Advertisement

ANSIBLE: A Virtual World Ecosystem for Improving Psycho-Social Well-being

  • Tammy Ott
  • Peggy WuEmail author
  • Jacki Morie
  • Peter Wall
  • Jack Ladwig
  • Eric Chance
  • Kip Haynes
  • Bryan Bell
  • Christopher Miller
  • Kim Binsted
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 9740)

Abstract

We describe preliminary results of ANSIBLE – A Network of Social Interactions for Bilateral Life Enhancement. ANSIBLE leverages virtual worlds to deliver evidence based wellness promoting strategies and virtual agents as tools to facilitate asynchronous human-human communication in order to counteract behavioral health challenges associated with prolonged isolation and deep space exploration. ANSIBLE was deployed in August 2015 in a 12 month study with six crew members in an isolation simulated Mars habitat facility. In this paper, we compare the data for the first five months of this mission to a previous control mission for which ANSIBLE was not used. We found initial support for ANSIBLE to increase perceptions of closeness and satisfaction with friend and family relationships (but not other crew members) during prolonged isolation as well as a trend in stress reduction and increased feelings of ANSIBLE usability over time.

Keywords

Virtual worlds Virtual agents Psychological support Communications Psychological health 

1 Introduction

The social origin of humans is deeply integrated into our existence through millions of years of evolution, and is a major contributor to our survival as a species [1]. The unequivocal consequences of higher quality and quantity of social relationships on health outcomes have been demonstrated and widely accepted [2]. For future astronauts undertaking the envisioned 2.5 year Mars mission, sensory monotony and extreme work and living conditions serve as constant stressors to behavioral health. Astronauts will not only be physically separated from their social support for prolonged periods of time, but due to communication latencies, their timelines will also be detached. This will likely result in the magnification of the effects of social isolation seen with geographically distributed groups on earth, which permeate both the physiological and psychological systems [3]. In addition to the well-known effects on morbidity and mortality [4], social isolation poses a significant threat to astronaut behavioral health and performance with research indicating loneliness is a risk factor for, and may contribute to, poorer overall cognitive performance, faster cognitive decline, poorer executive functioning, more negativity and depressive cognition, contagion that threatens social cohesion [5] and reduced physical activity [4].

Novel psychological support systems are crucial to help manage these stressors and help astronauts maintain their relationships with friends, family, and colleagues on Earth. Currently thousands of individuals all over the world are socializing, conducting commerce, collaborating, and essentially living their lives in a physically isolated environment, but are socially engaged through virtual environments. Digitally enabled connections are created independent of physical space and geography, forming genuine relationships at an unprecedented rate and scope. Research suggests that people derive the satisfaction they seek in the real world from their interactions in the virtual world [6]. Additionally, studies have shown that gameplay in virtual reality can influence prosocial behavior and traits [7] such as virtual “superpowers” leading to greater helping behavior in the real world, regardless of how participants used that power [8], as well as enhancing [9] and promoting empathy [10].

While the isolation experienced on a mission to Mars will likely surpass anything experienced on Earth, the positive effects of virtual environments for isolation on Earth provides hope that a similar approach will be effective to combat the expected negative effects of going to Mars. To meet this need we developed ANSIBLE - A Network of Social Interactions for Bilateral Life Enhancement. A virtual space where we are implementing evidence based strategies to promote social connectedness and psychological well-being while accommodating technical and environmental limitations of long duration space flight. Further details about ANSIBLE can be found in [11].

The purpose of the current paper is to determine ANSIBLE’s effectiveness at maintaining the social connection between crew and family in an analog environment. To maximize the scientific value of the data collected during the long duration analog we have planned for various statistical comparisons between our experimental and control group, within our experimental group, and also with data collected from the crews friends and family. The current paper’s focus is on comparisons between the first five months of our control group data and our experimental group data.

