1 Introduction

Videos have become more prevalent in delivering learning content in online instruction since the emergence of the COVID-19 pandemic. With the widespread use of videos in education, numerous design issues have been applied to make videos more appealing, engaging (Wilson et al., 2018) and effective for instruction (van Wermeskerken & van Gog, 2017). However, the intrinsic transactional distance generated by online environments is still a challenging problem for instruction (Wang et al., 2020b). A commonly emphasized design issue to overcome this problem and improve students’ engagement and learning performance is the on-screen presence of instructors in videos (van Wermeskerken & van Gog, 2017).

The literature suggests two general classifications to categorize instructor-presence videos (IPV): lecture and demonstration videos (Stull et al., 2018; van Wermeskerken et al., 2018). In lecture-style videos, the instructor explains a material by standing next to a board on which learning content (i.e., figures, drawings, etc.) is presented or a screen on which the slides are projected to audiences; in demonstration-style videos, on the other hand, the instructor demonstrates a learning task (van Wermeskerken & van Gog, 2017). The instructor might be either in full-body or talking-head forms (Stull et al., 2018) and is regarded as a social cue (Yi et al., 2019), who use facial or body expressions, gestures, and hand moves, or provides particular attentional cues.

Although various studies have examined the effect of the instructor’s presence on learning outcomes, the results are inconclusive and mixed (Wang et al., 2020a). While some studies have shown that learning from IPVs is more effective than learning from those without the instructor (Pi & Hong, 2016), others have found that learning from IPVs is ineffective (Kizilcec et al., 2014) or distracting for learners (van Wermeskerken et al., 2018). In addition to the cognitive learning outputs, there have been studies focusing various social and affective learning outcomes on IPVs, including emotion (Beege et al., 2020), satisfaction (Zhang et al., 2021), learning enjoyment (Wilson et al., 2018), self-efficacy (Hoogerheide et al., 2018), social presence (Ng & Przybylek, 2021), and parasocial interaction (Pi et al., 2021). However, it is unclear whether or not the instructor’s presence is an effective strategy for these outputs since the research has not been thoroughly synthesized (Henderson & Schroeder, 2021). Given the crucial role of both social and affective processes in digital learning environments (Schneider et al., 2022), it might be beneficial to review IPV literature from different learning aspects to better understand how these processes differ under various conditions.

2 Prior reviews, meta-analyses, and the present study

The instructor’s presence has been discussed in the literature; however, few reviews have guided the new research. The review conducted by Henderson and Schroeder (2021) examined the effect of on-screen instructors on learning, cognitive load, and social presence. They synthesized 12 peer-reviewed articles published from 2011 to 2021 and revealed descriptive trends of these videos (e.g., topic, year, and location). The review also reported neither positive findings in favor of the instructor’s presence nor convincing evidence for leaving instructors out of the videos for learning outcomes, cognitive load, and social presence. Similarly, Alemdag (2022) conducted a meta-analysis with 20 experimental studies to clarify the overall impact of IPVs on learning, cognitive load, motivation, and social presence. The study showed that the instructor’s presence did not influence learning and social presence but increased learners’ cognitive load and motivation. Additionally, the study focused on several moderators, including video length, learning domain, human embodiment, study setting, and study location. Findings indicated that the instructor’s hand in the videos and the setting where learners watch the videos increase the impact of IPVs.

Although the abovementioned studies have provided different perspectives, the literature lacks studies describing the link between on-screen instructors and different aspects of learning. In other words, the previous works (Alemdag, 2022; Henderson & Schroeder, 2021) focused on the variables such as learning, cognitive load, social presence, and motivation and revealed some descriptive trends and moderators for IPVs. However, the current study extends earlier reviews by including new affective, cognitive, and social learning outputs, and it seeks the design-related trends in instructor-present videos. First, it divides the learning outputs into three groups, which are retention, transfer, and overall learning performance. This is important because multimedia learning literature suggests focusing on learners’ retention and transfer rather than learning performance alone (Mayer, 2005). Based on empirical results in the literature, the current study may contribute to the existing knowledge of how instructors’ presence affects learners’ retention, transfer, and learning performance separately.

Second, the current study examines learners’ attention in the IPV literature, which is also missing in previous reviews. Attentional processes can play an important role in designing multimedia learning materials (Mayer, 2014a). It is known that the learner’s information processing behavior is greatly affected by the visual information on the screen (Moon & Ryu, 2020). Several eye-tracking studies, for example, have shown that the instructor’s presence considerably impacts the learner’s visual attention (Kizilcec et al., 2014; van Wermeskerken et al., 2018), which may distract learner from the learning content since they use the working memory resources to process the visual features of the instructor (Fiorella et al., 2019). In this regard, revealing how learners draw their attention to instructor-present videos from the literature may give insights into the design of instructional videos.

