Effective Mitigation of Cognitive Load in Complex Mixed Reality Tasks
Callum Smith
a
, Karen Rafferty
b
, Vishal Sharma
c
and Eben Rainey
d
School of Electronics, Electrical Engineering and Computer Science, Queen’s University Belfast, Belfast, Northern Ireland
Keywords:
Virtual Reality, Human-Computer Interaction, Mixed Reality, Cognitive Load, Electroencephalogram,
Galvanic Skin Response, Visual Effects, Haptics.
Abstract:
Little work has been undertaken to investigate the effects of a high cognitive load in Mixed Reality (MR) head-
sets or successful mitigation techniques to reduce its related cognitive burden, especially compared to strictly
fully immersive VR settings. We explore the measurement and mitigation of cognitive load in MR environ-
ments through the deployment and analysis of a novel set of visual and haptic interventions aimed at optimising
the user’s experience and performance in complex tasks involving the use of proprioception. We conducted
a study comparing fully immersive VR against pass-through enabled MR environments, employing focused
blur, targeted lighting, targeted shadows, and haptic feedback to reduce cognitive load. Participants performed
complex motor tasks in both environments for comparative measures, measuring cognitive load through stan-
dardized subjective scales, such as the NASA Task Load Index and Likert scales, mixed with electrodermal
activity and electroencephalogram sensors. Results indicate that MR environments, augmented with tailored
visual and haptic effects, demonstrate a reduction in cognitive load compared to their non-augmented states.
These findings suggest that tailored visual effects within MR can offer a more conducive environment for task
performance through the reduction of cognitive load.
1 INTRODUCTION
Virtual Reality (VR) and Mixed Reality (MR) show
significant potential in the fields of education, health-
care, gaming, and corporate training, owing to their
interactive and immersive capabilities not found in
traditional 2d displays. Even with modern advances
in headset capabilities, there remains challenges in
managing cognitive load in these environments. (Ju-
liano et al., 2020)
Headsets allow users to engage with digital envi-
ronments in ways traditional media have not been ca-
pable of, offering quantifiable benefits in task perfor-
mance and cognitive learning. (White et al., 2024).
Despite the potential of these technologies, the
cognitive load they impose on users, especially during
complex tasks, remains an important factor that can
influence performance, retention, and engagement.
Research undertaken in this domain space, primarily
looking at fully enclosed VR environments, has ex-
plored the impacts of cognitive load on user experi-
a
https://orcid.org/0009-0002-2333-644X
b
https://orcid.org/0009-0005-6889-8587
c
https://orcid.org/0000-0001-7470-6506
d
https://orcid.org/0009-0004-1151-3116
ences. (Belani et al., 2023)(Schrader and Bastiaens,
2012)
However, while much research has explored the
cognitive impacts of fully immersive VR(Gabriel
et al., 2018)(Dan and Reiner, 2017), there remains a
notable gap in understanding how MR environments,
leveraging new technologies such as full colour pass-
through, cable-free usage, and recent software im-
provements, affect cognitive load. Especially in tasks
requiring complex decision-making and motor coor-
dination. In addition, there’s been a distinct lack of
approaches within MR research into how to measure
both quantitative and qualitative methods to integrate
effective mitigation strategies that reduce cognitive
load and its associated cognitive burden. (Slater et al.,
2022).
Cognitive load theory describes the term cognitive
load. It suggests that the human brain is limited in its
working memory capacity, with cognitive load refer-
ring to the mental effort required to process informa-
tion, with excessive cognitive load impairing learning
and task performance. (Orru and Longo, 2019) This is
pertinent in VR and MR environments in which users
generally process large amounts of visual, auditory
and haptic information simultaneously, often over-
506
Smith, C., Rafferty, K., Sharma, V. and Rainey, E.
Effective Mitigation of Cognitive Load in Complex Mixed Reality Tasks.
DOI: 10.5220/0013373700003912
Paper published under CC license (CC BY-NC-ND 4.0)
In Proceedings of the 20th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2025) - Volume 1: GRAPP, HUCAPP
and IVAPP, pages 506-517
ISBN: 978-989-758-728-3; ISSN: 2184-4321
Proceedings Copyright © 2025 by SCITEPRESS – Science and Technology Publications, Lda.

whelming sensory input, hindering task performance
and learning. (Armougum et al., 2019) (Rebenitsch
and Owen, 2016)
Cognitive load is likely to be a consistent factor
to address in headset environments as users continue
to expand into increasingly complex and mentally
demanding tasks in VR and MR, such as corporate
training, education (Merchant et al., 2014), military
and medical simulations (Qiu et al., 2022) and more
ubiquitously data visualisation (Olshannikova et al.,
2015), which requires processing increasingly com-
plex forms of information and, therefore, puts further
strain on user cognitive resources.
The interactive nature of VR/MR environments,
whilst beneficial for presence and engagement, inher-
ently introduce challenges by presenting users a high
volume of sensory input to manage simultaneously. If
this is not mitigated, cognitive overload will continue
to impair performance, learning, and decision making
in future applications.
We aim to address these challenges through the
introduction of visual and haptic effects aimed at re-
ducing cognitive load and, therefore, improving task
performance. This experimental design represents a
novel contribution to the field, comparing both MR
and VR environments while exploring the effective-
ness of multimodal sensory interventions in real-time
mixed and virtual reality scenarios.
This research presents an innovative approach to
addressing the persistent issue of cognitive overload
in MR environments in an attempt to provide insight
and methods that future development in VR and MR
applications can leverage to aid in the development of
immersive interfaces and user-centered applications.
2 RELATED WORKS
Here we review the current landscape of VR and
MR respectively, looking into recent advancements in
both hardware and software as well as recent innova-
tions in assessing cognitive load and presence, identi-
fying gaps in current VR/MR research.
Virtual and Mixed Reality: Modern innovations in
headset development have started to reach a wider and
more receptive market, bolstered by recent advances
in headsets such as the Meta Quest Pro and Meta
Quest 3. Mixed reality headsets have begun to gain
popularity, market share, and general traction within
the industry, as well as seeing new major competitors
in the domain space, such as Apple with the launch of
their flagship mixed reality headset, the Apple Vision
Pro.
