Virtual Reality (2023) 27:941–966

https://doi.org/10.1007/s10055-022-00700-z

ORIGINAL ARTICLE

Optimized virtual reality‑based Method of Loci memorization

techniques through increased immersion and effective memory
palace designs: a feasibility study
Brigham Moll1 · Ed Sykes1

Received: 28 December 2020 / Accepted: 18 August 2022 / Published online: 6 October 2022
© The Author(s) 2022

Abstract
For most, an improvement in memory would always be desirable, whether from the point of view of an aging individual
with declining memory, or from the perspective of someone seeking to memorize large amounts of information in the short-
est period of time. One way for people to improve upon their memory performance is by using the Method of Loci (MoL),
a famously complex, ancient memorization technique for non-spatial information recall. With the use of virtual reality
technology, this technique can finally be easily taught to individuals for use in their daily lives. In this paper, we present
an exploration into this avenue of using MoL in virtual reality and report on the design and evaluation of our new virtual
memory palace that aims to prove the feasibility of improving upon designs from other studies to optimize memory recall
performance. An experiment was conducted to evaluate our VR MoL environment. The results from week 1 on the pre-test
(M = 62.55, SD = 24.01) and post-test (M = 82.91, SD = 15.99) memory task showed an increase in the number of words
remembered was statistically significant, t(20) = -2.34, p = 0.014 where participants were able to remember approximately
20.4% more non-spatial information, when compared to traditional memorization techniques. After a second use, participants
improved, remembering 22.2% more non-spatial information on the pre-test (M = 63.44, SD = 26.64) and post-test (M =
85.67, SD = 16.10) memory task, indicating that the increase in number of words remembered was statistically significant,
t(16) = -2.142, p = 0.024. The results suggest that the virtual memory palace experience could be optimized to help par-
ticipants learn the MoL technique with very little training time and potentially produce significant improvements in recall
performance as a result.

Keywords Virtual reality · Memory · Method of loci · Psychology · Cognitive science · Improving memory recall ·
Memorization

1 Introduction The MoL is an ancient memorization technique known by

many to be one of the most effective ways available for peo-
In this paper, we present the results from an investigation ple to memorize non-spatial information by making use of
involving the Method of Loci (MoL) memorization technique the powerful spatial memory abilities exhibited by humans
in an immersive Virtual Reality (VR) setting. Loci is the (Yates 1999). It was used by ancient Greek and Roman ora-
plural form of the Latin word for locus, which means place tors to remember lengthy portions of text. A sample visu-
(Yates 1999). alization of the technique is shown in Fig. 1. Usage of the
technique dates as far back in time as ancient myths such as
in the story of Simonides of Ceos in the 5th century (Yates
1999).
* Ed Sykes In that myth, Simonides uses the MoL to remember the
[email protected] faces of various recently deceased people by imagining each
Brigham Moll of their seats at a table of a banquet hall. The technique
[email protected] makes use of the ability for a human to remember things
1
Sheridan College, 1430 Trafalgar Road, Ontario, ON, based on places they have visited, and things they have seen,
Canada by asking the user of the technique to imagine a building in

13
Vol.:(0123456789)
942 Virtual Reality (2023) 27:941–966

Fig. 1  Depiction of a Memory

Palace from 1511 AD, by Giulio
Camillo Krokos et al. (2019)

their mind, like a palace or a place from their past. Then, giving participants the opportunity to reach out and place
they are told to ‘place’ the different things they need to objects with their actual arms, walk using their own legs,
remember throughout this virtual building. Using a path hear sounds in the virtual world and be fully immersed in
defined by the memorizer through the imagined structure, the virtual experience, we postulate that this will have a
the participant is able to remember everything s/he wanted positive impact on the effectiveness of the MoL such that
to by simply finding ‘where’ the memories were placed. it will result in better memory recall when memorizing
A significant number of studies have been conducted non-spatial information.
developing this traditional mnemonic technique into a vir- While the traditional MoL has been shown to increase
tual or augmented reality (VR or AR) experience (Fettke memory recall performance for quite some time (Yates
et al. 2019; Huttner et al. 2015; Liu et al. 2019; Huttner 1999), it has always been considered complex and requir-
et al. 2019; Krokos et al. 2019; Huttner and Robra-Bissantz ing much training for one to learn and use the technique
2017; Peeters and Segundo-Ortin 2019; Huttner et al. 2018; effectively (Mccabe 2015; Huttner et al. 2015; Huttner and
Vindenes et al. 2018; Bhandari 2019). In several papers, it Robra-Bissantz 2017; Reggente et al. 2019; Legge et al.
was shown that people remember things more effectively 2012; Huttner et al. 2019). Many attempts have been made
with the traditional MoL. Research has shown VR enhancing with and without Head Mounted Displays (HMDs) to try
the memorization process has significant potential, but more and virtually simulate the memory palace of the MoL in
research is needed in this area (Krokos et al. 2019; Reggente order to lessen cognitive load of memorizers and make the
et al. 2019; Peeters and Segundo-Ortin 2019; Huttner et al. technique more encouraging and accessible to all people.
2019; Huttner and Robra-Bissantz 2017). The use of Vir- It is however an outstanding question in the field of cogni-
tual Reality MoL has already been shown to encourage tive science, memorization, memory recall and VR as to
people and increase people’s confidence in using the tech- whether a virtual MoL technique can be impactful enough
nique (Peeters and Segundo-Ortin 2019; Perera et al. 2019; on recall performance in a short enough time for people
O’Grady and Yildirim 2019; Huttner et al. 2015). to consider seriously using the MoL in their daily lives
In this research, these ideas of exploiting VR and MoL (Huttner et al. 2015; Bhandari 2019; Huttner et al. 2019;
are taken a step further. Past studies have not, for the Putnam 2015). Increasing the use of MoL in VR would
most part, given participants full immersion into the MoL have some significant benefits such as mitigating memory
experience, so there was an aim to heighten immersion as issues caused by old age or disorders affecting cogni-
much as possible in this work. This increased immersion tion (Jjaz et al. 2019; Manivannan et al. 2019; Tuena et al.
is said to strengthen memory recall (Huttner et al. 2019; 2019; Sayma et al. 2019; Wiederhold and Riva 2019).
Krokos et al. 2019; Huttner and Robra-Bissantz 2017). By

13
Virtual Reality (2023) 27:941–966 943

1.1 Research goal et al. 2019; Sayma et al. 2019; Manivannan et al. 2019; Wie-
derhold and Riva 2019; Jjaz et al. 2019; Optale et al. 2010).
In this work, we designed a virtual memory palace for maxi- In a study on using VR environments for neurorehabilita-
mum memory recall performance based on suggestions from tion (Tuena et al. 2019), results seemed to point toward the
multiple other studies, engaged 11 participants, and evalu- conclusion that the more immersive an environment is, the
ated the effectiveness of our system. This study involved an better the environment is for spatial memory improvement
optimized VR MoL environment developed with heightened training. Sayma et al. in Sayma et al. (2019) also supports
immersion when compared to other VR MoL systems such this idea on the benefits of immersion.
as those in Huttner et al. (2015), Vindenes et al. (2018), The idea is further supported in a study on a VR mem-
Huttner et al. (2018), Liu et al. (2019), Huttner et al. (2019), ory testing tool for seniors (Jjaz et al. 2019), where results
and Krokos et al. (2019) via an Oculus Rift S with the addi- showed immersion leading to better memory recall. Optale
tion of sensors on the ankles and waist to simulate real walk- et al. in Optale et al. (2010) had a similar experiment that
ing movements (i.e., KAT VR’s ‘KAT Loco’ sensors). The led to similar observed outcomes.
system created allowed experimental information to be col- In the earliest example reviewed (Huttner and Robra-
lected regarding memory performance of the participants Bissantz 2017), recall results leaned toward VR via HMD
under differing sensory stimuli aimed to increase immersion being more effective than through a monitor display (Huttner
and thus strengthen memories in the virtual memory palace. and Robra-Bissantz 2017).
Due to limited participants, this study is to be considered In Krokos et al. (2019), virtual MoL recall performance
a feasibility study that has the goal of showing that opti- is compared between participants using a monitor display
mization of the MoL through VR to allow participants to and participants using an HMD. Through an experiment
effectively use the famously complex memorization tech- that asked participants to memorize the faces and names of
nique is in fact possible in a small amount of training time. people while only having the ability to look around and not
The study hypothesis (H0) is: ‘Participants undergoing VR move in the virtual environment, results were found to show
MoL will demonstrate better memory recall than traditional that even a small amount of immersion could improve upon
memorization techniques, and will be able to learn the MoL memory recall. The literature review included in the study
technique within a limited training time.’ also supported this conclusion.
Immersiveness is again praised as a factor for improved
memory recall in Huttner et al. (2018), where people are
2 Related work found to recall information better in virtual MoL when items
to be remembered include not only the text of a word to be
This section presents the relevant work in the context of remembered, but an image too, following the idea of dual
proposed research: Traditional MoL; Immersion Support- coding theory. It was recommended that future research
ing Recall in VR & MoL; VR Immersion; and VR Memory should delve into animating items to be placed, trying to
Palace Design Considerations. use audio cues and videos, or any other types of media, to
see how they affect memory recall performance.
2.1 Traditional MoL In Huttner et al. (2019), authors also come to the conclu-
sion that high immersiveness in environments leads to better
The Method of Loci technique involves picturing one’s self recall when using VR MoL after reviewing other research.
in a place (also known as the memory palace) that is known It is said here that a minimum of 40 words seems to be
particularly well by the memorizer. Once prepared, knowing a good number for showing the true potential of the MoL
what they want to memorize, they imagine moving through technique. There is also support from this research for the
the place in a particular path, placing things they wish to idea that longer training times are needed for virtual MoL,
remember in various places throughout the palace. After highlighting the importance of the environment being famil-
the exercise, when they wish to recall the information, the iar to participants.
memorizer merely recalls the place and walks through it the
same way again in their mind, looking to the places they 2.3 VR immersion
placed memories to remember each of them.
Examining past research in VR MoL experiments, there is
2.2 Immersion supporting recall in VR & MoL a significant number of researchers encouraging increased
immersion for better recall of information (Huttner
VR and the MoL have been shown to be useful in multiple et al. 2019; Krokos et al. 2019; Huttner and Robra-Bis-
studies for helping people recover from injuries or losses, santz 2017). Further review was done on two papers that
and overcome mental disabilities or deficiencies (Tuena investigated how to increase immersion in simulated

