You may notice some images loading slow across the Open Collections website. Thank you for your patience as we rebuild the cache to make images load faster.

Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The effects of immersion and increased cognitive load on time estimation in a virtual reality environment Ghomi, Mehrdad 2018

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata


24-ubc_2018_november_ghomi_mehrdad.pdf [ 2.21MB ]
JSON: 24-1.0372785.json
JSON-LD: 24-1.0372785-ld.json
RDF/XML (Pretty): 24-1.0372785-rdf.xml
RDF/JSON: 24-1.0372785-rdf.json
Turtle: 24-1.0372785-turtle.txt
N-Triples: 24-1.0372785-rdf-ntriples.txt
Original Record: 24-1.0372785-source.json
Full Text

Full Text

The Effects of Immersion and Increased Cognitive Loadon Time Estimation in a Virtual Reality EnvironmentbyMehrdad GhomiB.Sc, The University of British Columbia, 2015A THESIS SUBMITTED IN PARTIAL FULFILLMENTOF THE REQUIREMENTS FOR THE DEGREE OFMASTER OF APPLIED SCIENCEinTHE FACULTY OF GRADUATE AND POSTDOCTORALSTUDIES(Electrical and Computer Engineering)The University of British Columbia(Vancouver)October 2018c©Mehrdad Ghomi, 2018The following individuals certify that they have read, and recommend to theFaculty of Graduate and Postdoctoral Studies for acceptance, the thesis entitled:”The Effects of Immersion and Increased Cognitive Load on TimeEstimation in a Virtual Reality Environment”Submitted by Mehrdad Ghomi in partial fulfillment of the requirements forthe degree of Master of Applied Science in Electrical and Computer Engineer-ing.Examining Committee:Dr. Sid Fels (Co-supervisor, Electrical and Computer Engineering)Dr. Bernie Garrett (co-supervisor, Nursing)Dr. Mathew Yedlin (Committee Chair, Electrical and Computer Engineering)iiAbstractThe perceived duration of a time interval can seem shorter or longer relative to realtime (i.e., solar time or clock time) depending on what fills that time interval. Re-search has suggested that increased immersion alters a users ability to reproducea given duration whilst doing a simple task or playing a game in an ImmersiveVirtual Environment (IVE). Virtual Reality (VR) allows users to experience vir-tual environments similar to the real world. The contribution of this experimentalresearch is to explore the effects of undertaking a cognitive spatial task and immer-sion within a VR environment on a persons perception of time. A VR experienceusing a cognitive task (maze navigation) was compared with a non VR (control)experience of the same task to explore if the effects exist and if the effects are moresignificant in an IVE compared to a screen-based simple multimedia experience.Also, a VR experience of the environment without any task was compared to thesame environment with the cognitive task to establish the effect of a spatial cogni-tive task on temporal perception. More specifically, this study measured how muchtemporal distortion is achievable utilizing cognitive tasks in a VR experience. Inthis thesis the use of cognitive tasks and VR are the independent variables and theperceived duration of the experiment (time) is the dependent variable. Obtaineddata suggest that being immersed in a VR experience results in 16.10% underesti-mation of time, while a non-VR experience results in 7.5% overestimation of time.Moreover, navigating mazes that involve a high cognitive load results in 6.45%underestimation of time. Finally, the combination of VR and high cognitive load(navigating the mazes without guiding lines in a VR experience) result in 22.18%underestimation of time. Finally, the implications of this research are discussed atthe end of this thesis.iiiLay SummaryVirtual reality (VR) technologies have granted users the possibility of experiencingvirtual environments in similar ways as experiencing the real world. One areaof interest in the development of VR is the way in which people perceive timewithin a VR experience. The goal of this study is to identify whether immersionwithin VR (using head mounted devices) and cognition load (performing spatialcognitive tasks) result in the underestimation of the passage of time by the user. Thecontribution of this research is the investigation of the effects of cognitive load andimmersion on our perception of time. The results suggest that increased immersionin VR and cognitive activity (undertaking spatial tasks, such as navigation) reducesthe perceived duration of time.ivPrefaceThis thesis is original, unpublished, independent work by the author. In the secondchapter (Background Knowledge), sections 2.1 (Factors that influence temporalperception) and 2.2 (VR and related work on temporal perception) contain descrip-tions of other research, which are cited.Dr. Sidney Fels and Dr. Bernie Garrett were the supervisors of the research.The Statistical Opportunity for Students (SOS) program experts