Journal of Computer Assisted Learning

ORIGINAL ARTICLE OPEN ACCESS

Implementing Pretraining to Optimise Learning in

Immersive Virtual Reality
Cynthia Y. Delgado | Richard E. Mayer

Department of Psychological and Brain Sciences, University of California, Santa Barbara, California, USA

Correspondence: Cynthia Y. Delgado ([email protected])

Received: 8 May 2024 | Revised: 29 October 2024 | Accepted: 9 November 2024

Funding: This work was supported by Office of Naval Research, N000142112047.

Keywords: immersion | pipetting | pretraining | procedural training | virtual reality

ABSTRACT
Background: In recent years, immersive virtual reality in education has garnered attention, however, there have been mixed
findings on the efficacy of IVR in education. Thus, exploring which strategies are effective in transferring learning from IVR to
real-­world applications is imperative.
Objective: This study aims to investigate the efficacy of the pretraining principle for acquiring procedural knowledge and skills
in an IVR setting that will transfer to real-­world environments.
Methods: Ninety-­three participants were randomly assigned to either a pretraining or no-­pretraining group. The pretraining
group watched a video before the IVR lesson, providing the names and characteristics of the physical objects and actions of
a micropipette, while the no-­pretraining group did not receive this video. During the IVR lesson, participants completed a
training phase, followed by a four-­step serial dilution test. Afterwards, all participants completed a modified serial dilution test
in a real-­life setting, along with a knowledge test and assessment on cognitive load, presence, self-­efficacy and demographic
information.
Results and Conclusions: Analyses demonstrated the pretraining group scored significantly higher on the knowledge test and
committed fewer errors in the real-­life serial dilution task compared to the no-­pretraining group. The pretraining group also
reported lower cognitive load, with no observable differences in presence, self-­efficacy ratings or errors during the virtual serial
dilution task between groups. Theoretical and practical implications are discussed.

1   |   Objective and Rationale Learning in virtual reality can be both distracting and moti-
vating (Mayer, Makransky, and Parong 2022). Therefore, it is
Suppose we want to help someone learn a technical procedural crucial to explore effective strategies for acquiring new skills in
skill, such as how to use pipettes to transfer liquid from one con- IVR technologies that can be transferred to real-­world applica-
tainer to another, through training in immersive virtual reality tions. The development of information technology continues to
(IVR). How can we increase the chances that the learner will be challenge all facets of education and training as technology pro-
able to transfer what they learned in a virtual environment to gresses to evolve from traditional teaching pedagogies to more
perform in a real-­world environment? In the present study, we ex- technology-­based ways of teaching. The importance of investi-
amine the instructional effectiveness of providing pretraining in gating what is the best practice or guidelines in developing IVR
the names and characteristics of the physical objects and actions curriculum is crucial for enhancing learning outcomes. This
represented in the virtual environment through a short video. study aims to investigate the efficacy of the pretraining principle

This is an open access article under the terms of the Creative Commons Attribution-NonCommercial License, which permits use, distribution and reproduction in any medium, provided the
original work is properly cited and is not used for commercial purposes.

© 2024 The Author(s). Journal of Computer Assisted Learning published by John Wiley & Sons Ltd.

Journal of Computer Assisted Learning, 2025; 41:e13099 1 of 16

https://doi.org/10.1111/jcal.13099
 Store, Smutny (2023) located 171 educational VR applications
Summary in 2019 compared to 234 in 2021 with most available at no
cost. Research on learning in IVR also encourages the idea
• What is currently known about this topic
○
We know that virtual reality systems have been that learning in IVR can be an effective venue for learning
used to help learners obtain occupational safety be- academic content. For example, in a recent meta-­a nalysis,
haviours, STEM knowledge and have been used for Villena-­Taranilla et al. (2022) reported that K-­ 6 students
training purposes. learned academic material better in virtual reality than with
○
We know that the pretraining principle is effective conventional media, particularly when the lessons were short
in increasing declarative knowledge through nar- and presented in IVR. In media comparison research, such as
rated slideshows, narrated animations and video the Villena-­Taranilla study, researchers compare the learning
lectures. outcomes of students who learn in IVR versus with conven-
• What does this paper add tional media. However, Lawson et al. (2024) have shown that
○
This paper adds to the existing literature by empiri- there are serious methodological flaws in most media com-
cally evaluating the effectiveness of the pretraining parison studies pitting learning in IVR versus with conven-
principle, particularly within the context of teaching tional media.
procedural tasks in virtual reality environments.
• Implications for practice Although a primary objective is often to replicate identical
○
We suggest incorporating a pretraining video prior real-­world setting in IVR, the role of instructional methods
to immersing learners in virtual reality to optimise in these platforms is frequently overlooked (Dobrowolski
learning outcomes, especially for lessons involving et al. 2021; Gavish et al. 2013). Instead, the research focus
technical scientific procedures. is often on media comparison studies in which learning in
○
This study expands the application of the pretrain- IVR is compared with learning with conventional media.
ing principle from conceptual knowledge to the ac- For instance, Dobrowolski et al. (2021) conducted a study
quisition of procedural knowledge and skills. comparing the efficiency of IVR against conventional train-
○
Future research should explore the long-­term im- ing across different age groups. Their findings illuminated
pacts of the pretraining principle on knowledge re- age-­related variations in training efficiency; however, their
tention and skill transfer. inquiry primarily centred on efficiency metrics, without con-
sidering how instructional methods might optimise learning
outcomes. Similarly, Gavish et al. (2013) conducted a study in
for acquiring procedural knowledge in an IVR setting that will the field of industrial maintenance and assembly task train-
transfer to real-­world environments. ing. Intriguingly, they reported no significant performance
differences between the IVR and traditional training groups.
However, the authors did not explore the consideration that
1.1   |   Literature Review targeted instructional methods might be required when learn-
ing in IVR. In essence, these studies have mainly focused
In recent years, the incorporation of IVR into educational on efficiency metrics without considering how instructional
contexts has sparked widespread interest. IVR has emerged as methods might optimise learning outcomes.
a learning environment for various fields of study (e.g., mili-
tary, medical and education) and numerous studies have ven- The present takes a value-­added approach by pinpointing fea-
tured into the use of IVR to teach specialised skills (Gavish tures that can be added to a IVR lesson to increase its instruc-
et al. 2013; Dobrowolski et al. 2021; Tai et al. 2022). An im- tional effectiveness (Mayer, Makransky, and Parong 2022). In
portant consideration in research on learning in IVR concerns value-­added research, such as the present study, researchers
whether procedures learned in IVR can transfer to perfor- compare the learning outcomes of students who learn from an
mance in real-­l ife situations. In a recent meta-­a nalysis, Akgün IVR-­based lesson versus the same lesson with one feature added.
and Atici (2022) reported evidence that students' occupational For example, Lawson and Mayer (2024a) added pretraining to a
safety behaviours taught in IVR could effectively transfer to IVR-­based lesson on climate change, in which learners gained
real-­world settings (Cha et al. 2012; Fu and Li 2020; Zhang experience in using the controllers in an IVR environment, but
et al. 2017). Furthermore, the use of virtual environments this form of pretraining did not improve subsequent learning of
was found to be effective in enhancing STEM competen- the target lesson. In the present study, we example a different
cies (Chen et al. 2019; Rogers et al. 2017). Additional meta-­ form of pretraining, in which students are exposed to videos
analyses have corroborated the utility of IVR as an effective showing what key objects from the IVR environment look like
tool for simulation-­based learning (Chernikova et al. 2020; in the real world. Learning in IVR may be distracting, particu-
Wu, Yu, and Gu 2020) and training purposes (Jensen and larly for students with lower executive function skills (Lawson
Konradsen 2017; Radianti et al. 2020). This substantial body and Mayer 2024b) so there is a need to identify features that can
of evidence supports the integration of IVR in STEM-­based improve learning in IVR.
education, with an emphasis on advancing procedural knowl-
edge and skill development. In contrast to media comparison research, value-­added stud-
ies explore which features can be added to IVR training to im-
Opportunities to learn academic material in IVR are increas- prove student learning (Mayer, Makransky, and Parong 2022).
ing and becoming more easily accessible. For example, in a Grounded in the framework of generative learning theory,
recent market analysis of applications available on the Oculus meaningful learning is proposed to occur when learners

