Professional Documents
Culture Documents
The Role of Spatial Ability in Mixed Reality Learning With The Hololens
The Role of Spatial Ability in Mixed Reality Learning With The Hololens
DOI: 10.1002/ase.2146
RESEARCH REPORT
1
Department of Psychology, University
of British Columbia, Vancouver, British Abstract
Columbia, Canada
The use of mixed reality in science education has been increasing and as such it has
2
Department of Cellular and Physiological
Sciences, University of British Columbia,
become more important to understand how information is learned in these virtual
Vancouver, British Columbia, Canada environments. Spatial ability is important in many learning contexts, but especially
Correspondence
in neuroanatomy education where learning the locations and spatial relationships
Dr. Simon Ho, Department of Psychology, between brain regions is paramount. It is currently unclear what role spatial ability
University of British Columbia, 2136 West
Mall, Vancouver, BC V6T 1Z4, Canada.
plays in mixed reality learning environments, and whether it is different compared to
Email: simonho213@gmail.com traditional physical environments. To test this, a learning experiment was conducted
Funding information
where students learned neuroanatomy using both mixed reality and a physical plastic
Mitacs, Grant/Award Number: F18-02840 model of a brain (N = 27). Spatial ability was assessed and analyzed to determine its
effect on performance across the two learning modalities. The results showed that
spatial ability facilitated learning in mixed reality (β = 0.21, P = 0.003), but not when
using a plastic model (β = 0.08, P = 0.318). A non-significant difference was observed
between the modalities in terms of knowledge test performance (d = 0.39, P = 0.052);
however, mixed reality was more engaging (d = 0.59, P = 0.005) and learners were
more confident in the information they learned compared to using a physical model
(d = 0.56, P = 0.007). Overall, these findings suggest that spatial ability is more rel-
evant in virtual learning environments, where the ability to manipulate and interact
with an object is diminished or abstracted through a virtual user interface.
KEYWORDS
cognition, learning, mental rotation, mixed reality, neuroanatomy education, neuroscience,
spatial ability
1074 | © 2021 American Association for Anatomy wileyonlinelibrary.com/journal/ase Anat Sci Educ. 2022;15:1074–1085.
|
19359780, 2022, 6, Downloaded from https://anatomypubs.onlinelibrary.wiley.com/doi/10.1002/ase.2146 by Air University, Wiley Online Library on [01/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
HO et al. 1075
with which it allows learners to manipulate spatial data (Hanna et al., virtual models only improve learning in individuals with high spa-
2018). The learning benefit of mixed reality, however, is not entirely tial ability. This is known as the ability-as-enhancer hypothesis and
clear as some studies have shown that it yields the same student suggests that high spatial ability allows the learner to devote more
performance as traditional teaching methods (Moro et al., 2017; cognitive resources to building a mental representation of the object
Gnanasegaram et al., 2020; Henssen et al., 2020). It could even be (Huk, 2006; Lee et al., 2009). However, studies have also shown the
detrimental to performance: for example, in some studies, learning opposite effect and support the ability-as-compensator hypothesis,
pelvic anatomy in mixed reality led to poorer knowledge retention where the presence of a virtual model can compensate for lower
compared to the use of a physical model (Khot et al., 2013; Wainman spatial ability of the learner (Lee & Wong, 2014). A confound in these
et al., 2018, 2020). This has been attributed to the lack of stere- studies is that the virtual model is not true virtual reality, and instead
opsis in many mixed reality platforms, which can lead to impaired displayed on a computer screen, which removes the possibility of
depth perception, whereas stereo vision is retained when using true stereopsis. A recent study showed that stereopsis can interact with
physical models (Wainman et al., 2018, 2020). Overall, mixed reality spatial ability, improving performance of individuals with low base-
learners expect the technology to be more helpful than it actually line spatial ability (Bogomolova et al., 2020). Additionally, immersive
is in practice, and current implementations are not fully effective virtual or mixed reality (i.e., using a head-mounted display) allows for
replacements for traditional techniques (Zafar & Zachar, 2020). This increased interactivity and direct manipulation of the model, which
raises the question of why mixed reality leads to inconsistent effects has also been shown to interact with spatial ability (Sjölie et al., 2010;
on learning, and whether individual characteristics of the learners Jang et al., 2017). These features are potentially more important in
themselves might be affecting the relationship (Lee et al., 2009; the context of neuroanatomical learning due to the higher level of
Cheng & Tsai, 2013). One candidate for a learner characteristic that visual occlusion compared to, for example, pelvic anatomy. In fact, it
might moderate learning benefits is spatial ability, as it has been could be argued that the need for spatial ability is minimized when
shown to correlate with learning performance across a number of learning pelvic anatomy due to how capacious the region is (Khot
domains (Khot et al., 2013; Lee & Wong, 2014). Of particular interest et al., 2013; Wainman et al., 2018, 2020). Currently, it is unclear how
in the present study is how spatial ability might affect learning in spatial ability affects learning in an immersive environment when
mixed reality environments, specifically in the context of neuroana- high levels of visual occlusion are involved.