With these innovations, there have arisen new op-
portunities for expanding prior VR research into MR.
Improvements in field of view, display latency, and
resolution of headsets, paired with new software de-
velopment kits for MR development from Meta, Ap-
ple, and Unity, have led to a breakthrough point for
many in the industry, with previous limitations from
headsets in terms of compute power, camera resolu-
tion, and bulky form factors of headsets seeing signif-
icant improvements.
Cognitive Load: This leap has revealed a gap in un-
derstanding the cognitive implications of using MR
in complex motor tasks. Excessive cognitive load in
these environments is a reoccurring issue, shown to
be found in strictly VR environments through work
by Juliano et al. (Juliano et al., 2020). Differing tech-
niques have shown promise in measuring cognitive
load in VR. Work from Dan et al (Dan and Reiner,
2017) investigates electroencephalogram (EEG) mea-
sures of cognitive load within VR, calculating cogni-
tive load as the ratio of the average power of frontal
theta and parietal alpha events in the occipital and
temporal lobes.
Ahmadi et al. (Ahmadi et al., 2023) compounded
these methods, combining EEG data with galvanic
skin response (GSR) to record the cognitive load of
users in a fully immersed VR setting.
We see more qualitative measures being used
in the work of Armougum et al. (Armougum
et al., 2019), measuring the impact of cognitive load
through the recording of relevant factual and con-
textual information seen by the research participants,
with a combined approach that employs the NASA
Task Load Index with electrodermal activity data. The
work of Armougum et al. suggests VR as a promising
technique for cognitive load analysis.
Cognitive load is a term used in cognitive psychol-
ogy. Early research by Fred Paas describes it as a
multidimensional concept in which mental load and
mental effort can be distinguished (Paas, 1992). It is
a construct that refers to the amount of information
our working memory can process at any given time.
Research by Sweller et al. identifies three types
of cognitive load: intrinsic, germane, and extraneous
(Sweller et al., 1998). Recent research demonstrates
that when there is a high cognitive load in VR, there
is a hindrance to learning (Juliano et al., 2020). The
same research also posits that complex motor tasks
in immersive VR increase cognitive load, decreas-
ing motor performance compared to computer screen-
based tasks. Consequently, where you find ways to re-
duce cognitive load, there is benefit in terms of mem-
ory retention (Gabriel et al., 2018), engagement (Be-
lani et al., 2023), and presence (Slater and Wilbur,
1997).
Effective Mitigation of Cognitive Load in Complex Mixed Reality Tasks
507

Cognitive load is in relationship with the positive
feelings of long-term memory function and the feel-
ing of presence felt by a user (Huang et al., 2019).
Moreover, it shows that lack of immersion, poor in-
terface quality, high cognitive load, and overexerted
mental effort can all be negative predictors of cogni-
tive learning outcomes (Merchant et al., 2014). As
such, evaluating cognitive load goes hand in hand
with measuring presence when attempting to gain
insight into improving reactions, engagement, and
learning retention within an MR space.
Presence and Cognitive Load: Presence is an im-
portant concept in VR, it refers to the psychological
state or perception of feeling in which a headset user
feels as if they are actually “there” in the virtual or
mixed environment, feeling as if you are physically
present in a nonphysical world. It is a concept pri-
marily explored through one of VR’s highest cited pa-
pers by one of the areas most influential authors Mel
Slater (Slater and Wilbur, 1997), further explored and
evolved upon in their 2009 paper which argued that
presence consists of three illusions referred to as place
illusion (feeling as if you are in the location being de-
picted in VR), the plausibility illusion,(That events in
the virtual simulations are really happening) and the
illusion of ownership over the virtual body self rep-
resented by the participant (Slater, 2009). An update
to that research, published in 2022 discusses mixed
reality in the role of presence, stating that it is more
complex than in VR due to the introduction of real-
world objects.(Slater et al., 2022)
It directly mentions a need for visual software ef-
fects such as object occlusion, real world reflective
light, virtual light, and shadows in order to preserve
presence, which is argued to have no fundamental
conceptual difference from VR. We address this gap
in research through our experimental design and re-
sults. As a high feeling of virtual presence enhances
a user’s engagement, immersion, and learning out-
comes but increases cognitive load, it is necessary to
create mitigation strategies to ensure that such an in-
crease in cognitive load does not potentially hinder
learning.
There are natural human limitations to the amount
of information items that we can process consciously
in real time, long confirmed by historical research in
the field of psychology (Miller, 1956), attempting to
process items near or beyond this limit leads to an in-
crease in cognitive strain. Reducing the amount of
items in a users focus and using visual techniques to
lead them towards objects in advance should, in the-
ory, reduce the cognitive strain, and thus cognitive
load that they experience.
This should improve learning outcomes and task
performance through immersive environments and
experiences that are cognitively manageable.
It is the researchers hypothesis that users can ex-
perience a measurable reduction in their cognitive
load in MR and VR environments through the use
of focused blur, targeted shadows, passthrough re-
lighting, and general controller-based vibration haptic
feedback when used in combination. It is further the-
orized that mixed reality is a more ideal environment
to reduce cognitive load in complex tasks than fully
immersive virtual reality.
Prior research into cognitive load in the VR space
has led to a range of methods in order to capture and
evaluate qualitative data and quantitative data, respec-
tively, from experiments.
One of the more recent influential and cited papers
in VR is that of Mahdi Azmandian’s paper on Haptic
Re-targeting (Azmandian et al., 2016). In this paper,
valuable and insightful methods of measuring pres-
ence and engagement involve the use of VR-specific
methods, such as the Witmer and Singers 21 question
presence questionnaire (Witmer and Singer, 1998); as
well as the recording of Likert scale responses (Likert,
1932).
Whilst the presence questionnaire is the appropri-
ate and standard method for qualitatively measuring
VR presence, the Likert scales are more appropri-
ate to measure cognitive load and the feeling of en-
gagement. Research into the influence of virtual pres-
ence on learning outcomes shows that higher levels of
presence are associated with increased cognitive load,
suggesting that as you’re more immersed in the vir-
tual environment the mental effort required to process
and interact with the content also rises(Schrader and
Bastiaens, 2012).