Third, the extensions of cognitive learning theories emphasize that learning performance depends not only on the cognitive processing of information but also on affective and motivational factors (Schneider et al., 2022). The current study, therefore, reviews the learners’ affective states in IPV literature, which were not given much attention in the previous reviews. As an affective variable, we focused on the learners’ emotions, satisfaction, and enjoyment. These variables were selected in the current review to seek learners’ subjective reactions to the IPVs in the literature. It is considered that the result of this review may have a valuable contribution to the field by clarifying the role of IPVs on the affective aspect of learning.

Forth, the social cues provided by the instructor in videos lead to a sense of social presence in the learner, resulting in increased cognitive processing and more profound learning performance (Mayer, 2014b). Although previous reviews touched upon the learners’ social presence, little attention is paid to the parasocial interaction, a critical variable that researcher frequently discussed in the multimedia learning environments recently (e.g., Pi et al., 2021). Parasocial interaction is described as the affective, cognitive, and behavioral influence of the media figures (e.g., instructor in the video) on the media recipient (e.g., learner) and is worth examining since it can positively impact cognitive processing (Schneider et al., 2022). Understanding both learners’ social presence and parasocial processes in IPV settings may give insights into the design of instructional videos from a social learning dimension. In this regard, this study extends the findings of previous reviews regarding learners’ social presence and allows examining how the instructor’s presence affects parasocial interaction depending on the literature.

Last, the design-related issues of videos, other than the visual of the instructor, may also help to interpret the findings in IPV literature. Alemdag (2022) proposed five potential moderators for knowledge acquisition from IPVs: video length, domain, human embodiment, study setting, and study location. Apart from these variables, however, there are also additional factors that are manipulated in IPVs, such as knowledge type presented in the video (e.g., Hong et al., 2018), video style (e.g., lecture or demonstration; Hew & Lo, 2020), and social cues used by the instructor (e.g., Pi et al., 2021). Classifying these factors with their varieties may reveal the trends in the design of IPVs and guide further research. Therefore, the current study focuses on categorizing design-related issues of IPVs, including video length, knowledge type, video style, and social cues.

Consequently, to get a holistic picture of diverse learning outcomes from IPVs, it remains an open question concerning under which conditions and with which video type students could attain more positive learning outcomes from instructor-present videos (Wang et al., 2020a). The current study aims to systematically review the recent literature that explores how instructors’ presence affects affective, cognitive, and social aspects of learning in different conditions and with different video types. Furthermore, it also extends the results of previous works (Alemdag, 2022; Henderson & Schroeder, 2021) and addresses the following questions.

  1. 1.

    What is the influence of instructor-present videos on affective, cognitive, and social learning outcomes?

  2. 2.

    What are the design-related issues of instructor-present videos?

3 Method

A systematic literature review was conducted to review the recent research on instructor-present videos. This method allows rigorous and reliable knowledge organization (Denyer & Tranfield, 2009) and projects particular insights into the field through theoretical synthesis (Tranfield et al., 2003). It may be author-centric, which uses a chronological review to uncover the causes of a problem, or theme-centric, which shows readers how prior research has advanced understanding of themes and phenomena of interest (Linnenluecke et al., 2019). This study used the theme-centric approach based on the PRISMA principles (Moher et al., 2009).

3.1 Search strategy

The study included three steps to find the relevant studies respectively: (i) search in the Web of Science, ERIC, Scopus, and Education Source digital research databases in January 2021, (ii) an additional search in the same databases in July 2022, and (iii) final search in Google Scholar to access missing articles. The following search string was used in all fields of each database: (“instructor* face” OR “instructor* gaze” OR “instructor* head” OR “instructor* presence” OR “video modeling” OR “face in video” OR “social cue”) AND (“video”). The inclusion and exclusion criteria are presented in Table 1.

Table 1 Inclusion and exclusion criteria
Full size table

3.2 Study selection

As seen in Fig. 1, the initial search yielded 2731 articles at the beginning of 2021. The records were filtered based on publication date, language, and source type, and we reached a total of 1192 articles, 1539 of which were excluded because of duplicates. The remaining 607 articles’ titles and abstracts were examined, and irrelevant articles were excluded from the list, resulting in the selection of 22 relevant articles. The same procedure was applied for the second search, and 10 studies were reached. We also obtained nine articles from an additional search in Google Scholar. Consequently, 41 articles were taken into consideration.

Fig. 1
figure 1

PRISMA flow diagram for the study

Full size image

3.3 Identifications of categories and synthesis

The articles were examined depending on the categories presented in Table 2. The first category (research themes) was synthesized by mostly studying the articles’ results section. The last category (design-related issues) was captured by examining the articles’ design section. A descriptive analysis was undertaken to reflect the rates of categories, mainly frequencies and cross tables. The list of the reviewed articles and the categories for the synthesis were presented in Appendix 1.

