细反馈可能拥有相同的人际神经基础。在教育实践上,本研究可以为真实课堂中反馈环节的有效设计提供指导性的建议和启示。更多还原

【Abstract】 Humans’ learning mostly depends on direct or vicarious feedback.The former comes from experience learning,while the latter comes from vicarious learning by observing others’ experience.Teaching feedback is one of the important forms of feedback,which refers to the information about students’ performance or understanding provided by teachers to help students fill the gap between their current learning state and target state.According to its complexity,the information in feedback can be divided into verification and elaboration information: the former refers to the simple judgment of whether the current answer is correct or wrong,while the latter has many types,such as explaining the reasons for the correct or wrong answer,providing examples or repeatedly presenting learning materials,which are collectively defined as elaborated feedback.Most research carried out under the framework of Cognitivism learning theory presented feedback information to learners through instructional program,mainly focusing on how learners process feedback information;it has been consistently found that elaborated feedback is more effective than simple feedback,mainly in promoting deep-level learning,such as reasoning and transfer.However,in real-world learning,feedback is usually a process of interaction and dialogue between teachers and students in specific social situations.It is not limited to one-to-one teacher-student interaction mode,but more common in one-to-many teacher-student interaction.Then,how will the elaborated feedback in the real-world teacher-student interaction affect learning:(1)in one-to-one teacher-student interaction,does/how does the direct elaborated information feedback promote deep-level learning? With the information content unchanged,does/how does the feedback with elaborated presentation format,such as“chunking”,further promote learning? What is the neural basis?(2)In one-to-many teacher-student interaction,does/how does the vicarious elaborated feedback promote deep learning? Does/how does the model-observer similarity affect the learning effect d on vicarious elaborated feedback? What is the neural basis?Continuing to use the previous research paradigm(computer-based instruction or non real-time observation task)and technical method(single person behavior/brain data collection)can not satisfy the comprehensive investigation of the effects of elaborated feedback in real-world teacher-student interaction and its neural basis,because(1)neither computer-based instruction nor non real-time observation task can simulate the interaction scenario,lacking the interpersonal interaction and dynamic information exchange process in teacher-student interaction;(2)Based on the Interactive Brain Hypothesis,the neural processes involved in interpersonal interaction are essentially different from the neural activities associated with the individual behavior;therefore,it is necessary to collect and explore the brain activities of teachers and students who participate in or observe conversational elaborated feedback at the same time.Based on the above shortcomings,under the guidance of the Social cultural learning theory,this study adapted and developed a question-answer-feedback task based on teacher-student interaction.This task has both experimental control and ecological validity.It is suitable for exploring the role of direct and vicarious elaborated feedback in one-to-one or oneto-many teacher-student interaction,and combined with near-infrared multi brain simultaneous scanning technology to investigate the interpersonal neural basis.Study 1 used a one-to-one interactive question-answer-feedback task to investigate the effects of elaborated(information content,Experiment 1;presentation format,Experiment 2 and 3)direct feedback in teacher-student interaction and its interpersonal neural basis.The experiment was carried out over two visits to the laboratory,with the interval of several days.During visit 1,teachers and students were trained and pretested,respectively;During visit 2,10 social psychology concepts were instructed and learned,the flow relevant to one concept,that is one trial,could be split into teacher’s question,student’s answer and teacher’s feedback,with a total of 10 trials;After that,the students were tested for knowledge,including recognition and transfer.In Experiment 1,the teacher provided simple feedback(correct answer)and elaborated feedback(correct answer + example)to the two groups of students,respectively;According to previous studies,feedback information processing activated the learners’ and teachers’ frontal and parietal regions.Therefore,the above regions were regarded as regions of interest,and near-infrared light poles were placed to simultaneously record the teachers’ and students’ brain activities during the task,and data during the feedback period were selected according to the experimental video.Therefore,Experiment 1 used a univariate(feedback information: elaborated vs.simple)between-group design.In Experiment 2,the teacher provided two groups of students with chunking(two relative concepts)or separate(the same as Experiment 1)elaborated feedback(correct answer+ example),hereinafter referred to as chunking feedback and separate feedback;In addition,the level of students’ knowledge before receiving feedback was manipulated by setting whether there was a learning introduction phase;adding a second knowledge post-test after 7 days to investigate