基于脑电信号分析的消费者品牌延伸评估决策研究

【作者】 王小毅；

【导师】 马庆国；

【作者基本信息】 浙江大学 ， 管理科学与工程， 2008， 博士

【摘要】 决策神经科学(Decision Neuroscience)是快速成长的前沿交叉学科,也是管理科学的新生研究领域。它将管理科学延伸到神经科学层面,通过研究传统决策“黑箱”的神经活动,从而更好的构建决策行为模型,它可用于对消费者决策问题的研究。事件相关脑电位(ERP)是认知神经科学中最为成熟的脑扫描研究工具之一,它借助稳定的内源性脑电生理指标,结合行为学数据,能够研究传统的外显评估技术难以涉及的认知过程。运用ERP来研究消费者决策,不仅为消费者决策理论找到神经学证据,而且它所采取的基于过程测量的、多系统协作的视角,可弥补过去决策理论中基于结果测量的、单一系统框架不足。品牌延伸是常见的营销手段,主要用来降低推出新产品的成本,以及避免新品牌失败的风险。在过去二十多年里,学者们致力于探讨品牌延伸的评估决策过程,形成了围绕A&K模型(Aaker and Keller,1990)的主流学派,研究主题主要围绕影响品牌延伸态度形成的感知质量、契合性、类别相似性等变量;但是品牌延伸评估决策过程中的“黑箱”,即延伸评估的认知处理过程和伴随它的情感迁移过程却讨论不多,并由此带来一些学术争论。检验经典评估模型,弥补理论缺口,对关键性的学术争论给出解释,这是当前品牌延伸研究的主要任务。本研究以A&K模型等经典延伸评估理论为出发点,在研究方法上,以决策科学、消费者行为学和认知神经科学所共通的关联记忆网络为基础理论,并整合了品牌延伸理论、决策科学、认知科学、ERP研究以及最新的决策神经科学成果,从“认知—情感;自动—受控”四维度的决策加工处理视角,提出阶段性假设;并根据消费者通常进行延伸评估决策的过程特点设计了“刺激1启动(品牌名称)-刺激2目标(产品名称)”(prime—probe)的实验研究范式,通过ERP的过程测量来检验各阶段的评估决策变量,进而结合神经机制建立描述性评估决策模型,并以传统的消费者决策研究方法来对此进行检验;在内容上,建立了以母品牌“感知质量”为启动效应,通过“相似性匹配”的闸门作用来引导两类“分类处理”机制,再由“契合性判断”来完成协调与整合,最后进一步在“延伸态度”中完成情感评估,输出消费决策行为结果的阶段模型。本研究利用国际标准ERP/认知实验室,针对评估决策各阶段,历时2年,共设计并开展了包含不同控制条件的四项实验研究,找到了品牌延伸评估决策各阶段关键变量的脑电成分依据,从而支持了假设;并根据这些成分随各阶段时间进程的变化以及它们之间的数量关系,结合行为学数据,进一步探讨了品牌延伸评估决策的过程机制,最后形成了基于神经科学的品牌延伸五阶段评估决策模型。本研究还运用焦点小组座谈内容分析法在现实适用性上支持了“五阶段模型”;并用2448个样本规模的问卷数据对“五阶段模型”进行基于OLS回归和SEM路径分析的检验,发现不仅有较A&K模型更好的解释力,而且对于过去争论的焦点,如母品牌“感知质量”对品牌延伸评估决策的影响、“相似性”的作用、适度不“契合”更有利于评估、“契合性”与“延伸态度”如何导致最终决策等问题,给出统一的解释。在最后的展望中,本研究还分析了最近几年决策神经科学在国际顶级管理和科学期刊上的表现,并结合本研究提出的研究框架,设计了一些在该领域未来有望取得重大突破的研究方案。本研究提出的“五阶段模型”,包含的主要结论如下:(1)品牌延伸评估决策过程中存在母品牌感知质量的启动效应(阶段一)。它可由接受品牌名称刺激后诱发的LPC(晚期正成分复合波)脑电成分来表征,与记忆获取和情感(affect)处理有关;知名品牌能引起更大且偏脑区左侧化的LPC,说明营销中的品牌效应来源于知名品牌对正性情绪的激活。(2)品牌延伸评估决策过程中存在相似性匹配过程(阶段二)。它可由接受产品名称刺激后诱发的N270(270毫秒负波)脑电成分来表征,与基本属性的失匹配(mismatch)和冲突处理有关;与母品牌产品基本属性联系越少,N270越大。