from University ofBritish Columbia (UBC) Department of Statistics helped me with my data analysisin chapter 3.6 of this thesis.The necessary ethical review was requested from the UBC Behavioral ResearchEthics Board (BREB) and approval was obtained (H17-00106) (See Appendix C)before recruitment for the experiment.vTable of ContentsAbstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiiLay Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ivPreface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vTable of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viList of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ixList of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xGlossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiAcknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiiDedication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1 Problem Identification . . . . . . . . . . . . . . . . . . . . . . . 11.2 Statement of the Purpose . . . . . . . . . . . . . . . . . . . . . . 31.3 Research Questions . . . . . . . . . . . . . . . . . . . . . . . . . 41.4 Investigation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51.5 Significance of the Study . . . . . . . . . . . . . . . . . . . . . . 52 Background Knowledge . . . . . . . . . . . . . . . . . . . . . . . . . 72.1 Factors That Influence Temporal Perception . . . . . . . . . . . . 9vi2.2 VR and Related Work on Temporal Perception . . . . . . . . . . . 143 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183.1 Experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193.1.1 Experimental Design . . . . . . . . . . . . . . . . . . . . 203.1.2 Ethical Review . . . . . . . . . . . . . . . . . . . . . . . 203.2 Participant Recruitment . . . . . . . . . . . . . . . . . . . . . . . 213.2.1 Sampling Plan . . . . . . . . . . . . . . . . . . . . . . . 213.2.2 Inclusion and Exclusion Criteria . . . . . . . . . . . . . . 213.2.3 Recruitment Methods . . . . . . . . . . . . . . . . . . . . 223.3 Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223.3.1 Location . . . . . . . . . . . . . . . . . . . . . . . . . . 223.3.2 Hardware . . . . . . . . . . . . . . . . . . . . . . . . . . 223.3.3 Software . . . . . . . . . . . . . . . . . . . . . . . . . . 233.4 Instruments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243.4.1 Demographic Questionnaire . . . . . . . . . . . . . . . . 243.4.2 Post-experience Questionnaires . . . . . . . . . . . . . . 253.4.3 Terminal Interview . . . . . . . . . . . . . . . . . . . . . 253.4.4 Data Log . . . . . . . . . . . . . . . . . . . . . . . . . . 263.5 Procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263.6 Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293.6.1 Quantitative Analysis . . . . . . . . . . . . . . . . . . . . 293.6.2 Linear Mixed Effects Model . . . . . . . . . . . . . . . . 313.6.3 Qualitative Analysis . . . . . . . . . . . . . . . . . . . . 344 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 354.1 Quantitative Data . . . . . . . . . . . . . . . . . . . . . . . . . . 354.1.1 Descriptive Univariate Statistics . . . . . . . . . . . . . . 354.1.2 Inferential Statistics . . . . . . . . . . . . . . . . . . . . 364.2 Qualitative Data . . . . . . . . . . . . . . . . . . . . . . . . . . . 394.2.1 Demographic Questionnaires . . . . . . . . . . . . . . . . 394.2.2 Post-Experience Questionnaire Data . . . . . . . . . . . . 404.2.3 Terminal Interview Data . . . . . . . . . . . . . . . . . . 40vii5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 425.1 Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 476 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 496.1 Implications and Future Work . . . . . . . . . . . . . . . . . . . 50Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52Appendix A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57Appendix B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58Appendix C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61Appendix D . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65Appendix E . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67viiiList of TablesTable 3.1 Four Experiment Conditions . . . . . . . . . . . . . . . . . . . 19Table 3.2 Latin Square Design . . . . . . . . . . . . . . . . . . . . . . . 27Table 4.1 Results from the four runs of the experiment for participant 1 . 38Table 4.2 Results of the R Code for Various Experiment Conditions (ABolded P-value indicates Significant Effect) . . . . . . . . . . 38Table 4.3 Age Distribution of Participants . . . . . . . . . . . . . . . . . 38Table 4.4 Gender Distribution of Participants . . . . . . . . . . . . . . . 39Table D.1 Factors That Effect Perception of Time . . . . . . . . . . . . . 66Table E.1 User Study Data . . . . . . . . . . . . . . . . . . . . . . . . . 