the long-term effect of learning.Therefore,experiment 2 adopted 2(feedback presentation: chunking vs.separate)× 2(prior knowledge: high vs.low)between-group design.Based on the behavioral results of Experiment 2,in Experiment 3,the students’ prior knowledge before receiving feedback was manipulated to be low;during the task,teachers’ and students’ brain activities were simultaneously collected;in order to eliminate the interference of feedback delay caused by the chunking presentation,a false chunking feedback condition was added(the feedback timing is the same as that of chunking feedback,but for two unrelated concepts);Other settings are the same as experiment 2.Therefore,Experiment 3 used a univariate(feedback presentation: chunking vs.false chunking vs.separate)between-group design.The behavioral results showed that compared with the simple feedback group,the students in the elaborated feedback group had higher accuracy in transfer measurement(Experiment 1);the interaction between feedback presentation and prior knowledge was significant(Experiment 2);simple effect analysis revealed that under the condition of low prior knowledge,compared with the separate feedback group,the students in the chunking feedback group had higher accuracy in the transfer measurement after 7 days(Experiment 2);Under the condition of low prior knowledge,compared with the nonchunking(false chunking,separate)feedback group,the students in the chunking feedback group had higher accuracy in the transfer measurement after 7 days.There was no difference between the false chunking group and the separate feedback group(Experiment 3),which excluded the feedback delay interpretation of the results of Experiment 2.Mediation analysis found that under the condition of low prior knowledge,there was an indirect path from the presentation formats of elaborated feedback to the number of relevant concept-pair error correction,which affected the accuracy of transfer(Experiments 2 and 3),suggesting that chunking error correction may be one of the cognitive processes of elaborated presentation of feedback to promote transfer.The above results suggest that elaboration of direct feedback in teacher-student interaction,including information content and presentation formats,can bring deep learning gains,such as transfer,and possibly cognitive processes that are more effective error correction.Brain-brain results showed that compared with resting(no interaction),teacher students had significantly greater brain synchronization in frontal and parietal regions during elaborated feedback;Further,the elaborated feedback was divided into two parts:the correct answer and the example,which were compared with resting,respectively.It was found that the brain synchronization of the example part was significantly greater than that of resting and the correct answer part,suggesting that the brain synchronization may be specific to the elaborated information in the feedback;Based on the support vector machine regression algorithm,the brain synchronization between teachers and students in the parietal lobe region predicted the transfer performance of students,and the above prediction relationship was still valid when brain synchronization was teacher-leading or lagging(Experiment 1),suggesting that the brain synchronization in the parietal lobe may be the neural basis of elaborated information in feedback to deepen learning.Compared with non chunking(false chunking and separate)feedback,the brain synchronization of teacher student in frontal and parietal regions was greater in the process of chunking feedback;During the chunking feedback,the teacher student brain synchronization in the frontal region was positively correlated with the accuracy of transfer measurement and the number of relevant concept-pair error correction immediately after learning and 7 days after learning(Experiment 3),suggesting that the brain synchronization in the frontal lobe may be the neural basis for the elaborated presentation of feedback,promoting error correction and deepening learning.Considering that people are social,it is possible to learn not only from direct experience,but also to achieve vicarious learning based on the vicarious feedback by observing the experience of others(i.e.,models)without directly receiving feedback.Study 2(Experiment 4,Experiment 5a and Experiment 5b)added a new student subject on the basis of study 1 to form a teacher-student-student one-to-many interaction situation,simulated the simultaneous presence of teacher,model and observer in realtime observation learning,and mainly investigated the effects of vicarious elaborated feedback in one-to-many teacher-student interaction and its interpersonal neural basis.In Experiment 4,according to the difference of pre-test accuracy,the two students were divided into groups to manipulate whether the prior knowledge before receiving feedback was similar,and whether to directly conduct question and answer feedback with the teacher to manipulate the access to elaborated feedback;In Experiment 4,the roles of model and observer were fixed,that is,two students were designated to participate in or observe all question and answer feedback.Therefore,Experiment 4adopted 2(feedback access: direct vs.vicarious)× 2(knowledge similarity: high vs.low)mixed experimental design,the first is a within-group variable.Based on the behavioral results of Experiment 4,in Experiment 5a,the manipulative model-observer knowledge similarity before receiving feedback was high;During the