Table 2 Categories for synthesis
Full size table

4 Results

4.1 Research themes

4.1.1 Affective learning outputs

Emotion

Students had positive feelings towards instructor-present videos (Kizilcec et al., 2015; Wang et al., 2020a; Yuan et al., 2021), but not as much as in face-to-face instruction (Rosenthal & Walker, 2020), and showed stronger interpersonal attraction to human instructors than robot instructors on the screen (Li et al., 2016). Additionally, video style (Chen & Wu, 2015) and gestures (i.e., beat and deictic gestures; Beege et al., 2020) were not effective on learners’ emotions.

Satisfaction

Without considering the content difficulty (Wang et al., 2020a), learners’ satisfaction was higher while watching instructor-present videos (Wang & Antonenko, 2017), particularly when the instructor was presented on the right side of the screen (Zhang et al., 2021). Compared to the videos with the continuous instructor presentation, strategic presentation led to more learner satisfaction (Yi et al., 2019), but video style did not (Alasfor, 2021; Yuan et al., 2021).

Learning enjoyment

Instructor-present videos were more enjoyable than voiceover videos (Wilson et al., 2018), particularly learning from a male instructor was more enjoyable for male learners than females (Hoogerheide, Loyens, et al., 2016a). However, the instructor’s age, experience (Hoogerheide et al., 2016b), and gender (Hoogerheide et al., 2018) did not influence learning enjoyment.

4.1.2 Cognitive learning outputs

Attention

Except for the articles using questionnaires (Chen & Wu, 2015; Korving et al., 2016; Rosenthal & Walker, 2020) and EEG headsets (Lackmann et al., 2021), all the articles employed eye measurements to infer the learners’ attention. Additionally, learners’ visual attention distribution and fixation behaviors were measured with different eye tracking metrics, including fixation duration and count, saccades, first fixation time, and dwell time.

Some articles examined how learners’ attention changed in different video types, and mixed results were reported. The learners’ attention in the live composite video style was greater than in picture-in-picture and voiceover videos (Rosenthal & Walker, 2020). Sustained attention in the voiceover was higher than in picture-in-picture (Chen & Wu, 2015). In another study (Korving et al., 2016), however, picture-in-picture and voiceover conditions did not significantly differ in learners’ attention in the first watching of videos, but there was a difference in the second watch. The info-graphic-supported voiceover videos received more attention than lecture videos (Lackmann et al., 2021).

The instructor’s face (Ouwehand et al., 2015; Wang et al., 2020a; Wang & Antonenko, 2017) and the instructor with teaching enthusiasm (Qian et al., 2022) drew learners’ attention more to instructor, particularly in easy-topic videos (Wang & Antonenko, 2017), which decreased over time when they rewatched the videos (van Wermeskerken et al., 2018). The most attractive stimuli that caused learners to focus on the instructor’s face were their gaze behaviors, such as direct gaze (Pi et al., 2020) and guided gaze in demonstration videos (van Wermeskerken & van Gog, 2017). The other component was the whiteboard type used in video lectures. For instance, learners paid more attention to the instructor’s face when watching a video lecture with a transparent whiteboard than a conventional whiteboard (Stull et al., 2021). However, when the instructor used direct gaze, the whiteboard type did not affect the learner’s attention (Stull et al., 2018).

The guided gaze (Pi et al., 2020; Stull et al., 2021; Wang et al., 2018), pointing gesture (Pi et al., 2019), and gesture cues (Ouwehand et al., 2015) were superior at drawing attention to the learning content. Regardless of the instructor’s gaze behavior, learners focused more on the learning material in the traditional whiteboard condition than transparent whiteboard condition (Stull et al., 2018). In demonstration videos, learners paid more attention to the learning tasks in cases where the instructor used guided gaze (van Wermeskerken & van Gog, 2017), or the instructor was not presented (van Wermeskerken et al., 2018).

The instructor’s presence did not affect the fixation on learning content in picture-in-picture (Colliot & Jamet, 2018) and demonstration videos (Van Gog et al., 2014). However, when the instructor was on the right side of the screen (Zhang et al., 2021) and showed social cues, learners fixed more on the learning content. Additionally, the initial fixation on the learning content was longer, and the dwell time was shorter when the instructor used guided gaze with a surprised face (Pi et al., 2021). Furthermore, when the instructor showed direct gaze, learners focused more on the instructor’s face and less on the learning content (Pi et al., 2022a).