13
944 Virtual Reality (2023) 27:941–966

environments (Sanchez-Vives and Slater 2005; Kong et al. by participants to be useless. Rooms were all identical in
2017). Both studies expressed through experimental results size and there was no walking transition between rooms,
and review of other research that virtual body represen- so the design of the memory palace was found to be con-
tation is important. Not being able to see one’s self in a fusing by participants. Traditionally, the MoL has been
simulated environment can be strange to people, and it is seen to work best with unique environments that are non-
recommended that a virtual representation be synced with repetitive, and ones with an abundance of space between
the movements of a participant (Sanchez-Vives and Slater items to be remembered (Yates 1999).
2005; Kong et al. 2017). In Bhandari (2019), an experiment involving an envi-
When able to see one’s own hands or legs moving where ronment for participants to explore based upon the con-
they expect them to, and at the times they are expected to cepts of the MoL showed high compliance to use the
move, people feel a higher sense of ownership and agency method, supporting the idea that these VR MoL environ-
over their virtual bodies (Kong et al. 2017). This leads to ments are quite encouraging compared to traditional meth-
people feeling more like they are truly in the environment ods of learning (Bhandari 2019). Participants were given
as if it were real, and theoretically this should lead to better paths of coins to collect for exploration of the area, and
memory recall in a VR MoL scenario. asked to memorize large amounts of floating text. There
Other studies, such as Sanchez-Vives and Slater (2005), was also a 3D sound experience to further the immersive
supported 3D sound as an important factor to immersion. experience (Bhandari 2019). This simulated environment
Haptic feedback when touching virtual surfaces, realistic is shown in Fig. 2.
walking, meaningful movements of the actual body of a par- A study (Peeters and Segundo-Ortin 2019) that
ticipant, and inducing any intense emotion like fear are also reviewed various other VR MoL studies as well as more
mentioned by the authors as ways to improve upon making traditional MoL, investigated why the MoL is so effec-
the participant feel more like they are present in a virtual tive, and how they might come up with design advice for
environment. future VR MoL experiments where high memory recall
performance is the goal. The authors suggest increasing
2.4 VR memory palace design considerations immersion as much as possible, lighting up potential areas
of interest or using unique landmarks for placing things
In Liu et al. (2019), a pilot experiment is described where to be remembered in the palace, making participants use
people can walk naturally unlike most research done physical body movements to interact and place memorized
before, using an allotted space in real life that matches things as well as move about, and make it so that partici-
the size of the rooms in the virtual memory palace. They pants can choose or even better, create imagery for items to
could also reach out with controllers in each hand to inter- be remembered when tasked with memorizing them. They
act with the environment, but items to be remembered also express interest in future studies of animating things
were pre-placed throughout, so this interaction was seen to be remembered through moving objects or videos.

Fig. 2  Environment from Bhandari (2019) inspired by the MoL

13
Virtual Reality (2023) 27:941–966 945

3 Methodology environment has been shown to produce slightly better

memory recall results as opposed to other memory palace
This section describes various aspects of the methodology structure designs (Caplan et al. 2019). Also, multiple other
including: (i) Design of the Memory Palace; (ii) Materials; studies have made use of palaces designed based on the
(iii) Participants; (iv) Overview of the Experiment Design; structure of an apartment (Caplan et al. 2019; Huttner et al.
(v) High-level Architecture of the VR MoL Environment; 2019; Huttner and Robra-Bissantz 2017; Huttner et al. 2018,
(vi) Pilot Study; (vii) Main Experiment; (viii) Internal and 2015; Legge et al. 2012; Jund et al. 2016). The 3D model of
External Validity; and (ix) Analysis techniques. the apartment was visually designed and generated by Tyson
M., an architecture degree graduate. The furniture models
populating the apartment are free use license models used
3.1 Design of the memory palace for purely academic purposes, made by other creators. The
memory palace contains a kitchen, dining area, office, bath-
The simulated environment was designed according to the room, two bedrooms, a storage room, a living area, a recrea-
floor plan as shown in Fig. 3. An apartment was chosen tion room, and a balcony. Figure 4 presents two rooms in
for the palace because familiarity with the structure of an the memory palace. While an apartment’s general structure

Fig. 3  Virtual Memory Palace

Apartment Design

Fig. 4  a Bedroom of the Virtual Memory Palace Apartment; and b Balcony of the Virtual Memory Palace Apartment

13
946 Virtual Reality (2023) 27:941–966

is familiar, it has also been said traditionally that the MoL In the end, we were able to engage 11 participants (min.
should include usage of non-repetitive, distinct environments age: 14, max. age: 65, mean age: 35.9, std. dev. σ: 14.9;
for more effective use of the mnemonic technique (Yates eight males, three females; all adults had a post-secondary
1999). The apartment design was inspired by multiple high- degree [i.e., college or university]). Each participant was
end condo apartment layouts with the goal of creating an told that the experiment would involve them using a virtual
environment that looked distinct while not taking away from reality system in combination with an ancient memorization
the idea of having a familiar, building-like structure appear- technique to memorize large amounts of information, and
ance (Caplan et al. 2019). that their results would be compared against their typical
studying strategies.
3.2 Materials
3.4 Overview of the experiment design
We created the VR MoL memory palace system using Unity
2019.2 and C#. The Oculus Rift S was used extensively for This section presents an overview of the experiment design.
development, but our system can be generalized to most VR We used a quasi-experiment design with repeated meas-
headset technologies (e.g., HTC Vive, etc.). The control- ures using one group pre-tests and post-tests. Participants
lers associated with the headsets were integrated into the were exposed to two treatments, each involving a pre-test
application using the Unity XR plugin management system. and post-test, using the Method of Loci in our VR memory
All visualizations from the VR environment were able to palace environment.
be viewed on a monitor running the Unity editor, where the The experiment was conducted twice, on the same day of
researcher conducting the experiment was able to observe two consecutive weeks. This was done to see if there were
participants and record notes. Participants used the two any improvements in participant memory recall performance
touch controllers that accompany the Oculus Rift S for more from the first attempt at the experiment compared to the sec-
realistic use of their hands and arms, and they also used KAT ond, when participants would be more familiar with using
VR’s KAT Loco sensors to simulate realistic walking. The the VR MoL environment involved. Each week had entirely
sensor devices have an open-source Unity SDK for integrat- different word lists selected from the same overall pool of
ing realistic movement into software for added immersion. words discussed below. A few participants could not partici-
Lastly, the cable coming out of the HMD was hung from the pate in the second week of the experiment due to conflicts
ceiling to be out of the way of participants using Hyperkin’s in their schedules. We conducted the pre-test and post-test
Freestep VR Cable Management System. of this experiment on the same day, for each instance of the
The KAT VR’s Loco sensors attach to the participant’s experiment in weeks 1 and 2. It is possible that due to this
ankles and waist and are calibrated to ensure direct transla- there may be some interference from the prior condition
tion to the VR environment to interpret walking movements in the post-test. We hoped however that having completely
in as seamless and natural way possible. We incorporated separate word lists between the pre-test and post-test condi-
these sensors into our VR environment because there is tions would help to avoid this risk of interference.
ample evidence that suggest that by incorporating natu- Our experiment measured memory recall on words that
ral human movement (i.e., physical walking) into the VR had high imageability based on Madan et al. (2010). High
environment has shown to be an effective way to mitigate imageability (or concreteness) is present in words that are
VR cybersickness (e.g., nausea, headaches and disorienta- easy for one to imagine or visualize in their mind (Legge
tion) (Corriveau Lecavalier et al. 2018). et al. 2012). In Legge et al. (2012), using this same word
pool, high imageability words were shown to lead to bet-
3.3 Participants ter memory recall performance in a virtual memory palace
MoL scenario, as opposed to words of low imageability,
Due to the restrictions brought with the onset of COVID- and this encouraged our usage of the same word pool. For
19, convenience sampling was used to recruit participants both the pre-test and post-test phases for our experiment,
in this study. Only those that could be allowed to be in close the words were randomly selected from a total pool of 108
proximity of those running the experiment were permitted words (Madan et al. 2010). Sample word lists for pre-tests
to be included in the participant pool. The study was con- and post-tests are shown in Table 1.
ducted under an approved Research Ethics Board protocol For the pre-test, a list of words was given to each partici-
at the academic institution (REB Protocol #: 2018-12-001- pant to memorize within 15 minutes before they were asked
035). Furthermore, due to safety reasons, participants with to recall them. Each participant then entered the VR mem-
conditions such as epilepsy were excluded. The researcher ory palace and were given some time to become acquainted
was present and in close proximity to the participant at all with the virtual environment. Participants were then asked
stages of the experiment to ensure their safety and comfort. to place 30 images (i.e., see Fig. 5) in memorable locations

