Higher Cognitive Load Increases Unintended Positional Drift
during Navigation in Virtual Reality
Hugo Brument
VR & AR Research Unit
TU Wien
Vienna, Austria
hugo.brument@tuwien.ac.at
Martin Gerdenich
TU Wien
Vienna, Austria
e11776233@student.tuwien.ac.at
Hannes Kaufmann
VR & AR Research Unit
TU Wien
Vienna, Austria
hannes.kaufmann@tuwien.ac.at
Abstract
Steering techniques are common for traveling in Virtual Reality
(VR) applications. However, these techniques can generate Unin-
tended Positional Drift (UPD), a phenomenon characterized by the
user’s unconscious or unintentional physical movements of the
users in the workspace. Although some research work has already
investigated the influence of heading (i.e., how the virtual trajec-
tory direction is updated) and cognitive load on travel performance
using steering techniques, they did not investigate whether these
factors could increase UPD. In this paper, we thus replicated and
extended an existing user study to assess the influence of heading
and cognitive task (CT) on UPD while navigating with steering
techniques in VR. Participants had to perform a multidirectional
travel task (based on the ISO 9241-9) while doing a CT (two-back
task). We considered conditions with and without performing the
CT, where we varied the steering heading (head, hand, torso). We
compared our results with those of a previous study and found
that the heading did not influence users’ travel and cognitive per-
formance. However, we noticed the influence of workload on the
user’s UPD. This work contributes to understanding UPD, which
tends to be an overlooked topic, and the cognitive implications
of navigation in VR. We discuss the design implications of these
results for improving navigation in VR.
CCS Concepts
•Human-centered computing→ Virtual reality; User studies.
Keywords
Virtual Reality, Locomotion Technique, Cognitive Load, Unintended
Positional Drift
ACM Reference Format:
Hugo Brument, Martin Gerdenich, and Hannes Kaufmann. 2025. Higher
Cognitive Load Increases Unintended Positional Drift during Navigation
in Virtual Reality. In 31st ACM Symposium on Virtual Reality Software and
Technology (VRST ’25), November 12–14, 2025, Montreal, QC, Canada. ACM,
New York, NY, USA, 10 pages. https://doi.org/10.1145/3756884.3765968
This work is licensed under a Creative Commons Attribution-NonCommercial-
ShareAlike 4.0 International License.
VRST ’25, Montreal, QC, Canada
© 2025 Copyright held by the owner/author(s).
ACM ISBN 979-8-4007-2118-2/25/11
https://doi.org/10.1145/3756884.3765968
1 Introduction
Many locomotion techniques (LTs) have been proposed for navi-
gating in Virtual Reality (VR), which can be classified according
to the required motion involvement of the user [28]. Additionally,
numerous research works and user studies have investigated and
compared (LTs) for navigating in Virtual Environments (VEs) [48].
These studies usually assess some factors such as the travel per-
formances [11], their potential cybersickness issues [36], or their
usability [17]. However, the cognitive implications have been con-
sidered more for real LTs, such as walking [27, 44] or redirected
walking [4, 38], than virtual techniques, such as steering [26]. In
particular, some work compared the influence of cognitive load for
steering techniques using a dual paradigm task [25], where they
found that the heading (i.e., how the future direction is provided)
can increase the user’s cognitive load.
Users actively engage their entire bodies during VR interaction,
even when using virtual LTs. It can be either by involving conscious
physical movements, such as walking or reaching, or unconsciously,
as spontaneous reactions to virtual stimuli, without being able to
see the limits and obstacles in the actual workspace due to wearing
a Head-Mounted Display (HMD). These movements, from now
on referred to as Unintended Positional Drift (UPD), harm the VR
user experience but are typically disregarded in the design of VR
applications. UPD can negatively impact locomotion in VR with
an HMD [22, 23, 35], such as users reaching the boundaries of
the workspace or obstacles without noticing them. Studies have
shown that steering techniques yield UPD (i.e., moving physically
while virtually navigating the VE) [5, 6]. Still, the related work
investigating the cognitive demands of steering did not consider
whether steering techniques and cognitive load could increase UPD.
In this paper, we replicated the study presented by Lai. et al [25]
we extended it by adding a control condition, i.e., no cognitive task
(CT), to assess whether the CT will increase UPD during steering
navigation in VR. We confirmed the results from Lai. et al, where
the heading (head, hand, or torso) did not influence participants’
cognitive performance. However, we found additional results where
the CT increased UPD. Our results contribute to understanding the
cognitive demands of virtual LTs and UPD during navigation in
VR. We discuss the design of LTs and how to consider UPD in VR
literature better.
2 Related Work
In the related work, we focus on steering LT evaluations and UPD.
For further information about the related work on steering tech-
niques in general, please refer to [48]. Since our study includes a
CT, we briefly cover the influence of cognitive load on navigation
VRST ’25, November 12–14, 2025, Montreal, QC, Canada Brument et al.,
with steering techniques only. For further information about how
cognitive load can be assessed with different paradigms, please
refer to the following survey [24, 29] or the related work written
by Lai et al. [25, 26].
2.1 Evaluation of Steering Techniques in VR
The first evaluation of steering techniques was introduced by Bow-
man et al., where they showed that head and hand steering provided
similar task completion time, but hand steering provided better dis-
tance estimation than head steering [2]. In addition, they also did
not find any significant differences for time to navigate in virtual
corridors [3]. Since then, many other studies have investigated how
the design of steering techniques or external factors can influence
users’ navigation performance, cybersickness, and cognitive load.
2.1.1 Performance Evaluations. In terms of performance evalua-
tions, Suma et al. did not find an influence of heading (head, hand,
torso) on maintaining distance with a virtual target [41]. However,
Christou et al. found hand steering faster than gaze steering during
a wayfinding task [10]. Similar results were found in another study
in a point origin task [46]. In contrast, Suma et al. found that gaze
steering allowed users to navigate faster in a VE maze with fewer
collisions than hand steering [43]. In comparison, Zielasko et al.
showed that task torso steering reduced the number of collisions
with the VE compared to head or hand steering [49]. Yet, Gao et al.
showed that head steering resulted in the lowest trial failures and
torso the highest in a goal-directed locomotion task [15]. Overall,
while several studies showed the influence of heading in terms of
performance (time, collisions with the VE), there are conflicting
results. One reason for inconclusive travel performance results
is that most navigation studies are task-dependent, reducing the
generalization of findings.
2.1.2 Cognitive Evaluations. Previous studies have not found sig-
nificant differences in cognitive performance among steering tech-
niques. Using a scoring method that combined travel time with
post-task information recall, Bowman et al. reported no notable
differences across the three techniques [3]. Suma et al. found no
significant differences between gaze and head steering in cognitive
tasks involving sketch maps and object recall [42]. Another study
found no significant differences in dual-task cognitive performance
or post-task word recognition [41]. Brument et al. found that torso
steering combined with different speed updates can influence the
user’s perception of physical demand during a slalom task [7]. Re-
cently, Lai et al. compared the cognitive implications of head, hand,
and torso steering [25]. The three steering yielded comparable cog-
nitive performance during a dual task involving multidirectional
travel and a verbal memory task.