72ixList of FiguresFigure 3.1 HTC Vive Setup and Equipment . . . . . . . . . . . . . . . . 23Figure 3.2 Top View of the Tutorial Maze with Guiding Lines . . . . . . 28Figure 3.3 Top View of the Lobby (four Portals to the four main Mazes) . 29Figure 3.4 View of a Maze With Guiding Lines, Pictures and Voice Overs 30Figure 4.1 Error Distribution of all Trials (The Horizontal Axis Shows theAbsolute Error in Seconds, Vertical Axis Shows the ParticipantCount) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37Figure 4.2 Actual and Perceived Time on Various Experiment Conditions 40Figure 5.1 Actual, Perceived and Mean Error for all Four Mazes . . . . . 43Figure 5.2 Histogram of Perceived Over Actual Ratio Distribution . . . . 48Figure B.1 Top View of All 4 Identical Main Mazes (2 with guiding linesand pictures, 2 without lines and pictures) . . . . . . . . . . . 59Figure B.2 Players View of a Maze . . . . . . . . . . . . . . . . . . . . . 60Figure C.1 Demographic Questionnaire . . . . . . . . . . . . . . . . . . 62Figure C.2 Post Exposure VR Experience Questionnaire . . . . . . . . . 63Figure C.3 Post Exposure Semi-Structured Interview Questions . . . . . 64xGlossary3D Three-DimensionalADHD Attention Deficit Hyperactivity DisorderANOVA Analysis of VarianceBREB Behavioral Research Ethics BoardDF Degree of FreedomECE Electrical and Computer EngineeringGHZ GigahertzGVR Game VR EnvironmentsHMD Head Mounted DisplayIVE Immersive Virtual EnvironmentLME Linear Mixed EffectsRDLPFC Right Dorsolateral Prefrontal CortexRTMS Repetitive Transcranial Magnetic StimulationSOS Statistical Opportunity for StudentsUBC University of British ColumbiaVR Virtual RealityxiAcknowledgmentsI want to thank my parents for making it possible for me to be able to pursue myacademic career up to this point.I would like to show my gratitude to Dr. Sidney Fels and Dr. Bernie Garrettfor sharing their pearls of wisdom with me during the course of this research, andI thank all of the 34 participants who participated in my user study and gave mevery useful insights on the research. I want to thank Dr. Jim Little and Dr. DavidPoole for helping me with participant recruitment. I am also immensely grateful toDr. Joseph Anthony for his comments on earlier versions of this research. I wouldlike to mention that any errors are my own and should not tarnish the reputationsof these esteemed persons. I would also like to thank my colleagues from the Uni-versity of British Columbia, departments of Electrical and Computer Engineeringand School of Nursing who provided insight and expertise that greatly assisted thisresearch.Finally, I thank the Department of Statistics for assistance with data analysisand suggestion of the Linear Mixed Effects Model, and also for comments thatgreatly improved this research.xiiDedicationI dedicate all my work to my parents, who have supported me through life, withoutwhom none of my success would be possible.xiiiChapter 1IntroductionWhen a man sits with a pretty girl for an hour, it seems like a minute.Let him sit on a hot stove for a minute and it’s longer than any hour.— Albert Einstein1.1 Problem IdentificationThe experience of time pervades every aspect of our lives. Duration is an important,basic aspect of the temporal experience [16]. Although there are some indicationsthat being immersed within a virtual environment does alter a user’s perception oftime, the nature of time perception within virtual environments represents a fieldwith little published research, to date. The perceived duration of a time intervalcan seem shorter or longer relative to real time (i.e., solar time or clock time) de-pending on what fills that time interval. As an example, auditory impulses mayinteract with our bodys temporal oscillator that works like a pacemaker, and re-sult in perturbations in our estimation of time. Higher frequencies are known tobe associated with the overestimation of temporal duration (thinking an event hastaken longer than it actually has) while lower frequencies are associated with theunderestimation of temporal duration (thinking an event has been shorter than itactually has) [39].There are two paradigms to consider when researching the perception of time:Prospective, where a person is aware that they needed to make a time estimatebefore experiencing an event, and retrospective, where a person is unaware of the1need for a time estimate until after the event has passed. The prospective paradigmis known as experienced duration, while the retrospective paradigm is known asremembered duration.Researchers [34] illustrate that for prospective time estimation attention is nec-essary to monitor time passing. In some sense, the person is mentally counting theticking of some internal clock. However, the more a persons attention is requiredelsewhere, the more ticks they miss, and hence, they underestimate time [34]. Forretrospective time estimates, a person looks back over their memory of a specificduration and essentially counts the memories. The fewer contextual changes re-quiring distinct memories, the lower the estimated time is.One consistent finding across both paradigms is that if users are engaged inimmersive environments and are given two tasks to do, one temporal, and onenot, then there is a strong inference effect with the secondary task causing theestimation of time to become shorter, more variable, or more inaccurate. Researchhas also suggested that increased immersion alters a players ability to reproduce agiven duration whilst doing a simple task or playing a game in an immersive virtualenvironment [14].Virtual Reality (VR) allows users to experience virtual environments similar tothe real world. Nowadays different fields of applications such as training programs,immersive walk-throughs, architectural and industrial designs, and video gamesbenefit from VR technologies by creating virtual scenes. Head Mounted Display(HMD)s have allowed users to experience a sense of presence in virtual scenes bycombining motion-tracking and visual stimuli together, rendering graphical scenesin real time. A trend of the growing industry and market is observable due torelatively cheap hardware such as the Oculus Rift, HTC Vive or Samsung GearVR.With the rapid growth of VR and its ever-increasing number of applicationsin various fields, altering a user’s perception of time is a useful area for explo-ration. The alteration of perceived time could, for example, be beneficial duringlong flights where by distracting the brain, a person may underestimate the tem-poral duration of the flight. In general, shortening the perceived duration may bebeneficial for a variety of unpleasant circumstance.One area of interest in the development of VR is the way in which people2perceive time within VR itself. This research focuses on the implementation oftime in virtual environments and discusses how, by altering the implementationof the environment, it may be possible to affect a users perception of time andmanipulate it.1.2 Statement of the PurposeThe goal of this research is to explore if it is possible to alter users perception oftime while they are present in an immersive virtual environment (IVE). The timeperception distortion that we chose was underestimation of passage of time, mean-ing, users underestimated the amount of time they spent in the IVE. We decided toexplore the specific contribution of:1) Increased immersion within a virtual experience2) Increased cognitive load by undertaking a cognitive spatial taskA controlled experiment was designed and conducted in order to verify thelevel of contribution of immersion and cognitive spatial task on temporal under-estimation. Four virtual mazes were developed and 34 participants navigated themazes in order to find the exits of each maze. They solved the mazes with eitherthe presence or absence of a VR experience and spatial task (four runs were doneby each participant).The outcome of these 136 runs was analyzed using linear mixed effects modeland the quantitative results are reported in the fourth chapter of this thesis. More-over, demographic questionnaires, post-experiment questionnaires and terminal in-terviews were designed in order to obtain qualitative data as well. These results arealso reported in the fourth chapter of this thesis.Although overestimation of time has its own applications in areas such as videogame content development or general situations where it is preferable to stretchout the duration of a pleasant experience with minimal resources, the main focusof this thesis is to identify methods that can practically make VR technology usersunderestimate the passage of time while they are undergoing a virtual experience.Applications of time underestimation can include long travels on airplanesequipped with VR tools or shortening of any unpleasant experience. Implement-ing VR in a way that makes the user underestimate the passage of time can be3highly beneficial for uncomfortable medical treatments, such as for post-traumaticstress disorder or exposure therapies (fear of flying, claustrophobia, etc.). We canimprove the experience of VR pain treatments (such as burn victims) by reducingthe perceived passage of time in treatment periods. This can also help to shortenthe perceived duration of a workout on a treadmill or other exercising equipment.Further, this can be used through physical therapy sessions, in which the users canhave an enjoyable experience if the perceived duration of the session is reduced.The contribution of this experimental research is to explore the effects of im-mersion in a VR experience on a user’s perception of time while performing acognitive task in the environment.A VR experience using a spatial cognitive