2 of 16 Journal of Computer Assisted Learning, 2025

actively engage in generative processing—a cognitive process Furthermore, while studies have tailored instructional de-
that involves effectively integrating newly acquired knowledge sign principles for learning in IVR, the focus commonly has
with one's existing knowledge base (Mayer 2021). For instance, been on learning academic content, such as botany (Moreno
Parong and Mayer (2018) found that students who were tasked and Mayer 2002, 2004), biology (Meyer, Omdahl, and
with summarising after each segmented IVR lesson performed Makransky 2019; Parong and Mayer 2018, 2020), history (Parong
significantly better on the post-­test compared to those who ex- and Mayer 2021) and climate change (Petersen et al. 2020). More
perienced a continuous IVR lesson without summarisation. recently, studies have just begun applying these strategies to
Similarly, in a study by Makransky et al. (2021), students ex- helping people learn procedural skills (Johnson et al. 2022), sug-
posed to an IVR lesson incorporating the generative learning gesting that instructional design principles may hold significant
strategy of enactment (i.e., engaging in movements or object promise for enhancing skill acquisition in IVR environments.
manipulations that aid in mapping concepts onto meaningful
actions) outperformed their counterparts who received the same In a study aimed at applying an instructional design principle
IVR lesson without enactment on both a procedural knowledge to procedural learning in an IVR setting, Johnson et al. (2022)
test and transfer assessments. found that individuals with lower spatial abilities experienced
learned better when utilising physical arm movements while
Moreover, instructional design studies focus on creating mul- learning a procedural task within a virtual environment. The
timedia presentations that are designed to facilitate learning effectiveness of interacting with virtual objects for individuals
outcomes (Mayer 2021). One such effective instructional design with lower spatial abilities highlights the role of designing vir-
principle based on research with multimedia presentations is tual environments that allow learners to use their body to de-
the pretraining principle, which calls for providing learners with velop better mental models that aid in enhancing the learning
the names and characteristics of key elements before the lesson process. In contrast, individuals with higher spatial abilities
is presented (Mayer 2021; Mayer and Fiorella 2022). Although learned the procedural steps equivalently with or without being
most research on the pretraining principle has involved conven- asked to engage in physical arm movements during learning in
tional media such as narrated slideshows, narrated animations, IVR. This work encourages the idea that well-­designed virtual
or video lectures, research on applying the pretraining principle environments can serve as valuable learning aids for individu-
in IVR environments is in its initial stage (Meyer, Omdahl, and als requiring additional support, particularly when alternative
Makransky 2019; Petersen et al. 2020). The pretraining principle technologies, such as desktop computers might be less effective.
is that students learn better from a lesson when they have been
previously exposed to the names and characteristics of the key In a study conducted by Petersen, Klingenberg, and
components in the lesson. Makransky (2022), they examined the efficacy of teaching stu-
dents the procedural task of conducting a serial dilution in either
A successful example of this approach comes from Petersen an IVR setting or in a real-­life instructional setting. They found
et al. (2020), who implemented the pretraining principle with that students in the real-­life instructional setting outperformed
middle-­ school-­
age students during a climate change educa- the IVR group on a knowledge test, experienced less extraneous
tion programme. They divided students into pretraining and cognitive load and committed fewer errors in accomplishing a
no-­pretraining groups, where those in the pretraining group serial dilution task. Overall, they found that both groups per-
received a narrated introduction video providing background formed similarly in accomplishing a serial dilution, but those in
information on climate change before engaging in a IVR simu- the IVR group committed more errors in completing the task. A
lation. In contrast, students in the no-­pretraining group received challenge posed by this media comparison study concerns how
the same narration but during the IVR simulation itself. The to design IVR learning experiences for training procedural skills
results revealed that those in the pretraining group achieved that enhance learning outcomes. This challenge is addressed in
higher scores on a subsequent transfer test compared to their the present study by examining the effectiveness of adding pre-
counterparts (Petersen et al. 2020). training to an IVR lesson on serial dilution.

In another study, Meyer, Omdahl, and Makransky (2019) used In conclusion, there is encouraging evidence that value-­added
the pretraining principle in an IVR lesson in biology. They asked research can help identify a collection of features that aid learn-
some students to study a pretraining picture consisting of an ing in IVR, such as using spoken words rather than printed
educationally relevant biological cross-­sectioned cell presented words, using conversational language rather than formal lan-
with key terms of the parts of the cell for 1 min and 30 s whereas guage and asking learners to engage in generative learning activ-
others had no pretraining prior to engaging in either an IVR ities such as summarising during pauses in the lesson. Although
environment or before watching a video on a biology lesson on some forms of pretraining have been successful in IVR (Meyer,
cells. Their results indicated that the IVR group with pretraining Omdahl, and Makransky 2019; Petersen et al. 2020), other
scored significantly higher on the retention and transfer post-­ forms have not (Lawson and Mayer 2024a) and none have been
tests than the IVR group with no pretraining. However, they did used for learning procedural skills, so there is a need to better
not find a difference on post-­test scores between the video con- clarify which forms of pretraining are effective for learning a
ditions with and without pretraining, suggesting that pretrain- procedural skill. The present study fills this gap by examining
ing may be particularly helpful in learning academic content in a clearcut form of pretraining that has not yet been applied to
IVR. These findings are in line with studies (Mayer, Makransky, learning in IVR—exposing learners to video-­based images of
and Parong 2022; Parong and Mayer 2020) that demonstrate the what the key elements in the IVR lesson look like in the real
limitations of learning in IVR settings, such as the distractibility world. The reason why we are interested in whether pretraining
caused by excessive sensory stimuli that can hinder learning. could be helpful in the present study is that it has been shown

3 of 16
to be effective in helping students learn in multimedia learning Hypothesis 2. The pretraining group will perform better than
environments involving computer screens. the no-­pretraining group on the in-­person pipetting performance
test (based on the number of correctly completed tasks and num-
ber of errors committed).
2   |   Theoretical Framework and Hypotheses
Hypothesis 3. The pretraining group will perform better than
2.1   |   Learning Outcomes the no-­pretraining group on the pipetting knowledge test (based on
proportion correct).
According to the cognitive theory of multimedia learning
(CTML), learners have limited cognitive resources avail-
able for processing information in our working memory 2.2   |   Extraneous Cognitive Load
(Mayer 2021, 2022). This theory posits that for meaningful
learning to occur learners must engage in appropriate cog- When the learning experience is overwhelming, as can happen
nitive processing during learning: selecting the relevant pre- in IVR, learners may become distracted by extraneous aspects of
sented information for further processing in working memory, the experience (Parong and Mayer 2020; Petersen, Klingenberg,
mentally organising it into a coherent representation and inte- and Makransky 2022). For instance, participants who viewed a
grating it with relevant prior knowledge. However, processing biology lesson in IVR reported more extraneous cognitive load
capacity in working memory is limited so when the demands and showed less engagement based on EEG measures than
are too great, learners are hindered in their knowledge con- those who viewed a slideshow lesson (Parong and Mayer 2020).
struction. When a multimedia lesson is not designed to aid the Similarly, Petersen, Klingenberg, and Makransky (2022) found
learner in selecting, organising and integrating complex infor- that students learning with IVR reported higher levels of extra-
mation in a manner that facilitates effective learning, it can neous cognitive load when compared to students learning in a
lead to cognitive overload. Cognitive overload occurs when real-­life setting. When learners are distracted, this creates extra-
the task or information a learner attempts to process exceeds neous cognitive load—cognitive processing that does not serve
their working memory capacity. Several scenarios can lead to the instructional goal—that can waste the learner's limited
this situation, such as attempting to process an instructional processing capacity during learning. However, with pretrain-
message that contains an excessive amount of irrelevant in- ing, learners can be more focused on core material in the lesson
formation, trying to integrate novel and complex information and thereby reduce extraneous cognitive load. This leads to our
at an unmanageable pace, or encountering difficulty under- fourth prediction:
standing a topic due to a lack of prior knowledge. These types
of instances can lead to cognitive overload, which can lead to a Hypothesis 4. The pretraining group will produce a lower
lack of understanding of the material (Mayer 2021, 2022; Paas mean rating of extraneous cognitive load than the no-­pretraining
and Sweller 2022; Sweller, Ayres, and Kalyuga 2011). group.

One way to mitigate the potential of cognitive overload during

learning in IVR aimed at training a procedural skill is to pro- 2.3   |   Presence
vide pretraining in the names and characteristics of the key
components (Mayer 2021; Mayer and Fiorella 2022). Mayer, Witmer, Jerome, and Singer (2005) define presence as ‘a psy-
Mathias, and Wetzell (2002) proposed a two-­stage model of chological state of “being there” mediated by an environment
learning how a system works consisting of building compo- that engages our senses, captures our attention, and fosters
nent models of how each part works and building a system our active involvement.’ (298). There are four key aspects of
model of how the entire system works. When exposed to a presence, each contributing to the user's immersive experience
multimedia lesson, learners must engage in building both in a virtual environment. The first dimension, Involvement, is
kinds of models, which may lead to cognitive overload. In the defined by the psychological state in which individuals direct
present study, we posit that pretraining helps learners build their attention and actively engage with meaningful stimuli
component models—mental models of the key components or activities in a virtual setting. The second facet, Adaptation/
and their possible actions—so they allocate more cognitive Immersion, is characterised by the sense of complete integra-
resources to building a system model—an understanding of tion and active engagement within a virtual environment that
the step-­by-­step procedure for serial dilution—while they are closely resembles a desired scenario, facilitating an uninter-
training in IVR. In addition, we posit that learners who see rupted flow of stimuli and experiences. Sensory Fidelity, the
videos showing the actual objects and their actions can better third dimension, hinges on the degree of realism experienced
relate what they are learning in virtual reality to their relevant in virtual environments that can support a greater level of
prior knowledge. Based on this analysis, we offer three predic- involvement. Finally, Interface Quality, the fourth aspect, is
tions concerning learning outcomes: determined by the speed and efficiency with which users can
adapt to new surroundings while interacting with the virtual
Hypothesis 1. The pretraining group will perform better than environment, ultimately contributing to an elevated sense of
the no-­pretraining group on pipetting performance in IVR (based immersion. Collectively, these aspects combine to shape the
on the number of correctly completed tasks and number of errors overall sense of presence in virtual environments (Witmer,
committed). Jerome, and Singer 2005).