The same research also mentions that identify-
ing strategies to keep learners’ attention more on the
learning materials within, and reducing cognitive load
based on design without reducing virtual presence
was a good area of concern for researchers, conclud-
ing that the effectiveness of strategies to reduce cogni-
tive load without breaking virtual presence is an area
for deeper research. There’s also investigation into al-
ternative methods than presence questionnaire’s and
Likert scales, such as using similar subjective scales
such as the NASA Task Load Index (Hart, 1988) or
an adapted version of the 9-point symmetrical cate-
gory mental effort rating scale(Paas, 1992).
There is similarly further research on alternative
methods as seen in Kim Ouwehands paper “Measur-
ing Cognitive Load: Are There More Valid Alter-
natives to Likert Rating Scales?” (Ouwehand et al.,
2021). However, there remains a gap in combining
electrodermal, electroencephalogram, and subjective
HUCAPP 2025 - 9th International Conference on Human Computer Interaction Theory and Applications
508

scales simultaneously in mixed reality; as such, we
address this in our research through the combined ap-
proach of measuring in-software analytics, EEG and
GSR for quantitative data and using the NASA-Task
load index, Likert scales, and open user feedback for
qualitative data tracking.
3 ENVIRONMENT DESIGN
Here we go over apparatus, recording measures and
design of the virtual environment for our experiment.
3.1 Apparatus
We use a modern digital environment for the research,
developed within the Unity 2022.3.38f1 game engine,
rendered on a Meta Quest 3 headset in standalone
mode (Not requiring the pairing of a desktop PC). The
Quest 3’s standard handheld controllers were used
to facilitate interaction within the MR environment.
User’s have the ability to navigate the environment
through locomotion based steering and teleportation
based movement using the controller’s thumb-sticks,
with the direction of where the users head is facing
corresponding to the forward direction in the environ-
ment.
3.2 Virtual Environment Aim
The experiment was designed to measure cognitive
load and task performance through a motor task re-
quiring rapid responses to visual stimuli and the use
of proprioception. Within this environment, the user
presses randomly illuminated targets as swiftly as
possible with a number of visual or haptic interven-
tions chosen via a toggle-able setting within the en-
vironment. This allows for controlled manipulation
of task variables, allowing us to use different effect
combinations in order to attempt to mitigate cognitive
load experienced within the task.
3.3 Effects
To reduce extraneous cognitive load, the environment
uses a variety of visual effects. Focused blur is ap-
plied to areas outside the users focal object, ensuring
the illuminated target remains focused and blurring
virtual objects not relevant to the task. In addition, tar-
geted lights and shadows are used to further empha-
size the active target, creating a visual contrast draw-
ing attention to the desired focal object. Passthrough
re-lighting is applied, allowing digital lighting to in-
teract with the physical world seen by the participant,
projecting light and shadows onto real world surfaces
scanned via the headset. Haptic feedback, when en-
abled, is provided through controller vibrations when
targets are pressed in an aid to reinforce user actions
and helping make the experience more immersive.
The combination of visual and haptic feedback aims
to reduce distraction and thus cognitive load, allowing
users to perform better under cognitively demanding
situations.
3.4 Data Recording Measures
The Unity engine and experiment has been designed
to record response times, engagement metrics and
record results and interactions whilst the app is used
in real time. These are stored and analyzed for com-
parison with different runs of the experiment in or-
der to gleam insights into the differences in effi-
cacy of differing effects. Electrodermal and elec-
troencephalogram activity is recorded via PluxBiosig-
nals 4 channel sensor kit, with separate electro-
dermal and electroencephalogram sensors. This is
processed real time and exported post-experiment
for post-processing in PluxBiosignals “Open Signal”
software.
3.5 Virtual Environment
The environment itself consists of an application,
booted through the Meta Quest 3’s standard dash-
board interface. Upon initialisation, the user is pre-
sented with 4 distinct visual objects, a wooden board,
a large floating interactable red button that is con-
nected to the board, a floating data panel displaying
performance metrics and settings to the user and 3d
representations of their handheld controllers.
The software settings are modified through the use
of buttons on the handheld controller, allowing each
individual visual effect to be toggled on or off. In ad-
dition, there is a button to swap between passthrough
enabled MR mode and fully immersed VR mode.
To start a run of the experiment, the user physi-
cally pressed the central red button with their con-
troller, initiating a 5-second countdown before start-
ing a round of the experiment.
A round consists of 30 seconds, in a round, the
user in base difficulty will see 12 buttons spread
around the wooden board, with 11 buttons being
coloured grey to show they are in an inactive state.
The remaining button becomes a singular lit up green
button, this button is the target of most of our visual
effects and considered the focal object. It is the aim
of the user to press this green button, in which a ran-
dom button out of the 11 inactive buttons will become
Effective Mitigation of Cognitive Load in Complex Mixed Reality Tasks
509

green, and the previous green button turning to an in-
active grey button, this will increase the users score
by 1.
After 30 seconds are completed, the 12 buttons
disappear and the user is once again presented with
the familiar central red button, which can be used to
begin a new round. Difficulty level, amount of buttons
pressed, average response time and icons displaying
what effects are active are displayed on a floating data
panel to the left of the wooden board and within the
eyeline of the participant. Each round generates a
JSON file with the users performance metrics that is
saved to the internal storage of the headset for later
retrieval.
4 EXPERIMENT
This section includes details on the study design that
involves participants, methodology, effects, equip-
ment, and procedure.
Participants: In order to empirically assess the ef-
fects of cognitive load in both VR and MR, we re-
cruited participants among students and staff from
Queens University Belfast. A total of 15 participants
signed up for our study. We did not provide compen-
sation for completing the study and informed all par-
ticipants that they could withdraw voluntarily at any
time. The study was conducted after receiving ethical
approval from the university.
Prior to the task, we asked each participant using a
Likert scale questionnaire about how frequently they
use VR headsets, whether or not they agree VR has a
positive benefit to mankind, how knowledgeable they
considered themselves in the area of VR, their age
group and their gender identity.
Within the demographics of our participants, the
majority of participants are in the 25-34 age group,
representing 8 out of 15 participants, with the aver-
age experience score being 2.13 out of 5, indicating
that most participants rarely or only sometimes use
VR headsets. There is a fair gender balance of partic-
ipants with 9 Male participants, 5 Female participants
and one participant choosing “Prefer not to say”.