task was compared with a non-VRexperience (a simple screen-based multimedia experience using the same task, asa control) to explore if an effect on temporal perception existed. In addition, aVR experience of the environment without any task was compared to the sameenvironment with a spatial cognitive task, to establish any effect of the task ontemporal perception.More specifically, the study was designed to measure how much temporal dis-tortion was achievable utilizing a cognitive task in a VR experience. In this re-search, the use of VR and a cognitive spatial task were independent variables andthe perceived duration of the experiment (perceived time) was the dependent vari-able.1.3 Research QuestionsThis work sought to address the following questions:1) To what degree does immersion within a VR experience decrease the per-ceived passage of time (underestimation) compared to a non-VR multimedia screen-based experience?2) To what degree does undertaking a cognitive spatial task decrease the per-ceived passage of time (underestimation) in a VR experience?To address the above questions, the effects on temporal perception in a VRexperience with a simple screen-based multimedia version of the same experience(the control) were compared to test whether VR immersion indeed makes a dif-4ference. The effects on temporal perception of introducing a cognitive spatial taskinto the VR experience was also tested.1.4 InvestigationThe two main factors investigated in this study were the effects of VR immersionand cognitive load on temporal underestimation in virtual environments.In order to investigate whether immersion and cognitive load had a significanteffect on temporal underestimation in virtual environments, an experimental studywas conducted with 34 participants. The participants were asked to navigate fourmazes. Two of the mazes provided solution guidelines on the floor, showing thecorrect path (requiring a minimal cognitive load), while the other two were typicalmazes that the participants needed to navigate and solve (using a high spatialcognitive load). The participants were asked to solve one maze with the solutionand one without the solution in a VR experience, and another two mazes (one withsolution and one without solution) in a simple multimedia screen-based experience.After each maze the participants were asked to fill a questionnaire in whichone of the questions asked for their estimated duration for completing the maze.This number was compared with the actual logged duration. The Linear MixedEffects Model method was used to test whether immersion and cognition load hadany significant effects on temporal underestimation.1.5 Significance of the StudyOur gathered and analyzed data suggest that being immersed within a VR experi-ence results in 16.10% underestimation of time, while a non-VR experience resultsin 7.50% overestimation of time. Also, our data suggests that navigating mazes in-volving a high cognitive load results in 6.45% underestimation of time. Finally, thecombination of VR and high cognitive load (navigating the mazes without guidinglines in a VR experience) resulted in 22.18% underestimation of time in our study.The contribution of this thesis indicates that by using a combination of in-creased immersion and increased cognitive load, a significant underestimation oftime (18.11%) is achievable. These two factors are not only significantly effective,but also practical and feasible to implement.5To summarize the contributions of this thesis:1) It was found that being immersed within a virtual experience significantlyaffects perception of time and results in users’ underestimation of the passage oftime.2) It was found that undertaking a cognitive spatial task did not result in signif-icant underestimation of time on its own.3) The most underestimation occurred when both immersion within a VR ex-perience and a cognitive spatial task were present.4) Although undertaking a cognitive spatial task did not result in significantunderestimation of time on its own, when the task was performed in a VR environ-ment, users perception of time was significantly affected and underestimation wasthe result.5) It was found that factors such as user frustration can contribute to user over-estimation of time.6Chapter 2Background KnowledgePsychologists and neuroscientists believe that humans have several complementarysystems involving the cerebral cortex, cerebellum and basal ganglia that governour perception of time [31]. Some cell clusters appear to be capable of short-rangetimekeeping (ultradian rhythm), while the suprachiasmatic nucleus is responsiblefor the circadian rhythm (daily) [15]. On the other hand, in physics time is unam-biguous and is defined as what a clock reads [24]. Time can also be introducedusing an operational definition; where observing a certain number