介面配置對學生認知負荷和學習投入的影響：學習平台的眼動追蹤研究

Abstract

線上學習平台的評論區能促進師生間的交流，讓學習者透過互動深化理解知識，在學習平台的介面配置中評論的方式分為論壇和彈幕兩種，論壇是指將評論獨立呈現於影片右側評論框，而彈幕則是將評論由右至左滑動直接顯示在學習影片上。依據空間鄰近效應，當介面上資訊以更接近的方式呈現時，學習者更好整合相關資訊（Mayer, 2009），空間鄰近性的差異可能會影響學習者的學習狀態，其影響卻很少被研究。每位學習者先備知識差異和使用介面評論的不同方式，都可能會影響到學習者的認知負荷、學習投入程度、以及閱讀學習相關的評論，進而影響到學習者的學習表現。因此，本研究旨在探討空間鄰近性差異和先備知識如何影響學習者的認知負荷、學習投入、學習相關視覺閱讀行為、學習表現，以及各依變項之關係，藉此引發學習平台介面配置對於學習者學習重要性之關注。本研究探討以下六個問題：（一）空間鄰近性對學習者認知負荷的影響。（二）空間鄰近性對學習者學習投入的影響。（三）空間鄰近性對學習者閱讀到與學習相關或無關評論的影響。（四）不同空間鄰近性之下，先備知識程度不同的學習者對其學習表現的影響。（五）不同空間鄰近性之下，先備知識、學習表現、認知負荷、學習投入與發佈評論量的關係。（六）不同空間鄰近性之下，認知負荷和學習投入分別對先備知識影響學習表現之調節效果。研究對象為74位學生，使用資訊安全測驗、認知負荷量表和學習投入量表分別測量學習者的資訊安全、認知負荷與學習投入，加入Tobii 4C 眼動追蹤儀器測量學習者的認知負荷和視覺移動軌跡，並採用受試者間設計，將參與者隨機分為高空間鄰近性（彈幕）與低空間鄰近性（論壇）兩組，使其操作以資訊安全為影片主題的學習平台，透過分析其量表與眼動數據，比較空間鄰近性差異對學習者學習的影響。結果發現，空間鄰近性差異對學習者在認知負荷、學習投入和發佈評論量程度未達顯著差異，而在閱讀與學習相關或無關的評論數量達到顯著差異，表示當使用高空間鄰近性介面配置時，學習者較頻繁閱讀到與學習相關的評論。進一步分析發現，在高和低空間鄰近性之下，先備知識與學習表現具有中度到高度顯著正相關，並分析認知負荷和學習投入在其中的調節效果，結果指出認知負荷不具調節效果，而在高空間鄰近性介面配置中，學習投入與其子向度認知投入可分別調節先備知識與學習表現之間的關係，經單純斜率分析得知，在高空間鄰近性介面配置中，高先備知識組的學習者在認知投入較高時，學習表現較佳。上述結果豐富學習者在使用學習平台中，空間鄰近性介面配置的差異對認知負荷、學習投入和閱讀與學習相關的評論之影響。The comment area of an online learning platform can promote communication between teachers and students, allowing learners to deepen their understanding of knowledge through interaction. In the interface layout of the learning platform, the comment methods are divided into forum and danmu. Forum refers to the independent presentation of comments in the comment box on the right side of the video, while danmu refers to the direct display of comments on the learning video by sliding from right to left. According to the spatial proximity effect, when information is presented in a closer way on the interface, learners can better integrate relevant information (Mayer, 2009). Differences in spatial proximity may affect learners' learning status, but its impact has rarely been studied. The differences in each learner's prior knowledge and the different ways of using interface comments may affect learners' cognitive load, learning engagement, and reading learning-related comments, thereby affecting learners' learning performance. Therefore, this study aims to explore how differences in spatial proximity and prior knowledge affect learners' cognitive load, learning engagement, learning-related visual reading behavior, learning performance, and the relationship between the dependent variables, thereby raising attention to the importance of learning platform interface layout for learners' learning. This study explores the following six questions: (1) The impact of spatial proximity on learners' cognitive load. (2) The impact of spatial proximity on learners' learning engagement. (3) The impact of spatial proximity on learners' reading of comments related to or unrelated to learning. (4) The impact of different spatial proximity on learners' learning performance with different levels of prior knowledge. (5) The relationship between prior knowledge, learning performance, cognitive load, learning engagement and the number of posted comments under different spatial proximity. (6) The moderating effect of cognitive load and learning engagement on the impact of prior knowledge on learning performance under different spatial proximity. The research subjects were 74 students. The information security test, cognitive load scale and learning engagement scale were used to measure learners' information security, cognitive load and learning engagement respectively. The Tobii 4C eye tracking device was added to measure learners' cognitive load and visual movement trajectory. A between-subjects design was used to randomly divide the participants into two groups: high spatial proximity (danmu) and low spatial proximity (forum). They were asked to operate a learning platform with information security as the theme of the video. By analyzing their scales and eye movement data, the impact of spatial proximity differences on learners' learning was compared. The results showed that the differences in spatial proximity did not significantly affect learners' cognitive load, learning engagement, and the number of comments posted, but did significantly affect the number of comments related to or unrelated to learning, indicating that when using a high spatial proximity interface layout, learners read learning-related comments more frequently. Further analysis found that under high and low spatial proximity, prior knowledge and learning performance had a moderate to high significant positive correlation, and analyzed the moderating effects of cognitive load and learning engagement. The results showed that cognitive load had no moderating effect, while in a high spatial proximity interface layout, learning engagement and its sub-dimension cognitive engagement could moderate the relationship between prior knowledge and learning performance respectively. Simple slope analysis showed that in a high spatial proximity interface layout, learners in the high prior knowledge group had better learning performance when their cognitive engagement was higher. The above results enrich the impact of differences in spatial proximity interface layout on cognitive load, learning engagement, and reading learning-related comments when learners use learning platforms.