(3)品牌延伸评估决策过程中存在分类处理过程(阶段三)。它可由接受产品名称刺激后诱发的P300(300毫秒正波)脑电成分来表征。P300与类别处理、记忆更新、注意等有关;类别一致性延伸可以诱发更大的P300。(4)品牌延伸评估决策过程中存在契合性判断过程(阶段四)。它可由接受产品名称刺激后诱发的N400(400毫秒负波)脑电成分来表征,与大脑对冲突控制和高级协调有关;类别不一致性延伸可以诱发更大的N400。N400可由N270和P300振幅的线性函数来表示,表明契合性判断是一个由相似性匹配所引导(闸门作用),依靠分类处理机制所完成的一个自动加工处理结果。(5)品牌延伸评估决策过程中存在两种分类机制。N270对不同产品类别描述无显著差异,说明相似性匹配与分类处理是两个不同的过程;而P300和N400则均对产品类别描述敏感,且大类描述导致更大的P300,细类描述导致更大的N400,前者支持第一类基于类别的分类处理(category-based processing),而后者支持了第二类基于片断的分类处理(piecemeal integration processing)。(6)品牌延伸评估决策过程中存在延伸态度形成过程(阶段五)。它可由接受产品名称刺激后诱发的LPP(晚期正电位)脑电成分来表征,与情感评估有关;不一致性延伸可以诱发更大的LPP。同时品牌延伸接受率只有在控制了N400后才与LPP显著偏相关,且行为结果可由N400和LPP振幅的线性函数来表示,表明最终的决策行为可以通过契合性判断和延伸态度两个独立的系统共同作用而产生;前者以认知和自动加工过程为主,后者以情感和受控加工过程为主。本研究是决策神经科学研究的一次探索,在研究方法、研究范式、研究内容和结论上均具有交叉学科的创新之处,也具有较强的现实意义。主要有如下三个创新点:(1)紧跟当前国际上决策神经科学的发展趋势,采用脑扫描技术(事件相关脑电位,ERP)研究消费者决策问题。其意义有二:目前运用ERP研究决策问题较少,本研究是对近年来决策神经科学前沿研究成果的验证和补充;用ERP在毫秒级测量决策全过程的行为和脑电信号变化,从而在时间进程上研究品牌延伸评估决策机制,揭示传统研究难以涉及的决策过程黑箱。(2)在神经机制上支持了品牌延伸领域经典理论的基础假设,并完善了延伸评估决策模型。本研究以传统决策科学、行为决策和认知神经科学所共通的关联记忆网络理论作基础,针对以A&K模型(1990)为代表主流研究的核心问题,找到在决策过程中存在“感知质量”、“契合性”和“延伸态度”等关键变量的神经学证据;进一步通过对脑电信号动态变化的分析,提出了较A&K模型有更高解释力的“五阶段模型”。其中,“感知质量”的启动效应;以“相似性匹配”的闸门作用来引导两类“分类处理”机制并最终通过“契合性判断”完成冲突处理与协调;通过“契合性判断”的自动处理和“延伸态度”的受控处理双系统作用形成消费决策;等内容,均是对传统理论的有益补充。(3)“五阶段”模型针对理论争论焦点给出了基于神经机制的统一解释。本研究从决策神经科学的“认知—情感;自动—受控”四维视角出发,融合了消费者决策研究中长期对立并存的基于认知过程和基于情感过程的两大流派:而以神经机制为基础的阶段评估决策模型,对一些经典争论给出新的解释。如,发现母品牌的记忆和情感启动效应,解释了Aaker and Keller(1990)和后来Sunde and Brodie(1993)等研究关于实证中为何品牌感知质量不起作用的争论;通过对分类机制的探讨支持了Tauber(1988)情感迁移模型中以契合性作关键中介的假设,也强调了Bhat and Reddy(2001)属性联想模型中以认知评价为中介的合理性;以感知质量的启动效应和契合性判断的冲突处理机制来解释倒“U”型评估现象,支持了Mandler(1982)的图式调适解释和Martin andStewart(2001)的实证结果,也阐明了其他研究没有发现该现象的原因。更多还原