2.1.3 Cybersickness Evaluations. Overall, previous studies have
not found significant differences in simulator sickness across steer-
ing techniques. In two separate studies, Suma et al. reported no
differences between head and hand steering [41, 42]. Brument et al.
also did not find differences in cybersickness between head, hand,
and torso steering during repetitive curvilinear trajectories [8]. In
contrast, Lai et al. showed that head steering led to less simulator
sickness than hand or torso steering [25]. Langbehn et al. showed
that steering provided higher cybersickness than other LTs such as
WIP or teleportation [27]. Regarding the influence of body position,
Clifton and Palmisano found no difference between using steering
techniques while standing or sitting [12]. In contrast, Kim et al.
found that standing led to higher cybersickness than sitting [21].
The type of control can explain this contradiction during the task,
where Clifton et al. chose a passive task (automatic motion).
2.2 Unintended Positional Drift
2.2.1 Definitions. First introduced by Nilsson et al. [34], UPD is
a phenomenon where users move unintentionally in the RE using
a different technique than natural walking, like walking-in-place
(WIP), or steering. Although research explicitly addressing the prob-
lem of UPD is rare, UPD could have a high impact during prolonged
VR sessions due to the limited workspace in standard VR setups.
Nilsson et al. [34] split the potential approaches to reduce UPD
into two categories: focusing on the navigation control law’s input
modalities or the workspace’s physical constraints. They inves-
tigated the impact of input gestures during WIP locomotion on
UPD [32], showing that the difference in gesture impacted users’
UPD. They also obtained similar results in another study, in which
the difference in UPD was probably caused again by the leg move-
ments defined by commonWIP gestures [33]. More recently, a study
compared the influence of four different LT (WIP, Running in Place,
teleportation, and steering) on UPD. Results showed that it is easier
to stay within the center of the workspace with teleportation and
steering than with WIP or Running in Place [23].
2.2.2 Factors influencing UPD. While the current literature ad-
dressed UPD by understanding the factors that could influence it
for WIP [22], few contributions address virtual LTs. Brument et
al. found that trajectory curvature increased UPD during a slalom
task [5]. Additional studies showed that rotation gains applied to
steering navigation can influence the rotation performed and thus
reduce UPD [6]. Choudhary et al. showed they could induce UPD
by manipulating the virtual camera’s rotation magnitude and for-
ward translation flow during physical rotation [9]. Nilson et al.
assessed different modalities for minimizing UPD, including addi-
tional sensory feedback (auditory, visual, audiovisual, and passive
haptic) [35]. The results showed that both passive haptic feedback
and feedback types with gradual onset were the most efficient
at reducing UPD. The passive haptic feedback tended to be more
helpful and less distracting than some feedback with a gradual
onset. Finally, Montano et al. demonstrated that the UPD could also
occur using scale adaptive techniques (that dynamically adapt a
user’s displacements) during walking [31]. They noticed that due
to the mismatch between the computed position in VR and the
position in the workspace, these accumulated differences between
the user’s physical and (scaled) movements in VR resulted in UPD.
They quantified the UPD using scale techniques and proposed a
UPD correction model to minimize UPDwhile walking in VEs. They
demonstrated that UPD could be reduced by increasing the distance
traveled in the VE without being overtly redirected.
2.3 Summary and Contribution
The literature about evaluations of steering is extensive, since many
factors (e.g., human, task, and LT design) can influence users’ experi-
ence while navigating with them. Among these factors, the heading
Higher Cognitive Load Increases Unintended Positional Drift during Navigation in Virtual Reality VRST ’25, November 12–14, 2025, Montreal, QC, Canada
is an important factor when designing LTs as it directly influences
the travel. Moreover, cognitive load is also critical in LT evaluations
as users often interact more (e.g., selecting or manipulating objects)
while traveling. Yet there is a lack of research regarding how both
heading and cognitive load may influence UPD and travel behavior.
This has recently drawn attention during some evaluations of steer-
ing techniques in VEs [5, 23]. UPD can negatively impact steering
locomotion, which is one of the most common LT used in con-
sumer VR applications [40]. Due to eventual discrepancies between
the physical and virtual motion, users may not be aware of their
physical position in the workspace. They may reach workspace
boundaries or collide with obstacles. Thus, studying the cognitive
implications of steering techniques on UPD could help improve the
design of LTs for navigating in VR.
We used the experiment by Lai et al. [25] as a basis for our user
study. We did not have access to either the experimental platform
or their dataset; thus, we reproduced the experiment based on the
research paper, which had enough description to replicate it fully.
They compared how a dual task involving a multidirectional travel
task and a verbal memory task could influence users’ navigation
in VR using Head, Hand, or Torso steering. They showed that (1)
the heading in steering techniques does not influence cognitive
load during navigation in VR, where similar cognitive load metrics
were observed; (2) Travel performances (time to travel between
poles, misalignment with the poles) were similar between the LTs;
(3) Hand Steering led to higher cybersickness than Head or Torso
steering. Although their work provides insight regarding the influ-
ence of heading during navigation in VR, their study lacks a control
condition where no cognitive task is present to see whether the
cognitive load influences travel performance and cybersickness.
In addition, we are interested in how LT and cognitive load can
influence UPD, as we believe that considering UPD is fundamental
for improving the design of steering LTs. Since most VR consumers
may have a small workspace, understanding how UPD occurs may
help VR designers mitigate it to prevent users from reaching the
boundaries of their workspace.
In this paper, we only focused on the steering LTs study and
not teleportation-based LTs presented in the original paper, since
steering may influence more UPD than teleportation [23]. Thus, our
extension study includes a baseline condition without a cognitive
task as an independent variable and the computation of UPD as a
dependent variable. Our results can give insights into (1) validating
the results from the original work about the influence of heading
in multitasking in VR. (2) How can the combination of heading and
cognitive load tasks influence steering navigation in VR under a
control condition? (3) Improve our understanding of UPD, which is
still an overlooked topic in VR.
3 User Study
We adopted and replicated the same dual-task methodology with
standardized cognitive and travel tasks that Lai et al. used [25].
In this section, we describe the study design, including the travel
and cognitive tasks, new independent and dependent variables to
address the research gap from the initial user study, new hypotheses
that were missing, the apparatus and procedure of the study, and
finally, the data processing and analyses.
Figure 1: Top view of the multidirectional travel task used in
the user study. The yellow pole represents the current pole to
reach, and the number represents the order of the sequence.
Figure 2: The two-back CT used in the user study.
3.1 Study Design
3.1.1 Travel Task. The travel task is based on the ISO 9241-9 task,
where the user has to make several selections of objects displayed
around a ring, in which the target switches sides of the ring while
continuing clockwise. We used 12 semi-transparent gray poles
to indicate targets (Figure 1). The yellow column was the next
destination to reach. Participants were asked to stop and stay within
the yellow pole for two seconds. We displayed a visual timer in
front of the user’s viewport to provide feedback. We set the poles’
width and height, respectively, to 0.5 and 2.30 meters to guarantee
users do not feel constrained within the poles. The distance of poles
from the center is 5 meters.
3.1.2 Cognitive Task. We used a two-back working memory task,
where participants observed a sequence of letters and responded
by pressing the controller’s trigger only when the current letter
matched the one presented two stimuli earlier [45] (Figure 2). Let-
ters were randomly selected from a predefined set (A, B, C, D), each
displayed for 500 ms, with a pseudo-randomized inter-stimulus
interval ranging from 1100 to 1500 ms.