tomical learning. The present study addresses two main questions. First, does
Spatial knowledge can be represented using different frames of mixed reality confer any additional benefit over traditional physical
reference, for example, egocentric and allocentric (Easton & Sholl, model learning? Second, to what extent does spatial ability affect
1995; Mou & McNamara, 2002). Object locations in an egocentric performance in each of the two learning modalities? To test the
representation are stored relative to the observer (e.g., “where is the validity of the ability-as-enhancer and ability-as-compensator hy-
coffee shop from my current location?”), whereas allocentric repre- potheses, a learning experiment was conducted using a mixed real-
sentations are coded with respect to other objects in the environ- ity head-mounted display (displaying a virtual brain) and a physical
ment (e.g., “where is the coffee shop in relation to the hospital?”). plastic brain model. It was hypothesized that spatial ability would
Allocentric representations are particularly important in the case have a differential effect in each of these modalities. As mental
of neuroanatomical learning as learners are required to encode the rotation involves similar cognitive processes to physical object ro-
relative spatial locations and orientations of different brain regions. tation (Gardony et al., 2014), it was predicted that spatial ability
As it is concerned with object-to-object relations and transforma- would be unrelated to learning performance when physical rotation
tions, the allocentric system is highly related to mental rotation abil- of the model is possible (i.e., learning from a physical model). This
ity, which is the ability to mentally rotate an internal representation is supported by studies that show no relationship between spatial
of a two-dimensional (2D) or three-dimensional (3D) object (Wraga ability and academic course performance, where the course was
et al., 1999; Zacks & Michelon, 2005; Voyer et al., 2017). Findings taught using traditional methods like physical models (Sweeney
in the educational literature converge on the idea that mental rota- et al., 2014). However, when physical rotation is unavailable and ab-
tion ability is highly beneficial for performance on spatial learning stracted through a user interface, it was predicted that higher levels
tasks. For example, mental rotation ability has been found to cor- of spatial ability would be required for effective learning.
relate positively with spatial learning performance across different
learning modalities (e.g., mixed reality, virtual reality, and computer
screen), and can be a stronger predictor of overall performance than M ATE R I A L S A N D M E TH O D S
the learning modality itself (Khot et al., 2013; Wainman et al., 2018,
2020). These findings have also been observed in anatomical learn- Participants
ing (Garg et al., 2001; Guillot et al., 2007; Lee & Wong, 2014), and
participation in mental rotation training can lead to improvement on A two-tail power analysis was conducted to determine minimum
tests of anatomy (Hoyek et al., 2009). sample size for the study. In terms of effect size, previous studies
The literature is more divided, however, with respect to the role have shown that the relationship between mixed reality learning
of spatial ability in virtual environments. Some studies show that 3D and test performance ranges from 0.38 to 1.99 (Enyedy et al., 2012;
|
19359780, 2022, 6, Downloaded from https://anatomypubs.onlinelibrary.wiley.com/doi/10.1002/ase.2146 by Air University, Wiley Online Library on [01/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1076 HO et al.
Albrecht et al., 2013; Merchant et al., 2014); however, pilot stud- of these user interactions, as well as the position of the HoloLens
ies in the laboratory showed that the effect size could be as low as itself, were recorded to a log file on the device.
0.08 when learning neuroanatomy. Across these studies, the aver- HoloBrain consisted of two primary learning modes: education
age effect size was 0.64, which was used for the power analysis. room and magnetic resonance imaging (MRI) room. The purpose of
The analysis indicated that a minimum of 22 participants would be the education room was to teach the locations of different brain re-
needed to achieve 80% power when using a within-subjects design. gions in 3D space (Figure 1A). For the limbic system, HoloBrain con-
A total of 31 participants were recruited from the University of tained only the thalamus, amygdala, hippocampus, fornix, mammillary
British Columbia and received $30.00 CAD for their time. In terms of bodies, and the mammillothalamic tract. For the basal ganglia regions,
inclusion criteria and eligibility for the study, participants had to be HoloBrain contained only the caudate nucleus, putamen, globus
current undergraduate or graduate students with no prior participa- pallidus, thalamus, substantia nigra, and the subthalamic nuclei. The
tion in any neuroanatomy courses. Four participants were excluded names of the six brain regions were shown using pop-up labels that
due to computer problems that prevented data collection. The would appear whenever the user looked at a region. The education
final sample consisted of 27 participants (mean age = 20.93 years, room offered an “isolate” mode, where the user could choose the spe-
SD ±3.46, 10 males, 16 females, 1 other). cific brain regions they wanted to view. Upon selecting a region, an
audio recording was played that spoke the region name. An “expand”
mode increased the spacing between selected brain regions, which al-
HoloLens and HoloBrain lowed for easier visualization of the shape of the individual structures.
The second learning mode was MRI room, which was designed
The experiment used a Microsoft HoloLens (Microsoft Corp., to teach locations in 2D space (i.e., an MRI scan). Upon entering the
Redmond, WA) as the presentation device as it offers full stereopsis. MRI room, 2D MRI scans were displayed in a side panel (in a two-
For the neuroanatomy content, the research team developed an ap- by-t wo grid) next to the brain, and each scan had colored outlines
plication called HoloBrain (Holman et al., 2018), which is an educa- to indicate the different brain regions of interest (Figure 1B). The
tional tool designed to teach neuroanatomy by projecting a virtual side panel was the primary method by which 2D information was
interactive brain in front of the user. The brain could be locked in learned. Additionally, each scan was also shown intersecting the 3D
place and allowed for full exploration across all three spatial axes. brain model to show the location at which that scan was taken. The
Additionally, the user could also reposition, resize, and rotate the combination of side panel images and intersecting images was de-
virtual brain using a series of menu options and voice commands. All signed to provide a link between 2D and 3D spatial understanding.