Learning performance

The results for learning performance were inconclusive. Picture-in-picture and video lectures was more effective than voiceover videos (Chen & Wu, 2015). This result was extended by Kokoç et al. (2020), indicating that learners’ sustained attention could influence learning performance when studying videos in different styles. Additionally, the right of the screen was the most effective position to present the instructor on the screen (Zhang et al., 2021). However, some studies did not find meaningful difference between different video types (Alasfor, 2021; Wilson et al., 2018, Experiment 2,3 & 4). On the other hand, Lackmann et al. (2021) found the voiceover videos including infographics more effective than lecture videos. Additionally, the instructor’s presentation (i.e., either static or strategic) style (Kizilcec et al., 2015), and whiteboard type in lecture video (Stull et al., 2018) did not affect the learning performance.

Regarding how to present the learning content, the dynamic (drawing) presentation provided more promising results than the static one, particularly in transparent whiteboard conditions (Fiorella et al., 2019). For declarative video contents, both the instructor’s presence (Hong et al., 2018) and gaze behavior (Wang et al., 2018) had an impact on learning performance, while only the gaze behavior was superior for procedural knowledge (Wang et al., 2018).

Neither the gaze nor the gestures facilitated learners’ performance in a problem-solving task video (Ouwehand et al., 2015). However, the instructor’s teaching enthusiasm (Qian et al., 2022), the instructor’s pointing gesture in videos with complex contents (Pi et al., 2022b), gaze and facial expressions, particularly the interaction of direct gaze and happy face, influenced the learning performance (Pi et al., 2022a). The learning performance became weaker when the instructor performed guided gaze with a surprised face (Pi et al., 2021). Contrary to the pointing gesture condition, learners performed higher in the beat and depictive gesture condition (Pi et al., 2022b).

In the demonstration videos, the instructor’s gender (Hoogerheide et al., 2018; Hoogerheide et al., 2016a) and experience (Hoogerheide et al., 2016b) did not affect the learning performance, but the age did (Hoogerheide et al., 2016b). The instructor’s presence would be effective in learning performance in the second watch (Van Gog et al., 2014) as opposed to the first watch (Van Gog et al., 2014; van Wermeskerken & van Gog, 2017).

Retention

Learners’ retention was examined from different perspectives, including the instructor’s presence, content’s complexity, video type, gaze behavior, instructors’ body orientation, gestures, and handwriting. Results were inconclusive in general. While Colliot and Jamet (2018) found videos with the instructor more effective than those without the instructor, particularly when the instructor with a professional coherence (Beege et al., 2022), others (Ng & Przybylek, 2021; van Wermeskerken et al., 2018; Wang et al., 2020b) did not observe any differences, and Yuan et al. (2021) reported more positive results in favor of the instructor’s absence.

The instructor’s presence did not affect learners’ retention in videos covering complex topics (Wang et al., 2020a; Wang & Antonenko, 2017). However, mixed results were reported for easy topics; one study (Wang & Antonenko, 2017) found the instructor superior in retention, while another study (Wang et al., 2020a) presented non-significant findings.

In videos comparing the effect of video style on retention, no main difference was discovered (Hew & Lo, 2020; Rosenthal & Walker, 2020, Study 2). However, the way how the instructor is presented (e.g., continuous or intermitted; Yi et al., 2019), the whiteboard type (Fiorella et al., 2019, Experiment 2), and the instructor’s nature (e.g., human or animated pedagogical agent; Li et al., 2016) were the possible conditions that may affect the retention.

For gaze behavior, the guided gaze was more effective in retention than the direct and averted gaze (Pi et al., 2020). Furthermore, the direct gaze was more influential than the averted gaze (Pi et al., 2020), particularly when combined with the pointing gesture (Pi et al., 2019).

The results were inconclusive for the pointing gesture condition. Pi et al. (2019) reported a significant effect of the pointing gesture on retention compared to no social cue condition, but, Fiorella and Mayer (2016) found no difference (Experiment 1). Again, no significant difference was found between the videos, including pointing gestures and hand drawings (Fiorella & Mayer, 2016). However, deictic gestures were superior for retention (Beege et al., 2020).

Instructor-present videos containing drawings were reported to be more effective on retention than videos containing static information presentation (Fiorella et al., 2019, Experiment 1). However, neither the hand nor the instructor’s body significantly affected retention during the drawing (Fiorella & Mayer, 2016, Experiment 4). Similar results were obtained while drawing on a transparent whiteboard compared to a conventional whiteboard (Fiorella et al., 2019, Study 3).

Results regarding the instructor’s body orientation were likewise inconsistent. Beege et al. (2017, 2019) found the frontal body orientation more effective than lateral orientation in lecture videos; however, Pi et al. (2020) did not find a difference in picture-in-picture videos. Additionally, the instructor’s proximity to the camera while lecturing did not impact learners’ retention scores (Beege et al., 2017).