VRST ’25, November 12–14, 2025, Montreal, QC, Canada Brument et al.,
3.1.3 Steering Locomotion Technique. We implemented three steer-
ing LTs: Gaze, Hand, and Torso-directed steering. We reproduced
the approach from Lai et al. [25]: We used the Vive trackpad to
initiate the travel. Forward and backward motion was enabled by
pressing the trackpad forward or backward. The continuous motion
while pressing the trackpad was constant (1.44m/s), with maximum
and minimal linear acceleration set to respectively -1 and 1𝑚.𝑠−2,
as recommended to match comfortable real walking navigation [7].
The heading direction (i.e., travel direction) was defined by three dif-
ferent body segments (Figure 3). We projected the HMD’s forward
vector onto the ground in the head steering LT. We projected the
hand controller onto the ground in the hand steering LT. We pro-
jected the HTC Vive tracker onto the ground in the torso steering
LT. The tracker was attached to the front of the user’s chest using
a strap. The user could choose which hand to hold the controller
and all decided to use their right hand.
3.1.4 Independent Variables. Our study had a 3 LT-Heading (Head
Steering, Hand Steering, Torso Steering) x 2 CT (CT-With, CT-
Without) within-subject design (Table 1). We selected the CT (sub-
subsection 3.1.2) and the same steering techniques (subsubsec-
tion 3.1.3) as in Lai. et al [25]. In addition, we added a control con-
dition where users conducted the travel task (subsubsection 3.1.1)
without the CT. The experiment was split into six blocks, where
each combination of LT-Heading andCTwas tested once per block
and counterbalanced using a Latin square design. A block consisted
of 12 trials, each defined as the travel between two poles (Figure 1).
Table 1: Conditions tested in the replicated study.
Independent Variable Levels
LT-Heading Head, Hand, Torso
Cognitive Task (CT) With-CT, Without-CT
3.1.5 Dependent Variables. Regarding travel performance, we mea-
sured for each trial (i.e., travel between two poles) the time to per-
form the trial and the number of missed terminations (i.e., where
participants would terminate travel outside of the pole for at least
two seconds or leave the pole before two seconds).
Regarding CT metrics, we collected the number of true positive
(TP, i.e., pressing during a correct two back match), false positive
(FP, i.e., not pressing during a correct two backmatch), true negative
(TN, i.e., not pressing during an incorrect two back match) and false
negative (FN, i.e., pressing during an incorrect two back match).
We then calculated the accuracy, precision, recall, and F-measure
using Equation 1 Accuracy measures the overall correctness of
predictions without distinguishing between correct matches and
non-matches. F-measure balances precision and recall into a single
metric using a harmonic mean to evaluate the success of the user’s
performance when the match stimuli appear more often than the
non-match stimuli.
𝐴𝑐𝑐𝑢𝑟𝑎𝑐𝑦 = (𝑇𝑃 +𝑇𝑁 )/(𝑇𝑃 +𝑇𝑁 + 𝐹𝑃 + 𝐹𝑁 )
𝑃𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛 = 𝑇𝑃/(𝑇𝑃 + 𝐹𝑃)
𝑅𝑒𝑐𝑎𝑙𝑙 = 𝑇𝑃/(𝑇𝑃 + 𝐹𝑁 )
𝐹 −𝑚𝑒𝑎𝑠𝑢𝑟𝑒 = (𝑃𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛 ∗ 𝑅𝑒𝑐𝑎𝑙𝑙)/(𝑃𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛 + 𝑅𝑒𝑐𝑎𝑙𝑙)
(1)
In addition to the replicated study, we computed UPD metrics
to see how much users would drift in the workspace. We recorded
the highest absolute value of UPD on both X (UPD-ABS-X-MAX)
and Z (UPD-ABS-Z-MAX) axes (i.e., the displacement between
the beginning and the end of the block) to assess whether the
LT-Heading or CT could influence it. These metrics measure the
dispersion after performing the travel task. We also computed the
maximal norm of the UPD (UPD-Norm-Max) to assess whether
users would have more physical movement after one block.
Regarding subjective measures, we collected after each block
the answers from the Simulator Sickness Questionnaire (SSQ) for
each trial to assess any simulator sickness effects [20]. Additionally,
we decided to collect for each block the raw values of the NASA
Task Load Index (NASA-TLX) to have insights about the subjective
cognitive load perception of users [18].
3.2 Hypotheses
The initial study did not have hypotheses. Since this work replicates
it, we formulated the hypotheses based on their findings (RH, which
stands for replication hypotheses). We added new hypotheses (NH,
which stands for new hypotheses) based on the extension of the
dependent and independent variables tested. Regarding cognitive
load, the previous study did not reveal an effect of LT-Heading.
We suggest observing similar results regarding objective metrics.
In addition, since our study considers a control condition where
participants are performing the task without the verbal memory
task, we expect that perceived cognitive load will be lower in the
control condition:
• [RH1.1] - There will be an effect of LT-Heading on the
user’s accuracy and F-measure of the cognitive task.
• [NH1.2] - There will be an effect of LT-Heading on the
user’s raw TLX scores.
• [NH1.3] - RawTLX scoreswill be lower duringCT-Without.
Regarding travel performance, we expect a similar observation
where the LT-Heading will not affect the user’s time to perform
the task or missed termination. However, since our study considers
a control condition where participants are performing the task
without the verbal memory task, we expect that performance will
be higher in the control condition:
• [RH2.1] - There will be an effect of LT-Heading on
travel performance (time, missed termination).
• [NH2.2] - Travel performance will be better (faster time,
lower missed termination) during CT-Without.
Regarding cybersickness, the previous study showed that Hand
steering led to higher cybersickness scores than Head Steering.
We expect similar results since Hand Steering is a less common
technique and the Hand is detached from the body trunk, thus the
decoupling of heading compared to ourHead or Torso increases user
discomfort. Furthermore, we expect that an increase in cognitive
load will increase SSQ scores as a similar study showed [39]:
• [RH3.1] - Hand Steering will result in higher SSQ scores
than Head or Torso Steering. [NH3.2] - CT-With will in-
crease SSQ scores compared to CT-Without.
Last, our work investigates UPD, whichwas not considered in the
previous study. Since UPD occurs during steering, we suggest that
Higher Cognitive Load Increases Unintended Positional Drift during Navigation in Virtual Reality VRST ’25, November 12–14, 2025, Montreal, QC, Canada
Figure 3: Left - A participant wearing an HMD and a tracker
on the Torso. Right - Overview of the VE used in the study.
increased cognitive load will result in increased UPD, as users may
notice less unintentional drift in the workspace as they are more
focused on performing the verbal memory task.We also suggest that
LT-Headingwill influence UPD, where Hand steering will result in
higher UPD than Head or Torso, since the hand is decoupled from
the orientation process when navigating in VEs: [NH4.1] - Hand
Steering will increase UPD in the workspace. [NH4.2] - Increased
cognitive load will increase UPD in the workspace.
3.3 Participants and Apparatus
An a priori power analysis was conducted using G*Power version
3.1.9.7 [14] to determine the minimum sample size required to test
the study hypotheses. The results indicated that the sample size
required to achieve 80% power to detect a medium effect (as no
effect size was reported in the original study), with a significance
criterion of 𝛼 = 0.05, was N = 24 for ANOVAwith a 2x3 fully within-
subjects design. Thus, 24 participants (10 female) aged from 21 to 39
years old (30.10±5.86) participated in the experiment. The sample
was mostly students and researchers recruited from our faculty
and all had a right-hand dominance. 16 participants reported using
VR every week, and 8 only a few times or never. The standards of
the Declaration of Helsinki were followed when conducting the
study. They signed an informed consent form and were naive to
the purpose of the experiment.