13
Virtual Reality (2023) 27:941–966 947

Table 1  Sample of randomly selected word list for the experiment Training Phase and the VR MoL Phase. The Experiment-
(pre-test and post-test) Manager also tracks progress of collecting collectible
Sample pre-test word list Sample post-test word list objects in the area during the Participant Training Phase.
For the Training Phase, key locations in the palace were
Bolt Roast Ankle Canal Feast Gate
chosen for collectible floating objects (Collectibles) to be
Crowd Burial Riot Helmet Rose Hammer
spawned (CollectibleSpawnPoints). These objects disap-
Drive Stable Beard Chapel Cave Limb
peared when the participant touched them with either of
Twin Toilet Tank Bishop Troops Museum
their hands. The ExperimentManager collected metrics
Cigar Scarf Sponge Cart Dwarf Barrel
such as the number of collectibles remaining to be col-
Basket Lace Tail Onion Flock Blouse
lected, and the number and location of placed objects to
Limp Flame Rubber Drum Deer Infant
be remembered in the VR MoL Phase. This time and pro-
Veil Bill Meal Salt Tongue Button
gress information was visible to participants through pop-
Stain Card Crest Autumn Disc Wound
up dismissible displays known as UserProgressDisplays
Beam Bone Devil Essay Cherry Ladies
that could be presented and dismissed with a simple button
press on a controller. A UserProgressDisplay is shown
in Fig. 7. The DataTracker records the time required by
participants to place each object or collect each collectible,
the locations of the objects placed, which images for each
object were selected, and the order in which the collecti-
bles were collected.
An ImageSelector provided participants with the ability
to select an image from three presented (see Fig. 8) to them.
This enabled the participant in the VR environment to have
choice and presumably select the image that would be most
personally memorable for them during usage of the MoL
mnemonic technique.
Depending on which specific word was being placed next,
three relevant images were presented in front of participants,
and using their hands, they could reach out and touch one of
the pictures to select it for the item to be placed. ImageSelec-
tors spawned based on wherever participants were facing,
Fig. 5  A Placeable object that represents a word on a word list to be with some distance in front of them so that each ImageSe-
remembered by a participant
lectorComponent could be seen clearly. Each component
of an ImageSelector contained a single image that was one
of their choice while attempting to implement the Method of the three choices for a given word to be remembered.
of Loci as explained to them in a traditional manner prior. Objects to be remembered that were to be placed around
After exiting the VR environment, participants were then the palace were known as Placeables (see Fig. 5). After
asked to recall the words studied in the virtual environment. the image selection process was complete for an item to be
remembered, a chosen image was placed on a blank black
3.5 High‑level architecture of the VR MoL background of a flat object with the word associated with
environment it and this new Placeable object floated in front of the par-
ticipant until interacted with. The flat slate object was held
Figure 6 presents the high-level architectural model of the physically by the user and was able to be manipulated by the
VR software system that we created to support this research. participant using the held VR controllers. Placeables were
Unity’s XR management system and Oculus’ Integration able to be picked up and manipulated naturally. Pressing
package are used to integrate HMDs into the system. A Par- and releasing the grip trigger button on the controllers was a
ticipant Avatar is made up of two gloved hands, representing natural motion just like picking up or dropping a real object,
the participant and controlled entirely by the inputs from the and so it lent more immersion to the simulation. If at any
VR headset, controllers, and the worn ankle and waist sen- point a Placeable was dropped, thrown, or placed in some
sors used for realistic walking. accidental position, as long as the placement had not been
Spawning of participant avatars to begin the experi- confirmed, participants could use a certain button press on
ment are controlled with an ExperimentManager object. their controllers to bring the Placeable back to be floating in
This system keeps track of time limits in the Participant front of them once more, like when it spawned.

13
948 Virtual Reality (2023) 27:941–966

Fig. 6  High-level architecture of the VR MoL environment

Fig. 8  ImageSelector with three ImageSelectorComponent options for

the word Apple

Fig. 7  A UserProgressDisplay showing timing and progress informa-

sound in a 3D fashion coming from the direction where it
tion to a participant
had been placed in the virtual environment relative to the
participant’s position. Once placement was confirmed, the
Once the Placeable was placed where the participant Placeable could not be moved, and a new image selec-
wished it to be, a button on the controller was used to con- tion process for the next word in the word list to memorize
firm the placement. The Placeable then played a chiming began as managed by the ExperimentManager and a newly

13
Virtual Reality (2023) 27:941–966 949

spawned ImageSelector. Upon completion of the image After this test, 30 randomly selected words of high image-
selection process, a new Placeable spawned like the pre- ability were presented to each participant from a word pool
vious, in front of the user to be grabbed and placed. The provided in Madan et al. (2010). Participants were given 15
process repeated until the list of items to be remembered minutes to memorize as many words as possible, in order
was exhausted. if possible, using any memory strategy they wished. Either
when the time was up or they felt ready, participants were
3.6 Pilot study then moved to a separate testing area where they were asked
to write down as many words as they could recall, as closely
A pilot study was conducted to test all aspects of the vir- to the original ordering as possible. Participants were moved
tual memory palace system to prepare and refine the system to a separate area to avoid testing bias, as it has been shown
accordingly for the main experiment. The main reasons for that information can be more easily recalled in the location
conducting the pilot study were to: in which it was learned (Godden and Baddeley 1975). The
written recall test had a maximum time limit of 5 minutes,
• Determine the appropriate amount of time needed to and participants were not given any feedback on their per-
complete each section of the experiment, formance that might influence other parts of the experiment.
• Determine if the virtual memory palace is too large or Words spelled incorrectly were marked as correct.
small for participants to realistically traverse and com-
plete the tasks, 3.7.2 Participant training phase in the VR environment
• Determine if the architectural layout of the palace should
be altered in some way (e.g., new pathways where there “As the MoL is known to be a complex mnemonic tech-
are none, or to remove pathways through the memory nique (Yates 1999), it was important to allow participants to
palace that existed before, etc.), become familiar with the technique and the memory palace
• Verify that the size of the word lists in the pre-test and before performing tasks in the environment. To support this
post-test phases were appropriate for the tasks given to learning, each participant was introduced to what the MoL
participants, and entailed using an explanation of the technique from Huttner
• Ensure that the process of getting a new word, selecting and Robra-Bissantz (2017), paraphrased and used by Legge
an image for it, and placing its representative object in et al. (2012), and originally taken from Yates (1999). This
the memory palace, was intuitive, and straightforward explanation is as follows from Huttner and Robra-Bissantz
to allow participants to complete their tasks quickly and 2017; Yates 1999.
efficiently (i.e., without them getting distracted from their ‘In this method, memory is established from places
goal of memorizing the items). and images. If we wish to remember an object, we
must first imagine that object as an image, and then
3.7 Main experiment place it in a location. If we wish to remember a list of
objects, then we must make a path out the many loca-
The main experiment consisted of four phases: pre-test; par- tions. The easiest way would be to imagine a familiar
ticipant training phase; VR MoL phase; and post-test, as environment and place the imagined objects inside
described in the following sections. it. Then, you can pick up the objects as you imagine
navigating the environment, thereby remembering the
object list in order.”
3.7.1 Pre‑test
After being told about the technique, participants were
Using other similar studies as a framework, participants first encouraged to ask questions. Through encouraging partici-
completed a spatial cognitive ability test consisting of five pants to use the MoL technique with direct instruction and
training mental rotation tasks and ten testing mental rota- thorough explanation, it was hoped to increase compliance
tion tasks (Shephard and Metzler 2018). The percentage in participants, which has been shown to be a problem in
of correct answers and the average time taken per mental MoL experiments with particularly older people (Verhae-
rotation task are recorded. This testing was done to collect ghen and Marcoen 1996). When participants felt confident
information that might help explain differences in participant that they understood the MoL technique, they were intro-
performance during use of the MoL technique. In another duced to the virtual reality equipment prepared to simulate
study, it was suggested that those who are better at com- a virtual memory palace.
pleting mental rotation tasks also tend to have better recall Participants were given 10 minutes to learn how to use
(Vindenes et al. 2018), so the spatial cognitive ability of the virtual reality system, so that they were aware of how
participants was deemed important. to grab and place objects, as well as how to walk around in

13
950 Virtual Reality (2023) 27:941–966

the virtual environment and use the controllers. Time was in which a given participant does not need to explore any
allotted for participants to explore each of the rooms and given area twice.
areas in the memory palace. Participants were allowed to The phase ended when either time ran out or the par-
take more than 10 minutes for this part of the experiment if ticipant felt comfortable with the memory palace and the
they felt that they needed it. Familiarity with the memory controls of the virtual reality environment.
palace to be used in the MoL has been shown to be a sig- When exploration of the environment was complete, the
nificant factor for better performance in memory recall in participant was then placed in a very short version of the
several studies (Reggente et al. 2019; Caplan et al. 2019; main VR MoL Phase of the experiment. This shortened ver-
Jund et al. 2016). It is important to note that participants sion involved only 5 words (Placeables) (none of which were
in the experiment conducted were given very little time to on the lists of words to actually memorize). This portion of
train and prepare for using this technique in comparison with the training phase was intended to introduce participants to
other studies where multiple weeks or a few days were pro- the process of selecting images and manipulating Placeable
vided for getting familiar with the MoL (Huttner et al. 2019; items that represented the words to be memorized. The pur-
Bhandari 2019; Liu et al. 2019; Huttner et al. 2018; Huttner pose of this training was to ensure that each participant met
and Robra-Bissantz 2017). the required competency level in the VR environment and
As purposeful navigation has been suggested to be more no additional learning was required for them to perform the
memorable than simply wandering the memory palace with- main experiment.
out a specific aim (Reggente et al. 2019; Bhandari 2019),
15 floating collectible items were placed around the palace
for participants to find and collect by touching them (see 3.7.3 VR MoL phase
Fig. 9). With the items being scattered throughout the pal-
ace, collecting them allowed participants to explore every Similar to the way the pre-test memorization phase was
room thoroughly. Collectibles have been successfully used conducted, participants were given 15 minutes to memorize
for exploration of memory palaces in other studies (e.g., another set of 30 words randomly chosen from the same
Reggente et al. 2019; Bhandari 2019), and they follow the pool of high imageability words, without choosing any of
principle of trailblazing (Darken and Sibert 1993; Bhandari the words used in the pre-test memorization phase. However,
2019) by creating a specific path through the memory palace unlike in the pre-test memorization phase, participants made