4 of 16 Journal of Computer Assisted Learning, 2025

Research by Selzer, Gazcón, and Larrea (2019) supports the levels (Bandura 1977; Schunk and DiBenedetto 2016). These key
significance of these dimensions in understanding presence in sources of self-­efficacy can also be applied to learning in virtual
virtual environments. Their study revealed that participants as- reality environments.
signed to the VR groups reported experiencing higher levels of
presence compared to those assigned to a desktop-­based con- In previous studies, learners engaged in performance accom-
dition. Notably, these findings also underscored a positive cor- plishments by manipulating virtual objects and completing ac-
relation between one's sense of presence and learning outcomes, tions in a virtual environment (Johnson et al. 2022; Makransky
highlighting the impact of perceived virtual presence on the ef- et al. 2021; Petersen, Klingenberg, and Makransky 2022).
fectiveness of learning. Learners were also able to observe virtual agents as a method
of vicarious experiences in learning how to complete a set of
While VR systems offer immersive advantages, an excessive actions. Verbal persuasion can be facilitated in a virtual en-
degree of immersion has been associated with increased affec- vironment by allowing virtual agents to provide immediate
tive processing, resulting in diminished cognitive processing feedback of completed actions (Petersen, Klingenberg, and
and subsequent declines in learning outcomes (Parong and Makransky 2022). Finally, virtual environments have been sug-
Mayer 2021). In their study, Parong and Mayer (2021) observed gested to elicit positive affective states, such as feeling a sense
that participants exposed to the IVR condition led to higher of presence, increased motivation and engagement (Parong
levels of emotional and physical arousal in comparison to and Mayer 2018; Makransky et al. 2021). However, there have
those engaged with a video-­based lesson. The video condition, been instances when learning in virtual environments may not
in contrast, was discovered to elicit greater levels of cognitive always provide the most positive emotional and physiological
engagement compared to the IVR condition. These results states (Parong and Mayer 2020). Pretraining may contribute to
emphasise the suggestion that the sensory richness of IVR self-­efficacy by improving student performance during training
content may prompt learners to engage in extraneous process- IVR, yielding the following hypothesis.
ing, a phenomenon well-­documented for its adverse impact
on learning outcomes (Mayer, Makransky, and Parong 2022). Hypothesis 6. The pretraining group will produce a higher
However, it remains uncertain whether the absence of an in- mean rating of their performance on the self-­efficacy post-­test
structional design principle, such as the pretraining principle, than the no-­pretraining group.
will yield comparable levels of presence that enhance effec-
tive learning or if these elevated levels of presence might be
so pronounced that they become a distraction to the learning 2.5   |   Research Questions
process. To investigate this issue, we propose the following
hypothesis that aims to shed light on the relationship between As a preliminary exploration, we are interested in the mech-
instructional design principles and the potential distraction of anisms by which pretraining may affect learning outcomes.
heightened presence: Specifically, we explore whether extraneous cognitive load and
presence mediate the relationship between the pretraining con-
Hypothesis 5. The pretraining group will produce a higher ditions and learning outcomes.
mean rating of presence than the no-­pretraining group.
Research Question 1. Will extraneous cognitive load mediate
the relationship between pretraining conditions and learning out-
2.4   |   Self-­Efficacy comes. Specifically, will students who experience lower extrane-
ous cognitive load have better learning outcomes than those who
Self-­efficacy is defined as the belief of one's own capabilities to experience greater extraneous cognitive load?
learn and perform behaviours at designated levels. The theory
of self-­efficacy, originating from Bandura's (1977) work, has Research Question 2. Will presence mediate the relationship be-
continued to develop, and be applied across various academic tween pretraining conditions and learning outcomes? Specifically,
domains (Schunk and DiBenedetto 2016). Researchers have ex- will students who experience a higher sense of presence have bet-
plored its multifaceted role, demonstrating its positive impact ter learning outcomes than those who experience less sense of
on learning, motivation, achievement and self-­regulation subse- presence?
quent to Bandura's initial work (Schunk and DiBenedetto 2016).
Bandura (1977) proposed four key sources through which indi- These hypotheses and research questions are summarised in
viduals gauge their self-­efficacy: performance accomplishments, Table 1. The hypotheses are that the pretraining group will
vicarious experiences, verbal persuasion and physiological carry out the procedure better in IVR and in person, answer
states. Performance accomplishments reflect how individuals the questions more accurately on the pipetting knowledge test,
interpret their capabilities based on their own achievements, and experience lower extraneous cognitive load, higher pres-
serving as a fundamental gauge. Vicarious experiences involve ence and higher self-­efficacy as compared to the no-­pretraining
learning through observing others’ actions, particularly when group. The research questions concern whether the first three
performing threatening activities without adverse consequences. predictions apply more strongly to learners with lower rather
Verbal persuasion encompasses the influence of positive feed- than higher self-­reported extraneous cognitive load and higher
back and encouragement from others in shaping individuals’ rather than lower self-­reported presence. In this study, the hy-
beliefs about their capabilities. Lastly, physiological cues from potheses provide anticipated answers to research questions
emotional and physiological states pertaining to a task act as de- based on the learning theories, whereas research questions are
terminants that either constrain or augment one's self-­efficacy exploratory questions without expected results.

5 of 16
TABLE 1    |    Six hypotheses and two research questions. or Pacific Islander, 5 classified themselves as Middle Eastern or
North African, and 1 classified themselves as other and opted
Pretraining group
to identify as Indian. The ages of the participants ranged from
will perform Pretraining group
18 to 22 years (Mage = 19.00, SDage = 1.13). Twenty-­six students
No. better on will have
reported to be first-­generation college students. We originally
1 Carrying out More correctly completed recruited 104 participants but had to eliminate 11 participants
the pipetting steps and fewer due to technical difficulties (n = 7), simulation sickness in the
procedure in IVR errors committed virtual reality simulation (n = 1), missing data (n = 2) and not
being 18 years of age or older (n = 1), yielding a final sample of
2 Carrying out the More correctly completed
93 participants. The experiment employed a between-­subjects
pipetting procedure steps and fewer
design with 47 participants in the pretraining group and 46 in
in person errors committed
the no-­pretraining group. Participants were randomly assigned
3 Answering questions Higher proportion correct to either an experimental group (i.e., pretraining group) or a con-
on the pipetting on the knowledge test trol group (i.e., no-­pretraining group). The experimental group
knowledge test watched a pretraining video focused on the names and charac-
teristics of a pipette and a brief description of a completed serial
4 Experiencing lower Lower self-­rating of
dilution, while the control group received no pretraining. Our
extraneous cognitive cognitive load
dependent variables included knowledge retention, assessed by a
load during learning
knowledge test and pipetting procedural performance measured
5 Experiencing a higher Higher self-­rating through pipetting errors and accuracy in completing a four-­step
sense of presence of presence serial dilution task. We also assessed extraneous cognitive load
during learning and sense of presence as additional dependent variables, using a
post-­survey for both. The independent variable was the presence
6 Experiencing higher Higher self-­rating
or absence of the pretraining video.
self-­efficacy for the of self-­efficacy
pipetting procedure

3.3   |   Materials and Equipment

RQ1: Do Hypotheses 1–3 apply to learners with lower rather 3.3.1   |   Equipment

than higher extraneous load?
The equipment used in this experiment included a virtual re-
RQ2: Do Hypotheses 1–3 apply to learners with higher rather ality system, a laptop computer and pipetting equipment. The
than lower presence? virtual reality system was an Oculus Quest, from the company
MetaQuest, formally known as Oculus (https://​oculus.​com/​go),
used to present the training lesson. The laptop computer was a
3   |   Method MacBook Pro used to collect survey responses through the web-­
based platform, Qualtrics and to present an mp4 video of the
3.1   |   Transparency and Openness pretraining lesson. The pipetting equipment was used for the in-­
person serial dilution task. The materials consisted of an adjust-
The hypotheses, design and analytic strategy were preregistered able micropipette with a volume range of 100 to 1000 μL, a set of
during data collection, but before analysis were undertaken at sterile pipette tips, a set of beakers, four plastic test tubes housed
https://​osf.​io/​a s4j3/​? ​v iew_ ​only=​1d046 ​0 9fe4 ​c 4403 ​0 9f4f​efa6d​ in a rack, a disposable cup for the used pipette tips, cranberry
8dc9a82. We report how we determined our sample size, all juice and water used for the solution, professional white lab
data exclusions, all manipulations and all measures in the study. coats, a box of nitrile-­v inyl blend exam gloves and a document of
Data were analyzed using SPSS version 27. the pipetting instructions.