2 Materials and Methods

2.1 Participants

The current study is part of the overarching Hawaii Space Exploration Analog and Simulation (HI-SEAS, www.hi-seas.org), located on an isolated Mars-like site in the barren landscape of Mauna Loa, HI. Crews of six people (three female) live and work through long-duration exploration simulations with the aim of removing barriers to human exploration of Mars. Our research will span two missions. The first, HI-SEAS III, acted as our control group, was eight months long, and concluded on June 13, 2015. The second mission began on August 28, 2015 and will last 12 months. In addition to completing the same surveys as the control group, HI-SEAS IV will also utilize the ANSIBLE virtual world ecosystem. In addition to our tasks, participants in both missions performed science research projects inside the habitat as well as Extra Vehicular Activities (EVAs) in the form of geological surveys or equipment maintenance tasks outdoors while donning prototype space suits.

2.2 Surveys

Social isolation refers to the lack of contact an individual has with others [12]. In the context of space flight, social isolation includes the lack of human contact outside the crew. Perceived social isolation (loneliness) is more closely related to the quality than quantity of social interactions [13, 14]. Therefore we measured perceived loneliness through a modified Circles of Closeness (COC) questionnaire. This is a four item questionnaire provided a pictorial measure of closeness and satisfaction, separately for mission control (MC) and friends and family (FF). Lower ratings indicated more closeness and satisfaction with the relationship. Previous implementations of COC found high alpha reliability, ranging from high .50 s to .92 depending on scale length [15].

In addition to a survey directly measuring perceived satisfaction/closeness with social support we included a survey on a consequence of lack of social support: stress [16]. For this we used the Perceived Stress Questionnaire (PSQ). This is a 30 item measure of subjectively experienced stress independent of a specific and objective occasion. It uses a 5 pt. Likert scale with higher scores indicating greater stress. The PSQ showed adequate reliability > 0.9 and was correlated with trait anxiety, Cohen’s Perceived Stress Scale, depression, self-rated stress and stressful life events; with test-retest reliability varying by a factor of 1.9 over 6 months for past month ratings [17].

The ANSIBLE group completed additional surveys about their interactions with ANSIBLE. The first was the Short Feedback Questionnaire (SFQ) [18], a seven item measure of the sense of presence, perceived difficulty of the task and discomfort encountered during the experience. The first six questions are an abbreviated alternative to the Presence Questionnaire developed by Witmer and Singer [19]. This survey has been found to be suitable for various virtual environments and with different clinical populations [18] and has successfully tested other systems [20]. It uses a 5 pt. Likert scale with higher scores indicating higher levels of positive experience with the system. They also completed the System Usability Scale (SUS), a ten item measure providing a global view of subjective system usability [21]. It uses a 5 pt. Likert scale with higher scores indicating greater usability. Finally, we included seven 5 pt. Likert scale items tailored to specific ANSIBLE usability concerns: usefulness in social connection maintenance, well-being improvement, clearness of purpose, willingness to recommend to friend, working as wanted, wanting to use, and compatibility with schedule and workload.

2.3 Procedure

The subjects completed informed consent forms and explanations of the aims of our study prior to the commencement of the study. While in the Habitat all participants received self-paced training that described the procedures of completing the surveys. Those using ANSIBLE also received training on how to interact with the system. The PI was available to answer questions or concerns of participants. Participants completed the COC at least once a week throughout the study, with the option to complete it up to three times a week. The PSQ was completed once a week during the duration of the study. Questionnaires were completed on the same schedule for both groups, with the exception that the ANSIBLE group completed the COC before and after ANSIBLE interactions, hence within this paper the after ANSIBLE COC measures were used for comparisons to the control group. Interactions within ANSIBLE lasted 20–30 min, where participants were given the ability to explore and interact as they pleased. For the control group we conducted an outgoing debrief after they exited the habitat, where we asked them about their experiences in the study as well as changes in social connections due to their participation. When the experimental group exits the habitat we will ask similar questions and additional questions about positive and negative experiences with the ANSIBLE system.