Fig. 9  Positions of Collectibles

throughout the memory palace

13
Virtual Reality (2023) 27:941–966 951

use of the MoL in the virtual environment they had explored 3.7.4 Post‑test
in order to memorize the new list of words they were given.
Words were given to them one at a time to present them Finally, the participants were moved to the same testing area
in a specific order like in the pre-test, to encourage memoriz- that they were tested in before, and there they were asked to
ing how they were placed in the MoL as a path through the try to recall all 30 of the words that they tried to memorize
structure from the first word placed to the last word placed. in the memory palace, in order if possible. Like the pre-test,
Keeping track of a path when using the MoL to place and the post-test took a maximum of 5 minutes, and participants
recall memories is integral to the traditional technique, did not know their results afterward. When finished with
allowing people to theoretically memorize items in their the post-test, participants were given a closing questionnaire
original order with more accuracy than with other memori- (see Appendix A). These questions were chosen due to their
zation strategies (Yates 1999). similarity to those posed to participants in other similar stud-
When a word was presented to a participant to memo- ies (Legge et al. 2012; Huttner and Robra-Bissantz 2017;
rize, the text was shown in front of them with three different Caplan et al. 2019; Reggente et al. 2019).
images above it that matched the word (see Figure 8). Par-
ticipants were able to use one hand to select an image they 3.8 Internal validity
felt they would like to use to memorize a word by simply
moving their hand into the image. The other two images To ensure that the experiment was measuring what it was
would disappear, and the chosen image would be attached intended to effectively, various measures were taken:
to a newly spawned object that looked like a floating black
slate. Under the image, the new object also had the word in • Experimental data were recorded electronically by auto-
question to memorize, written on it. The participant could mated scripts. This was done to minimize human error
then reach out and grab the black slate object (Placeable), when recording information.
and place it in a memorable location in the memory palace. • Participants were given as much time as they required
The motivation behind allowing participants to choose to get comfortable with the VR environment, movement
an image that they feel matches with each given word is to using the motion sensors, and making use of the con-
allow personalization of items to be remembered. Studies trols in the simulated memory palace. This was done to
have shown that personalization of how memorized items bring all participants up to the same level of comfort,
are displayed can lead to improved recall (Bhandari 2019; competency and to reduce any bias toward participants
Peeters and Segundo-Ortin 2019). This is something often with more VR experience. Participants started the VR
done by memory champions but rarely or never done before MoL phase of the experiment only when they were fully
in virtual MoL research studies. comfortable and ready to do so.
When a participant placed an item, it played a 3D sound • The short training period participants experienced to
from the location it was placed in, sounding somewhat like place five items in the memory palace before commenc-
the ring of a bell. Sounds were played to increase sensory ing the VR MoL phase of the experiment aimed to reduce
involvement in the task of placement. It has been shown bias toward those that learned the VR environment’s
that providing more audio, body movement requirements, interface faster than others.
tactile feedback, and other sensory information to the user • A pilot study was conducted to ensure that the main
can increase immersion and presence in a simulated environ- experiment was designed properly to address the research
ment (Sanchez-Vives and Slater 2005; Kong et al. 2017), and question.
as shown in other studies, further immersion leads to bet-
ter memory recall (Huttner et al. 2019; Krokos et al. 2019; We believe that bias was reduced through these activities
Huttner and Robra-Bissantz 2017). to ensure clear alignment between the experiment and the
All throughout the virtual simulation, participants saw a results achieved.
simulated representation of their hands when they looked
at themselves, because certain research has shown this will 3.9 External validity
further increase immersion in the environment (Sanchez-
Vives and Slater 2005) and thus, increase memory recall The results of this work can be generalized to multiple
accuracy (Huttner et al. 2019; Krokos et al. 2019; Huttner domains in which a VR-based MoL technique could be use-
and Robra-Bissantz 2017). ful to memorize non-spatial information for a given purpose.
Participants were asked to leave the simulated environ- Other studies such as Huttner et al. (2018), Huttner et al.
ment when all items had been placed and the participants (2019), Huttner et al. (2015), have discussed the potential of
felt comfortable with remembering what had been placed, or an effective VR-based MoL technique before. Some example
the time to complete the memorization process had expired. potential uses of this system could be:

13
952 Virtual Reality (2023) 27:941–966

• In universities, and other educational institutions, where words were remembered in the correct order and is computed
students need to memorize large amounts of information using the following formula (Eq. 1):
(e.g., medical school). Professors could also use this system
lev(u, o)
to help them memorize lesson material; strict_score = 1 − (1)
• to help those who have lost some of their memorization max
abilities, either through aging or other reasons. Such a sys- where lev(u, o) returns the levenshtein costs of the user input
tem could be used to maintain and/or improve their memo- sequence u and the original sequence o. The value max rep-
rization abilities, especially in a long-term capacity. The resents the maximum amount of operations that might be
MoL has been suggested to be quite effective for long-term necessary to transform any given sequence of terms into the
memory; (Huttner and Robra-Bissantz 2017; Optale et al. original one Huttner et al. (2019).
2010). Standard descriptive statistics including t tests, boxplots,
• to help people remember particularly difficult to recall and a two-way ANOVA were also computed to determine the
memories, as suggested in Dalgleish et al. (2013); and performance of participants.
• to provide a powerful, easy to learn memorization tech-
nique that removes the original complexities of the MoL 4 Findings (analysis and evaluation)
such as extensive training. Our VR MoL system could be
useful to just about anyone who does not already use highly 4.1 Mental rotation task results
effective memorization techniques in their daily lives.
The first part of the experiment participants completed was the
3.10 Analysis mental rotation tasks to measure their spatial cognitive ability.
These results are shown in Table 2. The mean performance
As done with multiple other similar studies (Mccabe 2015; across all 11 participants was 82.73% with a SD = 8.06. The
Huttner et al. 2015, 2019; Liu et al. 2019; Huttner et al. 2018; average time taken per mental rotation task was 4.90 seconds
Huttner and Robra-Bissantz 2017), originally proposed and with a SD = 2.39.
used by the study (Legge et al. 2012), memory recall perfor- No significant patterns were observed to correlate the men-
mance for the pre-tests and post-tests were calculated by deter- tal rotation task results with how well a given participant did
mining lenient and strict scores of participants. The lenient with the VR MoL phase of the experiment. In contrast, higher
score is a percentage of simply how many words the partici- performance on mental rotation tasks, and therefore better spa-
pant recalled correctly divided by the total number of words tial cognitive abilities, were observed to generally lead to better
to remember. On the other hand, the strict score is calculated recall in participants in Vindenes et al. (2018). It is hypoth-
using the levenshtein distance, an algorithm for determining esized that if this experiment had been conducted with more
how many changes must be made to a list of recalled words to participants, a pattern may have emerged to follow this trend.
make the list identical to the original list (Huttner et al. 2019).
This algorithm calculates the minimum costs of transforming 4.2 Traditional studying (pre‑test) phase results
one sequence (e.g., a string or an array of terms) into an original
one (Huttner et al. 2019). The algorithm includes three basic After completion of their mental rotation tasks, partic-
operations: replace, delete and insert. Every time the algorithm ipants were given 15 minutes to memorize a list of 30
has to use one of them, a counter increments the costs of trans- words, using a single piece of scrap paper if they wanted
formation by one. In the end, the minimum costs are returned. to draw or write anything down to help with whatever
For instance, the original sequence is table, spoon, fork, apple, mnemonic technique they decided to use. Participants
banana, while the user’s input was spoon, fork, apple, banana, almost always used all the 15 minutes allotted to them
table. In this case, the order is almost perfect except for the term for studying the list. Various studying methods were used
table. The levenshtein distance then deletes table and adds it for this phase, as indicated by the participants through
at the beginning of the sequence. Hence, two operations were the scrap paper given to them and the questionnaire at the
performed (deletion and insertion) and the cost of transforming end of the experiment where people reported what strat-
the sequence is two. The strict score is a measure of whether egies they used. Participants P4, P6, P8, P10, and P11

Table 2  Participant mental Performance (correct mental rotation task answers (%)) Time taken per mental rotation task (seconds)
rotation task performance and
timings Mean Min Max Standard devia- Mean Min Max Standard
tion (σ) deviation (σ)

82.73 70 93 8.06 4.90 2.23 9.35 2.39

13
Virtual Reality (2023) 27:941–966 953

used repetition, trying to remember by repeating the words 4.3 VR MoL studying (post‑test) phase results
to themselves out loud and by writing them down over
and over. P1, P7, and P9 created a story out of the words When participants collected each of the Collectible objects in
instead. Figure 10a, b illustrate these strategies, respec- the memory palace, the order that they collected each of these
tively. P2, P3, and P5 used hybrid approaches by making items was recorded. Figure 11 illustrates an example path that
associations between words and images, linking two words a participant (P4) took. On average, it took approximately 3
into a single combined image, or creating an acronym out minutes and 3 seconds for participants to collect all of the Col-
of different words to associate. lectibles in the memory palace during the training phase of the

Fig. 10  a Participant P6’s

repetition strategy to memorize
the word list; and b Participant
P9’s drawing of small images of
each object which provided the
framework for a story to be cre-
ated that linked all the images
together

Fig. 11  Path taken by Partici-

pant P4 during the Collectibles
phase

13
954 Virtual Reality (2023) 27:941–966

Table 3  Summary statistics Time taken to collect all Collectibles (Week 1) (seconds) Time taken to collect all Placeables (Week 1)
for time taken for participants (seconds)
to collect all Collectibles and
Placeables in Week 1 Mean Min Max Standard Mean Min Max Standard
deviation (σ) deviation (σ)

206.12 117.10 360.40 103.72 637.47 567 729.7 61.79

Table 4  Summary statistics Time taken to collect all Collectibles (Week 2) (sec- Time taken to collect all Placeables (Week 2)
for time taken for participants onds) (seconds)
to collect all Collectibles and
Placeables in Week 2 Mean Min Max Standard Mean Min Max Standard
deviation (σ) deviation (σ)

154.30 95.20 283.50 87.73 429.22 429.22 516.90 106.33

Table 5  Week 1 Lenient and Week 1 Lenient Recall Scores Week 1 Strict Recall Scores
Strict Scores
Mean Min Max Standard Mean Min Max Standard
deviation (σ) deviation (σ)

Pre-test 62.55 27.00 100.00 24.01 pre-test 39.36 3.00 90.00 32.92
Post-test 82.91 53.00 100.00 15.99 post-test 23.18 3.00 77.00 23.89

Table 6  Week 2 Lenient and Week 2 Lenient Recall Scores Week 2 Strict Recall Scores
Strict Scores
Mean Min Max Standard Mean Min Max Standard
deviation (σ) deviation (σ)

