Investigating Behavioral and Neurophysiological Responses Across

Landslide Scenarios in Virtual Reality

Arjun Mehra

, Arti Devi

, Ananya Sharma

, Sahil Rana

, Shivam Kumar

K. V. Uday

and Varun Dutt

Applied Cognitive Science Lab, IIT Mandi, India

Geotechnical Engineering Lab, IIT Mandi, India

Keywords: Landslide Probability, Time of Day, EEG Measures, Collision Analysis, Alpha/Theta Ratio, Alpha/Gamma

Ratio, Beta/Gamma Ratio.

Abstract: The potential of virtual reality for disaster preparedness is enormous, but little is known about how different

landslide risks and environmental factors (day versus night) affect human reactions. Neurophysiological

(alpha/theta, alpha/gamma, and beta/gamma ratios from EEG) and behavioral (Euclidean distance, collisions,

and velocity around collisions) measures are combined in this study to investigate stress and cognitive

engagement in landslide simulations. In order to expose 80 participants to varying landslide probabilities, they

were randomly assigned to four groups with varying landslide risk and lighting conditions. Behavioral

deviations and cognitive workload were significantly influenced by perceived risk rather than lighting

conditions, according to the results. Electroencephalography (EEG) and behavioral outcomes were correlated,

which emphasized how crucial integral analysis is to comprehending disaster responses. These results

demonstrate how well virtual reality can develop cognitive resilience and offer guidance for creating training

plans that maximize performance in high-risk scenarios. This study develops dynamic, immersive VR-based

disaster preparedness apps.

1 INTRODUCTION

Landslides cause significant global infrastructure

damage and fatalities annually, often triggered by

human activities or natural disasters like earthquakes

and floods, compounding disaster management

challenges (Highland, 2008; Petley, 2012). Even with

improvements in early warning systems, disaster

response is still hampered by the unpredictable nature

and quick onset of landslides.

A revolutionary tool for researching human

behavior in dangerous, difficult-to-replicate disaster

scenarios is virtual reality (VR) technology, which

provides immersive and controlled environments

https://orcid.org/0009-0006-0175-3532

https://orcid.org/0000-0002-0171-297X

https://orcid.org/0009-0004-3000-8924

https://orcid.org/0009-0002-7545-5264

https://orcid.org/0009-0008-4816-6006

https://orcid.org/0000-0002-9579-5496

https://orcid.org/0000-0002-2151-8314

(Alene, 2023; Du, 2021). Research indicates that VR-

based training improves risk perception, memory, and

real-world applicability, frequently matching field

training results (Gross, 2023; Adami, 2021).

Prior work only focused on static models and used

traditional methods to navigate landslide-prone areas.

This research highlighted this issue and addressed this

gap by using VR simulations to investigate people's

actions and responses in landslide scenarios.

Using virtual reality driving simulations, this

study examines how people react physically and

behaviorally to landslide threats during the day and at

night. It provides a thorough understanding of

landslide responses by analyzing participant behavior

Mehra, A., Devi, A., Sharma, A., Rana, S., Kumar, S., Uday, K. V. and Dutt, V.

Investigating Behavioral and Neurophysiological Responses Across Landslide Scenarios in Virtual Reality.

DOI: 10.5220/0013253000003911

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 18th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2025) - Volume 1, pages 949-956

ISBN: 978-989-758-731-3; ISSN: 2184-4305

949

and neurophysiological data (such as EEG). Public

awareness campaigns, disaster training initiatives,

and the construction of safer infrastructure in

landslide-prone areas can all benefit from the

findings.

The findings suggest that people's physical and

behavioral reactions can be influenced by the level of

perceived danger and simulated environment.

The remainder of the paper is organized as

follows. First, prior work done in related areas was

discussed briefly followed by the research gap. The

expectation section included details about behavioral

indicators, measures and effects of illumination. In

material and methodology, the section briefly

discussed the simulation, participants and

experimental design. At last, the implications and

future research highlighted

2 BACKGROUND

Research on disaster management is being

revolutionized by virtual reality (VR), which makes it

possible to examine human behavior in risky

situations without actual dangers. According to

Petley's research, virtual reality can be used to study

how people react to landslides, a common and deadly

natural disaster that causes a lot of damage (Petley,

2012).