F I G U R E 1 Screen captures of the HoloBrain application. (A) HoloBrain main menu system; (B) HoloBrain MRI room
|
19359780, 2022, 6, Downloaded from https://anatomypubs.onlinelibrary.wiley.com/doi/10.1002/ase.2146 by Air University, Wiley Online Library on [01/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
HO et al. 1077
Plastic model fundamentally allocentric reference frame as the occipital lobe, for
example, will be posterior to the parietal lobe regardless of where the
The overall design goal of the plastic model modality was to ensure observer is standing. Although some egocentric ability is likely to be
that participants had access to the same information, and that it involved given the use of mixed reality (for example, HoloBrain users
was presented in a similar fashion, to HoloBrain. Some concessions need awareness of where their hand is in relation to the virtual brain
had to be made, however, as the two learning platforms are fun- in order to interact with it), this is beyond the scope of the current
damentally different (e.g., there are no buttons and menu systems research question.
in reality, physical disassembly of the brain is not possible in mixed To assess allocentric spatial ability, each participant completed a
reality, etc.). Participants were asked to learn region locations using computerized version of the mental rotations task (MRT; Shepard &
a physical plastic model of the brain (BS 25; SOMSO® MODELLE Metzler, 1971; Vandenberg & Kuse, 1978; Peters et al., 1995). During
GmbH, Coburg, Germany), which could be assembled and disassem- each trial of the task, a target image, consisting of a geometric shape,
bled by the user (Figure 2). The model was color coded, and partici- was presented along with four options. Two of the (correct) options
pants were provided with a separate sheet that showed the names were identical to the target but rotated around one of the axes. The
and locations of the specific regions they were asked to learn. To remaining two (incorrect) options were similar looking shapes but
minimize differences between learning modalities, the plastic model slightly modified such that they could never be rotated to match the
only contained labels for the six regions presented in the HoloBrain target. Participants were asked to select the two options that could
condition. Thus, the learning requirements were the same: to learn be correctly rotated to match the target. A single point was given for
the spatial locations of the same six basal ganglia regions or six limbic the trial if both shapes could be identified, and no point was awarded
system regions that were described in the previous section. if only one correct answer was chosen. There were a total of 24 trials,
In addition to the 3D model, participants were also provided with split into two halves of 12, and participants had a total of 6 minutes
a sheet of MRI images that was used to teach 2D spatial information. for the task (3 minutes per half). Each participant received a score out
To ensure plastic model learning was as similar to HoloBrain as pos- of 24 for this task. The version of the task used was identical to the
sible, the paper-based scans were the same as the ones used in the MRT (Peters et al., 1995) in terms of stimuli, task timings, and scoring
HoloBrain application and were also organized in an identical two- method, thus preserving the psychometric properties of the assess-
by-t wo grid. HoloBrain also contained scans that intersected the 3D ment. The full set of task stimuli can be found in Peters et al. (1995).
brain to signify the location at which the scan was taken. To create
an analogous demonstration on paper, a small brain diagram was dis-
played, next to each MRI scan, that contained an intersecting line to Knowledge test and survey
show the location from which that scan was taken.
A 40-item multiple choice knowledge test was designed to assess
retention of the learned material. Furthermore, the test showed
Spatial ability new/unseen images and contained novel application questions as
participants were expected to generalize their learning to unfamiliar
The present study focused only on allocentric spatial ability as the contexts; this was made clear to them at the beginning of the study.
research question was about learning the spatial relationships of ob- In terms of Bloom's taxonomy, the questions were designed to as-
jects to each other, rather than in relation to the observer. This is a sess the first three levels of the cognitive domain: (1) remember the
names and locations of brain regions, (2) understand their spatial re-
lationships with regards to the ventricular system and other learned
regions, and (3) apply that information to new images and unfamiliar
contexts (Anderson & Krathwohl, 2001). All knowledge test scores
were converted to percentages for ease of comparison.
The test itself consisted of 20 questions about the limbic system
and 20 questions about the basal ganglia. Half of the questions asked
about general 3D spatial understanding (e.g., “the posterior aspect
of the thalamus is <blank> to the hippocampus”), and the other half
tested the ability to translate those 3D relationships to 2D slices
through the brain (e.g., “if the brain was sliced at the line indicated
below, which one of these structures would you expect to see?”). For
each question, participants were also asked to state how confident
they were in their answer on an ordinal scale from 1 to 10, with 10
being the highest level of confidence. The full list of knowledge test
questions, along with a breakdown of basal ganglia/limbic system and
F I G U R E 2 Disassembled plastic model of the inner brain regions 2D/3D understanding, can be found in the Supporting Information.
|
19359780, 2022, 6, Downloaded from https://anatomypubs.onlinelibrary.wiley.com/doi/10.1002/ase.2146 by Air University, Wiley Online Library on [01/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1078 HO et al.
After the knowledge test, participants completed a quick survey central nervous system (Rothman, 2014) to provide everybody with
containing a couple of demographic questions designed to be used a basic familiarity with the content they will be asked to learn. The
as covariates in the analyses: (1) what is your level of proficiency with remainder of day one was spent teaching participants how to work
English? (1 to 6, with 6 being native level), and (2) have you used a with the plastic model and how to use the various functions available
virtual reality device before? (Yes/No). Participants were also asked in the HoloBrain application. The practice and training sessions were
to provide their general opinions on the HoloBrain learning method essential; otherwise, the novelty of both learning methods quickly
(“what did you like/dislike about the HoloBrain learning method?”). became daunting and overwhelming.
