Instructor Visibility in Educational Videos: The Effect of User-Controlled Visibility on Cognitive Load, Social Presence and Learning Performance-An Eye-Tracking Study

Abstract

Background: Educational videos are increasingly used in higher education as they provide simultaneous visual and auditory information. However, the role of instructor visibility in such videos remains debated. While constant visibility can foster social presence, it may also increase cognitive load, whereas lack of visibility can reduce distraction but limit social connectedness. User-controlled visibility is a relatively new approach with the potential to balance these effects, yet it has been scarcely examined with objective measures such as eye-tracking. Objectives: This study aimed to investigate the effects of different instructor visibility conditions (no instructor visibility, fixed instructor visibility and user-controlled instructor visibility) on learners' cognitive load, perceived social presence and learning performance. Additionally, eye-tracking data were employed to analyse learners' visual attention allocation and interaction behaviours in user-controlled environments. Methods: The study was conducted with 90 associate degree students randomly assigned to three groups: audio-only narration (IVN), fixed instructor visibility (IVY) and user-controlled instructor visibility (UC). Data collection instruments included an achievement test (KR-20 = 0.80), the Cognitive Load Scale (alpha = 0.93), the Social Presence Scale (alpha = 0.91) and eye-tracking metrics (fixation duration, fixation count, visit count, heat maps). Data were analysed using descriptive statistics, ANOVA, Tukey post hoc tests and eye-tracking visualisations. Results: Findings revealed significant differences in posttest achievement scores across groups (F(2, 87) = 12.013, p < 0.05, eta(2) = 0.216). The IVN and UC groups outperformed the IVY group, indicating that constant instructor visibility may hinder performance by dividing attention. Cognitive load analysis showed no significant differences in intrinsic and extraneous load but revealed higher germane load in the IVY group compared to IVN (p < 0.05). Social presence perceptions were significantly higher in IVY compared to IVN, with UC showing an intermediate effect. Eye-tracking data indicated that IVN participants concentrated primarily on content; IVY participants split attention between instructor and material, and UC participants flexibly adjusted attention by toggling visibility. Interaction logs demonstrated frequent use of the 'hide' button during visually dense content and 'show' button during complex tasks. Conclusions: The results highlight that user-controlled instructor visibility provides a balanced solution by reducing unnecessary cognitive load while sustaining social presence. Audio-only videos maximise content focus but weaken social connection, whereas constant instructor visibility increases social presence but may impair learning performance due to divided attention. User control empowers learners to regulate attention and interaction according to individual needs, offering a promising design strategy for educational videos. Future research should explore adaptive systems that automatically adjust instructor visibility based on learners' attention and content complexity.