Earthquakes, rain, or human activities like

deforestation can cause landslides, which are

complicated geophysical phenomena in which

material slides down slopes. Effective risk mitigation

and disaster response training are essential due to the

unpredictable nature of landslides and their abrupt

onset (Highland, 2008).

Researchers can evaluate stress levels, attention,

and cognitive engagement by examining these EEG

frequency bands during disaster simulations, giving

them a thorough grasp of how people react under

pressure. By pinpointing areas where participants

might need more assistance or practice, this method

improves the efficacy of training initiatives. EEG has

been shown to be useful in assessing mental stress and

cognitive workload in earlier research. For example,

studies have demonstrated that differences in Alpha,

Beta, Theta, and Gamma bands can be used as

markers of mental stress, underscoring the value of

these metrics in evaluating reactions in disaster drills

(Bakare, 2024).

By incorporating EEG analysis into disaster

preparedness training, customized interventions that

enhance cognitive resilience and performance under

stress can be created. This integration ultimately leads

to more effective disaster response strategies by

facilitating a more nuanced understanding of the

neural mechanisms underlying human behavior in

emergency situations.

Conventional crisis training is based on static

simulations or theoretical scenarios that don't

accurately represent the dynamics of actual events.

By producing dynamic, immersive environments,

virtual reality (VR) overcomes these drawbacks and

works well in disaster training scenarios. The

effectiveness of disaster preparedness is increased by

VR-based fire safety training, which performs better

than conventional approaches in terms of application

and retention (Smith, 2009).

According to Chittaro and Ranon's research,

virtual reality simulations can accurately evaluate and

get stakeholders ready for risks, producing outcomes

that are on par with field research (Chittaro, 2009).

These studies support the usefulness of VR in crisis

training, particularly in situations where testing in the

real world is risky or impractical.

Although there is evidence to support the use of

virtual reality (VR) in disaster training, the majority

of research ignores dynamic interactions, such as

navigating terrain that is prone to landslides, in favor

of static simulations (Takeda, 2005). Realistic

training is provided by dynamic scenarios, which

improve readiness and emergency response skills.

This study fills the gap by using VR-based

simulations to investigate reactions to dynamic

landslide scenarios.

This study simulates landslide accidents using

day-night conditions and probability distributions

(low: 0.2, high: 0.8). The study investigates how

participants react behaviorally and physiologically to

various situations. It uses neurophysiological

measurements (such as EEG) and self-reports to gain

a thorough understanding of how people react in risky

situations.

Through the extension of virtual reality to

dynamic, interactive scenarios, this study advances

our understanding of human behavior during

landslides. The results can be used to inform disaster

training programs, public awareness campaigns, and

safer infrastructure in landslide-prone areas.

3 EXPECTATIONS

We hypothesized that the perceived danger levels

associated with different landslide probabilities

would significantly alter the behavioral and

neurophysiological responses of research

participants. In particular, we anticipated that:

BIOSIGNALS 2025 - 18th International Conference on Bio-inspired Systems and Signal Processing

950

3.1 Behavioral Indicators

Participants exposed to high-probability landslide

scenarios would deviate more from their intended

driving path, as indicated by the Euclidean Distance

(ED) surrounding crashes. Because of the increased

perceived danger, this would imply more erratic

driving behavior and increased caution. More

collisions would occur in high-probability scenarios,

suggesting that participants would find it more

difficult to navigate the hazardous environment.

Participants would probably report faster velocities

close to collisions in high-risk situations, despite the

possibility that this would lead to less controlled

driving and more crashes. This could be a reflexive

attempt to quickly move away from perceived risks.

3.2 Measure of Neurophysiology

People would be more nervous and mentally focused

on the threat than they are at ease in high-risk

situations, according to the Alpha/Theta ratio, a

measure of cognitive engagement and alertness. The

measured Alpha/Gamma ratio, which is associated

with information flow and attention, would also

increase. This suggests that comprehending these

complex situations requires more mental work. In

high-risk circumstances, the Beta/Gamma ratio,

which gauges cognitive effort, would likewise rise,

reflecting the mental strain that the participants would

experience in riskier circumstances.