Pre-test 63.44 30.00 100.00 26.64 pre-test 42.56 10.00 100.00 30.10
Post-test 85.67 47.00 97.00 16.10 post-test 20.44 7.00 60.00 22.51

experiment. In the first week, this average time was 3 minutes 4.4 Overall mental recall results
and 26 seconds, and in the second week, it was 2 minutes and
34 seconds. Participants in the second week of the experiment Tables 5 and 6 present the lenient and strict score results
reported feeling much more comfortable with the system than for week 1 and week 2, respectively. These scores were
in the previous week and predictably took less time to complete calculated based on the lists of words written down by
the exploration of the memory palace through collecting Col- participants trying to recall the word lists they were asked
lectibles. Summary statistics are provided in Tables 3 and 4. to memorize in the pre-test and post-test phases.
On average, it took 10 minutes and 55 seconds for Figure 12a, b shows the distribution of test scores in
participants to place all of the Placeable items in the VR both pre-test and post-test for lenient and strict scoring,
MoL phase. In the second week, it took an average of 8 for the first week of the experiment, displaying a visible
minutes and 10 seconds. The decrease in time is likely due to increase in lenient scores from pre-test to post-test condi-
the participants getting more comfortable with the VR MoL tions, while displaying a decline in strict score from pre-
system, as they stated was the case after the experiments test to post-test. Similar trends, albeit more pronounced,
in the second week. Participants felt comfortable with the are presented in the boxplots in Figure 13a, b for week 2.
remaining time given to them for reviewing their placements Several t tests were conducted for lenient and strict
(the remainder of their 15 minute time limit to complete the recall scores for week 1 and 2 based on the pre-tests and
VR MoL phase), and no participant felt the need to rush the post-tests. The results from week 1 lenient recall scores on
procedure. Some participants did not need all of the allotted the pre-test (M = 62.55, SD = 24.01) and post-test (M =
time and left the environment early if they wished to. 82.91, SD = 15.99) memory task indicate that the increase

13
Virtual Reality (2023) 27:941–966 955

Fig. 12  a Week 1 Lenient Recall Scores and b Week 1 Strict Recall Scores

Fig. 13  a Week 2 Lenient Recall Scores; and b Week 2 Strict Recall Scores

in number of words remembered was statistically signifi- 4.4.2 Cohen’s d effect size for a t‑test
cant, t(20) = − 2.34, p = 0.014. The post hoc statistical
|̄x − x̄ 2 |
power for this t test was calculated which revealed 41.2% d= √ 1 ,
observed power (given the observed effect size [Cohen’s (3)
(𝜎12 + 𝜎22 )∕2
d: 0.98], probability level of 0.05, and a sample size of
11). The formulas presented in Eqs. 2, 3, 4, 5, 6, and 7 where x̄ 1 and x̄ 2 are the means of group 1 and group 2, and
were used in the calculation for the post hoc power for the 𝜎12 and 𝜎22 are the variances of group 1 and group 2.
t tests (Cohen 1988).
4.4.3 Gamma function
4.4.1 Beta function
∞

∫0
1 𝛤 (z) = tz−1 e−t dt (4)
∫0
B(x, y) = t x−1
(1 − t) y−1
dt (2)

4.4.4 Lower incomplete beta function

x

∫0
B (x; a, b) = ta−1 (1 − t)b−1 dt (5)

13
956 Virtual Reality (2023) 27:941–966

4.4.5 Noncentral t‑distribution noncentrality parameter was no statistically significant improvement. In summary,

√ the t test results indicate that in both weeks of the experi-
n1 n2
𝛿=d , (6) ment, there was a significant improvement in lenient scores
n 1 + n2 during the post-test.
A two-way ANOVA with repeated measures was also
where d is the Cohen’s d effect size, and n1 and n2 are the
performed to analyze the effect of Trial (week 1 vs. week
sample sizes for group 1 and group 2.
2) and Memorization Strategy (Traditional vs. VR MoL)
on memory recall (using lenient scoring). The following
4.4.6 Regularized lower incomplete beta function
assumptions were met for the ANOVA: (1) Independence
B(x;a, b) of variables: The two variables for testing are independent
Ix (a, b) = , (7) from each other; (2) Homoscedasticity: The variance in this
B(a, b)
two-way ANOVA was homogenous, that is, the variation
where the numerator is the lower incomplete beta function, around the mean for each set did not vary significantly for
and the denominator is the beta function. the groups; and (3) Normal distribution of variables: The
Similarly, the results from week 2 lenient recall scores participant’s scores followed a normal distribution pattern.
on the pre-test (M = 63.44, SD = 26.64) and post-test (M = The test for Homoscedasticity was conducted using Bar-
85.67, SD = 16.10) memory task indicate that the increase lett’s test for homogeneity of variances (Barlett 1937). The
in number of words remembered was statistically significant, formula is presented in Equation 8.
t(16) = -2.142, p = 0.024. The post hoc statistical power for
this t test was calculated which revealed 44.7% observed 4.4.7 Barlett’s test for homogeneity
power (given the observed effect size [Cohen’s d: 1.05],
∑k
probability level of 0.05, and a sample size of 11). (N − k)ln(Sp2 ) − i=1 (ni − 1)ln(Si2 )
However, the results from week 1 strict recall scores on B= 1 ∑k , (8)
1 + 3(k−1) ( i=1 ( n 1−1 ) − N−k
1
)
the pre-test (M = 39.36, SD = 32.92) and post-test (M = i

23.18, SD = 23.89) memory task indicate that there was

where n: the total number of observations across all groups,
not a statistically significant improvement or drop in perfor-
k: the total number of groups, ln: the ‘natural log,’ s2 : The
mance. This was also the case for week 2 strict recall scores
pooled variance, nj : The number of observations in group j,
on the pre-test (M = 42.56, SD = 30.10) and post-test (M =
s2j : The variance of group jThe results from Barlett’s test
20.44, SD = 22.51) memory task, which indicate that there

Fig. 14  Q-Q Plot

13
Virtual Reality (2023) 27:941–966 957

were B: 4.586 and the p-value of: 0.095, thus verifying the effect (RMSSE) (Steiger 2004), as presented in Equation 15,
assumption that the variances were equal across samples. the effect size was 0.727.
A Q-Q plot and a Shapiro–Wilk test were used to assess
the distribution of the participant’s scores. Figure 14 pre-
sents the Q-Q plot which visually shows a close approxi- ⎧ ⎫
⎪ 𝛥 ⎪
mation to a normal distribution. The Shapiro–Wilk test is Power = 𝜙⎨−Z1−a∕2 + � ⎬ (10)
commonly used to determine if the data is normally distrib- ⎪ 𝜎12 ∕n1 + 𝜎22 ∕n2 ⎪
⎩ ⎭
uted (Shapiro and Wilk 1965). The Shapiro–Wilk formula
is shown in Equation 9. The results were: W(472) = 0.952, � �
and p-value = 0.258 (i.e., p > 𝛼 (0.05)), indicating the par- 3
ticipant’s scores followed a normal distribution. =𝜙 −(1.96) + √ (11)
22 ∕11 + 22 ∕11
4.4.8 Shapiro–Wilk normality test
∑ = 𝜙{1.558} (12)
( ni=1 ai x(i) )2
W = ∑n , (9)
(x − x̄ )2
i=1 i = 0.94 (13)

where = 94%power (14)

x(i) is the ith order statistic;
x̄ = (x1 + ⋯ + xn )∕n is the sample mean, 4.4.9 RMSSE for measuring the ANOVA effect size
T −1
the coefficients ai are: (a1 , … , an ) = m CV , where �
√ ∑k 2
C = ��V −1 m�� = (mT V −1 V −1 m) d
d=
j=1 j
, (15)
and m = (m1 , … , mn )T k−1

The results of the ANOVA showed a significant main effect for where
uj −u x̄ j −̄x
dj = ≈ √
memorization strategy, F(1, 40) = 9.914, p = .003 indicating 𝜎 MSE
the VR MoL strategy performed well-above traditional memo-
rization techniques. Table 7 presents these results. The p-value These results indicate that memorization strategy was
for the interaction between trials (i.e., week 1 and 2) and the only factor that had a statistically significant effect on
memorization strategy (traditional and VR MoL) was 0.914. memory recall, and there were no interaction effects, that is,
This was not statistically significant at the 𝛼 = 0.05 level. The exposure to the trial in week 1 did not have a statistically sig-
p-value for trial was 0.450, which was also not statistically nificant impact on the participant’s performance in week 2.
significant at the 𝛼 = 0.05 level. The p-value for memoriza-
tion strategy was 0.003 which was statistically significant at 4.5 Qualitative findings based on the questionnaire
the 𝛼 = 0.05 level. A post hoc statistical power calculation was
performed on the ANOVA for main effect for memorization This section presents the findings from the questionnaire
strategy which revealed a 94% power for this study as shown each participant completed at the end of the experiment.
in Equations 10–14. Using the root-mean-square standardized Participants were asked to talk about their pre-test memori-
zation technique, their experience level in VR, whether they

Table 7  Two-factor ANOVA Source of Variation SS df MS F P-value F crit

with repeated measures by trial
and memorization strategy on Trial: week 1 or 2 250.568 1 250.568 0.583 0.450 4.085
memory recall
Memorization strategy: 4261.114 1 4261.114 9.914 0.003 4.085
Traditional or VR MoL
Interaction 5.114 1 5.114 0.012 0.914 4.085
Within 17191.636 40 429.791
Total 21708.432 43