Transfer

The instructor’s presence in demonstration videos did not affect the learners’ transfer scores (Van Gog et al., 2014; van Wermeskerken et al., 2018). Conflicting results, however, were obtained for picture-in-picture and voiceover video lectures. While Wang et al. (2020b) reported positive results favoring the instructor’s presence, some researchers (Colliot & Jamet, 2018; Hew & Lo, 2020, Study 2a) did not observe any differences, and Yuan et al. (2021) found voiceover videos more promising. Some studies examined how the content difficulty level affected transfer in instructor-present videos. The results were again inconclusive, indicating no effect, regardless of the content difficulty (Wang & Antonenko, 2017), but influential in videos with complex content (Wang et al., 2020a).

The instructor’s intermitted (i.e., a strategic view) presence (Yi et al., 2019) and professional coherence (Beege et al., 2022) was beneficial for transfer. However, the instructor’s gender (Hoogerheide et al., 2016a), proximity to the camera (Beege et al., 2017), and content presentation (i.e., static or drawing content) style and whiteboard type (Fiorella et al., 2019) had no meaningful effect on transfer scores. The results for body orientation were mixed; some (Beege et al., 2017; Van Gog et al., 2014) reported non-significant results, but Beege et al. (2019) found frontal-style more beneficial than lateral condition.

Some of the social cues shown by the instructors, such as guided gaze (Pi et al., 2020), deictic gestures (Beege et al., 2020, Experiment 2), and pointing gestures with direct gaze (Pi et al., 2019), were fruitful in getting better transfer scores. However, making direct gaze alone was not effective (Pi et al., 2019). It was also indicated that using gaze together with gestures does not necessarily increase the transfer scores (Ouwehand et al., 2015). Additionally, learners’ prior knowledge was noted as a crucial factor influencing the learning transfer (Fiorella & Mayer, 2016).

Cognitive load

Mixed results were reported regarding the participants’ cognitive load in the reviewed articles. No difference on cognitive load was observed between picture-in-picture and voiceover conditions (Ng & Przybylek, 2021). However, Kizilcec et al. (2015) showed that learners’ cognitive load in instructor-present videos was fewer than those without the instructor (Phase 1). While this result was confirmed for intrinsic and extraneous load, it was not affirmed for the germane and overall load (Wang et al., 2020b). A mediating effect of the learning style was also noticed on the cognitive load (Chen & Wu, 2015; Kizilcec et al., 2015, Phase 2).

Some articles indicated that the knowledge type, the content’s complexity, teaching enthusiasm, and instructor’s professional congruence might affect the cognitive load. For example, Hong et al. (2018) discovered that the instructor’s face could not affect the cognitive load in videos, including declarative knowledge (Phase 1) but could increase it in videos with procedural knowledge (Phase 2). Regarding the content’s complexity, the instructor’s presence did not influence the cognitive load in the easy-topic videos (Wang et al., 2020a; Wang & Antonenko, 2017); however, the cognitive load was less in the instructor-presence videos with complex contents (Wang & Antonenko, 2017). Supporting this result, the intrinsic and extraneous cognitive load was lower in videos with the difficult topic but was not different for mental effort and germane load (Wang et al., 2020a). Additionally, videos including instructor with higher teaching enthusiasm (Qian et al., 2022) and professional coherence (Beege et al., 2022) decreased learners’ cognitive load.

Regarding the model-observer similarity, the instructor’s gender (Hoogerheide et al., 2018) and experience (Hoogerheide et al., 2016b) did not affect the cognitive load, but the age (Hoogerheide et al., 2016b) and body-orientation with a professional dress did (Beege et al., 2019). Additionally, male learners’ cognitive load was less when they watched videos, including a male instructor (Hoogerheide et al., 2016a).

How the instructor is presented, content presentation style, whiteboard type, and the availability of nonverbal cues were other factors that might impact the cognitive load. Results were mixed for the instructor’s presentation style. While Yi et al. (2019) showed that continuous presence negatively affected cognitive load, Kizilcec et al. (2015) reported that cognitive load was higher in strategic view than in constant view condition. Furthermore, the content’s presentation style (i.e., drawing vs. static content) and the whiteboard type (i.e., transparent vs. conventional) (Fiorella et al., 2019), and nonverbal communication cues (i.e., gaze and gesture cues; Ouwehand et al., 2015) did not affect the cognitive load. However, the results for the beat and deictic gesture conditions were mixed (Beege et al., 2020).

4.1.3 Social learning outputs

Social presence

Regarding the social presence, the following conditions were examined in the reviewed articles: video type, instructors’ characteristics and presentation style, gaze behavior, and gestures. While Rosenthal and Walker (2020) demonstrated that the instructor’s presence was influential on the learner’s social presence (Study 2), some researchers (Alasfor, 2021; Colliot & Jamet, 2018; Ng & Przybylek, 2021; Yuan et al., 2021) did not find a considerable effect. Additionally, the instructor’s presence in the live composite videos was higher than in the picture-in-picture videos (Rosenthal & Walker, 2020, Study 2).