Finally, they were asked to state their level of agreement with spe- Day two was the main learning day. Participants were explicitly
cific attitude questions: “I found the <method> engaging” and “the informed that none of the images from the learning day would be
<method> helped me learn the concepts” (method was replaced by shown as-is on the knowledge test, but rather they were expected
both “HoloLens” and “plastic model” to create a total of four differ- to generalize their learning to unfamiliar contexts. For each learn-
ent question combinations). These questions were measured using a ing method, participants were provided with a list of six different
6-point Likert scale (1 = Strongly disagree, 2 = Disagree, 3 = Somewhat brain regions to be learned and 20 minutes to complete the task.
disagree, 4 = Somewhat agree, 5 = Agree, 6 = Strongly agree). The learning was entirely self-guided, and participants were re-
minded to focus on the four main learning objectives. Both learning
method and brain regions were counterbalanced across partici-
Experimental procedure pants to reduce the possibility of order and practice effects. For
example, one participant started by using the plastic model to learn
The experiment was spread across three consecutive days, with each the basal ganglia, followed by HoloBrain for the limbic system. The
session lasting a maximum of one hour (Figure 3). On day one, the presentation order was then switched for the next participant (i.e.,
experimenter explained the study and acquired written informed plastic model for limbic system and HoloBrain for the basal gan-
consent from the participant. The four main learning objectives were glia). Finally, participants were instructed not to review the material
then explained: (1) learn the correct anatomical orientation terms at home to ensure learning was limited to the session itself.
to describe locations (e.g., anterior, superior, and lateral), (2) learn Day three was spent on the computer task, knowledge test, and
the locations of brain regions in three dimensions, (3) learn the lo- survey. Participants started with the MRT followed by the 40-item
cations of brain regions in two dimensions from MRI scans, and (4) knowledge test. The questions on the knowledge test assessed un-
learn the locations of these structures relative to the ventricular sys- derstanding of both the limbic system and basal ganglia, and presen-
tem. These objectives, along with a diagram explaining anatomical tation order of all questions was randomized for each participant.
orientation, were displayed on a screen during days one and two. Knowledge test scores were calculated as percentages for each
Participants then watched a short introductory video about the learning method separately and used as the primary outcomes in the
19359780, 2022, 6, Downloaded from https://anatomypubs.onlinelibrary.wiley.com/doi/10.1002/ase.2146 by Air University, Wiley Online Library on [01/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
HO et al. 1079
analyses. At the end of the session, participants were debriefed on MRT = 𝛽 0 + 𝛽 1 Mixed reality + 𝛽 2 Plastic model
the purpose of the study and paid for their time. + 𝛽 3 English + 𝛽 4 Experience + 𝛽 5 Travel.
19359780, 2022, 6, Downloaded from https://anatomypubs.onlinelibrary.wiley.com/doi/10.1002/ase.2146 by Air University, Wiley Online Library on [01/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1080 HO et al.
Limbic system 47.59 (±17.12) travel distance were not significantly related to spatial ability. The
regression estimates for each predictor can be seen in the Model 2
2Da 50.19 (±16.49)
section of Table 2. Overall, these results suggest that spatial ability
3Db 46.30 (±12.37)
played a bigger role in a mixed reality environment than a real physi-
Mixed reality 51.85 (±15.82)
cal environment.
Plastic model 44.63 (±13.65)
Confidence ratings
Mixed reality 5.96 (±1.25) Qualitative responses
Plastic model 5.42 (±1.33)
Attitudes For the question regarding what participants liked about the
Mixed reality helpful 5.48 (±0.70) HoloBrain learning method, the thematic analysis revealed three
Mixed reality engaging 5.56 (±0.80) main themes: (1) facilitation of learning objectives, (2) effectiveness
Plastic model helpful 4.85 (±0.99) of the technology, and (3) enjoyable user experience. The most com-
Plastic model engaging 4.74 (±1.10) mon theme was facilitation of learning objectives, which made up
Spatial ability
50% of the coded responses. This theme was primarily composed of
response codes related to ease of manipulation of individual brain
Mental rotations task 11.78 (±5.74)
regions, which promoted easier memorization of the areas:
N = 27 for all measures.
a
Two-dimensional test items.
b
I enjoyed being able to make certain areas of the brain
Three-dimensional test items.
disappear or reappear, in order to form an accurate
and plastic model test performance, t(26) = 2.03, P = 0.052, d = 0.39. visual picture of where these areas are and how they
All of the participants performed at, or above, chance levels as can fit into the grander scheme of brain anatomy.
be seen in Figure 4 (top left). Four participants in the mixed reality
condition, and five in the plastic model condition, achieved a chance- The next most common theme was effectiveness of the technol-
level score on their test (25%), which is akin to guessing. However, ogy (making up 30% of coded responses), where participants com-
this is not too surprising given participants only had 20 minutes to plimented the novelty and ease of use of the HoloBrain application.
learn the material, and some low scores would be expected. Despite Finally, the theme of enjoyable user experience made up 20% of codes,
observing a non-significant difference in test performance, partici- where participants reported the HoloBrain as an interactive and en-
pants were more confident when answering mixed reality-learned gaging learning method. In general, these findings are in line with the
questions than plastic model questions, t(26) = 2.92, P = 0.007, quantitative results reported previously, where participants found the
d = 0.56 (Figure 4, top right). Descriptive statistics for mixed reality mixed reality method to be significantly more engaging and helpful
and plastic model conditions can be seen in Table 1. than the plastic model.