【Abstract】 Decision Neuroscience is a fast - growing interdisciplinary research area, including in management science. It can extend managemenr science into neural aspect with the studying of "black box" theory in traditional decision science, especially in consumer decision-making. Event-related Potential (ERP) is a mature brain scan technology in Cognitive Neuroscience. Its endogenous neural indicator is superior to explicit evaluation techniques in cognitive processing studies.Brand extension is one of common marketing methods and is used to reduce the cost of new products and the risk new brands. Over the past two decades and more, scholars made attempts at expounding the process of brand extension evaluation and consumer decision - making. Aaker and Keller’s model (1990) and the succession, the leading theory in brand extension, focused in the variables of "Brand Quality", "Fit", "Category" and so on, but little in the process of cognitive and affective processing of evalustion and decision-making. It is an urgent issue for many disputes in this area and a task to test the classic model, revamp the defects of leading theory and explain the conflict.In research methodology, this study based on the comman theory of associative memory model in decision science, consumer behavior and congnitive neuroscience, designed an experimental paradigms of "Stimulus 1brand name- Stimulus 2 product name" (prime-probe) , and build multistage hypothesis. With the process-based measurement of ERP, we test the the relationships of variables in the hypothesis and build a decriptive evalution and decision-making model and tested it with traditional methods. In research content, this study build a five-stage model: the priming of "brand quality", a dual process mechanism for "categorization" with the valve effect of "similarity" and be congruous in "fittness", be affective evaluated in "attitude" and get the decision output of behavior.With the ERP laboratory of international standard, this study designed and conducted four laboratorial study in two years, and found the ERP evidences for the critical variables in every stage of the brand extension evaluation and decision -making, so to support the hypothesis. What’s more, we analysised the ERP and behavior data in the time interval between stages and build "five-stage model " of brand extension evaluation and decision-making based on the neural mechanism. We also used focus-groups to get the practical support and used surveies with 2448 cases scale to test the model. With the analysis of OLS and SEM Path analysis, we found the "five-stage model" had a better explanatory ability than A&K model, and can also give a unified explaination to many disputes. In the last, this study also analysised the trends of decision neuroscience in top journals of management and science, and designed several study program which can get breakthrough of the frond domain.The conclusion of this study is as following:(1) There is a priming of "brand quality" (Stage 1). It can be indexed by LPC, which can be elicited by memory retrieval and incidental affect processing.(2) There is a "similarity match"(Stage 2). It can be indexed by N270, which can be elicited by mismatch processing.(3) There is a "categorization"(Stage 3). It can be indexed by P300, which can be elicited by categorization, working memory update and attention.(4) There is a "fit"(Stage 4). It can be indexed by N400, which can be elicited by conflict processing. N400 can also be regressed by N270 and P300. It means that "fit" is an automatic processing with the homing of "similarity match" and performing of "categorization".(5) There is a dual categorization process mechanism (category-based processing and piecemeal integration processing) for attitude formation for brand extension and only "categorization" and "fit" relate to it.(6) There is an "Attitude"(Stage 5). It can be indexed by LPP, which can be elicited by affective evaluation processing. The behavior result can be regressed by N400 and LPP. It means that there is a dual-attitude system in brand extension evaluation and decision-making: "fittness" is cognitive and automatic processing and "attitude" is affective and controlled processing.The innovation of this study is as following:(1) Following the trend of decision neuroscience, we used brain scan technology (ERP) to study consumer decision-making.(2) We supported the classic theory by neural evidence, and perfect the model of brand extension evaluation and decision-making.(3) "Five-stage model" gave a uniform explaination to the literatural disputes by neural mechanism. For example, the priming effect of brand can explain why "brand quality" was not be supported in A&K model; the categorization supports "fit" (Tauber, 1988) and "evaluation" (Bhat and Reddy, 2001) as the intervening variable; the interaction of the priming effect of brand and conflict congruent effect of "fit" can support the schema theory of Mandler (1982) and explain why Martin and Stewart(2001) could find the inverse "U" in brand extension evaluation but most literatures not.更多还原