数学应用题解决能力作为数学认知能力的重要组成部分，是研究者们广泛关注的问题之一。影响数学应用题解决的因素包括题目本身的客观因素和解答者的基本认知能力与其他主观因素。以往的研究发现多种基本认知能力对于数学应用题解决有重要影响，但没有研究表明这些基本认知能力作用于数学应用题解决的脑机制是什么。本论文从认知行为的角度探讨与数学应用题解决密切相关的各种基本认知能力，并利用磁共振成像技术从神经层面揭示数学应用题解决的大脑激活区域的认知成分，回答数学应用题解决的脑机制是什么这一问题。实验一以132名大学生为被试，考察了数学应用题解决与多种基本认知能力的关系，结果发现，数学应用题解决与数学计算能力、言语理解能力、空间能力和推理能力的相关很高，与数字大小比较能力和数目大小比较能力的相关较低。实验二采用磁共振成像技术考察了62名大学生在进行包括数学应用题解决的多种认知任务时的大脑激活。通过构建线性回归模型的来控制数学应用题解决任务中可以被其他认知任务解释的成分，结果发现，（1）在控制了数学应用题解决的大脑激活中可以被图形关系推理任务解释的部分之后，具有数学领域特异性区域—双侧顶叶的激活就不再显著；（2）在控制了数学应用题解决的大脑激活中可以被词语辨析任务解释的部分之后，具有语言领域特异性区域—左侧颞中回和左侧额下回的激活就不再显著；（3）数学应用题解决中前额叶区域的激活无法单独被某个任务完全控制掉，但可以被所有任务共同解释。本论文的主要结论是，数学应用题解决中双侧顶叶区域激活的脑机制是推理加工，左侧颞中回和左侧额下回区域激活的脑机制是言语加工，前额叶区域激活的脑机制代表一般任务加工成分。

As the important part pf mathematics cognition, word problem solving ability has been an important question for a long time. However most previous researches just reported those factors which have influence on word problem solving based on cognitive behavior experiment. The underlying cognitive mechanism were hardly discussed. Moreover, previous fMRI (functional magnetic resonance imaging) studies focused on the contrast of different condition or presentation format of word problem solving, lack of direct analysis of brain mechanism of problems solving ability and what was represented in each activated region. In this study, behavioral tests were used to investigate the cognitive mechanism of word problem solving, and fMRI was used to investigate the relationship of neural activation in word problem solving and other cognitive tasks. In first study, 132 college students were recruited to complete several cognitive tasks including word problem solving. We found that, there were significant high correlation between word problem solving ability and following abilities: arithmetic, language processing, visual-spatial and reasoning abilities, rather than number comparison and numerosity abilities. In second study, we used fMRI to investigate brain activation of 61 college students in several cognitive tasks including word problem solving. We found that, (1) in word problem solving task, the activation in parietal lobe, which was mathematics specific area, was totally controlled by nonverbal matrices reasoning task, brain mechanism of it was reasoning processing; (2) the activation in left middle temporal gyrus and inferior frontal gyrus, which was language specific area, was totally controlled by word semantic task, brain mechanism of it was language processing; (3) the activation in superior frontal gyrus was hardly controlled by any tasks, brain mechanism of which maybe was general processing ability.