3 Results

Due to the nature of our data (multiple measurements per participant, missing data, and uneven spacing of repeated measurements; participants sometimes missed surveys or completed them off the given schedule), we used Mixed-Effect Model Repeated Measures (MMRM) to analyze results. In addition to overcoming the above limitations MMRM also allowed us to model time (i.e. each measurement occasion) as a Level 1 random effect and group (i.e. ANSIBLE or control) as a Level 2 fixed effect. We began with an initial unconditional model (i.e. no predictors) to determine if sufficient variability existed within our data set. We also modeled for random intercepts and slopes, with our repeated time measure using a first order autoregressive covariance structure, as well as including a squared time variable to determine if a quadratic relationship held more explanatory power for our data than a linear relationship. The assumptions of independent observations at the level above nesting, bivariate normality, and random residuals were checked and met. To control for initial differences in survey responses we used the difference between initial response and current response over time in our analyses, thus higher COCdiff scores indicate greater closeness and satisfaction and greater PSQdiff scores indicate greater stress.

3.1 Circles of Closeness

COC Closeness.

For the “how close do you feel to your family and friends” COC question, the unconditional repeated-measures model revealed significant variability in the COCdiff measure, suggesting it would be worthwhile to examine a conditional model to explain this variability. Time as a predictor of COCdiff approached significance, F(1,126.72) = 3, p < .09, with a slight increase over time (β = .04, SE = .02), the quadratic trend was n.s. There was significant with-in time variance (Var = 9.9, SE = .38, Wald Z = 12.3, p < .001), between subjects variance (Var = .001, SE = .0004, Wald Z = 2.3, p < .05) and covariance between different times once individual differences were accounted for, rho = .7, SE = .03, Wald Z = 26.8, p < .001, making the autoregressive structure appropriate for our data.

The model with group added as a predictor explained significantly more variance than the unconditional model, −2LL χ Change 2 (1) = 4.6, p < .05. There was a significant effect of group, F(1,109.7) = 5.5, p < .05, with the ANSIBLE group reporting greater closeness to family and friends (M = 1.5) than Controls (M = −0.5). The linear [F(1,114) = 4.4, p < .05, β = .05, SE = .02] and quadratic [F(1,131.9) = 4.0, p < .05, β = −.0003, SE = .0001] trends both significantly describe the pattern of the data over time. There remains significant with-in time variance (Var = 9.4, SE = .78, Wald Z = 12.5, p < .001), between subjects variance (Var = .001, SE = .0006, Wald Z = 2.2, p < 05) and covariance between different times once individual differences were accounted for, rho = .7, SE = .03, Wald Z = 24.7, p < .001, see Fig. 1.
Fig. 1.

Variability in “How close do you feel to your family and friends” COC question for each participant during the first five months within the HI-SEAS habitat. Top shows control data. Bottom shows ANSIBLE group data. Quadratic trend lines are shown (Color figure online).

For the “how close do you feel to other crew members” COC question, the unconditional repeated-measures model revealed significant variability in the COCdiff measure, suggesting it would be worthwhile to examine a conditional model to explain this variability. Time as a predictor of COCdiff was n.s., but the quadratic trend was significant, F(1,104.6) = 6.2, p < .05, β = −.0002, SE = .00008. There was significant with-in time variance (Var = 4.6, SE = .34, Wald Z = 12.2, p < .001), between subjects variance (Var = 5.16, SE = 1.8, Wald Z = 2.9, p < .01) and covariance between different times once individual differences were accounted for, rho = .59, SE = .05, Wald Z = 13.2, p < .001 making the autoregressive structure appropriate for our data. Adding group as a predictor did not explain significantly more variance than the unconditional model, −2LL χ Change 2 (1) = .012, p > .05, see Fig. 2.
Fig. 2.

Variability in “How close do you feel to fellow crew” COC question for each participant during the first five months within the HI-SEAS habitat. Top shows control data. Bottom shows ANSIBLE group data. Quadratic trend lines are shown (Color figure online).

COC Satisfaction.

For the “how satisfied do you feel with your family and friends” COC question, the unconditional repeated-measures model revealed significant variability in the COCdiff measure, suggesting it would be worthwhile to examine a conditional model to explain this variability. Time as a linear and quadratic trend were n.s. There was significant with-in time variance (Var = 4.8, SE = .43, Wald Z = 11.2, p < .001), between subjects variance (Var = 12.8, SE = 4.1, Wald Z = 3.2, p < .001) and covariance between different times once individual differences were accounted for, rho = .66, SE = .04, Wald Z = 16.5, p < .001 making the autoregressive structure appropriate for our data.