In terms of attitudes and perceptions, participants found the With regard to disliked aspects of the HoloBrain application,
HoloLens to be a more engaging and helpful method of learning. three main themes emerged in the thematic analysis: (1) technologi-
Participants found mixed reality to be significantly more engaging cal bugs, (2) not user-friendly, and (3) physical discomfort. The most
than the plastic model, t(26) = 3.05, P = 0.005, d = 0.59 (Figure 4, common theme was technological bugs, which made up 45% of the
bottom left), and significantly more helpful, t(26) = 2.51, P = 0.019, d response codes. Participants reported many visual glitches in the
= 0.48 (Figure 4, bottom right). Descriptive statistics for each of the software and inaccuracies in hand gesture input, which disrupted
above attitude ratings can be seen in the attitudes section of Table 1. their learning experience:
19359780, 2022, 6, Downloaded from https://anatomypubs.onlinelibrary.wiley.com/doi/10.1002/ase.2146 by Air University, Wiley Online Library on [01/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
HO et al. 1081
F I G U R E 4 Box plots and distributions of (A) knowledge test score; (B) confidence; (C) engagement, and (D) helpfulness ratings (N = 27).
Mixed reality learning showed higher mean values across all four measures; however, only confidence, engagement, and helpfulness were
statistically significant
19359780, 2022, 6, Downloaded from https://anatomypubs.onlinelibrary.wiley.com/doi/10.1002/ase.2146 by Air University, Wiley Online Library on [01/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1082 HO et al.
19359780, 2022, 6, Downloaded from https://anatomypubs.onlinelibrary.wiley.com/doi/10.1002/ase.2146 by Air University, Wiley Online Library on [01/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
HO et al. 1083
One possible reason for the disparity between modalities is that coordinate system, or vice versa. The recommended approach for this
when using a physical model, the ability to mentally rotate the ob- relies on the use of specialized 3D motion capture cameras, which is
ject is not required because the model can be rotated physically. beyond the scope and capabilities of the present study (Bajana et al.,
However, in mixed reality, object rotation can only happen through 2016). As it stands, the relative importance of egocentric and allocen-
interactions with the user interface, which potentially increases the tric ability in virtual learning environments remains an open question
need to mentally rotate the object. Therefore, it is not necessarily and is an interesting direction for future research.
surprising that in a virtual mixed reality environment, individuals tend Finally, there were a few differences between the HoloLens and
to learn more if they have higher levels of spatial ability. This result plastic model learning conditions that may have confounded the
provides partial support for the ability-as-enhancer hypothesis (Huk, results. As mentioned previously, some of these differences were
2006; Lee et al., 2009) as spatial ability appeared to facilitate per- unavoidable due to the fact that learning in mixed reality is funda-
formance; however, the distinction between the ability-as-enhancer mentally different from learning with a physical model. For example,
and ability-as-compensator hypotheses may be more nuanced as it information in mixed reality is typically accessed through a menu
would seem to depend on whether physical object rotation is possi- system and requires navigation between different screens, which
ble. Overall, although the main finding of the present study is novel, limits how much information the user is exposed to at any one time.
future research should be conducted to confirm the result and pro- Another difference was the color/saturation/contrast of the brain
vide further insight into why such a modality difference might exist. regions. The HoloBrain application used higher contrast colors for
technology-related reasons. Virtual objects presented in mixed re-
ality are semi-transparent due to the display technology that is used,
Limitations and future directions which causes the objects to have a dynamic level of contrast with
the background. The unfortunate reality of working with mixed re-
Although the results of the present study provide novel insight into ality is that the perceptual qualities of virtual objects are not con-
how spatial ability affects both virtual and physical learning, there are stant and cannot be strictly controlled due to a constantly changing
some important limitations to note with regard to the findings. First, background behind the object (Lee et al., 2019). These differences
the reliability of the knowledge test was slightly lower than what is may have made the mixed reality condition easier and potentially
typically reported in the literature. The test reliability in the study was confounded the knowledge test finding. It is interesting to note,
within acceptable standards for a measurement instrument in science however, that mixed reality knowledge test performance was not
education; however, it is more common to see larger reliability values significantly better than the plastic model, despite potentially hav-
in the range of 0.7 to 0.8 (Taber, 2018). The primary driver behind ing these advantages. It is unlikely that this confound extended to
this disparity is likely the multidimensional nature of the knowledge the correlational spatial ability findings, as those were not direct
test. The test questions were designed to assess both 2D and 3D un- comparisons of knowledge test performance, however, that remains
derstanding; in multidimensional cases like this, Cronbach's alpha will to be tested. To overcome these issues, it would be prudent to rep-
underestimate the true reliability value as the method makes some licate the main findings using identical models in both learning mo-
assumptions about the item pool being unidimensional (Simms, 2008). dalities, or by overlaying the virtual brain regions over the physical
The limited sample size in the present study makes assessment of model to further equalize the conditions (McJunkin et al., 2018).