In two articles, the instructor’s characteristics were compared. Learner’s social presence was higher with the human instructor than with robot ones (Li et al., 2016). However, no significant difference was observed between the original and virtual instructor’s views (Yuan et al., 2021).

Regarding the instructor’s presence style in videos, no significant difference was found between continuous and intermitted presence (Yi et al., 2019). However, in one study (Kizilcec et al., 2015), strategic presence was superior to continuous presence. Finally, the instructor’s gaze guidance, regardless of knowledge type (Wang et al., 2018) and deictic gestures (Beege et al., 2020) increased learners’ social presence.

Parasocial interaction

The gaze behavior, facial expressions (Pi et al., 2021), and instructor’s proximity (Beege et al., 2017) did not have a meaningful effect on parasocial interaction. However, compared to the lateral orientation, the instructor’s frontal orientation (Beege et al., 2017) and professional coherence (Beege et al., 2019, 2022) influenced learners’ parasocial interaction. The deictic gesture was also effective on some dimensions of parasocial interaction (Beege et al., 2020).

4.2 Design-related issues of videos

The second research question focuses on the design-related issues of the videos in the reviewed articles (see Table 3). Regarding the video length, a total of 59 videos in 38 articles were considered since two articles (Alasfor, 2021; Hoogerheide et al., 2016a) did not report the video length, and one (Kizilcec et al., 2015) contained many videos with varying durations. Accordingly, the most frequently reported interval was 5 to 10 (42%), followed by 10 to 15 (29%), and lower than 5 min (24%).

As seen in Tables 3 and 65% of the reviewed articles focused on declarative knowledge. The percentage of procedural knowledge was 23, and %5 focused on both declarative and procedural knowledge.

The percentage of video styles were as follows picture-in-picture (52%), lecture-type (30%), and demonstration videos (4%). Additionally, %16 of videos was in voiceover format. In some articles, the instructor was presented strategically (3%), which means the instructor was not seen on the screen continuously. Lecture slides accompanied the instructor’s image in most picture-in-picture videos (46%).

Table 3 also shows the contexts where the video lectures were captured. These videos were generally recorded in a lecture hall, followed by a setting where a board or a screen exists. Demonstration videos included a video model in front of a desk.

Table 3 Design issues
Full size table

Although voiceover videos did not include the instructor’s presence, they were used to compare them with other video styles. Lecture slides were generally used to present the content in these videos.

Instructors used different cues in the videos, including gaze (51%), gesture (37%), and facial expressions (16%). Among the gaze behaviors, directed gaze was the most used, followed by guided or shifted gaze. Regarding the gestures, instructors generally used pointing gestures. Facial expressions such as neutral, surprised, and happy faces were also used in the reviewed articles.

5 Discussion

In this literature review, we examined the learning outcomes assessed in instructor-present videos and the design-related issues of these videos. We found that learners are more emotionally positive toward these videos. However, the results for cognitive and social aspects of learning are mixed. Additionally, design issues varied across video length, knowledge type, video style, and social or attentional cues the instructor used.

5.1 Affective, cognitive, and the social aspects of learning

Regarding the affective aspect of learning, learners’ emotions, satisfaction, and learning enjoyment were relatively higher in instructor-presence videos. However, we observed that the video style was not influential on learners’ emotions and satisfaction, and the instructor’s characteristics, such as age, gender, and expertise, were not sufficiently effective on learning enjoyment. A possible explanation for the positive results may closely be related to the social cues provided by the instructor (Wang et al., 2020a), indicating that social cues have the potential to trigger the affective responses of learners. However, the instructor’s age, gender, and expertise might not be equally valuable affective information or emotionally significant variables for learners in instructor-present videos. To test the assumption that higher similarity between the instructor and the learner enhances learners’ affective states (Hoogerheide et al., 2016a), further research needs to address different instructor characteristics.

Attention, overall learning performance, retention, transfer, and cognitive load were the selected cognitive variables in the current study. Regarding attention, mixed findings were obtained for different video types. The instructor’s face was a salient stimulus in the videos. Eye measurement data also showed that the instructor’s gaze behavior could be used to draw learners’ attention to the relevant learning element. These results seem consistent with other research that human faces attract a substantial amount of attention (Ouwehand et al., 2015) and theoretical inferences indicating that the instructor’s gaze behavior could be an attentional cue in multimedia learning settings (Mayer, 2005). Hence, it is plausible to assume that the instructor’s presence may affect how students distribute their visual attention in instructional videos (Wang & Antonenko, 2017).