task,teachers’ and students’ brain activities were simultaneously collected;The second post-test of knowledge was added after 7 days;Other settings are the same as experiment 4.Therefore,Experiment 5a adopted a univariate(feedback access: direct vs.vicarious)within-group design.Considering that there are both model and observer in real-time observation learning,and the imbalance caused by fixed roles may have a negative impact on their learning,the exchanged role of model and observer was used in Experiment 5b,that is,two students participated in and observed half of the questionanswer-feedback,and other settings are the same as experiment 5a.The behavioral results showed that the feedback access and knowledge similarity effect were not significant in the accuracy of post-learning recognition and transfer measurement(Experiment 4),indicating that vicarious elaborated feedback can also promote learning;In the accuracy of post-learning transfer measurement,the interaction between feedback access and knowledge similarity was significant;Simple effect analysis found that under the condition of vicarious feedback,compared with the low knowledge similarity group,the observers in the high knowledge similarity group had a higher accuracy in the transfer measurement after learning(Experiment 4),indicating that the model observer similarity modulates the observation learning based on vicarious elaborated feedback.In the case of high similarity of knowledge,the effect of feedback access was not significant in the accuracy of recognition and transfer measurement immediately and 7 days after learning(Experiments 5a and 5b),suggesting that vicarious elaborated feedback can also promote learning and long-term keep.Compared with observers who received vicarious feedback,the model that received direct feedback reported a greater task burden,there was a process effect in learning,and students’ test accuracy declined in the second half of the learning(Experiment 5a).The above results were not found in Experiment 5b,implying that both model and observer co-exist in real-time observational learning with fixed roles may not be conducive to the optimal effect of direct and vicarious elaborated feedback.Brain-brain results showed that compared with resting,in the process of direct/vicarious feedback,the model/observer and teacher had significantly greater brain-brain synchronization in frontal and parietal regions,and there was no effect of feedback access(Experiments 5a and 5b),suggesting that vicarious feedback and direct feedback may have the same neural basis;Compared with resting,the observer and model showed significantly greater brain-brain synchronization in the frontal lobe region(Experiment 5a);Compared with resting,the observer and model showed greater brainbrain synchronization in the parietal lobe region(Experiment 5b).In the process of vicarious feedback,the brain-brain synchronization between the observer and the teacher on the parietal lobe was positively correlated with the accuracy of the transfer measurement immediately and 7 days after the observer’s learning,and the brain-brain synchronization between the observer and the model on the parietal lobe was positively correlated with their mean transfer performance after 7 days(Experiment 5b),suggesting that the brain-brain synchronization on the parietal lobe may also be the neural basis of vicarious forms of elaborated feedback to deepen learning.Furthermore,the difference in brain-brain synchronization in the parietal lobe(observer&teachermodel&teacher)was positively correlated with the difference in transfer accuracy increase(observer-model)after 7 days of learning,and also with the difference in the number of error corrections(observer-model)on transfer measures 7 days after learning;the observer-model brain synchronization in the parietal lobe was negatively correlated with the absolute value of the difference in their accuracy on the transfer measure immediately after learning(Experiment 5b),the above results suggest that the brain synchronously reflects the learning/cognition alignment.In summary,this study revealed that direct and vicarious elaborated feedback in the interaction between teachers and students was to deepen learning and increase transfer,and the effect was long-term;the possible interpersonal neural basis was brainbrain synchronization in frontal and parietal regions between teachers and students,which predicted learning performance,indicated error corrections and suggested cognitive alignment.This study provides a feasible experimental paradigm and technical methods for carrying out educational psychology research in interactive situations under the guidance of the Social cultural learning theory.At the theoretical level,this study expands Cognitivism learning theory,and reveals the long-term benefit of elaborated information content and information presentation formats on deep-level learning such as transfer in the real-life teacher-student interaction;enriches the Interactive Brain Hypothesis,for the first time,it is revealed that the interpersonal neural basis of experience learning and real-time observational learning based on vicarious feedback may be brain-brain synchronization in the frontal and parietal regions,suggesting that direct and vicarious forms of elaborated feedback may have the same interpersonal neural basis.In educational practice,this research can provide instructive suggestions and inspirations for the effective design of feedback in real classrooms.更多还原