The data suggests a general sense of optimism
about the effects of VR on mankind in participants
and whilst perceptions do vary based on differing
VR experience, knowledge levels, gender identities
and age groups, no participant responded negatively
to the question relating to whether they believed VR
would have a positive beneficial effect on society.
Measuring Cognitive Load: This includes,
• Methods of qualitative measurement
– Cognitive Load : Likert scales, NASA Task
Load Index
– Mental Workload : NASA-TLX
– Mental Effort, difficulty & Task compre-
hension : Likert scale
– Overall Response and Additional Feed-
back : Likert Scale Questionnaire.
• Methods of quantitative measurement
– Cognitive Load : Electroencephalogram
(Dan and Reiner, 2017) (Root Mean Square
from frontal temporal lobe, µV). Electrodermal
Activity via Galvanic Skin Response (Power
Spectral Density, microsiemens/µS) (Buch-
wald et al., 2019).
The experiment investigates cognitive load and
overall performance when performing a complex
motor task in mixed reality; comparing it with
the same environment but in fully immersive VR
& then seeing how cognitive load can be reduced
through visual effects applied within the environment.
Visual effects: The details include,
• Focused Blur: A real-time blurring effect able to
blur specific objects so as to leave a “Focused ob-
ject” un-blurred, essentially blurring everything
you’re not supposed to be focusing on.
• Targeted Shadows: A real-time darkening effect,
able to darken the scene with shadows apart from
selected “Focused objects”, which remain unaf-
fected by the shadows.
• Targeted Lighting: A lighting system in which
the lights within the scene focus on the button
required to be pressed, similar to the targeted
shadows but using a virtual light system.
Non Visual Effects: This includes vibration haptic
feedback, which is vibration-based controller haptics,
with the ability to use custom profiles/haptics for
further customisation.
Equipment: Quest 3 Headset , Development PC,
Unity Engine 2022.3.30f1 , PluxBiosignals EEG
Sensor, PluxBiosignals EDA Sensor, Opensignals for
signal recording and post-processing.
Procedure: The task begins with the participant’s
pre-study questionnaire addressed in the participants’
section prior. After this questionnaire is complete,
the participant is provided with a brief training ses-
sion, involving 3 practice rounds of the task in order
to familiarise themselves with the controls and ensure
HUCAPP 2025 - 9th International Conference on Human Computer Interaction Theory and Applications
510

comfortable understanding of the task and environ-
ment at hand.
After completion of the training session, the EEG
and EDA sensors are attached. The EEG sensor con-
sists of 3 electrodes, 2 of which are attached to the
FP1 and FP2 locations using the international 10-20
electrode placement system, an internationally recog-
nized method to describe and apply the location of
scalp electrodes. The third electrode is placed at a lo-
cation with low muscular activity and is used to filter
movement-based artifacts from the output signal; this
electrode is placed behind either the left or right ear-
lobe.
The EDA sensor consists of two electrodes which
were attached to the dominant hand index and middle
finger of the participants. A brief signal test is per-
formed to ensure that the electrode and device con-
nectivity for both sensor types is valid and correct.
The mode by default is in MR mode with no visual
effects applied and we consider this our “Base” mode
for mixed reality.
Once the settings are confirmed and sensor con-
nectivity output checked, the participant is informed
that the researcher will press the record button
for the sensors, recording the sensor data to the
“OpenSignals” platform, followed by which they di-
rected to press the large red button visible within the
environment to start a 30 second round of the game
described as above in the environment design section
“E, Virtual Environment”.
Once the participants finish the round, the sensor
recording is stopped and saved as 2 files, a raw .txt file
output recording the raw values recorded by each sen-
sor suite and a .h5 file recording the converted units,
converted by the OpenSignals software to microvolts
(µV) or microsiemens ( µ S) for the EEG and EDA
sensors respectively.
Following this, the participant is asked a 2 part
per condition questionnaire, consisting of a 4 question
Likert scale and a 6 question NASA-TLX, within the
Likert scale participants are asked to rate the general
difficulty of the task, the level of mental effort during
the task, their confidence in understanding the objec-
tives of the task, and whether or not they consider the
task easier or harder than the base challenge for that
particular environment type between MR and VR.
Initial rounds of the experiment are performed in
MR mode without visual effects, followed by a round
within fully immersive VR mode, similarly without
visual effects. Subsequent rounds involved toggling 1
of 4 effect batches, each possible setting being tested
over 2 rounds, one in MR mode and one in VR. The
settings possibly chosen for toggle are the focused
blur, targeted shadows and lighting, vibration haptics,
and a full combination of each effect simultaneously.
In order to ensure that the order in which the con-
ditions presented did not introduce bias within the re-
sults, the settings chosen were counterbalanced via
randomized counterbalancing in order to ensure that
each setting was tested in different sequences between
trials. This allowed the researchers to ensure the con-
sistency of the output data.
After the participant has completed either the full
batch of available rounds, or in cases of limited par-
ticipation time, any batch of rounds including both a
MR and VR comparison, they are asked a final ques-
tionnaire as a simple Likert scale asking them to rate
the clarity of instructions provided before the task,
the perceived complexity of the game controls, the
responsiveness of the virtual environment to their ac-
tions, and their overall satisfaction with the experi-
ence. The majority of the participants responded pos-
itively to all categories in this questionnaire.
5 RESULTS
This section presents the evaluation in two parts, qual-
itative and quantitative. The details are as follows:
Qualitative Analysis: The Qualitative data from
the experiment is split into 3 categories, that being
the pre-study questionnaire, the per-condition Likert
scale, and the per-condition NASA-TLX.
Comparative analysis of the difference between
VR and MR modes without visual effects enabled
within these data revealed that participants generally
found the base MR environment without visual effects
to have a higher mental and physical demand com-
pared to its fully immersive VR counterpart, indicat-
ing an increase in cognitive load within the MR mode.
Participants also indicated higher levels of frustration
in MR mode suggesting that mixed reality integra-
tion of real world elements adds layers of cognitive
complexity to the task, this may introduce extraneous
stimuli requiring additional mental effort to concen-
trate and increasing overall cognitive load.