The results were inconclusive for learning performance, retention, and transfer in the reviewed articles. Many of the studies found instructor-present videos ineffective for learning-related variables. This finding is consistent with the results of previous works (Alemdag, 2022; Henderson & Schroeder, 2021) and the theoretical assumption of the image principle, stating that learning is not necessarily fostered by the instructor’s image in multimedia learning environments (Mayer, 2014b). However, different variables such as learners’ sustained attention (Kokoç et al., 2020), gestures (Beege et al., 2020), teaching enthusiasm (Qian et al., 2022), dynamic presentation on a board (Fiorella et al., 2019), instructor’s professional congruence (Beege et al., 2022), presentation style (Yi et al., 2019), guided gaze behavior (Pi et al., 2020), contents’ complexity (Wang et al., 2020a), and learner’s prior knowledge (Fiorella & Mayer, 2016) are the potential moderators for learning that need further examination in different conditions.

Contrary to findings reported by Alemdag (2022), we did not arrive at a definite conclusion about the learners’ cognitive load in instructor-present videos. This inconsistency may be related to the measurement of cognitive load in the reviewed articles (Kizilcec et al., 2015). The other reason might be the design-related issues of instructor-present videos. From a cognitive load perspective, although the literature informs that the visual of the instructor creates extraneous processing (Wang & Antonenko, 2017), this study suggests that the learners’ cognitive load can vary in instructor-present videos depending on the knowledge type (Hong et al., 2018), content’s complexity (Wang et al., 2020a), and instructor’s presentation style (Kizilcec et al., 2015; Yi et al., 2019).

The last learning dimension is the social learning outputs, which placed less attention in the reviewed articles than the affective and cognitive aspects of learning. Social presence was the most studied variable and was not relatively affected by the instructor’s presence, as noticed by Alemdag (2022). An explanation for this result might be the instructor’s voice, considered the most crucial social cue in learning environments (Colliot & Jamet, 2018). Alternatively, adding the instructor’s image to the video may not truly enhance social presence (Yuan et al., 2021). Based on the reviewed articles’ results, this study suggests that the instructor’s presentation style (Kizilcec et al., 2015) and gaze guidance (Wang et al., 2018) may play a critical role in the learner’s sense of social presence.

Regarding parasocial interaction, the reviewed articles reported inconclusive results for different conditions. Nevertheless, instructors’ professional coherence was found more influential on parasocial processes, which shows the importance of professionalism in video lectures (Beege et al., 2019). To get more generalizable conclusions on learners’ parasocial processes, more empirical research is required in which the instructor’s presence is controlled in various contexts (Beege et al., 2022).

5.2 Design-related issues of videos

Each design issues of the instructor-present videos (see Table 3) might be a moderating variable affecting the learning outputs. This review indicated that the length of the instructor-present videos varied widely from 5 to 10 and 10 to 15 min. These lengths might not be a favorable learning condition for learners since the optimal video length for engagement is lower than six minutes for instructional videos (Guo et al., 2014). Alemdag (2022) also provided empirical evidence on this direction for instructor-present videos. In a long video, the instructor’s dynamic visualization may increase the extraneous cognitive load and reduce learner engagement, leading to the consumption of working memory resources that need to be used for learning (Chen & Wu, 2015). This study, therefore, suggests that further empirical research is needed to determine the optimal video length conditions under which the instructor-present videos lead to better learning outcomes.

The knowledge type taught in the instructional videos is crucial for learning outcomes (Hong et al., 2018; Wang et al., 2020a). There may be changes in the instructional roles of instructors while presenting different knowledge. In declarative knowledge presentations, for example, the instructor may only explain the relevant concepts verbally; however, in procedural knowledge presentations, it may be necessary to show the procedure step-by-step. Almost all reviewed articles in this study included videos containing either declarative or procedural knowledge presentations. A few articles compared both knowledge types concurrently, which might be an obstacle to drawing a more generalizable conclusion about the topic. Further research, therefore, is needed to understand how the instructor’s presence affects learning outputs in different knowledge conditions.

The effect of the instructor’s presence can vary depending on the video style (Chen & Wu, 2015; Wang & Antonenko, 2017). This review revealed that the most preferred video style was picture-in-picture, followed by lecture and demonstration videos. This finding may indicate a trend in using picture-in-picture videos in education (Li et al., 2016). In picture-in-picture videos, lecture slides were often combined with a talking head to convey learning content. Although its number is limited, the strategic presentation was also used in some picture-in-picture videos. In this presentation, the instructor is hidden when explaining critical contents (Yi et al., 2019). This study suggests this presentation style as an alternative solution to guide learners’ attention and reduce the potential cognitive load during the instruction (Kizilcec et al., 2015).