13
958 Virtual Reality (2023) 27:941–966

Table 8  Summary of questionnaire results 5.2 Training in the VR MoL environment

Description %
In the first week of the experiment, participants needed some
Believed they used the MoL as instructed 100 time to get comfortable with the motion sensors from KAT
Would use the system in a real-life scenario 91 VR to learn how to walk in the simulation. In the second
Very immersed 82 week of the experiment, participants were more eager to
Very little/No experience with VR 73 move through the environment at faster speeds. Almost all
Very confident in post-test 72 participants seemed to experience increased immersion,
Very confident in pre-test 46 sometimes to the point of forgetting that they needed to walk
Used repetition to remember words in pre-test 45 in place rather than normally, leading to nearly bumping into
Had a little prior knowledge of the MoL 36 walls of the testing room.
Somewhat experienced with VR 27 During the Collectibles phase, participants seemed to
Somewhat confident in pre-test 27 move in mostly random paths, and no significant patterns
Not very confident in pre-test 27 were observed.
Created a story to remember words in pre-test 27
Used association to remember words in pre-test 27 5.3 Testing in the VR MoL environment
Not very confident in post-test 18
Somewhat immersed 18 The size of our word list was suitable for our study; however,
Somewhat confident in post-test 9 in a future study, it is recommended to follow the suggested
minimum word list size of 40 (or greater) from (Ross and
actually used the Method of Loci, their confidence in their Lawrence 1968). We selected 30 because we were concerned
answers in the pre-test versus the post-test, how immersive about nausea from the virtual reality environment if partici-
they found the environment, whether they knew about the pants were in it for too long. Fortunately, in our study, much
MoL beforehand, and whether they would use the system of this nausea seemed to be negated by the natural walking
in a real-life example to study something. The question- movements participants could use, as afforded by the KAT
naire was given out during the first week of the experiment VR motion sensors.
only. In the second week, participants were asked if they It was discovered that there were specific rooms in the
felt more comfortable using the system, to which all partici- memory palace that seemed to be either completely forgotten
pants agreed that they did. A summary of the questionnaire’s by participants or very seldom used. The least used parts of
results is presented in Table 8. the memory palace were the storage room, the smaller of the
two bedrooms, and the recreation room (refer to Figure 3).
Perhaps one reason why this happened is because the rooms
5 Discussion were not as visibly accessible from the large central space in
the apartment compared to the office, bathroom, and large
5.1 Traditional studying phase (pre‑test) bedroom. It seems that an optimal memory palace design
would have entrances to each available room always visible
Between weeks 1 and 2, some participants chose to change from the central space of the environment. This way people
the strategy they used to memorize the words. The change would be more encouraged to make use of all rooms avail-
in strategy often led to much improved results compared to able to them.
the week prior. For example, participant P3 used an acro- The image selector used in the experiment by participants
nym association technique in the first week, and achieved an to choose which image would be used with created Placeable
accuracy of 67% (lenient scoring) and in the second week, objects was a concept suggested by another study Peeters
they used an association between drawn images and the and Segundo-Ortin (2019) but entirely new to actual imple-
words that they represent, resulting in an accuracy of 93% mentation. It was well-received by participants, who enjoyed
(lenient scoring). Thus, P3’s performance improved by 26%. the degree of personalization in their Placeable items and
Other participant pre-test results also improved signifi- likely created stronger memory associations with the pic-
cantly by changing their strategies. Several mentioned after tures chosen and their paired words to memorize.
the experiment in the second week that they were inspired by Participant P4 also commented that in both weeks of
using the MoL in the first week and wanted to change their the experiment, the associations between the image and
traditional studying technique to include more connections the word given, in the virtual reality environment, helped
with pictures rather than just text like with repetition and a great deal in trying to memorize items. It was common
acronym studying strategies. among participants to say that in the recall test they remem-
bered the images of the things they placed long before they

13
Virtual Reality (2023) 27:941–966 959

remembered what word they were associated with. It is been suggested by multiple other studies such as Optale et al.
hypothesized that in an optimal scenario where participants (2010). We hypothesize this because in our own researcher
could use any image of their choosing for a given Place- observations, we noticed that participants, when recall-
able they are studying, that the effect of personalizing the ing words they were tested on from the previous week of
images on improving memory recall would become height- the experiment during the second week, believed that they
ened further. remembered more words from the VR MoL studying phase
than the traditional studying phase they had participated in
5.4 VR MoL studying phase (post‑test) the week before.
Through observing participants try to implement the
When using the Method of Loci in VR, on average, partici- Method of Loci in the memory palace on their given VR
pants had their lenient scores increase by over 20% in the MoL phases of the experiment, it became evident that there
first week of the experiment, as observed in the difference were different strategies in which one could use to complete
between the means of their pre-test and post-test lenient the task despite being told to use a specific mnemonic tech-
scores. This means that people usually remembered 20.4% nique. It is possible that the definition of the Method of Loci
more of the items from the list, and therefore their memory is perhaps too loosely defined. While almost all participants
recall was improved significantly compared to when they experienced improvement in their lenient scores when using
used traditional studying techniques to memorize words. the VR MoL environment to study, it seemed from observa-
In the second week, this lenient score improvement rose to tions that the strict score of participants depended greatly
22.2%. The t test computed on participant lenient scores on how they used the Method of Loci, in other words, what
indicated that the improvements were statistically significant was their strategy in using it. It was observed that the range
at the 𝛼 = 0.05 level. of strategies participants used fell into one of the following
It is important to note that participants were given very categories:
little time (less than half an hour), to familiarize themselves
with the MoL and the VR environment to virtually simu- 1. Associating items with each other in groups and thinking
late the technique. Nonetheless, it was shown that improved less about the environmental connections;
memory recall results can be achieved with very little train- 2. Associating items to their best locations in the environ-
ing and treatment time, at least with the small sample size ment while forgetting their order;
that was available. Perhaps if participants were given more 3. Placing items in a path through the environment without
time to become comfortable with the VR MoL system, then thinking about specifically where they are placed; and
it is postulated that there would be even further increased 4. Moving through the environment in a specific path and
performance results. Due to limited participants, it cannot be placing items in their most relevant locations along that
said without doubt that memory recall improvements were path to maintain both environmental associations and
accurate and unbiased, but it is believed that the results the order of the items.
at least show the feasibility of creating an optimized VR
MoL system to help participants learn and use the compli- These four categories of strategies are ordered from what
cated MoL technique with ease in a short time. The soft- seemed to be the least effective to the most effective with
ware developed for this study can also be used in the future regard to achieving the highest lenient and strict scores.
by other researchers to study with more participants and Participant P4 is a good example when discussing strat-
produce more reliable results that could possibly prove an egies for implementing the Method of Loci. P4 used the
improvement in memory recall is consistently experienced first strategy mentioned in the first week, then the last one
by participants using the given VR MoL system (or a further mentioned in the second week. When they changed their
optimized one). strategy, their strict and lenient score both improved sig-
There is also a point to be said about the potential nificantly. In the second week, P4 had a lenient score 17%
improvements to long-term memory as opposed to tradi- higher than the previous week, and a strict score 47% higher.
tional studying techniques. Perhaps in a future study, par- In the first week, P4 placed almost all of their items in the
ticipants could be asked if they can remember the words main room of the apartment with the living room, dining
from the pre-test and post-test phases of an attempt they room, and kitchen combined. The placements were not very
did several weeks ago. It is hypothesized that this would much inspired by their link to their placement location, as
show more favorable results for the MoL, and this idea of opposed to their link to each other. It was a way of associat-
long-term memory being improved through the MoL has ing the words together, forming them into related groups.

13
960 Virtual Reality (2023) 27:941–966

Fig. 15  Locations where

participant P4 placed their
Placeable items in the first week
of the experiment (as denoted
by the red spheres) (Color figure
online)

Also, being cramped into a single space, the participant did even stronger associations. The locations where P4 placed
not make use of the full memory palace. There was no vis- their Placeables in the second week can be seen in Fig. 16.
ible order to the placements either. The locations where P4 Participant results in this study may have become biased
placed items in the first week are shown in Fig. 15. due to a lack of counterbalancing from pre-test to post-test
However, in the second week of the experiment, par- to avoid things like learning effects and fatigue. We accepted
ticipant P4 decided to use nearly all of the rooms in the this possibility but designed the experiment as it is based
palace, walking in a specific path through the many rooms on the number of participants available to us, which was
and placing items as they walked, visibly in the order they admittedly quite small due to the pressures of COVID-19
were given to them. Rather than placing items according to and associated government restrictions. This is also why this
where they would best fit in the entire palace, instead par- study seeks only to prove the feasibility of optimizing the
ticipant P4 placed them where they seemed to best fit in the MoL through VR.
current position of their path through the memory palace. After completing our research, we came to realize that
For example, when the word ‘pole’ came up, it was placed a clear comparison between traditional MoL and our new
on a lamp that had a pole in its base next to P4’s current VR MoL environment would have produced more meaning-
location. Similarly, the word ‘smoke’ was placed off of the ful results. We believe that it is an excellent area for future
balcony’s edge when P4 walked past the balcony at the same research, even though in our time constraints, we did not
time as receiving the word. By doing this, P4 continued to believe we could sufficiently compare the two versions of the
make associations between the environment and the given MoL fairly, due to the large amount of time that is required
items, thinking carefully about the current location, while to learn the traditional MoL technique. The comparison in
still retaining the order of the items according to a walked how long it takes to learn each version of the MoL could
path. It may require more thinking and imagination to come have impactful results especially in encouraging more wide-
up with some kind of connection between a nearby location spread use of the MoL technique.
and the item to be placed (placing the word ‘bullet’ in the
bedroom for instance), but that extra thinking could lead to

13
Virtual Reality (2023) 27:941–966 961

Fig. 16  Locations where par-

ticipant P4 placed their Place-
able items in the second week
of the experiment (as denoted
by the red spheres) (Color figure
online)

5.5 Qualitative findings based on the questionnaire thinking about their position in the real world while in the
virtual one.
The qualitative findings were primarily based on the infor- All participants except one agreed that they would use the
mation participants provided in the questionnaire (see VR MoL environment system as a way to memorize things
Table 8). For example, participants were asked to rate the in a real-life situation. Participant P3 said that it was fun and
immersiveness they felt in the environment based on a scale slightly less stressful compared to traditional memorization.
from 1 to 5, where 1 was the least immersive. Two par- Participant P1, who used a MoL strategy that did not take into
ticipants rated it as a 3, three as a 4, and six as a 5. For account the order of the items, believed it would be effective as
most participants, the presence one felt in the experiment long as the memorized information did not need to be ordered.
seems to reflect how well they performed in the post-test Some participants agreed with this sentiment, while others
(as far as their lenient score) after using the VR MoL envi- believed it was effective for ordered memorization. Those who
ronment to study the word list given to them. Those who thought it was not as effective for ordered memorization did
rated the immersiveness lower (1, 2, or 3) tended to have tend to have worse strict scores compared to those who thought
worse performance when compared to those who rated the it would be effective for ordered memorization. Participant P7
immersiveness as high (4 or 5). Higher immersiveness lead- believed it would be more effective if they had a more familiar
ing to better recall is clearly shown to be the trend here, environment to use, such as a simulated version of their real
and it is a trend observed in multiple other studies such as home, or just being given more time to get comfortable with
Krokos et al. (2019), Huttner and Robra-Bissantz (2017), the current environment. Only participant P9, who had 100%
and Huttner et al. (2019). This further highlights the need on their lenient score for both pre-test and post-test in week
for a more immersive memory palace experience for better one when they did the questionnaire, stated that they would not
recall results, perhaps starting with using a frictionless plat- use the technique, saying that they did not think it was practi-
form or omnidirectional treadmill instead of motion sensors cal to use over traditional techniques. However, in the second
for more natural navigation that does not leave participants