The necessity for the participant to discern virtual
information amongst real-world surroundings could
explain the increased cognitive and emotional bur-
den reported by participants. The mean NASA-TLX
scores and the mean response scores of the Likert
scale for VR and Mr modes and their comparisons
can be seen in Table 1.
Comparison of the Likert scale scores show that
MR mode elicits a greater feeling of general difficulty
and mental effort, this is seen similarly within the data
from the NASA-TLX scores.
Comparison of the mean NASA-TLX scores be-
Effective Mitigation of Cognitive Load in Complex Mixed Reality Tasks
511

Table 1: Comparison of Mean Scores for BASE Mode be-
tween MR and VR
Measure MR Mean VR Mean Difference
Likert Scale Scores
Q1. Difficulty 1.64 1.50 +0.14
Q2. Mental Effort 2.62 2.36 +0.26
Q3. Confidence 1.07 1.00 +0.07
Q4. Comparison N/A N/A N/A
NASA-TLX Scores
Mental Demand 7.00 4.71 +2.29
Physical Demand 5.07 4.29 +0.78
Temporal Demand 9.36 8.79 +0.57
Performance 4.50 3.29 +1.21
Effort 7.43 6.93 +0.50
Frustration 5.64 3.14 +1.50
Table 2: Comparison of Mean Scores for FULL COMBI-
NATION Mode between MR and VR
Measure MR Mean VR Mean Difference
Likert Scale Scores
Q1. Difficulty 1.27 1.43 -0.16
Q2. Mental Effort 2.00 2.00 +0.00
Q3. Confidence 1.00 1.14 -0.14
Q4. Comparison 1.93 2.36 -0.45
NASA-TLX Scores
Mental Demand 3.80 4.71 -0.91
Physical Demand 4.20 4.43 -0.23
Temporal Demand 8.20 9.43 -1.23
Performance 1.93 2.79 -0.86
Effort 5.07 5.21 -0.14
Frustration 3.27 4.50 -1.23
tween VR and MR base modes shows a clear increase
in mental, physical, and temporal demands, perfor-
mance, effort, and levels of frustration when experi-
encing the task in mixed reality.
Similar trends are measured in runs including vari-
ations of all possible visual and haptic effects, this is
to the notable exclusion of the full combination of ef-
fects, which is found to be overwhelming in fully im-
mersive VR but positively mitigating cognitive load in
mixed reality; however, on task runs with effects en-
abled, the additional question “Do you consider this
task easier or harder than the base challenge” is put
forward to the participant, with available answers be-
ing either 1) Much easier, 2) Easier, 3) No difference,
4) Harder and 5) Much harder. As such, when cal-
culating the mean scores for this question, any mean
score below 3 indicates a positive reduction in effort
compared to no effects.
It can be seen in both comparison mean scores
that the full combination of visual effects reduces the
participants perception of difficulty, and by extension,
Table 3: Comparison of Mean Scores for BLUR Mode be-
tween MR and VR
Measure MR Mean VR Mean Difference
Likert Scale Scores
Q1. Difficulty 1.90 1.75 +0.15
Q2. Mental Effort 2.67 2.29 +0.38
Q3. Confidence 1.00 1.00 +0.00
Q4. Comparison 2.80 2.75 +0.05
NASA-TLX Scores
Mental Demand 7.80 7.63 +0.17
Physical Demand 6.40 6.62 -0.22
Temporal Demand 9.70 12.00 -2.30
Performance 5.30 3.75 +1.55
Effort 8.20 7.13 +1.07
Frustration 6.60 7.13 -0.53
cognitive load within the task environment, this trend
persists in both VR and MR mode indicating the ef-
fectiveness of a combined battery of visual effects in
increasing focus and task performance. This can be
observed in Table 2.
This is found to be the case for the majority of
effects when used in isolation; however, there is a no-
table exception to this rule, that being that the use of
the targeted blur effect in MR mode seemed to nega-
tively affect difficulty and mental effort and thus be-
ing found to be cognitively hampering performance.
The data in Table 3 indicate that blur mode was
significantly more challenging for participants in MR
compared to its VR counterpart, an increased cogni-
tive load was experienced in both NASA-TLX and
Likert scale responses.
The NASA-TLX assessment reveals a surge in
mental demand, physical demand, and effort, show-
ing greater processing demand and a higher percep-
tion of physical stress. The frustration experienced
by the participants more than doubled in VR (7.13),
an increase of 3.99 compared to the base, relative to
MR (6.60), an increase of 0.96.
MR environments inherently demand a higher
cognitive load than virtual reality environments due to
the need for people to continuously process and inte-
grate information from both real and virtual worlds si-
multaneously (Juliano et al., 2020), within the NASA-
TLX results, it is reported that the blur effect con-
tributes to overwhelming the participant in MR mode,
reducing overall performance in the task.
A breakdown of mean scores for the NASA-TLX
for each individual visual effect can be seen in the
chart below, an increase in effort for the blur mode is
observed; however, this is mitigated when all effects
are applied in combination, with the full combination
of effects showing a consistent lower cognitive load in
all categories reported by participants in NASA-TLX,
HUCAPP 2025 - 9th International Conference on Human Computer Interaction Theory and Applications
512

as shown in Figure 1.
Figure 1: An illustration of MR NASA TLX.
Individual effects have different rates of efficacy
in reducing cognitive load. The primary effect with
the largest reduction is targeted shadow’s and light-
ing, with a mean of 5.8, down from the base mean
of 6.5. Although, a greater overall reduction is con-
firmed when using a full combination of each vi-
sual effect, showing the largest reduction with a mean
NASA TLX score of 4.4, this trend continues within
the Likert scale responses, as shown in Figure 2.
Figure 2: An illustration of MR LIKERT Scale comparison.
Likert scale data confirm a reduction in the per-
ception of cognitive load with lower mean scores for
every visual effect setting; this is in contrast to the
NASA-TLX data on the targeted blur setting as it
shows a positive perception of decreased cognitive
load for all settings outside the base setting as well
as a consistent below 3 average for the question relat-
ing to whether or not the participant found that mode
easier or harder than the base task.