Video lectures in the reviewed articles were usually created by recording the instructor giving a lecture in a lecture hall or a class where learners could observe the learner-learner and learner-instructor interactions. In these videos, the instructor explained the learning content by standing adjacent to the board. Because of that, these video lectures are considered to have a higher degree of media richness (Chen & Wu, 2015). The combined use of many media elements in these videos has raised the concern of a possible split-attention effect (Ouwehand et al., 2015). However, it is possible to avoid such effect using social (e.g., guided gaze) or attentional cues (e.g., highlighting text). Therefore, the instructor can play a critical role in these videos.

In demonstration videos, the instructor usually stood behind a desk and solved a problem-solving task. Seeing the instructor’s face in these videos was considered a distractive stimulus that may impede learning (van Wermeskerken & van Gog, 2017). However, the instructor’s face might facilitate learning in a social interaction context where the instructor demonstrates the learning task (Van Gog et al., 2014). Hence, the instructor’s non-verbal communication cues in these videos might have a higher effect on learning. However, relatively little is known about how to design demonstration videos to foster learning (Van Gog et al., 2014); therefore, further research on social communication cues is suggested.

The voiceover videos, which combine the instructor’s narration with the learning content, were another frequently utilized video type in the reviewed articles. Examples of this type included the presentation of lecture slides, Khan-style, drawing on a board, and screencast. Researchers have often used this video to compare it with other instructor-present videos regarding various learning outcomes. The proponent of this format advocates that an instructor’s presence may not be needed since it may cause an extraneous cognitive load hindering learning (Kizilcec et al., 2015; Wang & Antonenko, 2017). However, learners do not have a positive attitude in this regard (Kizilcec et al., 2015; Wilson et al., 2018). Even some studies found instructor-present videos more effective than voiceover videos (Chen & Wu, 2015). It seems that the debate on the instructor’s presence over different video styles will not end in the short term, and new research will continue to shape the field.

In recent years, there has been a considerable increase in studies focusing on social and attentional cues in instructor-present videos. This study categorized these cues into three groups: gaze behaviors, gestures, and facial expressions. From the social agency perspective, learners may not learn deeply from instructor-present videos (Mayer, 2014b). However, social cues may enhance learners’ feeling of social presence (Mayer, 2014b) and motivate them to engage in generative processing for learning (Wang et al., 2020b), resulting in better learning performance. The literature suggests that as a social cue instructor’s gaze behavior (Wang et al., 2018), gestures, and facial expressions (Stull et al., 2018) can play a critical role in learning. Additionally, the attentional cues may reduce the unnecessary cognitive load and help to consume working memory resources effectively (Ozcelik et al., 2010). Therefore, it is reasonable to conclude that both social and attentional cues should be used carefully in videos to establish a connection with the learners and direct their attention to the relevant part of the learning content. Moreover, when used properly, these cues may lay a bridge between the learner and the learning content over the instructor.

5.3 Practical and theoretical implications

The literature review yielded inconclusive results regarding the instructor’s presence in videos. One of the reasons for these conflicting findings is likely the lack of detailed information describing the experimental conditions in the reviewed articles. In this regard, it becomes a requirement to collect more details about the instructors’ characteristics, the learning context in which the video was prepared, the video style (e.g., demonstration or picture-in-picture videos), learners’ characteristics who studied the videos, detail information about learning content, and the knowledge type being presented. Although some reviewed articles touched upon these factors, some neglected them. Therefore, the current analysis results might not be sufficient to make inferences for theory and practice related to the problem situation.

Nevertheless, the present study gives some insights into the design of instructor-present videos. Regarding the affective aspect of the learning, the instructor’s presence might be a preferable design component in videos. Based on the results, it seems reasonable to suggest that instructor-present videos increase learners’ positive feelings and satisfaction during the instruction. It is believed that this situation may have a favorable impact on learners’ behaviors in online courses.

Regarding the cognitive aspect of learning, the instructor’s presence may influence the learner’s visual attention. However, the visual of the instructor may not contribute to the sense of social presence and the learning outputs, including learning performance, transfer, and retention. Nevertheless, using social and attentional cues in instructor-present videos may foster learning and trigger social responses by learners. From a theoretical point of view, these cues might be suitable signaling components in video environments that lead learners to direct attention to the relevant learning content and help them allocate more cognitive resources to content. Furthermore, the results for learners’ cognitive load were also inconclusive. In cases where the instructor is perceived as an additional information source, split attention or redundancy effects may occur. This study suggests a strategic presentation of the instructor to avoid these effects.

6 Conclusion

The main conclusion that can be drawn from this review is that design-related issues (i.e., video length, knowledge type, video style, non-verbal cues) of instructor-present videos may affect the different aspects of learning. Although the instructor’s presence is favorable for affective learning outputs, the results for cognitive and social learning outputs are inconclusive and mixed. However, some social and attentional cues are suggested to benefit more from instructor-present videos in different learning conditions.