13
962 Virtual Reality (2023) 27:941–966

week, P9 recalled 10% more with VR MoL compared to their 6.2 Limitations

traditional studying technique.
The arrival of COVID-19 led to multiple limitations in this
research. Participants were selected through convenience
6 Conclusion sampling rather than random sampling, and the number
of available participants was limited because COVID-19
6.1 Summary restrictions regarding human contact was only permitted
within small social circles. This inevitably led to possible
In this study, a new virtual memory palace simulation was bias in results due to having a small and non-random sam-
developed for use of the Method of Loci mnemonic tech- ple size. The situation also led to the inability to acquire a
nique, using virtual reality technology. This simulation frictionless platform such as one of KAT VR’s KAT Walk
took into account numerous suggestions, predictions, and products, leading to the inclusion of KAT VR’s KAT Loco
past results from other studies of virtual memory palaces worn motion sensors instead. Originally, we had planned for
and improved upon the design. With the newly designed 40+ participants from Sheridan College. We understand that
memory palace, a group of participants tested whether they lacking participants has underpowered the potential conclu-
could use the system to memorize a list of words more effec- sions of this study, but simultaneously, we also believe that
tively than traditional studying techniques. It was found the study still holds great value in showing the feasibility
that participants on average seemed to be able to remember of VR MoL optimization through various techniques, and
approximately 20.4% more words when using the virtual additionally, there is value in the VR MoL environment we
memory palace for the first time, and various insights were built that is now open-source and available to other research-
observed that could potentially further boost memory recall ers in the field.
improvements in future VR MoL studies. When participants In our study there was no control group, as we believed
used the system a second time they appeared to remember teaching traditional MoL would be too time consuming to fit
approximately 22.2% more compared to traditional studying within our time constraints. Ideally, there should be a group
techniques. With such encouraging results, it is hoped that doing no intervention and a group using the MoL without
further studies will be conducted in the area of VR MoL, the virtual reality, to assess the real benefit of using the tech-
perhaps leading to further encouragement for people to make nology compared to traditional cognitive interventions (i.e.,
use of a VR MoL tool in their daily lives for general memo- teaching memory techniques in a classroom-type setting).
rization with the help of experiments involving a greater Due to concerns about first-time users of VR getting nau-
number of participants. There is great potential for this tech- sea, as well as time constraints, the word lists that partici-
nology especially among students studying new topics and pants were tested with were limited to 30 words rather than
those whose memory recall has worsened over time. We the minimum of 40 words suggested by Ross and Lawrence
believe that this study shows that VR MoL holds a potential (1968).
that applies to everyone, offering the chance for a boost in The limitations we faced in conducting this study made
memory recall for even those not particularly in need of it, it one that seeks to prove the feasibility of MoL optimiza-
or at the very least, the ability to feasibly learn and use the tion through VR rather than proving it has been done here.
complicated MoL technique in a short time. We acknowledge that there is not enough data to make a
In the spirit of furthering science and this work, the fol- conclusion about the true effectiveness of our VR MoL envi-
lowing resources are provided: ronment on participants as a whole, and we hope that other
researchers with the opportunity to study more thoroughly
• Source code for the VR MoL memory palace: https://​ with the environment will be able to make more robust
bitbu​cket.​org/​Brigh​amMol​l1/​thesis-​exper​iment-​vr-​mol/​ conclusions.
src/​master/
• The full Computer Science Thesis from which this work 6.3 Future work
is based: https://​bitbu​cket.​org/​Brigh​amMol​l1/​t hesis-​
exper​iment-​vr-​mol/​downl​oads/ There are a number of suggestions for future work with VR
• Collectibles Functionality Demo (YouTube): https://​ MoL that can be extended from this study. Greater immer-
www.​youtu​be.​com/​watch?​v =​ FGExh​xYDQog sion has been observed that results in better recall, as is
• Placeables Functionality Demo (YouTube): https://w ​ ww.​ evident from comparisons between how well people did and
youtu​be.​com/​watch?​v =​ MNo7H​8tN0uQ their rated level of immersion in the questionnaire at the end
of the experiment. This is also a trend that has been observed
We hope this will encourage other researchers to explore in other previous studies. We suggest that in a future study
and extend our work. a frictionless platform could be used for walking around the

13
Virtual Reality (2023) 27:941–966 963

memory palace, such as KAT VR’s KAT Walk products. nearby vicinity rather than just place the object in the best
Alternatively an omnidirectional treadmill could be used, perceived location within the whole palace. Participants
but it may not feel as smooth, and may make more noise, would be guided to use a well-defined, non-overlapping
than a truly frictionless platform, leading to less immersion path through the rooms that they choose for themselves.
for the participant; however, more research is required to Another area along this train of thought is to include
confirm this. Also in the realm of studying greater immer- traditional neuropsychological tests of memory with popu-
sion, it could be useful to study if there might be a relation- lation norms to measure if there’s a correlation between
ship between higher participant presence (people feeling like the performance in virtual reality and standardized pen
they are truly in the virtual environment) and memory recall and paper cognitive tasks. Such a study may further reveal
performance improvements. the potential impact VR has in the context of improving
Another suggestion for future studies would be to increase memory recall.
the length of the word lists used to test participants to at least Another enhancement that further supports the MoL
40 words, as suggested by Ross and Lawrence (1968), if not method is to strengthen the linkage between objects in the
more than 40. A word list of 30 words was suitable for our virtual environment by imposing an order on the words
research; however, increasing this size challenge for some as they are presented. Future work could explore adding
participants to clearly see the improvements in memory a sequence number on eachPlaceable as they are placed
recall due to using the Method of Loci versus traditional (i.e., ‘1,’ ‘2,’ etc.).
studying techniques. This was not done in this thesis due to It may also prove useful to study how well someone can
time constraints and concerns about feelings of nausea in remember words from the first time they used the system,
participants, but with the combination of a more immersive several minutes, days, or weeks later, to investigate the
walking method such as a frictionless platform it is believed power of VR MoL, in terms of long-term memory. Numer-
that longer word lists could be used without issue as long as ous studies have already shown the usefulness of the MoL
time permits. In this experiment, 30 seconds per word to be for long-term memory recall, but perhaps with further
memorized seemed like a comfortable amount of time for refinement of the virtual memory palace simulation,
participants to complete the VR MoL studying phase of the improvements to long-term memory could be increased
experiment, but the time required for studying may need to far more than with a traditional MoL approach.
be investigated further.
Further studies could also investigate the effects on
recall performance by providing participants more train- A. Appendix
ing time with our VR MoL palace (e.g., weeks or months).
We believe that with more chances to explore the system A.1 Questionnaire
and get familiar with both the VR equipment and the simu-
lated apartment palace, participants could improve on their Table 9 presents the survey questionnaire that was admin-
recall scores. Throughout the extended training period, istered for the qualitative investigation portion of the
participants could also be guided with a more detailed study.
description of the MoL technique from the start of the VR
MoL phase that outlines how they should look for loca-
tions that seem to memorably match their object in their

Table 9  Questionnaire

1. What strategy did you use during the first memory recall test to memorize the words shown to you before entering virtual reality?
2. How much experience did you have in virtual reality prior to this experiment? (Rating your experience from 1 (not at all experienced) to 5
(very experienced))
3. Did you use the Method of Loci (MoL) in the second memorization test, after using virtual reality, as instructed?
4. How confident were you with your answers in the first recall test (before VR)? (Rate your confidence from 1 (not confident at all) to 5 (very
confident))
5. How confident were you with your answers in the final recall test (after VR)? (Rate your confidence from 1 (not confident at all) to 5 (very
confident))
6. How immersive did you find this environment? Or, how much did you feel you were in the virtual world? (Rate from 1 (not immersed) to 5
(very immersed))
7. Before the experiment, did you have any knowledge of what the Method of Loci (MoL) was?
8. If you had the chance to use a system like this to memorize things in a real-life situation, would you? Why, or why not?

13
964 Virtual Reality (2023) 27:941–966

A.2 Experiment protocol (Script)

Table 10 presents the protocol that was used by the

researchers in conducting the main experiment.