From this we can see clearly that the full com-
bination of effects has a substantial reduction in the
perception of difficulty and mental effort. In being
asked to compare the full combination mode against
the base challenge in MR, no participants responded
harder or much harder, with 6 finding it easier, 5 find-
ing it much easier and 4 finding no difference, a clear
reduction in qualitative data-related cognitive load.
Delving into the comparison of the base mode
to full combination of effects further can be seen in
Table 4, within these data we can see in the Likert
scale data a statistical reduction in difficulty, men-
tal effort, and overall increase in confidence in ob-
jectives, as well as a significant reduction in mental
demand, temporal demand, and physical demand as
well as increased perception and confidence in per-
formance, a reduction in the perception of required
effort to achieve that performance and an overall gen-
eral feeling of lower frustration, showing clear bene-
fits in adopting visual focus aids and haptic feedback
in MR tasks.
Table 4: Comparison of Mean Scores for MR Runs
Measure Base Full Combo Difference
Likert Scale Scores
Q1. Difficulty 1.64 1.27 +0.38
Q2. Mental Effort 2.62 2 +0.62
Q3. Confidence 1.07 1 +0.07
Q4. Comparison N/A 1.93 N/A
NASA-TLX Scores
Mental Demand 7 3.8 +3.2
Physical Demand 5.07 4.2 +0.87
Temporal Demand 9.36 8.2 +1.16
Performance 4.5 1.93 +2.57
Effort 7.43 5.07 +2.36
Frustration 5.64 3.27 +2.38
Quantitative Analysis: Quantitative data from the
experiment consists of combined electroencephalo-
gram and electrodermal activity sensor data, recorded
in microvolts and microsiemens, respectively. Initial
results show for the most part similarity to the qual-
itative data, with trends across the data indicating a
reduced cognitive load for all effects and their combi-
nation in mixed reality.
Table 5: EEG average measures over all runs : Base vs Full
Combination
Measure Base Full Combo Difference
EEG (Channel 1) Measures
MEAN (µ v) 0.08 -0.0215 +0.1015
MIN (µ v) -35.85333333 -35.27541667 -0.5779
MAX (µ v) 35.83475 35.83375 +0.0010
STD (µ v) 13.89216667 12.93483333 +0.9573
RMS (µ v) 13.895 12.93566667 +0.9593
AREA (µv
2
) 9221.771833 4584.789333 +4636.9825
The electroencephalogram data are recorded in a
raw format, converted via bioplux opensignals soft-
Effective Mitigation of Cognitive Load in Complex Mixed Reality Tasks
513

ware into microvolts, and then processed into 6 met-
rics. These include the mean values, the minimum,
maximum, standard deviation, root mean square and
the area. All are in microvolts apart from the area,
being microvolts squared, as shown in Table 5.
Comparison between the mixed reality base mode
and the mixed reality mode with the full combination
of effects enabled shows a reduction in all metrics ex-
cluding the minimum, indicating that a reduced cog-
nitive load was achieved for participants, with root
mean squared values showing a reduction of 0.96, the
mean by 0.1 and the area by 4636.
Root mean squared values are used in analysis of
electroencephalogram data, they represent the magni-
tude or power of EEG signals; the higher the RMS,
the stronger the brainwave activity, allowing for di-
rect comparison of average signal power across the
different visual and haptic conditions. As microvolt-
age scales and vary between individuals, root mean
square values allow for the comparison of signals in
a normalised manner. A reduction in these values for
the full combination run in mixed reality indicates a
high probability that cognitive load has been reduced.
In contrast, in regards to our EEG signals (Fig-
ure 3), the area is the combined sum of signal values
of a period, a lowered area in the combined average
of experiment runs suggest brain activity was weaker,
less sustained or both, indicating reduced cognitive
load as conditions with a lower area involved will
have fewer or smaller peaks in signal over time.
Figure 3: Area over combined runs for EEG MR (µV
2
, N =
53).
We can see a reduction in area for the EEG data in
all experimental conditions excluding haptics. Elec-
troencephalogram signals oscillate between positive
and negative values; our area values for haptics in-
dicate high peaks and valleys in the oscillations of
the recorded signal data, however, our mean values
for haptics show a large reduction, significantly more
than any comparison mode, meaning the brain activity
recorded fluctuated more symmetrically around zero.
Figure 4: Mean over combined runs for EEG MR (µV, N =
53).
Haptic feedback (Figure 4) may produce short,
quick bursts of neural activity due to the cognitive
processing of physical vibrations felt, these bursts
could contribute to the increasing total area measured
while keeping the mean low. We can also see from the
mean values that all experiments that use individual or
combined visual and haptic effects result in reduced
brain activity and signal amplitude.
(a) Area over combined runs for EDA MR (µS
2
, N = 53).
(b) Mean over all runs for EDA MR (µS, N = 53).
Figure 5: EDA MR measurements shown as (top) Area and
(bottom) Mean over all runs.
In essence, the low mean with high area sug-
gests that the haptic feedback generates neural ac-
tivity that is found to be energetically intense, but
HUCAPP 2025 - 9th International Conference on Human Computer Interaction Theory and Applications
514

not a high continuous cognitive effort, if this is the
case, we would expect to see higher mean and aver-
age within the EDA data for haptic interventions, as
the EDA sensor records the participants’ physiologi-
cal response, we can see in the Figures 5a and 5b the
EDA data recorded that this is indeed the case.
There is a contrast between the lower mean EEG
found for the haptics condition in MR and its EDA
counterpart, suggesting that whilst the brain’s cogni-
tive response (EEG) may have been lower, the body’s
physiological response was heightened, this indicates
that vibration haptics provide participants with a more
engaging or stimulating on a sensory level without
being additionally cognitively demanding, indicating
potential benefits for increased presence and engage-
ment within mixed reality environments. These re-
sults are also found similarly in the root mean square
values over all combined runs in both EDA and EEG
data for mixed reality. All in all, the trends (Fig-
ures 6, and 7) are clear, consistent throughout the
recording measures, and suggest that sensible focus
enhancing effects in mixed reality can mitigate cog-
nitive load and burden, resulting in enhanced perfor-
mance in mixed reality tasks.
Figure 6: Root Mean Square over combined runs for EEG
MR (µV, N = 53).
Figure 7: Root Mean Square over combined runs for EDA
MR (µS, N = 53).