3.2   |   Participants and Design 3.3.2   |   Prequestionnaire

We used the software programme G*Power to conduct our power The prequestionnaire consisted of 11 items administered on
analysis. Our goal was to obtain a 0.80 power to detect a medium a laptop computer via the web-­based platform, Qualtrics. The
effect size of 0.6 at the standard 0.05 alpha error probability. Our first five questions assessed general interest and experience in
recommended sample size was 90 participants. The participants science, with response options of ‘Yes’ or ‘No’ (e.g., ‘Science
were 93 college students recruited from the Psychology Subject is my favourite subject in school. (a) Yes. (b) No.’; Parong and
Pool at a large university in California. They received credit to- Mayer 2018). Responses were coded as: 1 for ‘No’ and 2 for
wards fulfilling a course requirement. Concerning gender, 36 ‘Yes’. The scores from these five questions were averaged to
students identified as male, 55 identified as female and 2 iden- obtain a total mean rating for general interest and experience
tified as non-­binary. Concerning ethnic/racial category, 40 stu- in science. The next two questions asked about knowledge of
dents classified themselves as White, 18 classified themselves as micropipette and conducting a serial dilution. For the micro-
Latinx/Chicano/a, 6 classified themselves as Black, 36 classified pipette question, participants were asked to rate their knowl-
themselves as Asian, 4 classified themselves as Native Hawaiian edge of micropipetting on a scale from 1 ‘I don't know what a

6 of 16 Journal of Computer Assisted Learning, 2025

micropipette is’ to 6 ‘Very high’. The serial dilution question to help learners build component models of each of the key
asked participants to rate their knowledge of conducting a se- components of pipetting: the digital display, the volume ad-
rial dilution, using a similar scale, 1 for ‘I don't know what a justing knob, the tip cone, the plunger and the tip ejector but-
serial dilution is’ to 6 ‘Very high’. Each question was averaged ton. Specifically, the pretraining video provided the names
to produce a total mean rating score. The final four questions and characteristics of each part of a micropipette, including a
asked about self-­efficacy relating to pipetting. This four-­item description of the digital display and its purpose, description
self-­
efficacy questionnaire (α = 0.823) was a modified ver- and use of the volume adjusting knob, a description of the tip
sion of the self-­efficacy subscale in the Motivated Strategies cone and how to insert a pipette tip, description of two stops
for Learning Outcomes (MSLQ), as modified by Petersen, of the plunger for aspirating and dispensing and a description
Klingenberg, and Makransky (2022). It consisted of four of the tip ejector button and its purpose. During each expla-
statements, each with a five-­point Likert scale ranging from nation the instructor demonstrated the actions of these parts,
1 ‘strongly disagree’ to 5 ‘strongly agree’. The self-­efficacy ending with a brief description of what a serial dilution is. The
responses were summed to give us a total self-­efficacy score. pretraining video was stored as an mp4 file and displayed on
We administered the same self-­efficacy scale pre-­and-­post a MacBook Pro.
experiment.

3.3.4   |   Virtual Reality Lesson

3.3.3   |   Pretraining Lesson
The virtual reality lesson (or virtual simulation) used in this
The pretraining video, which we developed for the pretrain- study was designed to teach the technical scientific procedure of
ing group, was 4 min in length and was intended to show and pipetting, which involves using a pipette to measure and trans-
describe the physical objects and actions involved in the vir- fer a specific volume of liquid. The virtual simulation was de-
tual reality lesson, as seen in Figure 1. The video was intended veloped using Unity 2021 and was obtained from researchers in

FIGURE 1    |    Screenshots of the pretraining video: (a) introducing the micropipette. (b) explaining the parts and characteristics of a micropipette.
(c) demonstrating how to insert a tip to a micropipette. (d) demonstrating how to aspirate liquid. (e) demonstrate how to dispense liquid. (f) briefly
introduce what a complete serial dilution looks like.

7 of 16
the Psychology Department at the University of Copenhagen. participants could transfer their simulation-­based procedure
The virtual simulation contained two sections. The first section to the real world. To keep the materials consistent across tasks,
served as a tutorial for using a pipette to extract and transfer liq- we used physical materials that were closely similar to the ma-
uid from one test tube to another, in which the participant was terials in the virtual simulation. After the completion of the
prompted to carry out 15 steps. At every step, the virtual agent virtual simulation task, participants were asked to remove
first demonstrated the task, provided step-­by-­step instructions on the virtual headset and shown the real pipetting setup. The
how to complete it, and gave feedback at the end of every step. researchers showed the participants the physical materials
Each step needed to be completed accurately to move on to the to complete the task along with the printed instructions. The
following step. If participants were stuck on a particular step, printed four-­step serial dilution instructions were a modified
they were allowed to ask the researcher for help. The purpose version of the virtual four-­step serial dilution instructions, as
of the first section was to teach individuals on the basic safety seen in Figure 5.
guidelines when working in a laboratory and how to use a mi-
cropipette, which were necessary for conducting a serial dilution. A researcher was responsible for observing and evaluating the
participants performance during the completion of the serial
The second section served as the virtual pipetting test, where dilution on a printed assessment form. The scoring of the as-
participants were tasked to complete a four-­step serial dilu- sessment followed the same structure as the virtual assess-
tion by applying the skills they learned in the first section of ment, consisting of two sections. The first section was scored
the training tutorial. In the virtual pipetting test, participants as either True or False, with True scored as 1 and False scored
were shown an image of the instructions on a monitor and as 0. The minimum score a participant could receive was 0,
were told the virtual agent would no longer be available to whereas the maximum score a participant could receive was
help them. The participants were given the same materials 5. The researcher indicated whether the participant put on a
presented from the first section of the lesson but were tasked lab coat and gloves as True or False and whether the partici-
to independently complete the four-­step serial dilution. After pant correctly completed each of the four test tubes as True
the participant felt they completed the four-­step serial dilu- or False.
tion, they were instructed to push on a button in the virtual
lab to see their feedback from the test. The feedback was dis- The second section was scored as a frequency and was summed
played on the monitor and outlined whether they put on gloves together to produce a total score of pipetting errors committed
and a lab coat, the number of times they tilted the pipette, the in the in-­person serial dilution test. The minimum score one
number of times they did not use the first stop on the plunger, could receive was 0, whereas the maximum score one could
the number of times they did not use the second stop of the receive was unlimited. The researcher recorded the number of
plunger and the number of times they cross contaminated the instances in which a participant committed the following errors:
liquid from the beakers and the tubes. tilting the pipette tip, pressing the plunger after inserting the pi-
pette in liquid, dispensing liquid without reaching the second
The feedback shown to the participant was generated into stop and cross contaminating the pipette tip across all test tubes
a file in the virtual reality headset, wherein the researcher and beakers.
was able to transfer the participant data into a computer. The
scoring was calculated based on whether the participant ac-
curately transferred the required solution in each four test 3.3.6   |   Knowledge Test
tubes. If a test tube contained the correct amount of liquid
the virtual simulation marked it either as ‘True’ for correct The knowledge test, which was administered in Qualtrics, was
and ‘False’ for incorrect. The virtual simulation also indi- intended to assess participants' knowledge of the scientific pro-
cated whether the participant put on a lab coat and gloves as cedure of pipetting and consisted of 15 questions (i.e., four mul-
‘True’ for correct and ‘False’ for incorrect. During the com- tiple choice questions), 10 true-­false questions and one sorting
pletion of the serial dilution, the virtual simulation recorded question (α = −0.161). The low internal reliability reflects the
the number of instances in which a participant committed the construction of the test, which involves three different ques-
following errors: tilting the pipette tip, pressing the plunger tion formats and individual items that tap completely different
after inserting the pipette in liquid, dispensing liquid without pieces of information. The multiple-­choice items consisted of
reaching the second stop, using a contaminated pipette tip and four multiple-­choice options (e.g., ‘What happens during a se-
improperly mixing the beaker liquid. Figure 2 shows a series rial dilution? (a) The concentration of a substance is reduced at
of screenshots of the virtual pipetting simulation for the train- each step. (b) Water is distilled at each step to increase the con-
ing phase. Figure 3 shows a screenshot of the virtual pipetting centration of a substance. (c) Each step involves adding a new
instructions presented in the second section of the simulation substance. (d) Each step reduces the acidity of a substance’).
involving carrying out a serial dilution in IVR without guid- The true-­false questions consisted of 10 questions (e.g., ‘Push
ance. Figure 4 shows images of the real-­world lab setup during the plunger to the second stop during aspiration. (a) True (b)
the in-­person serial dilution task. False’). The sorting question consisted of eight statements dis-
played in a random order, which required the participant to
drag and drop each statement in the correct order of the pro-
3.3.5   |   In-­Person Pipetting Test cedure (e.g., ‘Please indicate which step each statement cor-
responds to from first to last step. (a) Put lab coat (b) Put on
The in-­person pipetting task was set up similarly to the virtual gloves (c) Pick up pipette (d) Adjust volume setting (e) Attach
pipetting task using concrete objects to determine whether pipette tip (f) Aspirate liquid (g) Dispense liquid (h) Discard

8 of 16 Journal of Computer Assisted Learning, 2025

FIGURE 2    |    Screenshots of virtual pipetting simulation during the training phase. (a) Instructions are to grab a pair of gloves. (b) Instructions are
to refer to the monitor to change the volume on the virtual pipette. (c) Instructions to hold the pipette in different angles to see what the appropriate
way is to hold a pipette. (d) Instructor demonstrating how to aspirate liquid from a beaker. (e) Instructors on how to dispense liquid into a test tube.

tip’). Participants who selected the correct answer or sorted

each statement in the correct order were coded as 1, while in-
correct responses and missorted statements were coded as 0. A
participant's post-­test score was determined by the number of
correct responses recorded. The total possible points a partici-
pant could receive was 22 points.