We used an HTC Vive Pro system (HMD, controllers, and one
tracker on the torso) with a 4m x 4m workspace. The controller
was used to navigate (trackpad) and answer the CT (trigger). We
suspended the cables above the user’s head to prevent any tripping
on them. The HTC Vive audio strap provided audio feedback for
the two-back task. The application was developed using Unity
(2020.3.30f1), and the VE consisted of a virtual storage building with
textures on the walls and floor to generate motion flow without any
salient features (Figure 3). We used gray poles to indicate the travel
zones and a yellow pole to indicate the current pole to reach. A black
and yellow platform was set at the center of the multidirectional
task for calibration. We guaranteed the application was run at the
HMD’s refresh rate (90Hz).
3.4 Procedure
At the beginning of the experiment, each participant signed a con-
sent form and completed a demographic form to collect information
about their background and experience with VR. Then, they filled
out a pre-SSQ questionnaire to assess their initial cybersickness be-
fore experiencing VR, and the experimenter explained the dual task
(LT usage and 2-back memory task). The users were equipped with
the HTC Vive Pro and the HTC Vive tracker to complete a training
including (1) performing the multidirectional task with each LT
without the memory task, (2) performing the memory task alone
without navigating and (3) a training of the study task involving
both navigation and memory task while reaching an accuracy of
at least 75%. After training, participants performed one block con-
sisting of an LT-Heading (Head, Hand, or Torso) and the presence
of the CT or not. The participant was instructed to complete the
dual task as fast and accurately as possible regarding travel and CT.
Once the task was done, the participant took off the HMD and filled
out the SSQ questionnaire. Participants had a three-minute break
to minimize cybersickness symptoms. They repeated the process
(performing the task and filling the SSQ) for the five remaining
blocks. Each participant took approximately 45 to 60 minutes to
complete the study procedure.
3.5 Data Analysis
Every participant completed the experiment and was considered
in the data analyses. We collected 144 block measures (6 per LT-
Heading and CT combination for 24 users), resulting in 1728 trials
(i.e., travel from one pole to another, done for 12 poles). For the
data analysis, we removed the first trial (i.e., the travel from the
calibration center to the first pole) as we were interested in the
continuous motion from one pole to another pole, resulting in 1584
trials to analyze (264 per condition).
For normally distributed metrics, assessed using the Shapiro-
Wilk test, we analyzed variance (ANOVA) with repeated measures
factors. Greenhouse-Geisser adjustments were applied to the de-
grees of freedom when the sphericity assumption was violated.
For metrics that deviated from a normal distribution, we used the
non-parametric Aligned Rank Transform (ART) test [47]. The post-
hoc analysis involved pairwise t-tests with Bonferroni corrections
for customarily distributed dependent variables or the multifactor
contrast test procedure presented in [13] for the non-normally dis-
tributed ones. SSQ and raw NASA-TLX scores were analyzed with
Friedman ANOVA with pairwise Wilcoxon post-hoc comparison
with Bonferroni corrections.
4 Results
4.1 Travel Performance
A block lasted around 115.90±12.36 second. For the time to perform
the trial (line 1 of Table 2), we did not observe an effect of LT-
Heading, but we did for the CT (medium effect size). Post-hoc
tests confirmed that participants took longer to perform the travel
task during CT-With than CT-Without. For the missed terminations
(line 2 of Table 2), we did not observe an effect of LT-Heading, but
we did for the CT (medium effect size). Post-hoc tests confirmed
that participants had more missed terminations (i.e., not staying
two seconds within the pole or leaving the pole before the end of
the countdown) during CT-With than CT-Without.
VRST ’25, November 12–14, 2025, Montreal, QC, Canada Brument et al.,
Table 2: Mean and standard deviation, reported as M±SD, for the metrics gathered during the study, grouped by LT-Heading
and CT. The effect columns report whether there was a significant effect of the independent variable. Post-hoc tests for main
effects are reported using superscripts. Two levels sharing the same superscript are not significantly different.
Head Hand Torso Effect CT-Without CT-With Effect
Time 9.75±1.39 9.77±1.29 9.76±1.30 𝐹2,1555 =0.77, 𝑝 ==.46 9.58±1.341 9.93±1.292 𝐹1,1555 =90.90, 𝑝 < .001, 𝜂2𝑝 = .05
Missed Termination 1.12±2.65 1.23±1.87 0.97±1.69 𝐹2,115 =1.33, 𝑝 ==.26 0.65±1.102 1.59±1.102 𝐹1,115 =7.65, 𝑝 < .01, 𝜂2𝑝 = .04
Accuracy 0.73±0.11 0.71±0.10 0.75±0.09 𝐹2,46 =2.08, 𝑝 ==.13 - - -
F-measure 0.65±0.17 0.63±0.18 0.62±0.17 𝐹1,46 =0.32, 𝑝 ==.72 - - -
UPD-ABS-X-MAX 0.50±0.23 0.46±0.21 0.49±0.29 𝐹2,115 =0.41, 𝑝 ==.66 0.42±0.151 0.55±0.302 𝐹1,115 =0.41, 𝑝 < .01, 𝜂2𝑝 = .08
UPD-ABS-Z-MAX 0.55±0.22 0.52±0.18 0.53±0.22 𝐹2,115 =0.07, 𝑝 ==.92 0.53±0.18 0.54±0.24 𝐹1,115 =0.01, 𝑝 ==.96
UPD-Norm-MAX 0.76±0.27 0.71±0.25 0.75±0.30 𝐹2,115 =0.87, 𝑝 ==.42 0.68±0.191 0.79±0.332 𝐹1,115 =0.07, 𝑝 = .01, 𝜂2𝑝 = .04
SSQ-Disorientation 27.26±47.26 22.91±30.31 30.74±56.92 𝐹2,115 =0.16, 𝑝 ==.84 27.64±51.10 26.29±40.50 𝐹1,115 =3.01, 𝑝 ==.08
SSQ-Oculomotor 14.52±31.53 15.79±26.93 23.37±39.98 𝐹2,115 =2.58, 𝑝 ==.08 16.21±36.501 19.58±29.822 𝐹1,115 =5.38, 𝑝 < .02, 𝜂2𝑝 = .04
SSQ-Nausea 18.08±34.99 14.11±26.47 25.63±45.26 𝐹2,115 =1.44, 𝑝 ==.23 19.61±40.91 18.94±31.68 𝐹1,115 =5.38, 𝑝 ==.06
SSQ-Total 17.84±33.31 15.66±24.28 23.76±40.23 𝐹2,115 =0.57, 𝑝 ==.56 18.80±36.721 19.37±29.382 𝐹1,115 =5.54, 𝑝 < .02, 𝜂2𝑝 = .05
TLX-Mental 44.79±33.72 43.22±32.18 43.12±32.18 𝐹1,115 =0.24, 𝑝 = .80 16.87±15.501 70.55±20.582 𝐹1,115 =319, 𝑝 < .001, 𝜂2𝑝 = .73
TLX-Physical 24.58±19.01 22.50±16.24 23.85±17.26 𝐹1,115 =0.09, 𝑝 = .90 18.47±14.251 28.81±18.842 𝐹1,115 =28.06, 𝑝 < .001, 𝜂2𝑝 = .19
TLX-Temporal 40.31±27.74 35.41±24.46 38.33±27.15 𝐹1,115 =1.91, 𝑝 = .15 17.43±13.501 58.61±18.972 𝐹1,115 =310.11, 𝑝 < .001, 𝜂2𝑝 = .20
TLX-Performance 36.35±28.96 35.93±28.96 38.02±28.39 𝐹1,115 =0.35, 𝑝 = .70 16.87±19.721 56.66±20.172 𝐹1,115 =164.10, 𝑝 < .001, 𝜂2𝑝 = .42
TLX-Effort 47.18±30.99 44.68±31.47 48.85±28.39 𝐹1,115 =0.61, 𝑝 = .54 30.90±29.131 62.91±23.052 𝐹1,115 =84.44, 𝑝 < .001, 𝜂2𝑝 = .58
TLX-Frustration 34.06±27.61 34.06±24.40 37.08±25.69 𝐹1,115 =0.30, 𝑝 = .74 19.23±17.931 50.90±22.582 𝐹1,115 =107.53, 𝑝 < .001, 𝜂2𝑝 = .48
TLX-Total 37.88±23.00 35.97±22.47 38.21±22.12 𝐹1,115 =0.96, 𝑝 = .38 19.96±13.851 54.74±14.332 𝐹1,115 =284.85, 𝑝 < .001, 𝜂2𝑝 = .04
Figure 4: The cognitive load F-measures of LT-Heading with
standard deviation bars. Yellow contains the results of our
study, and black represents the original study’s results (stan-
dard deviation due to lack of report on the data).