multidimensionality difficult and moving forward it is recommended
to replicate the study using more participants so multidimensional re-
liability can be more accurately assessed. The larger data set could CO N C LU S I O N S
also then be used to refine the knowledge test, resulting in an im-
proved measurement instrument that can be used in future studies. Overall, the present study showed that although there was no test
Second, the present study focused on allocentric spatial ability; performance difference between the two modalities, mixed real-
however, this is only part of the picture when working in a virtual en- ity was seen as a more helpful and engaging way of learning neu-
vironment. When using virtual user interfaces, users are often inter- roanatomy than traditional methods involving physical models.
acting with menu systems that exist in 3D space, which requires some Furthermore, spatial ability played a bigger role when learning in vir-
degree of egocentric spatial ability in order to know where those tual environments compared to physical environments. As immersive
items are in relation to their hand. Egocentric performance would be virtual reality and mixed reality usage increases in science education,
interesting to incorporate; however, there are technical challenges it is becoming increasingly important to understand the role spatial
associated with its measurement, specifically in the recording of ability plays in that learning, as existing literature on spatial perfor-
hand locations relative to virtual menus. A few methods for mixed- mance might not necessarily generalize to these virtual environments.
reality hand tracking have been proposed in the literature, but they
require either a fixed recording location (Lee et al., 2017) or the use AC K N OW L E D G E M E N T S
of markers and additional cameras and sensors (Piekarski et al., 2004). The authors wish to thank Young Ji Tuen for her work on the the-
Furthermore, neither of these approaches address the central prob- matic analysis of the qualitative responses. The research project
lem, which is the translation of virtual coordinates into a real-world was supported through Mitacs Accelerate Internships in partnership
|
19359780, 2022, 6, Downloaded from https://anatomypubs.onlinelibrary.wiley.com/doi/10.1002/ase.2146 by Air University, Wiley Online Library on [01/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1084 HO et al.
with Microsoft Canada. Equipment (HoloLenses) was donated by Fairén González M, Farrés M, Moyes Ardiaca J, Insa E. 2017. Virtual Reality
Microsoft; however, Microsoft was not involved in the study design, to teach anatomy. In: Proceedings of the 38 Annual Conference of the
European Association for Computer Graphics (Eurographics 2017);
data collection, or data analysis.
Lyon, France, 2017 April 24-28. p 51–58. European Association for
Computer Graphics (Eurographics). Geneve, Switzerland.
C O N FL I C T O F I N T E R E S T Gardony AL, Taylor HA, Brunyé TT. 2014. What does physical rotation
The authors declare no financial or non-financial competing interests. reveal about mental rotation? Psychol Sci 25:605–612.
Garg AX, Norman G, Sperotable L. 2001. How medical students learn
spatial anatomy. Lancet 357:363–364.
E T H I C S S TAT E M E N T Gnanasegaram JJ, Leung R, Beyea JA. 2020. Evaluating the effectiveness
Ethical approval was received through the University of British of learning ear anatomy using holographic models. J Otolaryngol
Columbia's Behavioral Research Ethics Board (protocol #: H19- Head Neck Surg 49:63.
Guillot A, Champely S, Batier C, Thiriet P, Collet C. 2007. Relationship
02328), written informed consent was obtained from each partici-
between spatial abilities, mental rotation and functional anatomy
pant prior to the start of the study, and research was performed in learning. Adv Health Sci Educ Theory Pract 12:491–507.
accordance with ethics board guidelines. Hackett M, Proctor M. 2016. Three-dimensional display technologies for
anatomical education: A literature review. J Sci Educ Tech 25:641–654.
F U N D I N G I N FO R M AT I O N Hackett M, Proctor M. 2018. The effect of autostereoscopic holograms on
anatomical knowledge: A randomised trial. Med Educ 52:1147–1155.
This work was supported by MITACS, Inc., Vancouver, BC, Canada
Hanna MG, Ishtiaque A, Nine J, Prajapati S, Pantanowith L. 2018.
Grant number: F18-02840 Augmented reality technology using Microsoft HoloLens in ana-
tomic pathology. Arch Pathol Lab Med 142:638–6 44.
ORCID Henssen DJ, van der Heuvel L, De Jong G, Vorstenbosch MA, van
Cappellen van Walsum AM, Van den Hurk MM, Kolos JG, Bartels
Simon Ho https://orcid.org/0000-0003-3080-8900
RH. 2020. Neuroanatomy learning: Augmented reality vs. cross-
sections. Anat Sci Educ 13:353–365.
REFERENCES Holman P, Bodnar T, Ghomi M, Choi R, Moukhles H, Krebs C. 2018. 3D
Akçayır M, Akçayır G. 2017. Advantages and challenges associated with Neuroanatomy: Using the HoloLens for an augmented reality ap-
augmented reality for education: A systematic review of the litera- proach in neuroanatomy education. FASEB J 32:S1.25.3.
ture. Educ Res Rev 20:1–11. Hoyek N, Collet C, Rastello O, Fargier P, Thiriet P, Guillot A. 2009.
Albrecht UV, Folta-Schoofs K, Behrends M, von Jan U. 2013. Effects of Enhancement of mental rotation abilities and its effect on anatomy
mobile augmented reality learning compared to textbook learning on learning. Teach Learn Med 21:201–206.
medical students: Randomized controlled pilot study. J Med Internet Huk T. 2006. Who benefits from learning with 3D models? The case of
Res 15:e182. spatial ability. J Comput Assist Learn 22:392–4 04.