Earlier trends found in the qualitative data are
once again asserted in the quantitative, the full com-
bination of visual and haptic effects reduce cognitive
load more than any singular individual effect with
reductions consistently below cognitive load experi-
enced within a base non-augmented mode, although
whereas in the qualitative data the blur effect was sub-
jectively experienced by participants as being cogni-
tively overbearing, the quantitative data show it does
reduce cognitive load compared to the base modes al-
beit at a reduced rate compared to shadow and light,
haptics, and the full combination of effects.
When comparing the efficacy of modes between
mixed reality and fully immersive virtual reality, the
full combination of visual and haptics effects shows
an increased reduction in cognitive load in mixed re-
ality compared to their VR counterpart. This is sim-
ilar to the Likert scales and Nasa-TLX scores, which
also show a reduction in cognitive load experienced
by participants, as observed in Table 6.
Table 6: EEG average measures in Full Combination Con-
dition: MR vs VR.
Measure MR Full VR Full Difference
EEG (Channel 1) Measures
MEAN (µV) -0.0215 0.3015 -0.3230
MIN (µV) -35.2754 -35.8532 +0.5778
MAX (µV) 35.8338 35.8343 -0.0005
STD (µV) 12.9348 14.2459 -1.3111
RMS (µV) 12.9357 14.2543 -1.3186
AREA (µV
2
) 4584.7893 12881.8604 -8297.0711
All in all, the most consistent effect in reducing
cognitive load individually across all data measures
recorded is the targeted shadow and lighting effects,
demonstrating an effective focus effect, reducing cog-
nitive load and thus improving task performance and
outcome. Shadow and lighting when used in combi-
nation with focused blurring and haptic interventions
show a marked reduction in both measurable and sub-
jective feelings of intellectual load, mental burden and
cognitive demand.
6 LIMITATIONS
Our study investigated complex mixed reality tasks
involving the use of proprioception, cognitive pro-
cessing and fast reactions, future work will conduct
studies investigating different tasks, testing recall, in-
terpretation and comprehension of virtual data pre-
sented within mixed reality environments. Whilst our
study provides valuable insights into developing com-
fortable and manageable mixed reality environments,
the limitations in sample size provide constraints in
making generalisations out of the findings.
Effective Mitigation of Cognitive Load in Complex Mixed Reality Tasks
515

Further limitations come down to the sensitiv-
ity of EEG and EDA data to movement-based arti-
facts. Whilst the sensor set up from Plux biosignals
takes measures to remove movement-based artifacts
through the use of additional sensors and software-
based pattern recognition in post-processing, further
research could be undertaken in order to investigate
how to reduce movement artifacts in VR and MR
EEG or EDA signal requisition.
Future studies should broaden the range of tasks
focus effects are applied to and measured within, as
well as approach novel ways of inducing high focus
states of person through the integration of artificial
intelligence assistance, such as through Meta’s voice
SDK suite or take experimental measures to integrate
real-world use cases into task environments in order
to provide tangible benefits to work in the field. De-
velopment of a modern framework for cognitive load
assessment could assist in opening new avenues of
novel research, now that a variety of qualitative and
quantitative measures exist with efficacy in examin-
ing cognitive load.
7 CONCLUSION AND FUTURE
WORK
This work has examined the effectiveness of visual
and haptic interventions as a method of reducing
cognitive load in mixed reality tasks and environ-
ments. Using combined qualitative and quantitative
data gathering, we measured cognitive load and effec-
tively reduced that load in MR through the combined
use of targeted blurring, directed shadows, and light-
ing and vibration haptics.
Investigation into the individual efficacy of singu-
lar effect based interventions show positive reductions
in cognitive load for directed shadow and lighting as
well as vibration haptics, indicating that these are ap-
propriate methods in reducing the cognitive burden
experienced by MR users in complex tasks. How-
ever, in isolation, the usage of targeted blurring in MR
has been shown to be an area that requires further re-
search, with results suggesting mild quantitative ef-
ficacy but with results suggesting that participants
feel cognitively overwhelmed qualitatively. Results
indicate that whilst mixed reality users experience a
higher cognitive load comparative to fully immersive
VR users, that there are means to mitigate and reduce
this load with the correct software and hardware fo-
cus enhancing measures, usage of targeted shadow
and lighting effects, combined with vibration based
haptics and blurring of non focused objects leads to
a reduced cognitive load in mixed reality propriocep-
tion tasks, leading to experiences with enhanced task
performance for participants as well as offer a more
comfortable and easier to engage with environment.
These findings give room for further investigation. In-
vestigations into the efficacy of visual and haptic in-
terventions can be extended to different types of tasks
such as precision based tasks, memory, engagement
or retention based tasks or similar appropriate objec-
tives.
In future work our research team expects to obtain
the answer to such questions; if reductions in cogni-
tive load are able to be replicated in other types of
task, then there is benefit to improving and imple-
menting focus-enhancing effects within a variety of
potential use cases across the industry.
REFERENCES
Ahmadi, M., Bai, H., Chatburn, A., Najatabadi, M. A.,
W
¨
unsche, B. C., and Billinghurst, M. (2023). Com-
parison of physiological cues for cognitive load mea-
sures in vr. In 2023 IEEE Conference on Virtual Real-
ity and 3D User Interfaces Abstracts and Workshops
(VRW), pages 837–838.
Armougum, A., Orriols, E., Gaston-Bellegarde, A., Marle,
C. J. L., and Piolino, P. (2019). Virtual reality:
A new method to investigate cognitive load during
navigation. Journal of Environmental Psychology,
65:101338.
Azmandian, M., Hancock, M., Benko, H., Ofek, E., and
Wilson, A. D. (2016). Haptic retargeting: Dynamic
repurposing of passive haptics for enhanced virtual re-
ality experiences.
Belani, M., Singh, H. V., Parnami, A., and Singh, P. (2023).
Investigating spatial representation of learning content
in virtual reality learning environments.
Buchwald, M., Kupi
´
nski, S., Bykowski, A., Marcinkowska,
J., Ratajczyk, D., and Jukiewicz, M. (2019). Elec-
trodermal activity as a measure of cognitive load: a
methodological approach. In 2019 Signal Processing:
Algorithms, Architectures, Arrangements, and Appli-
cations (SPA), pages 175–179.