4.2 Cognitive Load
Regarding objective cognitive measurements, Figure 4 shows the
bar plot with the mean and standard deviation of our results and
the initial study. Based on a visual inspection, the magnitude of the
results is similar. In our study, we also did not observe an effect of
LT-Heading on the accuracy and F-measure (respectively line 3
and 4 of Table 2). Regarding the raw NASA-TLX scores, the lines
12 to 20 from Table 2 report each subscale and total scores. We
did not observe an effect of LT-Heading on any NASA-TLX raw
scores, but we found an effect of the CT on every subscale and
total scores of the NASA-TLX (high effect size). Post-hoc analysis
showed that subjective cognitive load perception was significantly
higher during CT-With than CT-Without (Figure 5).
Figure 5: Boxplot of NASA-TLX raw scores during CT-With
(green) and CT-Without (blue) for each subscale.
4.3 Cybersickness
Overall, no participants reported strong cybersickness during the
experiment. Lines 8 to 11 of Table 2 report the SSQ scores per LT-
Heading and CT. We did not find an order effect of SSQ subscales
and Total scores (Figure 6); thus, we did not consider them in the
analysis.We did not find any effect of LT-Heading on any subscales
or the total scores of the SSQ. However, we found an effect of CT
on the Oculomotor subscales, where post-hoc analysis showed that
Oculomotor scores were higher during CT-With than CT-Without
(small effect size). Total SSQ scores were also higher during CT-With
than CT-Without (small effect size).
Higher Cognitive Load Increases Unintended Positional Drift during Navigation in Virtual Reality VRST ’25, November 12–14, 2025, Montreal, QC, Canada
Figure 6: Total SSQ score per condition order.
4.4 Unintended Positional Drift
Figure 7 shows the highest UPD recorded during a block between
CT-With and CT-Without. We computed a fit ellipse using a 2D
Gaussian distribution and a 95% confidence interval, and the ellipse
parameters are reported in Table 3. Visually, we can observe that the
spread of UPD (i.e., width and height of the ellipse, at the end of the
trial is bigger during CT-With than CT-Without. These observations
were confirmed when computing the UPD metric statistically (lines
5 to 7 in Table 2). We found that the CT increased the absolute
maximum UPD on the X axis during a block (medium effect size),
which was confirmed with post-hoc analyses. However, we did
not observe increased maximum UPD on the Z axis. In addition,
we found that the maximal norm of the UPD increased during
CT-With at the end of the block (small effect size).
Table 3: Parameters of the 95% Ellipse fit for UPD distribution
after a block for CT-With and CT-Without.
CT-Without CT-With
Centroid [-0.03;0.07] [-0.02;0.07]
Width/Height 1.67 1.14
Height 1.14 1.42
Angle 48.27 -169.83
Area 1.50 2.64
5 Discussion
Overall, we mostly confirmed every result found in the original
study, strengthening the design guidelines provided previously,
and emphasizing that the LT-Heading does not matter in terms of
performance, usability, and cognitive load during repetitive steering
navigation tasks. We discuss in further detail the new implications
of our replicated study.
5.1 Comparison with the Original Study
Regarding cognitive performance, we found a F-measure similar
to the initial study, rejecting [RH1.1] (Figure 4). Yet, it is worth
noticing that by comparing the variability of their study and ours
in terms of F-measure, our participants had a higher variability
Figure 7: Maximum UPD position recorded during CT-
Without (blue) and CT-With (green) blocks. Ellipse fits are
computed at a 95% confidence interval.
in terms of performance. VR practitioners could consider any LT-
Heading when designing VR applications without being worried
about the influence on cognitive performance. However, other fac-
tors such as the transfer function [15] or the orientation of the body
segment [8] can be influenced by the design of LTs.
Similarly to the original study, we did not find an effect of
LT-Heading on time to travel and missed termination, rejecting
[RH1.2]. Our participants were one second faster on average than
in the initial study. No implications of LT-Heading need to be con-
sidered in terms of navigation performance, where performance
seems to be more affected by the transfer function (i.e., how the
speed is updated), as already observed in past studies [7, 15].
Regarding cybersickness, Lai et al. found that Head steering
may induce significantly lower SSQ scores than Hand or Torso
steering [25]. Overall, our cybersickness scores were low, andwe did
not find an effect of LT-Heading on SSQ scores, rejecting [RH3.1].
Our results might be due to hardware factors, such as using a newer
HMD (HTC Vive Pro) than theirs (HTC Vive), which provides a
higher field of view. Second, our sample may have a larger number
of younger VR-experienced users, which could explain why some
participants did not feel sick. Last, one issue might be about the SSQ
scoring, where Lai et al. only administered the SSQ after performing
the condition, without considering pre-SSQ as a baseline to calibrate,
which could create higher changes in cybersickness scores.
Table 4 shows the differences between our study sample and the
one from the original study. Both studies had the same range of
experienced VR users. Yet, our study had a better gender distribu-
tion than the original study, but it covered a higher range of age
distribution. Overall, we found similar results in terms of travel
and cognitive performance across LT, which shows that cognitive
load may not influence experienced user navigation with steering
techniques in VR.
VRST ’25, November 12–14, 2025, Montreal, QC, Canada Brument et al.,
Table 4: Sample description between both user studies.