3.3 Effect of Illumination (Day Versus

Light)

We expected that responses would differ substantially

depending on the probability of landslides, but that

daytime versus nighttime lighting would have less of

an impact on the behavioral and neurophysiological

measurements. This would imply that the main factor

affecting participants' responses in the simulation is

perceived danger rather than environmental visibility.

These expectations were based on the knowledge

that people's thoughts and actions during disasters are

greatly influenced by their perceptions of risk. In

order to gain insight into the potential for VR-based

training programs to improve disaster preparedness,

the study was created to investigate how these

elements interact within a controlled virtual reality

environment.

4 MATERIALS AND METHODS

4.1 Participants

As seen in Figure 1, 80 participants (60 men and 20

women, mean age = 22, SD = 1) were chosen at

random through an advertisement and asked to take

part in the study. Informed consent was acquired

through a Google form, and participation was entirely

voluntary. The only prerequisite for being on the

shortlist was being older than eighteen. In addition to

having educational backgrounds at least as high as a

high school diploma, each participant had a scientific

background and obtained intermediate school,

bachelor's, master's, and even doctoral degrees. The

study's participants received INR 100 in

compensation for their involvement.

4.2 Virtual Reality Simulation for

Landslide

A virtual reality landslide simulation was created

using Unity 2021.3.30f1 with Gaia Pro, Easy Roads

3D Pro, and Realistic Car Controller V3 (Cai, 2023).

XR plugins were added to the simulation to enable

compatibility with Virtual Reality HMDs, and the

Logitech SDK was installed and incorporated into the

simulation to enable compatibility with the Logitech

steering wheel. According to the manually set

landslide probability, which was either 0.2 or 0.8

prior to the participant entering the play mode, the

simulation used three colliders to create a landslide

when a car entered the landslide-prone area. Gaia

Runtime was also used to set the day and night lights

before players switched to play mode. The night

mode can be accessed with reduced visibility and

time perception by adjusting the lighting settings'

time, which in turn modifies the position of the Sun,

the directional light source. To ensure that noise won't

impair participants' ability to drive, the hardware is

installed in a quiet lab space with air conditioning.

The participant's goal was to keep the vehicle on

the left side of the road and drive it to the end of the

road twice. Two C# scripts, positional and collision

data recorders, were added to the simulation in order

to gather the x, y, and z coordinates of the car's

position and collision with a landslide in CSV format.

In the parent folder where the project was stored,

these scripts stored the files in CSV format.

4.3 Experimental Design

The study complied with the Declaration of Helsinki

and received approval from the Institutional Ethics

Investigating Behavioral and Neurophysiological Responses Across Landslide Scenarios in Virtual Reality

951

Committee (World Medical Association, 2013).

Participants (N = 20 each) were split into four groups

at random:

Group 1: High probability of landslides, during

the day

Group 2: High probability of landslides, during

the night

Group 3: Low probability of landslides, during

the day

Group 4: Low probability of landslides, during

the night

Each experiment lasted 10–15 minutes,

depending on driving speed of the participant. The

collected data encompassed demographic

information, an Indian personality questionnaire, risk

assessment, evaluation of driving skills and

experience, terrain familiarity, weather condition

experience, night driving experience, accident

history, risk perception, and landslide perception, all

gathered subsequent to obtaining consent. Post-

simulation evaluations included VR experience

quality, driving behavior, realism, situational

awareness, confidence, and emotional response. Post-

simulation assessments encompassed the quality of

the VR experience, driving conduct, realism,

situational awareness, confidence, and emotional

response. The analyses were performed utilizing an

interactive landslide perception and education

program executed via a Google form.

Figure 1: A VR landslide sim setup (Night condition).

Following the acquisition of all pre-experimental

details, pre-experimental EEG data with eyes open

were concurrently collected for six minutes utilizing

a 4-channel headband (four channels: TP9, TP10,

AF7, and AF8), as illustrated in Figure 1.

Subsequently, participants donned a VR headset in

conjunction with a 4-channel headband to facilitate

the collection of EEG data while driving.

Subsequently, participants were instructed to

operate a vehicle under one of the four randomly

assigned conditions depicted in Figure 2. The

perspective is from the driver's seat, featuring a

movable steering wheel and unobstructed views of

the surroundings, simulating the experience of

operating a vehicle. The landslide is observable and

transpired in front of the vehicle. EEG data, positional

data, and collision data from the participants were

collected during the experiment, followed by a post-

experience assessment utilizing questionnaires.

