Virtual Maze Navigation Task

By July 19, 2019 No Comments



Maze navigation tests are utilized in the assessment of exploration, path planning, and navigation which rely on learning and memory capacities to form cognitive maps. Traditional animal navigation mazes such as the Morris Water Maze and the Radial Arm Maze have seen the translation for human application as virtual mazes (see Simian Virtual Reality Mazes); However, these lack the complexity that humans encounter in their everyday life. While it is possible to create large scale labyrinths for human testing, these are often cost-ineffective and lack flexibility for adaptations.

The Virtual Maze Navigation Task (VMNT) overcomes the limitations of physical mazes and in the long run, proves to be cost-effective. Using virtual environments, the experimenter has the flexibility to change the maze configurations/environments, insert or remove cues and easily monitor and record responses and behaviors. Additional complexity can be introduced in the task by the introduction of delays within the maze, altering the visibility and changing the starting pointing. The Virtual Maze Navigation Task is usually applied as a labyrinth with right/left turns, dead-ends, start area and end area.

Training Protocol

Participants are informed of the experimental process beforehand. Participant’s comfortability with the virtual reality technology used is also noted as this could also be a potential influencer on the performance. Ancillary tests may also be part of the investigation.

The general procedure of the Virtual Navigation Maze Task usually involves the following tasks.

  • Exploration Trial: This initial trial is performed to familiarize the participants with the technology and the virtual environment and usually lasts about 5 minutes.
  • Virtual Navigation Maze Training: Participants are instructed to explore the maze and find the exit. Start entries can be randomized to encourage the use of spatial strategies. Participants may also be given instructions regarding any cues that may be included during the trial.
  • Retention Trial: Participants are introduced to the same maze with corresponding start entries after a period of delay or after treatment and evaluated for their ability to recall the route to the exit/goal.

Behavioral Observations and Task Data

Depending on the design of the virtual maze in the Virtual Maze Navigation Task, participants may use different strategies. Primarily, navigation in mazes can be either allocentric or egocentric and can be affected by the presence or absence of cues. Performance analysis can also include responses to different set-ups such that induce stress and fear in the participant.

In general, behavioral measures can include the following,

  • Latency to initiate the task
  • Percentage of correct choices
  • Percentage of incorrect choices
  • Navigational strategy used
  • Navigation accuracy
  • Navigation speed
  • Distance traveled
  • Trial duration
  • Response to stimulus
  • Frequency of backtracking

Based on the requirements of the investigation EEG data may also be recorded. Other measures (relevant to the investigation) may include assessment of stress, anxiety and heart rate levels, among others. Ancillary questionnaires may also be used to further refine the data and the understanding of the task performance.

Literature Review

Evaluation of the effect of a daytime nap on VMNT performance

Objective:Wamsley, Tucker, Payne, and Stickgold (2010) assessed the effect of a post-learning nap on complex route-learning in humans. The study was aimed to overcome issues of discrepancy related to circadian time when testing sleep-deprived versus well-slept participants.
Participants:Participants included 53 college students (34 females and 19 males, age

18 to 30 years) with no known psychiatric and sleep disorders or on any medication that interferes with sleep. Participants were divided into nap group and wake group.

Participants were additionally instructed to abstain from caffeine, alcohol, and drug use for 24 hours prior to the maze task. Additionally, all participants were required to maintain a regular sleep schedule (3 nights preceding the study).

Maze Design:The virtual maze was composed of abstract patterned walls and black-white checkered floors. In order to encourage hippocampus-dependent spatial strategies, rather than response-based navigation strategy, the maze was created to be visually sparse, including only a few object types

repeated throughout the landscape. A tree served as the goal point in the maze. All start positions in the maze were equidistant from the goal.

The maze was implemented on a personal computer and participants controlled their movement in the maze using a keyboard.

Ancillary Questionnaires, Protocols & Recordings:·         Three-day retrospective sleep log

·         Stanford Sleepiness Scale (SSS)

·         Polysomnography (PSG)

·         Electroencephalography (EEG)

·         Electrooculography (EOG)

·         Electromyography (EMG)

·         3D gameplay experience level (self-assessed on a five-point scale)

Procedure:At around 12:30 pm, participants were trained on the VMNT. Training included a 20-minute pretest to assess the skill level of the participants based on which they were allocated to one of the 4 mazes that varied in complexity and size. Following the pretest, participants were allowed to explore novel maze starting at the tree object in the maze, which matched their skill level, for 5-minutes. On completion of exploration trials, participants were tested in a series of test trials with 3 unique, equidistant start positions relative to the tree (goal). Trials were ended when the participants reached the exit of the maze, or a maximum of 10 minutes had elapsed. The nap group was then allowed to nap for an hour and a half in a darkened room and were awoken at first sign of REM sleep. During the nap interval, the wake group began quiet wakefulness wherein they sat silently and did not engage in any activity. Following this period, the group was allowed to watch videos to ensure they remained awake. At around 5:30 pm, both groups were retested again as in the test trials.
Results:Performances were observed to have significantly improved across both test and retest trials with no significant difference between the groups. However, a correlation between baseline performance in the final test trial and prior 3D gameplay experience could be observed. Post-learning sleep improved performances relative to wakefulness in experienced players, with the wake group exhibiting performance deterioration in the maze task. Additionally, it was observed that experienced players with greatest stage 2 delta power displayed a significant sleep-dependent improvement in performances. However, such improvement was not observed among the novice participants.