The model with group added as a predictor explained significantly more variance than the unconditional model, −2LL χ2Change(1) = 4.5, p < .05. There was a significant effect of group, F(1,24.1) = 4.9, p < .05, with the ANSIBLE group reporting greater satisfaction to family and friends (M = 3.1) than Controls (M = −1.3). Time as a linear and quadratic trend were n.s. There remains significant with-in time variance (Var = 4.8, SE = .43, Wald Z = 11.1, p < .001), between subjects variance (Var = 10.46, SE = 3.4, Wald Z = 3.1, p < 01) and covariance between different times once individual differences were accounted for, rho = .66, SE = .04, Wald Z = 16.5, p < .001, see Fig. 3.
Fig. 3.

Variability in “How satisfied do you feel with your family and friends” COC question for each participant during the first five months within the HI-SEAS habitat. Top shows control data. Bottom shows ANSIBLE group data (Color figure online).

For the “how satisfied do you feel with other crew members” COC question, the unconditional repeated-measures model did not reveal significant variability in the COCdiff measure, suggesting it is not worthwhile to examine a conditional model. Time as a linear and quadratic trend were n.s. There was significant with-in time variance (Var = 6.9, SE = .78, Wald Z = 8.8, p < .001) and covariance between different times once individual differences were accounted for, rho = .66, SE = .05, Wald Z = 13.4, p < .001 making the autoregressive structure appropriate for our data, but the between subjects variance was not significant (Var = 2.8, SE = 2.0, Wald Z = 1.4, p < .05). As expected, adding group as a predictor did not explain significantly more variance than the unconditional model, −2LL χ Change 2 (1) = .043, p > .05. See Fig. 4.
Fig. 4.

Variability in “How satisfied do you feel with your fellow crew” COC question for each participant during the first five months within the HI-SEAS habitat. Top shows control data. Bottom shows ANSIBLE group data (Color figure online).

PSQ.

The unconditional repeated-measures model revealed significant variability in the total PSQdiff measure, suggesting it would be worthwhile to examine a conditional model to explain this variability. Time as a linear and quadratic trend were n.s. There was significant with-in time variance (Var = 22.5, SE = 2.3, Wald Z = 9.8, p < .001), between subjects variance (Var = 39.6, SE = 12.7, Wald Z = 3.1, p < .01), but covariance between different times once individual differences were accounted for, rho = −.24, SE = .22, Wald Z = −1.1, p > .05 was not significant making the autoregressive structure unneeded for our data. However removing it did not significantly change model fit, −2LL χ2Change(1) = 0.57, p > .05.

The model with group added as a predictor explained significantly more variance than the unconditional model, −2LL χ Change 2 (1) = 3.85, p < .05. The effect of group was marginally significant, F(1, 24.9) = 3.16, p < .09, with the ANSIBLE group reporting less stress (M = .1) than Controls (M = 6.7). Time as a linear trend was n.s and the quadratic trend approach significance, F(1, 200) = 2.84, p < .09, β = −.0003, SE = .0002. There remains significant with-in time variance (Var = 22.3, SE = 2.2, Wald Z = 10, p < .001) and between subjects variance (Var = 35.9, SE = 11.6, Wald Z = 3.1, p < 01), see Fig. 5.
Fig. 5.

Variability in Perceived Stress Questionnaire for each participant during the first five months within the HI-SEAS habitat. Top graph shows control data. Bottom graph shows ANSIBLE data. Quadratic trend lines are shown (Color figure online).

ANSIBLE Usability.

The ANSIBLE system is undergoing constant development and improvements as content is added. To determine the success of these improvements under continued system use we performed Pearson’s Correlation analyses between questionnaires related to ANSIBLE use and time in the HI-SEAS habitat. We found a significant positive correlation between time and SUS total, r = .38, p < .01, and a sum of our ANSIBLE specific usability questions, r = .37, p < .01. We also found a negative correlation between SFQ total and time, r = −.42, p < .01, see Fig. 6.
Fig. 6.