Figure 2: Participant’s point of view in Day condition.

The gathered data was processed using a number

of methods. Python was used to process the EEG data

for the pre-experimental and collision-area segments

in order to calculate the Alpha/Theta, Alpha/Gamma,

and Beta/Gamma measures. In order to determine the

Euclidean distance, the number of collisions, and the

velocity around the collisions, position and collision

data are also processed using a Python script. Two-

way ANOVAs were used on all behavioral and EEG

measures to examine the effects of various between-

subject training conditions.

5 RESULTS

5.1 Behavioral Measures

5.1.1 Ed Around Collision

ED around collisions was higher in high-probability

scenarios (M = 219.75; SE = 26.50) than in low-

probability ones (M = 82.16; SE = 26.85; F(1, 75) =

13.30, p < 0.001, ηp² = 0.15; Figure 3). This indicates

greater spatial navigation variation under higher

collision risks. Lighting conditions (Day: M =

134.36; Night: M = 167.55) had no significant effect

(F(1, 75) = 0.51, p = 0.48, ηp² = 0.01). There was no

significant interaction between landslide probability

and lighting (High Probability - Day: Mean = 220.91;

SE = 37.15, High Probability - Night: Mean = 218.59;

SE = 37.13; Low Probability - Day: Mean = 47.81;

SE = 37.15, Low Probability - Night: Mean = 116.50;

SE = 37.13) (F(1, 75) = 1.51, p = 0.22, ηp² = 0.02). 

BIOSIGNALS 2025 - 18th International Conference on Bio-inspired Systems and Signal Processing

952

5.1.2 Number of Collisions

Collisions were more frequent in high-probability

scenarios (M = 4.05; SE = 0.57) than in low-

probability ones (M = 0.62; SE = 0.57; F(1, 75) =

18.17, p < 0.05, ηp² = 0.20; Figure 3). This result

indicates that if participants thought there was a

greater chance of landslides, they were more likely to

collide with falling rocks. Collisions were unaffected

by lighting (Day: M = 2.45; Night: M = 2.22; F(1, 75)

= 0.51, p = 0.48, ηp² = 0.00) or probability-lighting

interactions. (High Probability - Day: Mean = 4.62;

SE = 0.81, High Probability - Night: Mean = 3.48; SE

= 0.81; Low Probability - Day: Mean = 0.28; SE =

0.81, Low Probability - Night: Mean = 0.96; SE =

0.81) (F(1, 75) = 1.51, p = 0.22, ηp² = 0.00). 

5.1.3 Velocity Around Collision

Collision velocity was higher in high-probability

scenarios (M = 411.67; SE = 52.57) than in low-

probability ones (M = 109.94; SE = 53.26; F(1, 75) =

16.26, p < 0.001, ηp² = 0.18; Figure 3). The results

indicate that when participants were in more

hazardous situations and faced collisions, they

reacted faster, probably as a reflex to avoid the

obstacles. Both lighting conditions had no significant

effect on collision velocity (Day: M = 265.80; Night:

M = 255.81; F(1, 75) = 0.22, p = 0.64, ηp² = 0.03).

Additionally, the interaction between landslide

probability and lighting conditions was also not

significant (High Probability - Day: Mean = 413.08;

SE = 75.98, High Probability - Night: Mean = 410.25;

SE = 75.98; Low Probability - Day: Mean = 118.53;

SE = 75.98, Low Probability - Night: Mean = 101.36;

SE = 75.98) (F(1, 75) = 2.20, p = 0.14, ηp² = 0.03).

Figure 3: Graph of the three behavioral measures.

Three behavioral metrics are displayed in Figure

3: the Euclidean Distance (ED) around collision, the

frequency of collisions, and the velocity around

collision. The mean values (along with standard

deviations) of these three metrics are displayed in the

bar chart for both high and low landslide likelihood

scenarios. The results showed that people displayed

significantly higher ED around collisions, more

collisions, and increased velocity around collisions in

high-probability scenarios compared to low-

probability scenarios. This suggests that participants

were more likely to navigate erratically, collide with

barriers, and move more quickly when they perceived

a greater risk of landslides.