Investigation of the effect of sleep in enhancing performances in a VMNT

Objective:Nguyen, Tucker, Stickgold, and Wamsley (2013) evaluated the effects of overnight sleep on the ability to skillfully navigate a maze. The study was undertaken to investigate whether sleep affected non-hippocampal navigation parameters such as navigation speed in addition to the consolidation of cognitive maps.
Participants:Participants included 30 college students (19 females and 11 males, age

19.6 ± 2.0 years) with no known psychiatric and sleep disorders. Subjects were pseudo-randomly divided into 3 groups; sleep group, sleep + awakenings group and wake group.

Participants were additionally instructed to abstain from caffeine, alcohol, and drug use for 24 hours prior to the maze task. Additionally, all participants were required to maintain a regular sleep schedule (nights preceding the study).

Maze Design:The virtual maze was a complex maze composed of stone walls and floors. The maze was also equipped with salient landmarks. The maze exit was depicted as a glowing door. In order to encourage the use of proximal visual information, fog was introduced into the maze to reduce linear visibility. All start positions in the maze were equidistant from the exit.

The maze was implemented on a personal computer and projected on the wall of a darkened testing room (60″ × 44″ viewing area). Participants controlled their movement in the maze using 4-key modified number pad that permitted forward, backward, left and right movements.

Ancillary Questionnaires, Protocols & Recordings:·         Demographic questionnaire

·         Three-day retrospective sleep log

·         Epworth Sleepiness Scale (ESS)

·         Stanford Sleepiness Scale (SSS)

·         Polysomnography (PSG)

·         Electroencephalography (EEG)

·         Electrooculography (EOG)

·         Electromyography (EMG)

Procedure:Sleep and sleep + awakenings groups reported for the VMNT training at 10:30 pm after which they were given an 8-hour sleep opportunity. While the sleep group was not disturbed during their sleep, the sleep + awakenings group was awakened several times to report dream experiences. The wake group reported for training at 10:00 am and were prohibited from sleeping or taking a nap until retesting. VMNT training included maze exploration for 5 minutes followed by 3 training trials using a unique starting point for every trial. Following training, groups were retested to evaluate their memory in 3 trials with starting points the same as the training sessions. The wake group was evaluated on the same day at 9:00 pm while the sleep and sleep + awakenings groups were tested 30 minutes after waking at approximately 9:00 am.

Trials were ended when the participants reached the exit of the maze, or a maximum of 10 minutes had elapsed.

Results:Minor differences in the sleep architecture of the sleep and sleep + awakenings group were observed. However, these differences did not influence the task performances in either group. Additionally, both groups displayed the same equivalent level of alertness as measured by the SSS in the morning.

Initial training observation did not reveal any significant differences among the three groups. However, the effect of sleep could be observed during the retention trials. Both sleep groups displayed significantly improved completion times, and reduced distance traveled as opposed to the wake groups. Further, backtracking in the sleep groups was also improved in comparison to the wake group. However, all groups exhibited a reduction in navigation speed, while navigation accuracy was improved in the sleep groups.


Investigation of the influence of personality and individual differences in VMNT

Objective:Walkowiak, Foulsham, and Eardley (2015) explored the relationship between personality traits, individual differences, and wayfinding using an immersive virtual navigation task.
Participants:Participants included thirty-three female undergraduate or postgraduate students in the age range 19 to 35 years old (mean age 23.03).
Maze Design:The virtual maze was composed of concrete walls and bricked floors with tufts of grass. The maze consisted of start, exit (large black hole in the wall), perpendicular turns and included seven local landmarks (such as a tree, table, vase, and TV) fixed in different regions of the maze to facilitate training. Additionally, the maze also consisted of red arrows to direct the participants only in the first phase of route learning stage.

The maze was implemented on a personal computer and projected on a 2000 × 1800 mm projection screen. Joysticks were used by the participants to control their movement in the maze. Participants were seated 200 cm away from the screen to enhance the immersive experience.

Ancillary Questionnaires, Protocols & Recordings:·         Demographic questionnaire

·         Computer Experience Questionnaire

·         State-Trait Anxiety Inventory

·         Wayfinding Anxiety Scale

·         International Wayfinding Strategy Scale

·         Eysenck Personality Questionnaire Revised-Short Form (EPQR-S)