Anderson LW, Krathwohl DA (Editors). 2001. A Taxonomy for Learning, Jang S, Vitale JM, Jyung RW, Black JB. 2017. Direct manipulation is better
Teaching, and Assessing: A Revision of Bloom's Taxonomy of Educational than passive viewing for learning anatomy in a three-dimensional
Objectives. 1st Ed. New York, NY: Longman. 352 p. virtual reality environment. Comput Educ 106:150–165.
Anderson SJ, Jamniczky HA, Krigolson OE, Coderre SP, Hecker KG. 2019. Khot Z, Quinlan K, Norman GR, Wainman B. 2013. The relative effec-
Quantifying two-dimensional and three-dimensional stereoscopic tiveness of computer-based and traditional resources for education
learning in anatomy using electroencephalography. NPJ Sci. Learn in anatomy. Anat Sci Educ 6:211–215.
4:10. Kurul R, Ögün MN, Neriman Narin A, Avci Ş, Yazgan B. 2020. An alter-
Bajana J, Francia D, Liverani A, Krajčovič M. 2016. Mobile tracking sys- native method for anatomy training: Immersive virtual reality. Anat
tem and optical tracking integration for mobile mixed reality. Int J Sci Educ 13:648–656.
Comput Appl 53:13–22. Kyaw BM, Saxena N, Posadzki P, Vseteckova J, Nikolaou CK, George PP,
Bogomolova K, Ham IJ, Dankbaar ME, Broek WW, Hovius SER, Hage Divakar U, Masiello I, Kononowicz AA, Zary N, Car LT. 2019. Virtual
JA, Hierck BP. 2020. The effect of stereoscopic augmented real- reality for health professions education: Systematic review and
ity visualization on learning anatomy and the modifying effect of meta-analysis by the digital health education collaboration. J Med
visual-spatial abilities: A double-center randomized controlled trial. Internet Res 21:e12959.
Anat Sci Educ 13:558–567. Lee EA, Wong KW. 2014. Learning with desktop virtual reality: Low
Braun V, Clarke V. 2006. Using thematic analysis in psychology. Qual Res spatial ability learners are more positively affected. Comput Educ
Psychol 3:77–101. 79:49–58.
Cheng KH, Tsai CC. 2013. Affordances of augmented reality in sci- Lee EA, Wong KW, Fung CC. 2009. Educational values of virtual reality:
ence learning: Suggestions for future research. J Sci Educ Technol The case of spatial ability. World Acad Sci Eng Technol 30:1157–1167.
22:449–462. Lee SC, Fuerst B, Tateno K, Johnson A, Fotouhi J, Osgood G, Tombari
Duncan-Vaidya EA, Stevenson EL. 2021. The effectiveness of an aug- F, Navab N. 2017. Multi-modal imaging, model-based tracking, and
mented reality head-mounted display in learning skull anatomy at a mixed reality visualisation for orthopaedic surgery. Healthc Technol
community college. Anat Sci Educ 14:221–231. Lett 4:168–173.
Easton RD, Sholl MJ. 1995. Object-array structure, frames of reference, Lee YH, Zhan T, Wu ST. 2019. Prospects and challenges in augmented
and retrieval of spatial knowledge. J Exp Psychol Learn Mem Cognit reality displays. Virtual Real Intell Hardw 1:10–20.
21:483–500. Maniam P, Schnell P, Dan L, Portelli R, Erolin C, Mountain R, Wilkinson
Enyedy N, Danish JA, Delacruz G, Kumar M. 2012. Learning physics T. 2020. Exploration of temporal bone anatomy using mixed reality
through play in an augmented reality environment. Int J Comput (HoloLens): Development of a mixed reality anatomy teaching re-
Support Collab Learn 7:347–378. source prototype. J Vis Commun Med 43:17–26.
Erolin C, Reid L, McDougall S. 2019. Using virtual reality to complement Marks S, White D, Singh M. 2017. Getting up your nose: A virtual re-
and enhance anatomy education. J Vis Commun Med 42:93–101. ality education tool for nasal cavity anatomy. In: Proceedings of
|
19359780, 2022, 6, Downloaded from https://anatomypubs.onlinelibrary.wiley.com/doi/10.1002/ase.2146 by Air University, Wiley Online Library on [01/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
HO et al. 1085
the Special Interest Group on Computer Graphics and Interactive computer presentations in anatomy learning. Med Educ 52:
Techniques (SIGGRAPH Asia 2017) Symposium on Education; Bangkok, 1138–1146.
Thailand, 2017 November 27-3 0. Art 1. Association for Computing Wraga M, Creem SH, Pro DR. 1999. The influence of spatial reference
Machinery (ACM), New York, NY. frames on imagined object-and viewer rotations. Acta Psychol
Martin S, Diaz G, Sancristobal E, Gil R, Castro M, Peire J. 2011. New (Amst) 102:247–264.
technology trends in education: Seven years of forecasts and con- Zacks JM, Michelon P. 2005. Transformations of visuospatial images.
vergence. Comput Educ 57:1893–1906. Behav Cogn Neurosci Rev 4:96–118.
McJunkin JL, Jiramongkolchai P, Chung W, Southworth M, Durakovic Zafar S, Zachar J. 2020. Evaluation of HoloHuman augmented reality
N, Buchman CA, Silva JR. 2018. Development of a mixed re- application as a novel educational tool in dentistry. Eur J Dent Educ
ality platform for lateral skull base anatomy. Otol Neurotol 24:259–265.