5.2 EEG Measures

5.2.1 Alpha/Theta

The landslide probability had a significant impact on

the Alpha/Theta ratio. High-probability scenarios had

higher ratios (Mean = 2.69; SE = 0.43) compared to

low-probability scenarios (Mean = 1.37; SE = 0.43)

(see Figure 4; F(1, 76) = 4.62, p < 0.05, ηp² = 0.56).

Participants' perception of a greater risk of landslides

was associated with increased cognitive engagement

and decreased relaxation. There was no significant

effect of lighting conditions on the Alpha/Theta ratio

(F(1, 76) = 2.38, p = 0.13, ηp² = 0.03), suggesting that

the day or night setting did not have a significant

impact on Alpha/Theta ratio. In addition, the

interaction effect between the probability of

landslides and lighting conditions was not significant

(High Probability - Day: Mean = 3.70; SE = 0.61,

High Probability - Night: Mean = 1.68; SE = 0.61;

Low Probability - Day: Mean = 1.31; SE = 0.61, Low

Probability - Night: Mean = 1.42; SE = 0.61) (F(1,

76) = 2.99, p = 0.09, ηp² = 0.04).

5.2.2 Alpha/Gamma

There were higher Alpha/Gamma ratios seen in high-

probability scenarios (Mean = 1.91; SE = 0.14) than

in low-probability scenarios (Mean = 0.97; SE = 0.14)

(see Figure 4; F(1, 76) = 22.25, p < 0.001, ηp² = 0.23).

The results imply a higher level of focus, attention,

and cognitive processing in reaction to more

dangerous conditions. The primary impact of lighting

conditions on the Alpha/Gamma ratio was found to

be insignificant (Day: Mean = 1.68; SE = 0.14, Night:

Mean = 1.20; SE = 0.14) (F(1, 76) = 3.00, p = 0.09,

ηp² = 0.04). There was also no significant interaction

effect between landslide probability and lighting

conditions (High Probability - Day: Mean = 2.37; SE

= 0.20, High Probability - Night: Mean = 1.45; SE =

0.20; Low Probability - Day: Mean = 0.99; SE = 0.20,

Low Probability - Night: Mean = 0.96; SE = 0.20)

(F(1, 76) = 2.25, p = 0.14, ηp² = 0.03).

Investigating Behavioral and Neurophysiological Responses Across Landslide Scenarios in Virtual Reality

953

5.2.3 Beta/Gamma

The main impact of landslide probability on the

Beta/Gamma ratio was significant, where high-

probability scenarios had higher ratios (Mean = 1.36;

SE = 0.06) compared to low-probability scenarios

(Mean = 0.85; SE = 0.06) (see Figure 4; F(1, 76) =

31.05, p < 0.001, ηp² = 0.29). This implies a rise in

cognitive workload and stress as a response to a

higher perceived risk. The main impact of lighting

conditions on the Beta/Gamma ratio was found to be

insignificant (Day: Mean = 1.11; SE = 0.06, Night:

Mean = 1.10; SE = 0.06) (F(1, 76) = 0.16, p = 0.70,

ηp² = 0.00). Lastly, there was also no significant

interaction effect between landslide probability and

lighting conditions (High Probability - Day: Mean =

1.36; SE = 0.08, High Probability - Night: Mean =

1.36; SE = 0.08; Low Probability - Day: Mean = 0.87;

SE = 0.08, Low Probability - Night: Mean = 0.83; SE

= 0.08) (F(1, 76) = 0.49, p = 0.49, ηp² = 0.01).

Figure 4: Graph of the three EEG measures.

The three EEG measures—Alpha/Theta,

Alpha/Gamma, and Beta/Gamma ratios—mean

values (with standard errors) were also compared as

shown in Figure 4 for both high and low landslide

probability scenarios. The findings show that in high-

probability scenarios, all three ratios were

considerably higher, suggesting higher levels of

stress, focused attention, and cognitive involvement.

This suggests that when individuals believed that

there was a larger chance of landslides, they felt more

stressed and under cognitive stress.

6 CONCLUSIONS

The study's findings provide intriguing new insights

into how people respond to varying degrees of danger

in landslide simulations, both behaviorally and

neurophysiologically. The findings suggest that

perceived risk, rather than ambient lighting

conditions, is the primary factor influencing

participants' cognitive and physiological responses.