·         Immersive Tendencies Questionnaire

Procedure:Participants were allowed 3 minutes to explore the maze and familiarize with the virtual reality set-up. Following the exploration trials, participants were trained and evaluated in a route-learning task which was divided into two phases. The first phase involved the participants following red arrows in the maze, which guided them to the exit. During this phase, participants were not allowed to backtrack. Once the training was complete, participants were asked to trace the exact path back to the start in a single trial (no guide arrows were present). At the end of the VMNT, participants filled battery of self-report personality and individual differences questionnaires.
Results:Participants with higher psychoticism scores (Eysenck Personality Questionnaire) as well as those with higher levels of wayfinding anxiety (Wayfinding Anxiety Scale) were observed to cover a significantly longer distance than those that scored lower on the corresponding tests. Additionally, participants with high wayfinding anxiety also spent significantly more time to reach the goal. Stronger overall immersive tendencies and higher involvement scores (Immersive Tendencies Questionnaire) resulted in participants covering significantly shorter distances and making a significantly lower number of errors than those that scored low. Participants that relied more on survey strategy (International Wayfinding Strategy Scale) were also observed to travel significantly shorter distances in addition to completing the task faster. Moreover, a significant effect of computer experience (Computer Experience Questionnaire) was observed on the task completion time and traveled distance. Participants with high computer experience as well as video game experience traveled faster and shorter distances in the task.


Evaluation of the effect of stress on VMNT performances

Objective:Delahaye et al. (2015) investigated the effect of stress on participant’s learning aptitude, spatial orientation and cognitive flexibility in a decision-making task.
Participants:Thirty-one healthy, unrelated Swiss-German descent individuals participated in the study (12 females and 19 males; age 25.1 ± 6.6).
Maze Design:The virtual maze was composed of concrete walls and floors with the ceiling covered. The maze design included alleys, crossings, and dead ends. The maze also included geometric icons (different colors and shapes) that served as information signs at the doors of the 24 rooms. Additionally, a stress element was also included in the maze. The maze was continuously flooded with virtual water that was accompanied by a loud sound. However, the water level stopped at 1.4 meters to eliminate the impression of a possible drowning.

All virtual reality tasks were implemented using a 3D Powerwall that created an approximately a 3 m cubed area providing a stereo view. Participants stood approximately 2 meters away from the three screens. Navigation in the virtual environments was controlled using a gamepad by the participants.

Ancillary Questionnaires, Protocols & Recordings:·         Virtual Trier Social Stress Test (TSST)

·         Virtual Cognitive Flexibility Labyrinth (Based on Wisconsin Card Sorting Test)

·         Physiological stress level measures

·         Subjective Stress level

·         Electroencephalography (EEG)

Procedure:Participants were first subjected to Virtual TSST to induce stress via public speaking. Participants were given 2 minutes to prepare before presenting themselves (5 minutes) before a virtual audience that imitated the typical public audience. The task started off with low levels of audience noise and gestures which were heightened towards the end to increase stress in the participants. Following stress induction, participants were evaluated in a virtual memory maze that required them to not only remember the route but also learn and associate right and wrong choices associated with geometrical icons in the maze. Additional stress and urgency were introduced in this navigational task by virtually flooding the maze. Finally, following the spatial orientation and learning aptitude test, subjects were evaluated in another virtual maze that consisted of rooms, doors and geometric icons to assess cognitive flexibility.
Results:Virtual TSST successfully induced stress in the participants. Participants exhibited relatively higher heart rates as compared to the post-stress period when tested in both the virtual mazes. As opposed to participants that reported TSST as less stressful, stressed participants made significantly more errors in the memory maze as observed by the lack of strategy or application of learned concepts. The cognitive flexibility test revealed that participants that reported TSST as unexpected displayed significantly fewer arbitrary category changes.


  1. Andreano, J. M., & Cahill, L. (2009). Sex influences on the neurobiology of learning and memory. Learning & Memory, 16(4), 248–266. doi:10.1101/lm.918309
  2. Castelli, L., Corazzini, L. L., & Geminiani, G. C. (2008). Spatial navigation in largescale virtual environments: Gender differences in survey tasks. Computers in Human Behaviour, 24, 1643–1667.
  3. Delahaye, M., Lemoine, P., Cartwright, S., Deuring, G., Beck, J., Pflueger, M., … Hachtel, H. (2015). Learning aptitude, spatial orientation and cognitive flexibility tested in a virtual labyrinth after virtual stress induction. BMC Psychology, 3(1). doi:10.1186/s40359-015-0080-5
  4. Gillner, S., & Mallot, H. A. (1998). Navigation and Acquisition of Spatial Knowledge in a Virtual Maze. Journal of Cognitive Neuroscience, 10(4), 445–463. doi:10.1162/089892998562861
  5. Nguyen, N. D., Tucker, M. A., Stickgold, R., & Wamsley, E. J. (2013). Overnight Sleep Enhances Hippocampus-Dependent Aspects of Spatial Memory. Sleep, 36(7), 1051–1057. doi:10.5665/sleep.2808
  6. Walkowiak, S., Foulsham, T., & Eardley, A. F. (2015). Individual differences and personality correlates of navigational performance in the virtual route learning task. Computers in Human Behavior, 45, 402–410. doi:10.1016/j.chb.2014.12.041
  7. Wamsley, E. J., Tucker, M. A., Payne, J. D., & Stickgold, R. (2010). A brief nap is beneficial for human route-learning: The role of navigation experience and EEG spectral power. Learning & Memory, 17(7), 332–336. doi:10.1101/lm.1828310

About Maze Engineers

Figuring out the future of neuroscience, one lab mouse at a time.