3.4   |   Postquestionnaire

The postquestionnaire, which was administered in Qualtrics, was

intended to assess cognitive load and presence. We assessed par-
ticipants' extraneous cognitive load by administering two cogni- FIGURE 3    |    Screenshot of the instructions for the four-­step serial di-
tive load questionnaires. We administered the Multidimensional lution task presented during the testing phase.
Cognitive Load Scale for Virtual Environments (Andersen and
Makransky 2020), which is composed of 17 items divided into
3 categories: intrinsic load (α = 0.80), extraneous load (α = 0.86) 3.5   |   Procedure
and germane load (α = 0.91). It is anchored by a five-­point Likert
scale ranging from 5 ‘strongly agree’ to 1 ‘strongly disagree’. We Participants were tested individually and randomly assigned to
also administered the Cognitive Load Questionnaire (Parong either a pretraining group or a no-­pretraining group. Participants
2019), which consisted of 9 items divided into 3 categories: ex- were seated at a workstation facing a computer screen. Upon ar-
traneous load (α = 0.91), essential load (α = 0.64) and generative rival, the participant was asked to review a paper version of the
load (α = 0.66). This scale ranged from 7 ‘strongly agree’ to 1 consent form. If they agreed to participate, they were asked to
‘strongly disagree’. However, due to its lower Cronbach's alpha sign and date the consent form. Following this, the participant
and the similarity of results, we decided to only include the re- completed the prequestionnaire administered on a computer via
sults of the Multidimensional Cognitive Load Scale for Virtual Qualtrics. After completing the prequestionnaire, those assigned
Environments. To assess a participant's sense of presence we to the pretraining condition watched an instructional video that
administered a modified version of the Presence Questionnaire explained the parts and functions of a micropipette, how to hold
containing 7 items (α = 0.77; Parong 2019), in a seven-­ point and use a micropipette and a brief explanation of what a serial
Likert scale, ranging from 7 ‘strongly agree’ to 1 ‘strongly dis- dilution is. Participants assigned to the no-­pretraining group did
agree’. Following this, participants were asked the same four not watch the pretraining video. Next, the participant was in-
questions about their self-­efficacy relating to pipetting from the structed to stand on a designated marked spot to be prepped on
prequestionnaire. Finally, participants were asked to provide how to use the Oculus Quest virtual reality (VR) head-­mounted
demographic information concerning their age, gender cate- display (HMD). The participant was shown how to use the hand
gory, ethnic/racial category and first-­generation college student controllers and headset. Next, the participant put on the VR
status. headset and was briefed on the safety features in the headset,

9 of 16
FIGURE 4    |    Pictures of the real-­world lab setup of the in-­person serial dilution testing phase. (a) Lab setup for the in-­person serial dilution task.
(b) Completed 4-­step serial dilution.

serial dilution task, a researcher evaluated the accuracy on ap-

plying the correct amount of liquid in each tube (e.g., did the
participant accurately complete test tube A) and collected the
frequency of errors committed while completing the serial dilu-
tion task. After completing the in-­person serial dilution task, the
participant was directed back to the original computer to com-
plete the knowledge test and postquestionnaire via Qualtrics.
Once the participants completed the postquestionnaire they
were thanked for their time and were informed they would re-
ceive credit for their participant through the SONA participant
pool system. We adhered to guidelines for treatment of human
subjects and obtained IRB approval.

4   |   Results

4.1   |   Do the Groups Differ on Basic

Characteristics?
FIGURE 5    |    Instructions for the in-­person serial dilution task.

A preliminary issue concerns whether random assignment re-

then instructed to input their participant ID number using the sulted in groups that differed on basic characteristics. A chi-­
hand controllers. Both groups were instructed to complete the square analysis demonstrated that the proportion of men and
virtual simulation, which was guided by a virtual agent that women did not differ significantly among the two groups, 𝜒 2 (2,
taught the technical procedure of pipetting. The simulation N = 93) = 0.053, p = 0.876. The two groups also did not differ sig-
contained two sections. The first section served as a practical nificantly in mean age, t(91) = −0.731, p = 0.467, d = −0.152, mean
tutorial that taught lab safety procedures and step-­by-­step ac- science-­knowledge score, t(91) = 0.294, p = 0.770, d = 0.061, prior
tions on how to use a micropipette. The second section served as knowledge with using a micropipette, t(91) = 1.634, p = 0.106,
an assessment of the skills learned from the first section, where d = 0.339 and prior knowledge with conducting a serial dilu-
learners were prompted to complete a four-­step serial dilution tion, t(91) = 0.391, p = 0.696, d = 0.081. We conclude that the two
task independently and received immediate feedback from the groups did not differ on basic characteristics.
virtual agent.
Hypothesis 1. Does the pretraining group perform better
After the completion of the second section, the participant than the no-­pretraining group on pipetting performance during
removed the VR headset and was directed to complete an in-­ learning in IVR?
person four-­step serial dilution task in the lab. The instructions
for the in-­person serial dilution task were based on the VR sim- The first hypothesis predicted that the pretraining group will
ulation instructions. A section in the lab was designated for the perform better than the no-­ pretraining group on pipetting
in-­person task. While the participant completed the in-­person performance during learning in IVR. Table 2 shows the mean

10 of 16 Journal of Computer Assisted Learning, 2025

TABLE 2    |    Overview of results for all six hypotheses.

Hypotheses Condition Mean SD p-­value Effect size

H1: Errors on the serial dilution task in IVR Pretraining 11.68 9.07 0.298 d = 0.217
No pretraining 13.63 8.90
H1: Correct solutions in IVR Pretraining 5.65 1.94 0.210 d = −0.261
No pretraining 5.17 1.77
H2: Errors on serial dilution task in-­person Pretraining 6.30 6.26 < 0.001 d = 0.784
No pretraining 12.43 9.18
H2: Correct solutions in-­person Pretraining 7.09 1.38 0.306 d = −0.213
No pretraining 6.76 1.65
H3: Knowledge test Pretraining 18.65 1.76 0.001 d = −0.729
No pretraining 17.23 2.12
H4: Extraneous cognitive load Pretraining 2.11 0.440 0.048 d = 0.420
No pretraining 2.37 0.773
H5: Presence Pretraining 37.48 6.22 0.888 d = 0.029
No pretraining 37.69 7.85
H6: Self-­efficacy Pretraining 15.77 2.52 0.196 η2p = 0.018
No pretraining 16.54 2.87
Note: Bold indicates statistically significant values.

and standard deviation for both the number of errors and cor- significant interaction between the pipetting tasks and condition,
rect solutions in the virtual simulation across the two groups. F(1, 91) = 5.130, p = 0.026, η2p = 0.053. Specifically, individuals in
Contrary to predictions, no significant difference was found be- the pretraining group committed approximately half the number
tween groups in the number of errors produced during the serial of errors (M = 6.30, SD = 6.26) during the in-­person pipetting task
dilution task in the virtual simulation, t(91) = 1.046, p = 0.298, when compared to the virtual task (M = 11.68, SD = 9.07), whereas
d = 0.217, 95% CI [−0.191, 0.624], Similarly, the groups did not individuals in the no-­pretraining group demonstrated a relatively
differ significantly on the proportion of correct solutions on consistent number of errors on the virtual pipetting task (M = 13.63,
the serial dilution task in the virtual simulation, t(91) = −1.263, SD = 8.90) and the in-­person pipetting task (M = 12.43, SD = 9.18).
p = 0.210, d = −0.261, 95% CI [−0.668, 0.148]. We conclude that
Hypothesis 1 was not supported. Hypothesis 3. Does the pretraining group perform better
than the no-­pretraining group on the knowledge test?
Hypothesis 2. Does the pretraining group perform better
than the no-­pretraining group on pipetting performance in the in-­ The third hypothesis predicted that the pretraining group will per-
person pipetting test? form better than the no-­pretraining group on the pipetting knowl-
edge test. Table 2 shows the mean number of correct responses on
The second hypothesis predicted that the pretraining group will the knowledge test (and SD) for the two groups. The groups signifi-
perform better than the no-­pretraining group on the in-­person cantly differed on the number of correct responses on the pipetting
pipetting test. Table 2 shows the mean and standard deviation knowledge test, t(91) = −3.773, p = 0.001, d = −0.729, 95% CI (−1.15,
for the number of errors and correct solutions in the in-­person −0.31), indicating that those in the pretraining group scored higher
pipetting test for both groups. The groups significantly differed on the knowledge test than those in the no-­pretraining group. We
in the number of errors produced during the serial dilution task, conclude that Hypothesis 3 was supported.
t(91) = 3.773, p < 0.001, d = 0.784, 95% CI [0.35, 1.20], indicating
that those in the pretraining group produced fewer pipetting er- Hypothesis 4. Does the pretraining group produce a lower
rors than those in the no-­pretraining group. In contrast to our mean rating of extraneous cognitive load than the no-­pretraining
predictions, the groups did not significantly differ on the pro- group?
portion of correct solutions on the in-­person serial dilution task,
t(91) = −1.029, p = 0.306, d = −0.213, 95% CI [−0.621, 0.195]. The fourth hypothesis predicted that the pretraining group will
Thus, we conclude that Hypothesis 2 was partially supported. produce a lower mean rating of extraneous cognitive load than
the control group. Table 2 shows the mean number of ratings for
In addition to our above analysis, we conducted a 2 × 2 ANOVA extraneous cognitive load (and SD) for the groups. A Levene's test
to investigate whether the number of pipetting errors changed for equality of variances was found to be violated for the pres-
across pipetting tasks and conditions. Our results indicated a ent analysis, F(1, 91) = 9.804, p = 0.002. Given the violation of the

11 of 16
assumption of equal variances, our degrees of freedom were ad- In contrast, as summarised in Table 4, we did find that Condition
justed from 91 to 71. As seen in Table 2, the groups significantly was a significant predictor of extraneous cognitive load, indicat-
differed on the number of ratings for extraneous cognitive load, ing those in the pretraining condition had a 0.263 reduction of
t(71) = 2.012, p = 0.048, d = 0.42, 95% CI [0.007, 0.830]. Specifically, extraneous cognitive load, when compared to the no-­pretraining
those in the pretraining group reported experiencing less extra- condition (b = −0.263, SE = 0.130, p = 0.046). Additionally, as
neous cognitive load than did those in the no-­pretraining group. summarised in Table 5, we found that extraneous cognitive
These results demonstrate support for Hypothesis 4. load was also a significant predictor of total pipetting errors,
indicating a one-­unit increase in extraneous cognitive load re-
Hypothesis 5. Does the pretraining group produce a higher sulted in an increase of 3.450 in committing an error (b = 3.450,
mean rating of presence than the no-­pretraining group? SE = 1.563, p = 0.029) (see Table 5).

