KDI: Visualization and Spatial Reasoning: Cognitive Models, Skill Acquisition and Intelligent Tutors

可视化和空间推理是智能系统的重要组成部分。它们构成了理解科学、数学和工程中各种主题的基础，包括分子结构、拓扑、运动和力以及制造过程。从历史上看，许多学生，特别是女学生，在获得可视化和空间推理技能方面存在困难，这可能会阻碍他们在科学、数学和工程领域的进步。在工程学领域，教员们发现，教授需要强大的可视化和空间推理技能的主题，如产品设计、制造和工程建模与分析，既具有挑战性，又耗时。同样，工科学生发现这些话题没有动力，也很难理解。随着复杂的计算机图形和动画的出现，人们可能会认为，对人类可视化技能的需求已经消除。但事实并非如此。计算机不能取代科学和工程中对这些技能的需求，就像计算器没有取代对定量技能的需求一样。因此，本项目有三个目标：L)增进我们对人类可视化和空间推理的理解；2)利用这些知识来开发基于计算机的可视化教学；以及3)将这种教学融入到智能多媒体教师中，使其对广泛的学生群体具有最大的有效性，同时最大限度地减少教师的开发时间和成本。要实现这些目标，我们需要组建一支具有心理学、教育学、工程学和计算机科学背景的研究团队。虽然可视化和空间推理是基本的认知技能，但人们对支配它们的认知过程知之甚少。因此，作为我们的第一个目标，我们将在L年期间在我们的眼动实验室进行一系列实验，旨在测试关于个人如何在心理上和空间上对3-D物体及其转换进行推理的替代理论。我们将使用详细的眼动数据作为了解潜在认知过程的窗口。我们在阅读、视觉搜索和场景感知方面也使用了类似的数据(Rayner，1992,1998；Rayner&Amp；Pollatsek，1992)。我们期望这些数据能够显示出个体之间巨大而稳定的差异，不仅是空间能力低的参与者和高空间能力的参与者之间的差异，也包括空间能力相似的参与者群体之间的差异。可视化和空间推理技能对于理解科学和工程中的许多概念至关重要。然而，我们对如何才能最好地教授这些技能知之甚少。因此，作为我们的第二个目标，我们将在第二年根据我们在第一年所学到的关于个人存在的问题和成功运作的策略的知识，开发基于计算机的可视化技能教学模块，这些模块将利用当前教学理论和技术的进步。在开发了模块后，我们将在第二年进行一系列实验，旨在测试提供可视化教学的理论上有动机的方法，这些方法可以改善向空间能力高和低的学习者提供的教学内容，优化部分和整体任务培训的组合，并最大限度地增加发展专业知识的个人数量。

Visualization and spatial reasoning are integral components of intelligent systems. They form the basis for understanding a wide variety of topics across science, mathematics and engineering, including molecular structures, topologies, motion and forces, and manufacturing processes. Historically, many students, especially female students, have had difficulty acquiring visualization and spatial reasoning skills, creating potential barriers to advancement in science, mathematics and engineering. Within engineering, faculty have found it both challenging and time consuming to teach topics that require strong visualization and spatial reasoning skills, topics such as product design, manufacturing, and engineering modeling and analysis. Similarly, engineering students have found these topics unmotivating and difficult to comprehend. With the advent of sophisticated computer graphics and animation, one might expect that the need for human visualization skills has been eliminated. But this is not the case. Computers cannot replace the need for these skills in science and engineering just as calculators have not replaced the need for quantitative skills. Thus this project has three goals: l) to advance our understanding of human visualization and spatial reasoning; 2) to use this knowledge to develop computer-based visualization instruction; and 3) to incorporate this instruction into intelligent multimedia tutors in ways that maximize their effectiveness for a broad mix of students while minimizing the development time and cost for the faculty involved. The achievement of such goals has required that we put together a team of researchers with backgrounds in psychology, education, engineering and computer science. Although visualization and spatial reasoning are fundamental cognitive skills, the cognitive processes that govern them are poorly understood. Thus, as our first goal, we will undertake during year l a series of experiments in our Eye Movement Laboratories designed to test alternative theories of how individuals represent mentally and reason spatially about 3-D objects and their transformations. We will use the detailed eye movement data as a window on the underlying cognitive processes. We have made similar use of such data in reading, visual search and scene perception (Rayner, l992, l998; Rayner & Pollatsek, l992). We expect these data to reveal large, stable differences among individuals, not only between low and high spatial ability participants, but also within groups of participants of similar spatial abilities.Visualization and spatial reasoning skills are critical to the understanding of many concepts within science and engineering. Yet, we have little understanding of how we can best teach these skills. Thus, as our second goal, we will develop during year 2 computer-based visualization skills instruction modules based on what we have learned during the first year about the problems that individuals have and the strategies that work successfully, modules that will take advantage of current advances in instructional theories and technologies. Having developed the modules, we will then conduct a series of experiments in the second year designed to test theoretically motivated methods for delivering visualization instruction that improve the content of the instruction delivered to high and low spatial ability learners, optimize the mix of part- and whole-task training, and maximize the number of individuals that develop expertise.