Metric Our study Lai et .al [25]
Participants (male,female) 24 (14,10) 30 (25,5)
Age range (min-max) 21-39 18-66
Age (Mean;SD) 30.1;5.86 25.10;No SD reported
Regular Experience with VR 16 17
5.2 Consider Cognitive Load during VR
Navigation
In addition to the LT-Heading condition tested in the initial study
of Lai et al. [26], we added a control condition to assess how users
subjectively perceive cognitive load, but also whether the absence
of the CT would influence the metrics. As observed in Table 2, we
did not find differences in raw NASA-TLX scores between Head,
Hand, and Torso Steering, rejecting [NH1.2]. Yet, we showed that
every subscale and total raw scores of the NASA-TLX question-
naire were higher when performing the cognitive task (Figure 5),
confirming [NH1.3]. This additional subjective questionnaire we
administered helped us to confirm the objective measurements in
terms of cognitive performance ([RH1.1]). The LT-Heading can be
thus chosenwithout concerns regarding any cognitive load increase.
Still, we gathered some post-experiment feedback that participants
preferred more hand or torso steering, since the letters displayed
were in front of the user’s head, which was also used as a heading
direction, leading to some confusing situations at the beginning.
It is worth noticing that some techniques, such as teleportation,
may help decrease cognitive load [26] and could be considered an
alternative option if spatial awareness is not required.
Table 2 shows that users were faster to perform the travel task
and had fewer missed terminations, which confirmed [NH2.2].
While past research work showed the cognitive demands of LT dur-
ing navigation in VR [26], the medium effect sizes observed in our
study prevent us from fully claiming that, in this situation,CT had a
substantial impact on users’ performance overall. Nevertheless, VR
practitioners should carefully consider cognitive load for ecological
applications. We suggest that the repetitiveness of the task may
prevent participants from performing poorly, where a low variabil-
ity was observed in terms of time to travel between two poles. Yet,
users reported sometimes having a short break between two poles
to try to "reset" the sequence of letters to maximize the accuracy
of the memory task as instructed. As we expect VR practitioners
to be aware of the cognitive implications when designing naviga-
tion tasks, our results confirm this design guideline, where we may
expect longer execution time as the cognitive load increases [30].
Last, we noticed that SSQ Oculomotor and SSQ Total scores were
higher during CT-With than CT-Without, confirming [NH2.3]. Due
to the small and medium effect sizes and the slight differences in
scores, it is hard to conclude. The small effect on the Oculomotor
subscale might be due to the repetitiveness of the letters being
displayed, creating some blinks over time, which could induce
discomfort. Overall, cybersickness was not an issue and was mainly
mitigated through block breaks. While our work was not targeted
at mitigating cybersickness, VR practitioners should still be aware
and take care of the cognitive load and task attention correlated
with cybersickness [39].
5.3 Consider UPD during VR Navigation
We extended the replication of the study by including UPD metrics.
Unfortunately, we could not get access to the dataset from Lai
et al. [25] to compute UPD metrics and compare them with our
work, which motivated us even more to replicate the study. UPD is
still overlooked in the literature, even though it is a side effect of
navigating with virtual LTs that must be addressed. On one hand,
the accumulation of UPD can pose safety risks—such as reaching
the boundaries of the physical workspace or colliding with real-
world obstacles—and can also negatively impact user experience, as
resets may interrupt presence. Overlooking these side effects may
not be optimal if the goal is to deliver a high-quality VR navigation
experience using virtual LTs.
We had two hypotheses related to the influence of LT-Heading
and CT on UPD. We did not observe an increased UPD between
Head, Hand, and Torso steering, thus rejecting [NH4.1]. We be-
lieved that using the hand to define the future direction during the
travel task may be more confusing to participants as the hand is
not part of the top-down body reorientation mechanisms during
rotations (i.e., we need to rotate the head and the torso to turn, but
the hand position and orientation do not matter) [16]. One reason
may be that users were cautious when using the Hand steering,
where they had to point with their arm fully extended to perform
the task, and put the controller to their chest or waist to perform
similarly to the torso steering. While other factors related to the
design of LTs could influence UPD, this first study showed that the
LT-Heading does not increase or decrease UPD.
On the contrary, we noticed that the CT increased UPD, which
validated [NH4.2]. As shown in Figure 7, a higher cognitive load led
to an increase in UPD, especially on the X axis (medium size effect),
confirming that UPD spreads more on the lateral side than for-
ward and backward sides [5]. Moreover, some participants already
exceeded the minimum workspace requirements for room-scale
VR (2×1.5m for SteamVR and 2×2m for Meta Quest). This raises
concern regarding the design of LTs, as such a study in a minimal
workspace would have required interrupting participants or enforc-
ing movement boundaries through guardian systems. While we
were able to observe and report the UPD effect during short-term
interactions, developing generalizable theories about UPD remains
challenging due to the sparsity of the design space—both in terms
of influencing factors and task variety. Promising directions for
future research are to explore the long-term effects of UPD and
the usage of virtual avatars to provide better spatial coordination
between virtual and real environments [1]. Over time, users may
become more aware of UPD and adapt their movement to stay
centered within the workspace, and adapting to the cognitive load
may also reduce UPD over time. In addition, the type of CT may
also influence UPD, where considering audio-only feedback [19]
or vibrations [37] to minimize visual distractions and enable better
navigation.
Thus, we argue that UPD must be a metric more considered and
reported in studies, to understand this phenomenon better, but also
extend the concept of minimizing UPD to optimizing workspaces,
where the potential users’ drift could be taken into account to
prevent them from hitting obstacles in the workspace. Gathering
daily data of regular VR users over time would be the best way to
Higher Cognitive Load Increases Unintended Positional Drift during Navigation in Virtual Reality VRST ’25, November 12–14, 2025, Montreal, QC, Canada
investigate the relationship between VR users and their workspace,
particularly how they get accustomed to UPD.
5.4 Limitations and Future Work
This replicated study confirmed previous findings, and our exten-
sion allowed us to provide further insight about how cognitive load
and heading in steering techniques can influence UPD. Yet, future
work should investigate additional research to address our limited
knowledge regarding UPD.
First, future research should explore a broader range of tasks and
environments beyond repetitive trajectories. Since UPD is likely
task-dependent, our study limits the generalization of our find-
ings. In addition, UPD has been mostly assessed in controlled and
repetitive trajectories. We must investigate more ecological tasks
to understand how UPD occurs in VR applications.
Second, the long-term effects of UPD also need to be explored.
We believe that UPD diminishes with increased user experience
or awareness of the phenomenon. Future studies examining user
adaptation to UPD could provide valuable insights into effective
mitigation strategies.
Finally, additional human factors might influence UPD. We are
aware that our study lacks diversity regarding other demographic
factors, such as cultural background, age distribution, and prior
knowledge or experiencewith cognitive tasks. Still, in future studies,
we aim to consider gender effects and beginner users as UPD might
also be user-dependent and more likely to occur for first-time VR
users, who may be even more immersed and not notice UPD. More-
over, we would like to investigate more cybersickness-inducing
tasks to eventually observe a correlation between cybersickness
and UPD, as it remains unclear whether they could influence each
other.
6 Conclusion
Considering cognitive load during locomotion in VR is essential, as
most VR applications require simultaneously performing one-to-
many interactions and navigating. Previous VR studies have only
looked at LT factors, overlooking cognitive load. In addition, the
analysis of UPD in the literature is limited, particularly in finding
the main reasons for such a phenomenon. We were interested in
seeing whether cognitive load could influence UPD by replicating
a user study that did not analyze it. Our work aims to provide new
insights regarding the limited knowledge of UPD with LTs, par-
ticularly that a higher cognitive load will increase UPD. However,
this represents only a small step toward a comprehensive under-
standing of UPD, with limitations including its generalizability to
prolonged VR exposure and human factors such as learning effects.