The fifth hypothesis predicted that the pretraining group Finally, we found that Condition was still a significant predic-
will produce a higher mean rating of presence than the no-­ tor of total pipetting errors when including presence and ex-
pretraining group. Table 2 shows that mean number of ratings traneous cognitive load in the model (b = −5.202, SE = 1.656,
for presence (and standard deviation) for the groups. The groups p = 0.002); this suggests that those in the pretraining condition
did not significantly differ on mean rating for sense of presence, had a 5.202 reduction in committing a pipetting error. However,
t(91) = 0.141, p = 0.888, d = 0.029, 95% CI (−0.377, 0.436). The as summarised in Table 6, after analysing the indirect effects, re-
findings do not provide support for Hypothesis 5. sults reveal Presence (ab = −0.025, 95% CI [−0.756, 0.567]) and
Extraneous Cognitive Load (ab = −0.908, 95% CI [−2.650, 0.044])
Hypothesis 6. Does the pretraining group produce a higher do not significantly mediate the relationship between Condition
mean rating of self-­efficacy post-­survey than the no-­pretraining and total pipetting errors (see Table 6).
group?

The sixth hypothesis predicted that the pretraining group will 4.2.2   |   Knowledge Test Score
produce a higher mean rating on the self-­efficacy post-­survey
than the no-­pretraining group. Table 2 shows the mean ratings A mediation analysis was conducted on SPSS PROCESS macro
on the self-­efficacy post-­survey (and standard deviation) for the extension version 4. Our independent variable Condition was
two groups. We conducted an ANCOVA, with rating on the self-­ coded as 0 for the no-­pretraining condition and 1 for the pre-
efficacy pre-­survey as a covariate. Overall, the groups did not training condition. As shown in Table 7, our results indicated
differ significantly, F(1, 90) = 1.696, p = 0.196, η2p = 0.018. We that Condition was not a significant predictor of presence
conclude that Hypothesis 6 was not supported. (b = −0.206, SE = 1.46, p = 0.888); and as shown in Table 9,
presence was not a significant predictor of the total score in the
knowledge test.
4.2   |   Research Questions: Does Extraneous
Cognitive Load and Presence Mediate TABLE 3    |    Condition predicting presence.
the Relationship Between the Pretraining
Condition and Learning Outcomes? β b SE t p
Condition −0.029 −0.206 1.460 −0.140 0.888
Learning outcomes were operationalised using two separate as-
sessments: the number of pipetting errors committed during the
in-­person pipetting task (i.e., a behavioural measure) and the
TABLE 4    |    Condition predicting extraneous cognitive load.
score received on the knowledge test (i.e., a knowledge-­based
measure). We chose to keep these two assessments distinct, as β b SE t p
they represent different dimensions of learning, one focused on
practical skill execution and the other on cognitive understand- Condition −0.412 −0.263 0.130 −2.023 0.046
ing. One set of research questions investigates whether presence Note: Bold indicates statistically significant values.
or extraneous cognitive load mediates the relationship between
pretraining conditions and learning outcomes, based on pipet-
ting errors and score on knowledge test scores. TABLE 5    |    Condition, presence and extraneous cognitive load
predicting total pipetting errors.

4.2.1   |   In-­Person Pipetting Errors β b SE t p

Constant −0.491 8.021 −0.061 0.951
A mediation analysis was conducted on SPSS PROCESS macro
extension version 4. Our independent variable Condition was Condition −0.620 −5.202 1.656 −3.140 0.002
coded as 0 for the no-­pretraining condition and 1 for the pre- Presence 0.105 0.125 0.138 0.905 0.367
training condition. As summarised in Table 3, our results indi-
cated that Condition was not a significant predictor of presence Extraneous 0.262 3.450 1.563 2.207 0.029
(b = −0.206, SE = 1.46, p = 0.888); and as summarised in Table 5, cognitive
presence was not a significant predictor of total pipetting errors load
(b = 0.125, SE = 0.138, p = 0.367). Note: Bold indicates statistically significant values.

12 of 16 Journal of Computer Assisted Learning, 2025

In contrast, as shown in Table 8, we did find that Condition was in-­person pipetting test) and in answering questions about pipet-
a significant predictor of extraneous cognitive load, indicating ting (by increasing correct responses on the knowledge test). Thus,
those in the pretraining condition had a 0.263 reduction of ex- although pretraining did not help students perform better during
traneous cognitive load, when compared to the no-­pretraining the simulation-­based training, it did allow them to learn in a way
condition (b = −0.263, SE = 0.130, p = 0.046). However, as shown that promoted transfer beyond the simulation. Our results are sim-
in Table 9, we did not find that extraneous cognitive load was ilar to the findings of Meyer, Omdahl, and Makransky (2019), in
a significant predictor of one's score on the knowledge test which pretraining was beneficial for participants in the IVR group
(b = −0.469, SE = 0.395, p = 0.238). on retention and transfer in a different domain. Thus, while pre-
training in the form of gaining experience with being in other IVR
Finally, we found that Condition was still a significant predictor environments was not effective in previous research (Lawson and
of total score on the knowledge test when including presence and Mayer 2024a), pretraining that focused on exposing learners to
extraneous cognitive load in the model (b = 1.294, SE = 0.419, the names and characteristics of the objects before the lesson was
p = 0.002); this suggests that those in the pretraining condi- effective in this study. Thus, the present study shows how it is pos-
tion had a 1.294 increase in their total score on the knowledge sible to extend the pretraining principle from multimedia lessons
test as shown in Table 9. After analysing the indirect effects, (Mayer 2021) on a computer screen to IVR lessons.
results reveal Presence (ab = 0.002, 95% CI [−0.122, 0.126]) and
Extraneous Cognitive Load (ab = 0.123, 95% CI [−0.100, 0.470]) In addition, as predicted, pretraining reduced learners' feel-
do not significantly mediate the relationship between Condition ings of extraneous cognitive load; but in contrast to predictions,
and total score in the knowledge test, as shown in Table 10. pretraining did not affect ratings of presence or self-­efficacy.
Mediation analyses revealed that extraneous cognitive load par-
Overall, we conclude that the effects of pretraining on promoting tially mediated the relationship between instructional condition
transfer from the simulation to pipetting tasks outside the simula- and in-­person pipetting errors committed during the in-­person
tion are not mediated by the presence or extraneous cognitive load. serial dilution task. However, based on bootstrap measures,
extraneous cognitive load did not fully mediate this relation-
ship, indicating that Research Question 1 was only partially
5   |   Discussion supported. Further mediation analyses indicated that presence
did not contribute significantly to the relationship between in-
5.1   |   Empirical Contributions structional condition and knowledge test scores or in-­person pi-
petting errors during a serial dilution task. Our results do not
The purpose of this study was to examine whether learning a tech- support our Research Question 2, that a sense of presence would
nical procedural task in IVR could be facilitated through pretrain- increase learning outcomes. The results concerning presence
ing, which involved seeing the relevant physical objects and actions are in line with Parong and Mayer (2020, 2021) who found no
in a short video. Overall, both groups performed at equivalent lev- significant difference in sense of presence between those who
els during training in IVR, but the pretraining group showed an actively answered questions during pauses in an IVR lesson and
advantage in transferring what they learned in virtual reality to
pipetting performance in the real world (by reducing errors on the

TABLE 9    |    Condition, presence and extraneous cognitive load

TABLE 6    |    Condition on presence and extraneous cognitive load predicting total score on knowledge test.
confidence intervals and bootstrap analysis for the indirect effects.
β b SE t p
Indirect Boot
effects Boot SE Boot LLCI ULCI Constant 18.833 2.029 9.279 < 0.001
Condition −0.627 1.294 0.419 3.088 0.002
Presence −0.025 0.295 −0.756 0.567
Presence −0.043 −0.012 0.035 −0.363 0.717
Extraneous −0.908 0.690 −2.650 0.044
cognitive Extraneous −0.145 −0.469 0.395 −1.186 0.238
load cognitive load
Note: Bold indicates statistically significant values.

TABLE 7    |    Condition predicting presence.

β b SE t p TABLE 10    |    Condition on presence and extraneous cognitive load

confidence intervals and bootstrap analysis for the indirect effects.
Condition −0.029 −0.206 1.460 −0.140 0.888
Indirect Boot
effects Boot SE Boot LLCI ULCI
TABLE 8    |    Condition predicting extraneous cognitive load.
Presence 0.002 0.056 −0.122 0.126
β b SE t p Extraneous 0.123 0.144 −0.100 0.470
Condition −0.412 −0.263 0.130 −2.023 0.046 cognitive
load
Note: Bold indicates statistically significant values.