Dan, A. and Reiner, M. (2017). Eeg-based cognitive load
of processing events in 3d virtual worlds is lower than
processing events in 2d displays. International Jour-
nal of Psychophysiology, 122:75–84. Neural Patterns
of Learning, Cognitive Enhancement and Affect.
Gabriel, P., Furtado, F., Hirashima, T., and Hayashi, Y.
(2018). Reducing cognitive load during closed con-
cept map construction and consequences on reading
comprehension and retention.
Hart, S. G. (1988). Development of a multi-dimensional
workload rating scale : Results of empirical and theo-
retical research. Human Mental Workload, 1:39–183.
Huang, C., Luo, Y.-F., Yang, S., Lu, C., and Chen, A.-
S. (2019). Influence of students’ learning style,
HUCAPP 2025 - 9th International Conference on Human Computer Interaction Theory and Applications
516

sense of presence, and cognitive load on learning out-
comes in an immersive virtual reality learning en-
vironment. Journal of Educational Computing Re-
search, 58:073563311986742.
Juliano, J. M., Schweighofer, N., and Liew, S.-L. (2020).
Increased cognitive load in immersive virtual reality
during visuomotor adaptation is associated with de-
creased long-term retention and context transfer. Jour-
nal of NeuroEngineering and Rehabilitation, 19:106.
Likert, R. (1932). A Technique for the Measurement of At-
titudes. Number nos. 136-165 in A Technique for the
Measurement of Attitudes. Archives of Psychology.
Merchant, Z., Goetz, E., Cifuentes, L., Keeney-Kennicutt,
W., and Davis, T. (2014). Effectiveness of virtual
reality-based instruction on students’ learning out-
comes in k-12 and higher education: A meta-analysis.
Computers & Education, 70:29–40.
Miller, G. A. (1956). The magical number seven, plus or
minus two: Some limits on our capacity for processing
information. Psychological Review, 63(2):81–97.
Olshannikova, E., Ometov, A., Koucheryavy, Y., and Ols-
son, T. (2015). Visualizing big data with augmented
and virtual reality: challenges and research agenda.
Journal of Big Data, 2.
Orru, G. and Longo, L. (2019). The evolution of cognitive
load theory and the measurement of its intrinsic, ex-
traneous and germane loads: A review. pages 23–48.
Ouwehand, K., Kroef, A. v. d., Wong, J., and Paas, F.
(2021). Measuring cognitive load: Are there more
valid alternatives to likert rating scales? Frontiers in
Education, 6.
Paas, F. (1992). Training strategies for attaining transfer
of problem-solving skill in statistics: A cognitive-
load approach. Journal of Educational Psychology,
84:429–434.
Qiu, R., Xu, W., Wang, B., and Shen, Q. (2022). A re-
view of research on virtual reality technology based
on human-computer interaction in military. In 2022
6th Asian Conference on Artificial Intelligence Tech-
nology (ACAIT), pages 1–5.
Rebenitsch, L. and Owen, C. (2016). Review on cybersick-
ness in applications and visual displays. Virtual Real-
ity, 20.
Schrader, C. and Bastiaens, T. J. (2012). The influence
of virtual presence: Effects on experienced cognitive
load and learning outcomes in educational computer
games. Computers in Human Behavior, 28(2):648–
658.
Slater, M. (2009). Place illusion and plausibility can lead to
realistic behaviour in immersive virtual environments.
Philosophical Transactions of the Royal Society B:
Biological Sciences, 364(1535):3549–3557. PMID:
19884149.
Slater, M., Banakou, D., Beacco, A., Gallego, J., Macia-
Varela, F., and Oliva, R. (2022). A separate reality:
An update on place illusion and plausibility in virtual
reality. Frontiers in Virtual Reality, 3:914392.
Slater, M. and Wilbur, S. (1997). A Framework for Im-
mersive Virtual Environments (FIVE): Speculations
on the Role of Presence in Virtual Environments.
Presence: Teleoperators and Virtual Environments,
6(6):603–616.
Sweller, J., Van Merrienboer, J. J. G., and Paas, F. (1998).
Cognitive architecture and instructional design. Edu-
cational Psychology Review, 10:251–.
White, M., Kupin, A., Inzerillo, S., Banerjee, N. K., and
Banerjee, S. (2024). Evaluating the effectiveness of
vr classrooms as a replacement for traditional asyn-
chronous video-based learning environments. In 2024
IEEE International Conference on Artificial Intel-
ligence and eXtended and Virtual Reality (AIxVR),
pages 147–156.
Witmer, B. G. and Singer, M. J. (1998). Measuring Pres-
ence in Virtual Environments: A Presence Question-
naire. Presence: Teleoperators and Virtual Environ-
ments, 7(3):225–240.
APPENDIX
Table 7: Comparison of Mean Scores for SHADOW &
LIGHT Mode between MR and VR.
Measure MR Mean VR Mean Difference
Likert Scale Scores
Q1. Difficulty 1.33 1.10 +0.23
Q2. Mental Effort 1.78 1.70 +0.08
Q3. Confidence 1.00 1.00 0.00
Q4. Comparison 1.67 1.60 +0.07
NASA-TLX Scores
Mental Demand 5.89 5.70 +0.19
Physical Demand 5.89 5.00 +0.89
Temporal Demand 7.33 8.90 -1.57
Performance 2.78 2.30 +0.48
Effort 5.56 5.50 +0.06
Frustration 4.00 3.40 +0.60
Table 8: Comparison of Mean Scores for HAPTICS Mode
between MR and VR.
Measure MR Mean VR Mean Difference
Likert Scale Scores
Q1. Difficulty 1.44 1.22 +0.22
Q2. Mental Effort 2.00 1.67 +0.33
Q3. Confidence 1.00 1.00 0.00
Q4. Comparison 2.22 2.11 +0.11
NASA-TLX Scores
Mental Demand 7.11 6.22 +0.89
Physical Demand 6.44 5.11 +1.33
Temporal Demand 9.89 9.22 +0.67
Performance 4.00 2.33 +1.67
Effort 7.78 6.22 +1.56
Frustration 6.44 3.67 +2.77
Effective Mitigation of Cognitive Load in Complex Mixed Reality Tasks
517