39:e1137–e1142.
Merchant Z, Goetz ET, Cifuentes L, Keeney-Kennicutt W, Davis TJ. 2014.
Effectiveness of virtual reality- based instruction on students' AU T H O R B I O G R A P H I E S
learning outcomes in K-12 and higher education: A meta-analysis.
Comput Educ 70:29–4 0. Simon Ho, Ph.D., is a researcher in the Department of
Moro C, Štromberga Z, Raikos A, Stirling A. 2017. The effectiveness of
Psychology at the University of British Columbia in Vancouver,
virtual and augmented reality in health sciences and medical anat-
omy. Anat Sci Educ 10:549–559. Canada. His research interest is in executive function and the
Mou W, McNamara TP. 2002. Intrinsic frames of reference in spatial use of emerging technologies for the measurement of cognitive
memory. J Exp Psychol Learn Mem Cogn 28:162–170. function.
Pan Z, Cheok AD, Yang H, Zhu J, Shi J. 2006. Virtual reality and mixed
reality for virtual learning environments. Comput Graph 30:20–28.
Peters M, Laeng B, Latham K, Jackson M, Zaiyouna R, Richardson C. Pu Liu, Ph.D., is a postdoctoral fellow in the Department of
1995. A redrawn Vandenberg and Kuse mental rotations test: Cellular and Physiological Sciences at the University of British
Different versions and factors that affect performance. Brain Cogn Columbia in Vancouver, Canada. Her research interest is in ex-
28:39–58.
tended reality and human-computer interaction.
Piekarski W, Avery B, Thomas BH, Malbezin P. 2004. Integrated head
and hand tracking for indoor and outdoor augmented reality. In:
Proceedings of the Institute of Electrical and Electronics Engineers Daniela J. Palombo, Ph.D., is an assistant professor in the
Conference on Virtual Reality (IEEE Virtual Reality 2004); Chicago, IL, Department of Psychology at the University of British Columbia
2004 July 27-31. p 258–276. Institute of Electrical and Electronics
in Vancouver, Canada. She teaches classes on memory and cog-
Engineers (IEEE), Piscataway, NJ.
nition and her research focuses on episodic autobiographical
Rothman Z. 2014. Introduction to the Central Nervous System -UBC
Neuroanatomy Season 1 -Ep 1. YouTube video. 14:46 min. URL: memory.
https://www.youtube.com/watch? v=xB7rXw_3gVY [accessed 8
August 2021]. Todd C. Handy, Ph.D., is a professor in the Department of
Shepard RN, Metzler J. 1971. Mental rotation of three-dimensional ob-
Psychology at the University of British Columbia in Vancouver,
jects. Science 171:701–703.
Simms LJ. 2008. Classical and modern methods of psychological scale Canada. He is the director of the Attentional Neuroscience Lab
construction. Soc Personal Psychol Compass 2:414–433. and teaches cognitive neuroscience. His research focuses on at-
Sjölie D, Bodin K, Elgh E, Eriksson J, Janlert LE, Nyberg L. 2010. Effects tention, mind wandering, and issues of cognitive health.
of interactivity and 3D-motion on mental rotation brain activity
in an immersive virtual environment. In: Proceedings of the 28th
International Conference on Human Factors in Computing Systems (CHI Claudia Krebs, M.D., Ph.D., is a professor of teaching in the
2010); Atlanta, Georgia, 2010 April 12-15. p 869–878. Association Department of Cellular and Physiological Sciences at the
for Computing Machinery (ACM), New York, NY. University of British Columbia in Vancouver, Canada. She teaches
Sweeney K, Hayes JA, Chiavaroli N. 2014. Does spatial ability help the
gross anatomy and neuroanatomy to the health professions. Her
learning of anatomy in biomedical science course? Anat Sci Educ
7:289–294. research interest is around the development and integration of
Taber KS. 2018. The use of Cronbach's alpha when developing and re- technology in the classroom.
porting research instruments in science education. Res Sci Educ
48:1273–1296.
S U P P O R T I N G I N FO R M AT I O N
Tolentino L, Birchfield D, Megowan-Romanowicz C, Johnson-Glenberg
MC, Kelliher A, Martinez C. 2009. Teaching and learning in the Additional supporting information may be found in the online
mixed-reality science classroom. J Sci Educ Technol 18:501–517. version of the article at the publisher’s website.
Vandenberg SG, Kuse AR. 1978. Mental rotations, a group test of three-
dimensional spatial visualization. Percept Mot Skills 47:599–604.
Voyer D, Jansen P, Kaltner S. 2017. Mental rotation with egocentric and How to cite this article: Ho S, Liu P, Palombo DJ, Handy TC,
object-based transformations. Q J Exp Psychol 70:2319–2330.
Krebs C. 2022. The role of spatial ability in mixed reality
Wainman B, Pukas G, Wolak L, Mohanraj S, Lamb J, Norman GR. 2020.
The critical role of stereopsis in virtual and mixed reality learning learning with the HoloLens. Anat Sci Educ 15:1074–1085.
environments. Anat Sci Educ 13:401–412. https://doi.org/10.1002/ase.2146
Wainman B, Wolak L, Pukas G, Zheng E, Norman GR. 2018. The supe-
riority of three-dimensional physical models to two-d imensional