6.1 Behavioral Responses to Perceived

Risk

High-probability scenarios increased participants’

course deviations, as shown by the rise in ED around

crashes. This suggests cautious yet erratic navigation

under pressure. Increased accidents suggest perceived

threats led to more mistakes, aligning with higher

stress and cognitive load.

High-risk conditions also saw increased crash

velocities. According to this, Participants may have

reflexively increased speed to evade perceived

danger. However, higher speeds likely reduced

control, leading to more crashes. Since lighting

conditions did not significantly alter any of these

behavioral measures, it is further evidence that

perceived danger level—rather than visibility—was

the primary factor influencing participants' behaviors.

6.2 Neurophysiology of Stress and

Cognitive Engagement

These findings from behavior are supported by

neurophysiological evidence. In high-probability

circumstances, the Alpha/Theta ratio significantly

increases, indicating increased cognitive involvement

and attention. In order to prepare for the impending

threat, participants were probably less at ease and

more mentally concentrated. This fits with the

observed propensity of behavior to increase velocity

and diverge more from the route surrounding

collisions.

A state of concentrated attention and cognitive

processing is reflected in the Alpha/Gamma ratio,

which also rose in high-risk circumstances. Not only

were participants eager, but they were also focusing

their mental energies on comprehending the intricate

and ever-changing simulation environment. Although

this enhanced amount of cognitive processing was

required to navigate the dangerous environment, it's

possible that it increased the number of crashes

because it put more strain on participants' minds than

they could handle.

In high-probability situations, there was a

substantial increase in the Beta/Gamma ratio, a

measure of cognitive strain and stress. This result

emphasizes the elevated psychological stress that

participants felt in reaction to the anticipated

landslide danger. The observed behavioral

consequences can be explained by the individuals'

BIOSIGNALS 2025 - 18th International Conference on Bio-inspired Systems and Signal Processing

954

reduced capacity to conduct precise, controlled

moves due to higher stress levels. Once more, this

EEG evidence does not show a substantial effect from

lighting conditions, indicating that perceived danger

was the only factor influencing cognitive and

emotional states.

6.3 Consequences of VR-Based

Emergency Education

These results have major implications for VR-based

disaster training design and implementation.

Perceived risk strongly affects tension, cognitive

engagement, and attention, as per the study. VR

training simulations should prioritize realistic risk

scenarios to evoke authentic cognitive and emotional

responses.

We would project that the total cost of

implementing VR-based training or education would

not exceed £2000 (roughly ₹200000 INR). A

performance desktop computer, a steering wheel with

pedals, a virtual reality headset, and a multi-channel

EEG band (for gathering EEG data) make up the

majority of the hardware in a lab room that can

accommodate all of this. If we were to train 80

participants, we would only need to spend 25 pounds

per person, which could be as little as 2 pounds per

person if there were 1000 participants. Mock drills

involving real vehicles and controlled rockfall on

such a large scale, in a large environmental setting,

would be more costly in real life.

Furthermore, the findings imply that although

greater cognitive involvement and processing help

individuals become more adept at navigating

dangerous situations, these advantages need to be

counterbalanced by techniques for handling the stress

and cognitive effort that come with it. VR disaster

training programs may be more successful if they

include instruction on stress management strategies,

making decisions under duress, and building

cognitive resilience.

6.4 Limitations and Prospects for

Further Study

Although the study offers insightful information,

there are several drawbacks. To improve the

simulation's ecological validity, other environmental

factors that are frequently connected to landslide-

prone places, including rain, fog, or thunderstorms,

might be added. Subsequent investigations may

examine the impact of these supplementary

environmental elements on the behavior and

cognitive reactions of participants.

Furthermore, the findings' generalizability could

be improved by increasing the sample's relative

homogeneity. Thus, including a wider variety of

backgrounds in the participant pool in the future may

yield more complex insights and improve the

robustness of the findings.