Table 10  Experiment protocol

1. Request participant to do five training mental rotation tasks, and ten testing rotation tasks. Record their results into a spreadsheet
2. Give 15 minutes (max) for participant to study a list of 30 words and memorize as many as possible, in order if possible. (Pre-test word list
print off)
3. Give 5 minutes (max) for participant to write down every word from the list, in order if possible, in another room. (Pre-test testing paper
print off)
4. Explain the ‘Method of Loci’ using the following description: ‘In this method, memory is established from places and images. If we wish
to remember an object, we must first imagine that object as an image, and then place it in a location. If we wish to remember a list of
objects, then we must make a path out the many locations. The easiest way would be to imagine a familiar environment and place the
imagined objects inside it. Then, you can pick up the objects as you imagine navigating the environment, thereby remembering the object
list in order’
5. Ask if the participant has any questions about the technique or if they understand. Answer any questions about how it works. Instruct them
that they will be using the technique in the post-test
6. Give roughly 10 minutes for the participant to get into the VR equipment and explore the environment. At this point, brief the participant
on what controls there are on the VR controllers, and which buttons do what in the VR environment. This way when they put the headset
on, they won’t be searching for the buttons too much. Once within the simulation, show them how to show and hide the progress display.
Instruct them to stand somewhere in the center of the play space available to them in the real world, and try to correct their position
whenever they get too close to a real wall. Have them walk around physically using the worn sensors, and collect each of the collectibles
in the environment as a way to explore it (Press ‘1’ on the keyboard to initiate this phase.), using their hands without pressing any but-
tons. Use the training object list (Press ‘2’ on the keyboard to initiate this phase.) to have them practice selecting and placing 5 Placeables
before the real test. Ensure the participant knows they are not meant to memorize these training words. Let them explore if they feel the
need to (within the 10 minute time limit), until they are ready for the final test. If they need more than 10 minutes for this, allow it if pos-
sible
7. In the test, have them place several Placeables and try to use the Method of Loci to remember all the items they place, and in what order
they were placed. Tell the participant to try to place each item in a place in the palace that you will remember well as matching with
the item. Also, make sure you pick pictures for each item that match what you think would be most memorable to you. To remember
the order of the items, you can try to remember the path you took through the palace when you placed each item. If you have any spare
time after placing all the items during the test, you can use that time to explore the environment and try to remember where you placed
everything”
8. When they are ready, begin the Placeables (main experiment) phase. Allow the participant to place every object until there are none left,
within the 15 minute time limit. Afterward, if they have spare time, they can explore and try to remember where everything is
9. When finished, press ‘9’ on the keyboard to save the participant’s results
10. Next, send the participant back to the testing room, where they will write down as many of the 30 items as they can, and if possible, in the
order from the palace, using the Method of Loci. They have 5 minutes to do this, maximum, as in the pre-test
11. Finally, have the participant answer the questionnaire for the experiment

A.3 VR controls

Table 11 presents the controls on the VR Controllers to

perform actions in the memory palace.
Table 11  VR controls

Controller Button Description

‘A’ Button Show progress menu

‘Y’ Button Confirm Placeable placement. Will freeze the Placeable in place and generate the next image selector if
there is another word to place in the word list.
‘X’ Button Return Placeable to participant (So that if a glitch happens or the Placeable gets thrown somewhere it can
quickly be brought back
Middle-finger(trigger/grip) Button Press these buttons and put the correct hand next to a Placeable to grab it, release button to drop the item.

13
Virtual Reality (2023) 27:941–966 965

A.4 Keyboard controls Darken RP, Sibert JL (1993) A toolset for navigation in virtual envi-
ronments. In: Proceedings of the 6th annual ACM symposium on
User interface software and technology, pp 157–165
Table 12 presents the keyboard controls for the experiment. Fettke P, Raso R, Lahann J, Loos P (2019) Walkable graph: An immer-
sive augmented reality interface for performing the memory pal-
ace method. In: AMCIS 2019 proceedings
Table 12  Keyboard controls Godden DR, Baddeley AD (1975) Context-dependent memory in
two natural environments: on land and underwater. Br J Psychol
Key Description
66(3):325–331
1 Collectibles phase Huttner J-P, Robra-Bissantz S (2017) An immersive memory palace:
supporting the method of loci with virtual reality. In: Association
2 Training Placeable phase for information systems AIS electronic library
3 Testing Placeable phase Huttner J-P, Pfeiffer D, Robra-Bissantz S (2018) Imaginary versus
9 Save participant data virtual loci: evaluating the memorization accuracy in a virtual
0 End current experiment phase memory palace. In: Proceedings of the 51st Hawaii international
conference on system sciences
Huttner J-P, Qian Z, Robra-Bissantz S (2019) A virtual memory palace
and the user’s awareness of the method of loci. In: AIS e-Library
research papers, 05
Acknowledgements We would like to thank our family and friends Huttner J-P, Robbert K, Robra-Bissantz S (2019) Immersive ars memo-
who helped with gathering experimental results during the unprec- ria: evaluating the usefulness of a virtual memory palace. In: Pro-
edented time of COVID-19, and specifically, Tyson M., for creating the ceedings of the 52nd Hawaii international conference on system
3D architecture required for the apartment used in the virtual memory sciences
palace. Jjaz K, Ahmadpour N, Naismith SL, Calvo RA (2019) An immersive
virtual reality platform for assessing spatial navigation memory
Declarations in predementia screening: feasibility and usability study. JMIR
Ment Health 6(9):e13887
Conflict of interest The authors declare that they have no conflict of Jund T, Capobianco A, Larue F (2016) Impact of frame of reference on
interest. memorization in virtual environments. In: 2016 IEEE 16th inter-
national conference on advanced learning technologies (ICALT),
Open Access This article is licensed under a Creative Commons Attri- pp 533–537
bution 4.0 International License, which permits use, sharing, adapta- Kong G, He K, Wei K (2017) Sensorimotor experience in virtual reality
tion, distribution and reproduction in any medium or format, as long enhances sense of agency associated with an avatar. Conscious
as you give appropriate credit to the original author(s) and the source, Cogn 52:115–124
provide a link to the Creative Commons licence, and indicate if changes Krokos E, Plaisant C, Varshney A (2019) Virtual memory palaces:
were made. The images or other third party material in this article are immersion aids recall. Virtual Real 23(1):1–15
included in the article's Creative Commons licence, unless indicated Legge E, Madan C, Ng E, Caplan J (2012) Building a memory palace
otherwise in a credit line to the material. If material is not included in in minutes: equivalent memory performance using virtual versus
the article's Creative Commons licence and your intended use is not conventional environments with the method of loci. Acta Psychol
permitted by statutory regulation or exceeds the permitted use, you will 141(3):380–390
need to obtain permission directly from the copyright holder. To view a Liu AC, Lee BH, Kopper R (2019) Towards a virtual memory palace.
copy of this licence, visit http://​creat​iveco​mmons.​org/​licen​ses/​by/4.​0/. In: 2019 IEEE conference on virtual reality and 3D user interfaces
(VR), pp 1046–1047
Madan CR, Glaholt MG, Caplan JB (2010) The influence of item prop-
erties on association-memory. J Mem Lang 63(1):46–63
References Manivannan M, Al-Amri M, Postans LJ, Westacott W, Gray M, Zaben
M (2019) The effectiveness of virtual reality interventions for
Barlett MS (1937) Properties of sufficiency and statistical tests. Proc improvement of neurocognitive performance after traumatic brain
R Stat Soc Ser A 160:268–282 injury. J Head Trauma Rehabili 34(2):225
Bhandari I, Pal S (2019) Designing a virtual environment in vr Mccabe J (2015) Location, location, location! demonstrating the mne-
space using memory enhancement techniques: introducing a monic benefit of the method of loci. Teach Psychol 42(2):169–173
vr approach to gdpr awareness training. MSc thesis, University O’Grady T, Yildirim C (2019) The potential of spatial computing to
of Dhaka augment memory: Investigating recall in virtual memory palaces.
Caplan JB, Legge ELG, Cheng B, Madan CR (2019) Effective- In: Constantine S (ed) HCI International 2019-Posters, pp 414–
ness of the method of loci is only minimally related to factors 422, Cham. Springer International Publishing
that should influence imagined navigation. Q J Exp Psychol Optale G, Urgesi C, Busato V, Marin S, Piron L, Priftis K, Gamberini
72(10):2541–2553 L, Capodieci S, Bordin A (2010) Controlling memory impair-
Cohen J (1988) Statistical power analysis for the behavioral sciences, ment in elderly adults using virtual reality memory training: a
2nd edn. Lawrence Earlbaum Associates, Hillsdale randomized controlled pilot study. Neurorehabil Neural Repair
Corriveau LN, Ouellet Émilie BB, Belleville S (2018) Use of immer- 24(4):348–357 (PMID: 19934445)
sive virtual reality to assess episodic memory: a validation study Peeters A, Segundo-Ortin M (2019) Misplacing memories? an enac-
in older adults. Neuropsychol Rehabili 30(3):462–480 tive approach to the virtual memory palace. Conscious Cogn
Dalgleish T, Navrady L, Bird E, Hill E, Dunn BD, Golden A-M 76:102834
(2013) Method-of-loci as a mnemonic device to facilitate access Perera RM, Priyanga EAI, DeSilva BP, Karunarathne KRR, Jayarathna
to self-affirming personal memories for individuals with depres- HMPP, Senanayake SHD (2019) A customizable virtual reality
sion. Clini Psychol Sci 1(2):156–162 application for enhancement of method of loci. UWU eRepository

13
966 Virtual Reality (2023) 27:941–966

Putnam AL (2015) Mnemonics in education: current research and Tuena M, Cipresso SR (2019) Neurorehabilitation of spatial memory
applications. Transl Issues Psychol Sci 1(2):130 using virtual environments: a systematic review. J Clini Med
Reggente N, Essoe JKY, Baek HY, Rissman J (2019) The method of 8(10):1516
loci in virtual reality: explicit binding of objects to spatial contexts Verhaeghen P, Marcoen A (1996) On the mechanisms of plasticity in
enhances subsequent memory recall. J Cognit Enhanc 5:448 young and older adults after instruction in the method of loci:
Ross J, Lawrence KA (1968) Some observations on memory artifice. evidence for an amplification model. Psychol Aging 11:164–178
Psychonomic Sci 13(2):107–108 Vindenes J, de Gortari AO, Wasson B (2018) Mnemosyne: adapting
Sanchez-Vives MV, Slater M (2005) From presence to consciousness the method of loci to immersive virtual reality. In: International
through virtual reality. Nat Rev Neurosci 6(4):332–339 conference on augmented reality, virtual reality and computer
Sayma M, Tuijt R, Cooper C, Walters K (2019) Are we there yet? graphics, pp 205–213. Springer
immersive virtual reality to improve cognitive function in demen- Wiederhold BK, Riva G (2019) Virtual reality therapy: emerging topics
tia and mild cognitive impairment. The Gerontologist 10:gnz132 and future challenges. Cyberpsychol Behav Soc Netw 22(1):5508
Shapiro SS, Wilk MB (1965) An analysis of variance test for normality Yates FA (1999) Art of memory, vol 3. Routledge, Frances Yates
(complete samples). Biometrika 52:591–611
Shephard RN, Metzler J (2018) Mental rotation task Publisher's Note Springer Nature remains neutral with regard to
Steiger JH (2004) Beyond the f test: effect size confidence intervals and jurisdictional claims in published maps and institutional affiliations.
tests of close fit in the analysis of variance and contrast analysis.
Psychol Methods 9:164–182

13