【作 者】：
李志河， 王元臣， 陈长玉， 董熔娟， 徐玉微
【关 键 词 】：
深度学习； 多模态； 生成性； 知识结构； 深度具身认知学习环境
【栏 目】：
课程与教学
【中文摘要】：
第一代认知科学因桎梏于“身心二元论”而陷入“认知科学危机”，但随着认知科学研究取向由“离身”转向“具身”，认知心理学获得新发展。具身思想为破解深度学习困境提供新的视角。文章首先采用文献分析法，厘清深度学习与浅层学习、具身学习和离身学习、具身学习环境的内涵与特征；其次，基于离身隐喻视域，透视深度学习面临的困境：“理性主义”知识观、“计算机隐喻”学习观、程式化课程观、预成性教学观、“离身传播”媒介观和机械化环境观；再次，透过具身隐喻视角，提出化解困境的新路向，即境脉化知识观、“多模态”隐喻学习观、具身化课程观、生成性教学观、“具身传播”媒介观和生态化环境观，并得出结论：深度学习要从“离身窘境”脱困，必须向“具身学习”进行转变；最后，从宏观情境与脉络、中观教学与学习、微观观念与知识三个层次，构建了深度具身学习环境模型，为具身学习环境设计以及促进学习者深度学习提供一种新的思路。
【英文摘要】：
The first generation of cognitive science was caught in the "crisis of cognitive science" due to the "body-mind dualism", but with the research orientation of cognitive science shifting from "disembodiment" to "embodiment", cognitive psychology has gained a new development. The idea of embodiment provides a new perspective for cracking the dilemma of deep learning. Firstly, this paper adopts literature analysis to clarify the connotation and characteristics of deep learning and shallow learning, embodied learning and disembodied learning, and embodied learning environment. Secondly, based on the perspective of disembodied metaphor, this paper analyzes the dilemma faced by deep learning, including the "rationalism" view of knowledge, the "computer metaphor" view of learning, the programmed curriculum view, the pre-established pedagogical view, the "disembodied communication" media view and the mechanized environment view. Then, through the perspective of embodied metaphor, this paper puts forward new ways to solve the dilemma: the contextualized knowledge view, the "multimodal" metaphor learning view, the embodied curriculum view, the generative pedagogical view, the "embodied communication" media view, and the ecological environment view. It is concluded that deep learning must be transformed into "embodied learning" in order to get rid of the "disembodied dilemma". Finally, a Deep-Embodied Learning Environments(DELEs) model is constructed from three levels of macro-situation and context, meso-teaching and learning, and micro-concept and knowledge, which provides a new way of thinking for the design of embodied learning environments and the promotion of learners' deep learning.