13 of 16
those who experienced the IVR lesson without answering any constructed pretraining in the names and characteristics of the
questions. key objects and actions. This can be accomplished by showing a
video of the key parts and their possible actions.
Overall, the main contribution of this study to the knowledge
base is that pretraining in the names and characteristics of the
key objects in the lesson through short video was effective in 5.4   |   Limitations and Future Directions
promoting transfer of learning of a technical procedural skill
from IVR to tasks outside of IVR including performing the tech- As an early study of how to improve learning of technical pro-
nical procedure with real objects and answering questions about cedural skills in IVR, this work is subject to several limitations.
the procedure as well as reducing extraneous cognitive load This was a short-­term lab study with immediate tests, so it
during instruction. would be useful to examine whether pretraining is effective in
more realistic classroom environments over longer time periods.
This study involved a narrow learner population of young adults
5.2   |   Theoretical Contributions at a highly selective university and focused on a single learning
topic, so it would be useful to examine whether pretraining is
The results from this study indicate the value-­added of watch- effective with other kinds of learners and with other kinds of
ing a pretraining video before learning in IVR, specifically for technical procedural tasks. Although this work encourages the
transferring learning of a technical procedure skill performed use of pretraining for training of technical procedures in IVR,
in a virtual reality simulation to performing and using the skill work is needed to determine the robustness of the effects.
in the real world. The results of this study are in line with the
pretraining principle developed from the CTML, which states While the gross motor actions were designed to replicate real
that people learn more deeply from a multimedia lesson when micropipette handling, the execution of fine motor actions was
they know the names and characteristics of the main concepts subject to some limitations. Students received visual guidance
before experiencing the multimedia lesson (Mayer 2021; Mayer on micropipette use within the IVR lesson; however, discrepan-
and Fiorella 2022). Most of the previous research investigating cies occurred between the visual demonstrations and the pre-
the pretraining principle have been conducted on conceptual cise fine motor skills required for task completion. Moreover,
learning (Mayer 2021; Mayer and Fiorella 2022). However, this the task demanded more than replicating movements. Students
is the first study to incorporate the pretraining principle in the had to remember specific buttons and their corresponding ac-
acquisition of technical procedural skills. The proposed mech- tions, potentially imposing additional cognitive demands. For
anism is that pretraining reduces extraneous cognitive load instance, while the tutorial demonstrated using the thumb to
during training by helping learners know the names and charac- push the plunger to the first stop, students were required to use
teristics of the key components discussed in the lesson, so learn- their index finger and press halfway down the button to achieve
ers can allocate more of their cognitive processing resources to the same result. Subsequently, after completing the tutorial
learning the procedure, which is a form of essential processing phase, students had to recall which actions aligned with specific
(Mayer 2021, 2022). In addition, pretraining activates relevant buttons during the virtual serial dilution task. Research com-
prior knowledge about the physical objects and actions that are paring the usage of classic VR controllers and ergonomic con-
represented in the simulation, so that learners can connect what trollers, such as Valve's Knuckles presents encouraging results
they are doing in the simulation to concrete knowledge in their (Noblecourt et al. 2021). After familiarity with Knuckles in an
long-­term memory, which is a form of generative processing initial study, participants demonstrated quicker task completion
during learning (Mayer 2021, 2022). and increased manipulation abilities at a subsequent task, show-
ing potential advantages of using more natural-­like controllers,
such as Knuckles. Despite evidence supporting the pretraining
5.3   |   Practical Contributions video's effectiveness in reducing errors during the in-­person
serial dilution task, future studies might benefit from aligning
A practical question is whether and how IVR should be used fine motor actions more closely with natural actions. This ad-
for training of technical procedural skills. Based on the results justment could potentially yield stronger effects.
of this study, it is possible to recommend using IVR to enhance
learning outcomes, particularly for lessons that involve learning Another limitation pertained to the awareness of researchers
a technical scientific procedure. Our results are consistent with regarding participants' assigned conditions. Researchers were
the findings of Petersen, Klingenberg, and Makransky (2022) tasked with capturing real-­time data as participants completed
that found learning a procedural lesson in IVR to predict real-­ the task. While instructed to observe and record participants'
life performance on a transfer test. Participants in the pretrain- actions during the in-­person serial dilution task, the possibility
ing condition made fewer pipetting errors on an in-­person task exists that some data might have been missed, or certain antici-
and scored higher on the knowledge test compared to those as- pated behaviours might have influenced observations based on
signed to the no pretaining condition, suggesting the potential participants' assigned conditions. All researchers underwent
benefits of including a pretraining video before immersing stu- extensive training with the lead researcher to standardise data
dents in a training simulation. This study extends the pretrain- recording and action and error identification. However, human
ing principle beyond learning conceptual knowledge to training error remains a possibility. Particularly, our study relied on a
of procedural knowledge. When the goal is to help students single researcher for data collection during the in-­person serial
transfer what they have learned in a training simulation to per- dilution task. Future studies could mitigate potential errors by
formance in the real world, we recommend including carefully employing two researchers for data collection and conducting

14 of 16 Journal of Computer Assisted Learning, 2025

inter-­rater reliability checks to ensure consistency. Alternatively, Bandura, A. 1977. “Self-­ Efficacy: Toward a Unifying Theory of
including a separate researcher to assess the in-­person serial di- Behavioral Change.” Psychological Review 84, no. 2: 191–215. https://​
lution task could help minimise bias. doi.​org/​10.​1037/​0 033-­​295X.​8 4.2.​191.
Cha, M., S. Han, J. Lee, and B. Choi. 2012. “A Virtual Reality Based Fire
Training Simulator Integrated With Fire Dynamics Data.” Fire Safety
6   |   Conclusion Journal 50: 12–24. https://​doi.​org/​10.​1016/j.​f ires​af.​2 012.​01.​0 04.
Chen, J. C., Y. W. Huang, K. Y. Lin, et al. 2019. “Developing a Hands-­On
The main contribution of this study to the knowledge base Activity Using Virtual Reality to Help Students Learn by Doing.”
is to demonstrate the effectiveness of the pretraining prin- Journal of Computer Assisted Learning 36, no. 1: 46–60. https://​doi.​org/​
ciple in enhancing the transfer of procedural knowledge 10.​1111/​jcal.​12389​.
and skills taught in an immersive virtual environment to a Chernikova, O., N. Heitzmann, M. Stadler, D. Holzberger, T. Seidel,
real-­life setting. Our findings demonstrated the pretraining and F. Fischer. 2020. “Simulation-­Based Learning in Higher Education:
group demonstrated increased retention on the knowledge A Meta-­A nalysis.” Review of Educational Research 90, no. 4: 499–541.
https://​doi.​org/​10.​3102/​0 0346​5 4320​933544.
test and demonstrated fewer pipetting errors during the real-­
life transfer test compared to the group without pretraining. Dobrowolski, P., M. Skorko, G. Pochwatko, M. Myśliwiec, and A.
Additionally, the pretraining group experienced lower levels of Grabowski. 2021. “Immersive Virtual Reality and Complex Skill
Learning: Transfer Effects After Training in Younger and Older
extraneous cognitive load. While we did not find extraneous
Adults.” Frontiers in Virtual Reality 1: 1–8. https://​doi.​org/​10.​3389/​f rvir.​
cognitive load and presence to a significant mediate the rela- 2020.​6 04008.
tionship between instructional conditions and learning out-
Fu, Y., and Q. Li. 2020. “Immersive Virtual Reality for Fire Safety
comes, nor did we find an increase in post-­self-­efficacy scores Behavioural Skills Training via Gesture-­ Based Technology.” IOP
among the pretraining group, the implications are promising. Conference Series 825: 012015. https://​doi.​org/​10.​1088/​1757-­​899x/​825/1/​
Virtual reality environments, when aligned with effective in- 012015.
structional design principles, show potential in enhancing re- Gavish, N., T. Gutiérrez, S. Webel, et al. 2013. “Evaluating Virtual
tention and facilitating transfer learning. Reality and Augmented Reality Training for Industrial Maintenance
and Assembly Tasks.” Interactive Learning Environments 23, no. 6: 778–
798. https://​doi.​org/​10.​1080/​10494​820.​2 013.​815221.