To obtain a more thorough knowledge of how

various characteristics of physiological stress interact

with behavioral outcomes in high-risk circumstances,

future studies might also include additional

physiological measurements, such as blood pressure,

heart rate variability (HRV), and blood oxygen levels,

along with neurofeedback. While the current study

provides separate analyses of behavioral and EEG

measures, future work may aim to establish integral

analyses by correlating EEG metrics directly with

behavioral outcomes. Specifically, metrics such as

the Alpha/Theta and Beta/Gamma ratios, which

indicate stress and cognitive engagement, can be

examined in relation to behavioral markers like

Euclidean distance and collision frequency. This

approach would offer a deeper understanding of how

cognitive and emotional states influence real-time

decision-making and performance in high-risk virtual

environments.

6.5 Conclusions

This study demonstrates that perceived danger, not

lighting, drives behavioral and neurophysiological

responses in simulated landslides. By simulating

actual crises and evoking genuine emotional and

cognitive responses, virtual reality (VR) holds

promise as a disaster training tool. Future research

may also provide more thorough insights into

participants' preparedness and reactions in such

circumstances by integrating behavioral and EEG

data. Integrating stress and cognitive effort

management can enhance VR-based disaster training

programs.

ACKNOWLEDGEMENTS

The authors would like to thank the Indian Institute of

Technology Mandi, for providing the hardware and

lab space. We acknowledge the support from

National Mission on Himalayan Studies project

(IITM/NMHS/VD/499) to Prof. Varun Dutt.

Investigating Behavioral and Neurophysiological Responses Across Landslide Scenarios in Virtual Reality

955

REFERENCES

Alene, G. H., Vicari, H., Irshad, S., Perkis, A., Bruland, O.,

& Thakur, V. (2023). Realistic visualization of debris

flow type landslides through virtual reality. Landslides,

20(1), 13-23.

Chittaro, L., & Ranon, R. (2009, March 23-24). Serious

games for training occupants of a building in personal

fire safety skills. In 2009 Conference in Games and

Virtual Worlds for Serious Applications (VS-GAMES

2009), Coventry, UK. https://doi.org/10.1109/VS-

GAMES.2009.8

Du, J., et al. (2021). Virtual Reality in Disaster Simulation:

A Review. IEEE Access, 9, 60373-60389.

Highland, L. M., & Bobrowsky, P. (2008). The landslide

handbook-A guide to understanding landslides (No.

1325). US Geological Survey.

Petley, D. (2012). Global patterns of loss of life from

landslides. Geology, 40(10), 927-930.

Smith, S., & Ericson, E. (2009). Using immersive game-

based virtual reality to teach fire-safety skills to

children. Virtual reality, 13, 87-99.

Gross, G., Ling, R., Richardson, B., & Quan, N. (2023). In-

person or virtual training?: Comparing the effectiveness

of community-based training. American Journal of

Distance Education, 37(1), 66-77.

Adami, P., Rodrigues, P. B., Woods, P. J., Becerik-Gerber,

B., Soibelman, L., Copur-Gencturk, Y., & Lucas, G.

(2021). Effectiveness of VR-based training on

improving construction workers’ knowledge, skills, and

safety behavior in robotic teleoperation. Advanced

Engineering Informatics, 50, 101431.

Cai, X., Wang, Z., Li, S., Pan, J., Li, C., & Tai, Y. (2023).

Implementation of a Virtual Reality Based Digital-

Twin Robotic Minimally Invasive Surgery Simulator.

Bioengineering, 10(11), 1302.

World Medical Association. 2013. WMA Declaration of

Helsinki – Ethical Principles for Medical Research

Involving Human Subjects. Retrieved September 12,

2024 from https://www.wma.net/policies-post/wma-

declaration-of-helsinki-ethical-principles-for-medical-

research-involving-human-subjects/., last accsessed

2016/11/21.

Takeda, T., & Kinoshita, Y. (2005). Lessons Learned from

Landslide Disaster Due to Torrential Rainfall in

Minamata, Japan. Prehospital and Disaster Medicine,

20, 67 - 67. https://doi.org/10.1017/S1049023X00013

418.

S. Bakare, S. Kuge, S. Sugandhi, S. Warad and V.

Panguddi, "Detection of Mental Stress using EEG

signals - Alpha, Beta, Theta, and Gamma Bands," 2024

5th International Conference for Emerging Technology

(INCET), Belgaum, India, 2024, pp. 1-9, doi:

10.1109/INCET61516.2024.10592994.

BIOSIGNALS 2025 - 18th International Conference on Bio-inspired Systems and Signal Processing

956