Nevertheless, this work may contribute to developing adaptive
methods to decrease users’ cognitive load while navigating, which
may mitigate UPD without compromising user experience.
References
[1] Alex van den Berg, Bart de Vries, Zoë Breedveld, Annelouk van Mierlo, Marnix
Tijhuis, and Laura Marchal-Crespo. 2023. Embodiment of virtual feet correlates
with motor performance in a target-stepping task: a pilot study. Frontiers in
Virtual Reality 4 (2023), 1104638.
[2] D. A. Bowman, D. Koller, and L. F. Hodges. 1997. Travel in immersive virtual
environments: an evaluation of viewpoint motion control techniques. In Proc. of
IEEE Annual International Symposium on Virtual Reality. 45–52, 215.
[3] D. A. Bowman, D. Koller, and L. F. Hodges. 1998. Amethodology for the evaluation
of travel techniques for immersive virtual environments. Virtual Reality 3, 2
(June 1998), 120–131. https://doi.org/10.1007/BF01417673
[4] G. Bruder, P. Lubos, and F. Steinicke. 2015. Cognitive Resource Demands of
Redirected Walking. IEEE Transactions on Visualization and Computer Graphics
21, 4 (April 2015), 539–544. doi:10.1109/TVCG.2015.2391864
[5] Hugo Brument, Gerd Bruder, Maud Marchal, Anne Helene Olivier, and Ferran
Argelaguet. 2021. Understanding, modeling and simulating unintended posi-
tional drift during repetitive steering navigation tasks in virtual reality. IEEE
Transactions on Visualization and Computer Graphics 27, 11 (2021), 4300–4310.
[6] Hugo Brument, Arthur Chaminade, and Ferran Argelaguet. 2024. Influence of
Rotation Gains on Unintended Positional Drift during Virtual Steering Navigation
in Virtual Reality. In Proceedings of the 30th ACM Symposium on Virtual Reality
Software and Technology. 1–10.
[7] Hugo Brument, Anne-Hélène Olivier, Maud Marchal, and Ferran Argelaguet.
2020. Does the control law matter? characterization and evaluation of control
laws for virtual steering navigation. In ICAT-EGVE 2020-International Confer-
ence on Artificial Reality and Telexistence & Eurographics Symposium on Virtual
Environments. 1–10.
[8] Hugo Brument, Iana Podkosova, Hannes Kaufmann, Anne Hélène Olivier, and
Ferran Argelaguet. 2019. Virtual vs. physical navigation in vr: Study of gaze and
body segments temporal reorientation behaviour. In 2019 IEEE Conference on
Virtual Reality and 3D User Interfaces (VR). IEEE, 680–689.
[9] Zubin Datta Choudhary, Ferran Argelaguet, Gerd Bruder, and Greg Welch. 2025.
Inducing Unintentional Positional Drift (UPD) in Virtual Reality via Physical
Rotations and the Illusion of Leaning. In 2025 IEEE Conference on Virtual Reality
and 3D User Interfaces Abstracts and Workshops (VRW). IEEE, 1198–1199.
[10] C. Christou, A. Tzanavari, K. Herakleous, and C. Poullis. 2016. Navigation in
virtual reality: Comparison of gaze-directed and pointing motion control. In 2016
18th Mediterranean Electrotechnical Conference (MELECON). 1–6. doi:10.1109/
MELCON.2016.7495413
[11] Chris G Christou and Poppy Aristidou. 2017. Steering versus teleport locomotion
for head mounted displays. In Proc. of International Conference on Augmented
Reality, Virtual Reality and Computer Graphics. Springer, 431–446.
[12] Jeremy Clifton and Stephen Palmisano. 2020. Effects of steering locomotion and
teleporting on cybersickness and presence in HMD-based virtual reality. Virtual
Reality 24, 3 (2020), 453–468.
[13] Lisa A Elkin, Matthew Kay, James J Higgins, and Jacob O Wobbrock. 2021. An
aligned rank transform procedure for multifactor contrast tests. In The 34th
annual ACM symposium on user interface software and technology. 754–768.
[14] Franz Faul, Edgar Erdfelder, Albert-Georg Lang, and Axel Buchner. 2007. G*
Power 3: A flexible statistical power analysis program for the social, behavioral,
and biomedical sciences. Behavior research methods 39, 2 (2007), 175–191.
[15] BoYu Gao, Ziiun Mai, Huawei Tu, and Henry Been-Lirn Duh. 2021. Evaluation
of Body-centric Locomotion with Different Transfer Functions in Virtual Reality.
In 2021 IEEE Virtual Reality and 3D User Interfaces (VR). IEEE, 493–500.
[16] Renato Grasso, Yuri Ivanenko, JosephMcIntyre, Isabelle Viaud-Delmon, and Alain
Berthoz. 2000. Spatial, not temporal cues drive predictive orienting movements
during navigation: A virtual reality study. Neuroreport 11 (2000), 775–778.
[17] Jan P Gründling, Nico Feld, Daniel Zielasko, and Benjamin Weyers. 2023. Corre-
lations of Flow, Usability, Workload, and Presence with Task Performance in a
Spatially Distributed Memory Task. In International Conference on Virtual Reality
and Mixed Reality. Springer, 153–165.
[18] Sandra G. Hart. 1986. NASA Task Load Index (TLX). Volume 1.0; Computerized Ver-
sion. Technical Report. NASA. https://ntrs.nasa.gov/search.jsp?R=20000021487
[19] Omar Janeh, Nikolaos Katzakis, Jonathan Tong, and Frank Steinicke. 2019. Infinity
Walk in VR: Effects of Cognitive Load on Velocity during Continuous Long-
Distance Walking. In ACM Symposium on Applied Perception 2019 (Barcelona,
Spain) (SAP ’19). Association for Computing Machinery, New York, NY, USA,
Article 9, 9 pages. doi:10.1145/3343036.3343119
[20] Robert S. Kennedy, Norman E. Lane, Kevin S. Berbaum, and Michael G. Lilienthal.
1993. Simulator Sickness Questionnaire: An Enhanced Method for Quantifying
Simulator Sickness. The International Journal of Aviation Psychology 3, 3 (1993),
203–220.
[21] Aelee Kim, Jeong-Eun Lee, and Kyoung-Min Lee. 2023. Exploring the relative
effects of body position and locomotion method on presence and cybersickness
when navigating a virtual environment. ACM Transactions on Applied Perception
21, 1 (2023), 1–25.
[22] Donghyeon Kim, Hyeong-geon Kim, and Myungho Lee. 2021. Effects of User’s
Gaze on the Unintended Positional Drift in Walk-in-Place. In Proceedings of the
27th ACM Symposium on Virtual Reality Software and Technology. 1–2.
[23] YongMin Kim, Yushin Lee, Ilsun Rhiu, andMyung Hwan Yun. 2021. Evaluation of
locomotion methods in virtual reality navigation environments: An involuntary
position shift and task performance. International Journal of Human-Computer
Studies 155 (2021), 102691.
[24] Thomas Kosch, Jakob Karolus, Johannes Zagermann, Harald Reiterer, Albrecht
Schmidt, and PawełWWoźniak. 2023. A survey onmeasuring cognitive workload
in human-computer interaction. Comput. Surveys 55, 13s (2023), 1–39.