Author Contributions Jensen, L. X., and F. Konradsen. 2017. “A Review of the Use of Virtual
Reality Head-­Mounted Displays in Education and Training.” Education
Cynthia Y. Delgado: conceptualization, data curation, formal anal- and Information Technologies 23, no. 4: 1515–1529. https://​doi.​org/​10.​
ysis, writing – original draft, methodology, investigation, supervision, 1007/​s1063​9 -­​017-­​9676-­​0.
writing – review and editing, software, project administration. Richard
E. Mayer: conceptualization, data curation, writing – original draft, Johnson, C. I., S. K. Bailey, B. L. Schroeder, and M. D. Marraffino.
methodology, investigation, supervision, writing – review and editing, 2022. “Procedural Learning in Virtual Reality: The Role of Immersion,
funding acquisition, resources, project administration. Interactivity, and Spatial Ability.” Technology, Mind, and Behavior 3, no.
4: 146. https://​doi.​org/​10.​1037/​tmb00​0 0087​.
Lawson, A. P., A. M. Martella, K. LaBonte, et al. 2024. “Confounded
Acknowledgements
or Controlled? A Systematic Review of Media Comparison Studies in
We appreciate that Guido Makransky and Gustav Peterson graciously Immersive Virtual Reality Research in STEM Education.” Educational
provided the pipetting simulation. This project was supported by Grant Psychology Review 36: 69.
N000142112047 from the Office of Naval Research.
Lawson, A., and R. E. Mayer. 2024a. “Individual Differences in
Executive Function Affect Learning With Immersive Virtual Reality.”
Disclosure Journal of Computer Assisted Learning 40, no. 3: 1068–1082.
The authors have nothing to report. Lawson, A., and R. E. Mayer. 2024b. “Effect of Pre-­Training and Role of
Working Memory Characteristics in Learning With Immersive Virtual
Reality.” International Journal of Human Computer Interaction: 1–18.
Conflicts of Interest
Makransky, G., N. K. Andreasen, S. Baceviciute, and R. E. Mayer. 2021.
The authors declare no conflicts of interest. “Immersive Virtual Reality Increases Liking but Not Learning With
a Science Simulation and Generative Learning Strategies Promote
Data Availability Statement Learning in Immersive Virtual Reality.” Journal of Educational
Psychology 113, no. 4: 719–735. https://​doi.​org/​10.​1037/​edu00​0 0473​.
The data that support the findings of this study are available from the
corresponding author upon reasonable request. Makransky, G., T. Terkildsen, and R. E. Mayer. 2019. “Adding Immersive
Virtual Reality to a Science Lab Simulation Causes More Presence but
Less Learning.” Learning and Instruction 60: 225–236. https://​doi.​org/​
References 10.​1016/j.​learn​instr​uc.​2 017.​12.​0 07.
Mayer, R. E. 2021. Multimedia Learning. 3rd ed. New York: Cambridge
Akgün, M., and B. Atici. 2022. “The Effects of Immersive Virtual Reality
University Press.
Environments on students' Academic Achievement: A Meta-­A nalytical
and Meta-­T hematic Study.” Participatory Educational Research 9, no. 3: Mayer, R. E. 2022. “Cognitive Theory of Multimedia Learning.” In The
111–131. https://​doi.​org/​10.​17275/​​per.​22.​57.9.​3. Cambridge Handbook of Multimedia Learning, edited by R. E. Mayer
and L. Fiorella, 3rd ed., 57–72. Washington, DC: Cambridge University
Andersen, M. S., and G. Makransky. 2020. “The Validation And Further
Press.
Development of a Multidimensional Cognitive Load Scale for Virtual
Environments.” Journal of Computer Assisted Learning 37, no. 1: 183– Mayer, R. E., A. Mathias, and K. Wetzell. 2002. “Fostering
196. https://​doi.​org/​10.​1111/​jcal.​12478​. Understanding of Multimedia Messages Through Pre-­
Training:

15 of 16
Evidence for a Two-­ Stage Theory of Mental Model Construction.” Rogers, C., H. El-­Mounaryi, T. M. Wasfy, and J. Satterwhite. 2017.
Journal of Experimental Psychology: Applied 8: 147–154. “Assessment of STEM e-­L earning in an Immersive Virtual Reality (VR)
Environment.” Computers in Education Journal 8: 4.
Mayer, R. E., and L. Fiorella. 2022. “Principles for Managing Essential
Processing in Multimedia Learning: Segmenting, Pre-­Training, and Schunk, D. H., and M. K. DiBenedetto. 2016. “Self-­Efficacy Theory
Modality Principles.” In The Cambridge Handbook of Multimedia in Education.” In Handbook of Motivation at School, edited by K. R.
Learning, edited by R. E. Mayer and L. Fiorella, 3rd ed., 243–260. New Wentzel and D. B. Miele, 2nd ed., 34–54. New York: Routledge.
York: Cambridge University Press.
Selzer, M. N., N. F. Gazcón, and M. L. Larrea. 2019. “Effects of Virtual
Mayer, R. E., G. Makransky, and J. Parong. 2022. “The Promise and Presence and Learning Outcome Using Low-­ End Virtual Reality
Pitfalls of Learning in Immersive Virtual Reality.” International Journal Systems.” Displays 59: 9–15. https://​doi.​org/​10.​1016/j.​displa.​2 019.​
of Human Computer Interaction 39, no. 11: 2229–2238. https://​doi.​org/​ 04.​0 02.
10.​1080/​10447​318.​2 022.​2108563.
Smutny, P. 2023. “Learning With Virtual Reality: A Market Analysis
Meyer, O. A., M. K. Omdahl, and G. Makransky. 2019. “Investigating of Educational and Training Applications.” Interactive Learning
the Effect of Pre-­Training When Learning Through Immersive Virtual Environments 31, no. 10: 6133–6146.
Reality and Video: A Media and Methods Experiment.” Computers &
Sweller, J., P. Ayres, and S. Kalyuga. 2011. Cognitive Load Theory. New
Education 140: 103603. https://​doi.​org/​10.​1016/j.​compe​du.​2 019.​103603.
York: Springer.
Moreno, R., and R. E. Mayer. 2002. “Learning Science in Virtual Reality
Tai, K., J. Hong, C. Tsai, C. Lin, and Y. H. Hung. 2022. “Virtual Reality
Multimedia Environments: Role of Methods and Media.” Journal of
for Car-­Detailing Skill Development: Learning Outcomes of Procedural
Educational Psychology 94, no. 3: 598–610. https://​doi.​org/​10.​1037/​
Accuracy and Performance Quality Predicted by VR Self-­Efficacy, VR
0022- ­​0 663.​94.3.​598.
Using Anxiety, VR Learning Interest and Flow Experience.” Computers
Moreno, R., and R. E. Mayer. 2004. “Personalized Messages That & Education 182: 104458. https://​doi.​org/​10.​1016/j.​compe​du.​2 022.​
Promote Science Learning in Virtual Environments.” Journal of 104458.
Educational Psychology 96, no. 1: 165–173. https://​doi.​org/​10.​1037/​
Villena-­Taranilla, R., S. Tirado-­Olivares, R. Cózar-­Gutiérrez, and J.
0022-­​0 663.​96.1.​165.
A. González-­ Calero. 2022. “Effects of Virtual Reality on Learning
Noblecourt, S., G. Bourgoin, V. Havard, and D. Baudry. 2021. “Evaluating Outcomes in K-­6 Education: A Meta-­A nalysis.” Educational Research
the Influence of Interaction Technology on Procedural Learning Review 35: 100434.
Using Virtual Reality.” In Proceedings of the 27th ACM Symposium on
Witmer, B. G., C. J. Jerome, and M. J. Singer. 2005. “The Factor
Virtual Reality Software and Technology. New York, NY: Association for
Structure of the Presence Questionnaire.” Presence Teleoperators and
Computing Machinery. https://​doi.​org/​10.​1145/​3 4898​49.​3 489918.
Virtual Environments 14, no. 3: 298–312. https://​doi.​org/​10.​1162/​10547​
Paas, F., and J. Sweller. 2022. “Implications of Cognitive Load Theory 46053​23384654.
for Multimedia Learning.” In The Cambridge Handbook of Multimedia
Wu, B., X. Yu, and X. Gu. 2020. “Effectiveness of Immersive Virtual
Learning, edited by R. E. Mayer and L. Fiorella, 3rd ed., 73–81. New
Reality Using Head-­Mounted Displays on Learning Performance: A
York: Cambridge University Press.
Meta-­A nalysis.” British Journal of Educational Technology 51, no. 6:
Parong, J. A. N. 2019. “Cognitive and Affective Mechanisms of 1991–2005. https://​doi.​org/​10.​1111/​bjet.​13023​.
Immersive Virtual Reality Learning Environments (Dissertation).”
Zhang, K., J. Suo, J. Chen, X. Liu, and L. Gao. 2017. “Design and
In Cognitive and Affective Mechanisms of Immersive Virtual Reality
Implementation of Fire Safety Education System on Campus Based on
Learning Environments. Santa Barbara, CA: University of California.
Virtual Reality Technology.” In 2017 Federated Conference on Computer
Parong, J., and R. E. Mayer. 2018. “Learning Science in Immersive Science and Information Systems (FedCSIS). Prague, Czech Republic:
Virtual Reality.” Journal of Educational Psychology 110, no. 6: 785–797. IEEE. https://​doi.​org/​10.​15439/​​2 017f376.
https://​doi.​org/​10.​1037/​edu00​0 0241​.
Parong, J., and R. E. Mayer. 2020. “Cognitive and Affective Processes for
Learning Science in Immersive Virtual Reality.” Journal of Computer
Assisted Learning 37, no. 1: 226–241. https://​doi.​org/​10.​1111/​jcal.​12482​.
Parong, J., and R. E. Mayer. 2021. “Learning About History in Immersive
Virtual Reality: Does Immersion Facilitate Learning?” Educational
Technology Research and Development 69, no. 3: 1433–1451. https://​doi.​
org/​10.​1007/​s1142​3 -­​021-­​0 9999​-­​y.
Petersen, G. B., S. L. Klingenberg, and G. Makransky. 2022. “Pipetting
in Virtual Reality Can Predict Real-­ Life Pipetting Performance.”
Technology, Mind, and Behavior 3, no. 3: 1–18. https://​doi.​org/​10.​1037/​
tmb00​0 0076​.
Petersen, G. B., S. L. Klingenberg, R. E. Mayer, and G. Makransky. 2020.
“The Virtual Field Trip: Investigating How to Optimize Immersive
Virtual Learning in Climate Change Education.” British Journal of
Educational Technology 51, no. 6: 2099–2115. https://​doi.​org/​10.​1111/​
bjet.​12991​.
Radianti, J., T. A. Majchrzak, J. Fromm, and I. Wohlgenannt. 2020.
“A Systematic Review of Immersive Virtual Reality Applications for
Higher Education: Design Elements, Lessons Learned, and Research
Agenda.” Computers & Education 147: 103778. https://​doi.​org/​10.​1016/j.​
compe​du.​2 019.​103778.

16 of 16 Journal of Computer Assisted Learning, 2025