Correlation between time in the HI-SEAS habitat and questionnaires related to ANSIBLE use. Linear trend lines are shown (Color figure online).

4 Summary and Conclusions

A main focus of the ANSIBLE ecosystem is to maintain the social connection between crew and family while crew members are participating in an isolated, long duration environments. Given this we expected ANSIBLE to impact the perception of friend and family relationships and not crew relationships. Accordingly we found ANSIBLE increased feelings of closeness and satisfaction with friends and family but not crew members, according to responses on the Circles of Closeness Questionnaire. Additionally we found ANSIBLE marginally decreased feelings of stress, a known consequence of social isolation. Given the significant trends of increased usability as ANSIBLE is used and improved upon, we expect the benefits of ANSIBLE to continue to grow.

This study is a first step in quantifying the considerable potential of communications through virtual environments in aiding isolated populations, particularly over prolonged durations such as future space exploration missions. However, the results are preliminary and data collection is ongoing. Future analyses will further examine the effects of ANSIBLE use on social connections compared to controls as well as deeper investigations into the within ANSIBLE group effects and ramifications of ANSIBLE use on friends and family. Further, additional analyses will examine the impact of the system to aid other expected consequences of the sensory monotony and extreme work and living conditions anticipated during long duration space missions are also planned, such as interventions to combat sensory and social monotony, social behavior modification, and social anxiety countermeasures.

Notes

Acknowledgments

The above work was sponsored by NASA’s Human Research Program under contract #NNX14CJ06C. We would like to thank NASA personnel Lauren Leventon, Laura Bollweg, Jason Schneiderman, Diana Arias, Brandon Vessey, Al Holland, and Ron Moomaw for their oversight and direction. We would also like to thank the HI-SEAS crew members and their family and friends for their support.