VRST ’25, November 12–14, 2025, Montreal, QC, Canada Brument et al.,
[25] Chengyuan Lai, Xinyu Hu, Afham Ahmed Aiyaz, Ann Segismundo, Ananya
Phadke, and Ryan P McMahan. 2021. The cognitive loads and usability of target-
based and steering-based travel techniques. IEEE Transactions on Visualization
and Computer Graphics 27, 11 (2021), 4289–4299.
[26] Chengyuan Lai and Ryan P McMahan. 2020. The cognitive load and usability of
three walking metaphors for consumer virtual reality. In 2020 IEEE International
Symposium on Mixed and Augmented Reality (ISMAR). IEEE, 627–638.
[27] Eike Langbehn, Paul Lubos, and Frank Steinicke. 2018. Evaluation of locomotion
techniques for room-scale vr: Joystick, teleportation, and redirected walking. In
Proc. of the Virtual Reality International Conference-Laval Virtual. ACM, 4.
[28] Joseph J LaViola Jr, Ernst Kruijff, Ryan P McMahan, Doug Bowman, and Ivan P
Poupyrev. 2017. 3D user interfaces: theory and practice. Addison-Wesley Profes-
sional.
[29] Tiffany Luong, Anatole Lecuyer, Nicolas Martin, and Ferran Argelaguet. 2021.
A survey on affective and cognitive VR. IEEE transactions on visualization and
computer graphics 28, 12 (2021), 5154–5171.
[30] Tiffany Luong, Nicolas Martin, Ferran Argelaguet, and Anatole Lécuyer. 2019.
Studying the Mental Effort in Virtual Versus Real Environments. In 2019 IEEE
Conference on Virtual Reality and 3D User Interfaces (VR). IEEE, 809–816.
[31] Roberto A Montano-Murillo, Patricia I Cornelio-Martinez, Sriram Subrama-
nian, and Diego Martinez-Plasencia. 2019. Drift-correction techniques for scale-
adaptive vr navigation. In Proceedings of the 32nd Annual ACM Symposium on
User Interface Software and Technology. 1123–1135.
[32] Niels Christian Nilsson, Stefania Serafin, Morten Laursen, Kasper Pedersen, Erik
Sikström, and Rolf Nordahl. 2013. Tapping-in-place: Increasing the naturalness
of immersive walking-in-place locomotion through novel gestural input. In 2013
IEEE symposium on 3D user interfaces (3DUI). IEEE, 31–38.
[33] Niels Christian Nilsson, Stefania Serafin, and Rolf Nordahl. 2013. The perceived
naturalness of virtual locomotion methods devoid of explicit leg movements. In
Proceedings of Motion on Games. 155–164.
[34] Niels Christian Nilsson, Stefania Serafin, and Rolf Nordahl. 2013. Unintended
positional drift and its potential solutions. In 2013 IEEE Virtual Reality (VR). IEEE,
121–122.
[35] Niels Christian Nilsson, Stefania Serafin, and Rolf Nordahl. 2014. A comparison
of different methods for reducing the unintended positional drift accompanying
walking-in-place locomotion. In 2014 IEEE Symposium on 3D User Interfaces
(3DUI). IEEE, 103–110.
[36] Lisa Rebenitsch and Charles Owen. 2016. Review on cybersickness in applications
and visual displays. Virtual Reality 20, 2 (June 2016), 101–125. doi:10.1007/s10055-
016-0285-9
[37] Isabel S Schiller, Carolin Breuer, Lukas Aspöck, Jonathan Ehret, Andrea Bönsch,
Torsten W Kuhlen, Janina Fels, and Sabine J Schlittmeier. 2024. A lecturer’s
voice quality and its effect on memory, listening effort, and perception in a VR
environment. Scientific Reports 14, 1 (2024), 12407.
[38] Patric Schmitz, Julian Hildebrandt, André Calero Valdez, Leif Kobbelt, and Mar-
tina Ziefle. 2018. You spin my head right round: Threshold of limited immersion
for rotation gains in redirected walking. IEEE transactions on visualization and
computer graphics 24, 4 (2018), 1623–1632.
[39] Nathan C Sepich, Angelica Jasper, Stephen Fieffer, Stephen B Gilbert, Michael C
Dorneich, and Jonathan W Kelly. 2022. The impact of task workload on cyber-
sickness. Frontiers in Virtual Reality 3 (2022), 943409.
[40] Anthony Steed, Tuukka M Takala, Daniel Archer, Wallace Lages, and Robert W
Lindeman. 2021. Directions for 3D user interface research from consumer VR
games. IEEE Transactions on Visualization and Computer Graphics 27, 11 (2021),
4171–4182.
[41] E. Suma, S. Finkelstein, M. Reid, S. Babu, A. Ulinski, and L. F. Hodges. 2010.
Evaluation of the Cognitive Effects of Travel Technique in Complex Real and
Virtual Environments. IEEE Transactions on Visualization and Computer Graphics
16, 4 (July 2010), 690–702. doi:10.1109/TVCG.2009.93
[42] E. A. Suma, S. Babu, and L. F. Hodges. 2007. Comparison of Travel Techniques in
a Complex, Multi-Level 3D Environment. In Proc. of IEEE Symposium on 3D User
Interfaces. doi:10.1109/3DUI.2007.340788
[43] E. A. Suma, S. L. Finkelstein, S. Clark, P. Goolkasian, and L. F. Hodges. 2010.
Effects of travel technique and gender on a divided attention task in a virtual
environment. In 2010 IEEE Symposium on 3D User Interfaces (3DUI). 27–34. doi:10.
1109/3DUI.2010.5444726
[44] E. A. Suma, S. L. Finkelstein, M. Reid, A. Ulinski, and L. F. Hodges. 2009. Real
Walking Increases Simulator Sickness in Navigationally Complex Virtual Envi-
ronments. In 2009 IEEE Virtual Reality Conference. 245–246. doi:10.1109/VR.2009.
4811037
[45] Nikolai von Janczewski, Jennifer Wittmann, Arnd Engeln, Martin Baumann, and
Lutz Krauß. 2021. A meta-analysis of the n-back task while driving and its effects
on cognitive workload. Transportation research part F: traffic psychology and
behaviour 76 (2021), 269–285.
[46] T. Weißker, A. Kunert, B. Fröhlich, and A. Kulik. 2018. Spatial Updating and Simu-
lator Sickness During Steering and Jumping in Immersive Virtual Environments.
In 2018 IEEE Conference on Virtual Reality and 3D User Interfaces (VR). 97–104.
doi:10.1109/VR.2018.8446620
[47] Jacob OWobbrock, Leah Findlater, Darren Gergle, and James J Higgins. 2011. The
aligned rank transform for nonparametric factorial analyses using only anova
procedures. In Proc. of the ACM SIGCHI conference on human factors in computing
systems. 143–146.
[48] M. Al Zayer, P. MacNeilage, and E. Folmer. 2018. Virtual Locomotion: a Survey.
IEEE Transactions on Visualization and Computer Graphics (2018), 1–1.
[49] Daniel Zielasko, Yuen C Law, and Benjamin Weyers. 2020. Take a look around–
the impact of decoupling gaze and travel-direction in seated and ground-based
virtual reality utilizing torso-directed steering. In 2020 IEEE Conference on Virtual
Reality and 3D User Interfaces (VR). IEEE, 398–406.