References

  1. 1.
    Brewer, M.B.: Taking the social origins of human nature seriously: toward a more imperialist social psychology. Pers. Soc. Psychol. Rev. 8, 107–113 (2004)CrossRefGoogle Scholar
  2. 2.
    House, J.S., Landis, K.R., Umberson, D.: Social relationships and health. Science 241(4865), 540–545 (2004)CrossRefGoogle Scholar
  3. 3.
    Otto, C.: South Pole Station: an analogue for human performance during long-duration missions to isolated and confined environments: neurobiology, neurochemistry, and neurostructural changes in humans during prolonged isolation and confinement (White Paper). NASA Johnson Space Center, Houston, TX: NASA JSC Behavioral Health and Performance Human Research Program and Space Medicine Division (2007)Google Scholar
  4. 4.
    Hawkley, L.C., Cacioppo, J.T.: Loneliness matters: a theoretical and empirical review of consequences and mechanisms. Ann. Behav. Med. 40(2), 218–227 (2010)CrossRefGoogle Scholar
  5. 5.
    Cacioppo, J.T., Hawkley, L.C.: Perceived social isolation and cognition. Trends Cogn. Sci. 13(10), 447–454 (2009)CrossRefGoogle Scholar
  6. 6.
    Castronova, E., Wagner, G.G.: Virtual life satisfaction. Int. Rev. Soc. Sci. 64(3), 313–328 (2011)Google Scholar
  7. 7.
    Gentile, D.A., Anderson, C.A., Yukawa, S., Ihori, N., Saleem, M., et al.: The effects of prosocial video games on prosocial behaviors: international evidence from correlational, longitudinal, and experimental studies. Pers. Soc. Psychol. Bull. 35, 752–763 (2009)CrossRefGoogle Scholar
  8. 8.
    Rosenberg, R., Baughman, S., Bailenson, J.N.: Virtual superheroes: using superpowers in virtual reality to encourage prosocial behavior. PLoS ONE 8(1), e55003 (2013)CrossRefGoogle Scholar
  9. 9.
    Greitemeyer, T., Osswald, S., Brauer, M.: Playing prosocial video games increases empathy and decreases schadenfreude. Emotion 10, 796–802 (2010)CrossRefGoogle Scholar
  10. 10.
    Gillath, O., McCall, C.A., Shaver, P., Blascovich, J.: Reactions to a needy virtual person: using an immersive virtual environment to measure prosocial tendencies. Media Psychol. 11, 259–282 (2008)CrossRefGoogle Scholar
  11. 11.
    Wu, P., Morie, J.F., Chance, E., Haynes, K., Hamell, J., Wall, P., Ladwig, J., Ott, T.: Maintaining psycho-social health on the way to mars and back. In: Virtual Reality International Conference (VRIC), 8–10 April, Laval, France (2015)Google Scholar
  12. 12.
    Cacioppo, J.T., Patrick, W.: Loneliness: Human Nature and the Need for Social Connection. WW Norton & Company, New York (2008)Google Scholar
  13. 13.
    Hawkley, L.C., Hughes, M.E., Waite, L.J., Masi, C.M., Thisted, R.A., Cacioppo, J.T.: From social structural factors to perceptions of relationship quality and loneliness: the Chicago health, aging, and social relations study. J. Gerontol. B Psychol. Sci. Soc. Sci. 63(6), S375–S384 (2008)CrossRefGoogle Scholar
  14. 14.
    Palinkas, L.A., Johnson, J.C., Boster, J.S.: Social support and depressed mood in isolated and confined environments. Acta Astronaut. 54(9), 639–647 (2004)CrossRefGoogle Scholar
  15. 15.
    Uleman, J.S., Rhee, E., Bardoliwalla, N., Semin, G., Toyama, M.: The relational self: closeness to ingroups depends on who they are, culture, and the type of closeness. Asian J. Soc. Psychol. 3(1), 1–17 (2000)CrossRefGoogle Scholar
  16. 16.
    Bachman, K.R.O.B., Otto, C., Leveton, L.: Countermeasures to Mitigate the Negative Impact of Sensory Deprivation and Social Isolation in Long-Duration Space Flight. NASA/TM-2012-217365 (2012)Google Scholar
  17. 17.
    Levenstein, S., Prantera, C., Varvo, V., Scribano, M.L., Berto, E., Luzi, C., Andreoli, A.: Development of the Perceived Stress Questionnaire: a new tool for psychosomatic research. J. Psychosom. Res. 37(1), 19–32 (1993)CrossRefGoogle Scholar
  18. 18.
    Kizony, R., Katz, N., Rand, D., Weiss, P.L.T.: Short Feedback Questionnaire (SFQ) to enhance client-centered participation in virtual environments. Cyberpsychology Behav. 9(6), 687–688 (2006)Google Scholar
  19. 19.
    Witmer, B.G., Singer, M.J.: Measuring presence in virtual environments: a presence questionnaire. Presence Teleoperators Virtual Environ. 7(3), 225–240 (1998)CrossRefGoogle Scholar
  20. 20.
    Kizony, R., Raz, L., Katz, N., Weingarden, H., Weiss, P.L.T.: Video-capture virtual reality system for patients with paraplegic spinal cord injury. J. Rehabil. Res. Dev. 42(5), 595–608 (2005)CrossRefGoogle Scholar
  21. 21.
    Brooke, J.: SUS-a quick and dirty usability scale. Usability Eval. Ind. 189(194), 4–7 (1996)Google Scholar

Copyright information

© Springer International Publishing Switzerland 2016

Authors and Affiliations

  • Tammy Ott
    • 1
  • Peggy Wu
    • 1
    Email author
  • Jacki Morie
    • 2
  • Peter Wall
    • 1
  • Jack Ladwig
    • 1
  • Eric Chance
    • 2
  • Kip Haynes
    • 2
  • Bryan Bell
    • 1
  • Christopher Miller
    • 1
  • Kim Binsted
    • 3
  1. 1.SIFT, LLCMinneapolisUSA
  2. 2.All These Worlds, LLCLos AngelesUSA
  3. 3.Univerisity of HawaiiHonoluluUSA

Personalised recommendations