of repetitionsof one cyclical event, such as the passage of a free-swinging pendulum, creates astandard unit such as a second.The term specious present refers to the time duration wherein one’s perceptionsare considered to be in the present. This term was introduced by E.R Clay (E.Robert Kelly) in 1882 and further developed by William James in 1890 [2]. Inrelation, time perception refers to the sense of time, which differs from other sensesas time cannot be directly perceived but must be reconstructed by the brain.Time is understood by one’s own perception of the duration of events unfold-ing. The perceived time interval between two successive events is referred to as”perceived duration”. Another person’s perception of time cannot be directly ex-perienced or understood, but it can be objectively studied and inferred throughscientific experiments. Time perception is, therefore, a construction of the brainthat is manipulable and distortable under certain circumstances. Such temporal il-lusions help to expose the underlying neural mechanisms working together in order7to let us perceive the passage of time.In the field of psychology, experimental studies of time perception have wellestablished that estimates of the duration of a stimulus do not always match its ob-jective time interval, and can be affected by a variety of factors. Since time cannotbe directly measured at a given moment in the brain, the mind is often assumed toestimate time based on internal biological or psychological events, or external sig-nals. The effect of exogenous cues (i.e. Zeitgebers or time symbols) from the localenvironment on endogenous biological clocks (e.g. circadian rhythms) is studiedin the field of chronobiology.It is possible that differences in exogenous time cues between those occurringin real world phenomena and those in virtual environments have an effect on inter-nal human time perception. In particular, system latency is known to change theperception of sensory synchronicity and can degrade the perceptual stability of theenvironment. Space and time are interdependent phenomena not only in physics,but also in human perception [5].Further essential background knowledge for this research is how VR, immer-sion and presence are defined in the context of this research. In literature, univer-sally accepted definitions for these phenomenons are yet to emerge, but here arethe definitions we found suitable for the purpose of this research from a reviewarticle [11]:VR involves an artificial 3-dimensional (3D) environment that is experiencedby a person through sensory stimuli (usually visual, aural, and often touch) deliv-ered by a computer and in which ones actions partially determine what happens inthe environment [11].The sense of immersion in an immersive virtual reality (IVR) environment isachieved through visual and auditory stimuli that simulate 3D visual and auditorycues available in the real world. Visually, this is delivered to the user with a head-mounted display (HMD), which presents the computer generated imagery (CGI)of the VR scene from the perspective of each of the users eyes. The HMD usuallydisplays stereoptic (3D) imagery and tracks head motion so that the user seems tomove naturally around the virtual space and observe it in a natural manner. Thus,stereoscopic imagery is presented with the 3D visual depth cues of occlusion, per-spective, motion parallax, and natural surface textures, all updated interactively8in real time. Audio is also simulated in 3D with a head-related transfer function,which enables the HMD wearer to locate simulated sound at a real location inspace. Together, this enables the user to gain a sense of immersion inside the 3Dvirtual world, by presenting the illusion of a 3D scene everywhere the user looks[11].Presence refers to the sense of being within an environment that is generatedby technically mediated means. VR involves human experience in which 2 tech-nological dimensions are considered to contribute to a sense of presence. The firstdimension is vividness, or the production of a sensorially richmediated environ-ment. The second is interactivity, defined as a users ability to engage with theenvironment and modify its form or alter events through interaction with it. Animmersive environment is considered to be a computed environment that elicits ausers sense of presence or ”being there” [11].2.1 Factors That Influence Temporal PerceptionResearchers have identified various environmental and non-environmental factorsthat may alter ones perception of time.Psychoactive Drugs: These drugs can alter the judgement of time. Stimulantsare known to produce overestimates of time duration, whereas depressants andanesthetics produce underestimates of time duration [46]. Psychoactive substancesinclude traditional psychedelics such as LSD, psilocybin, and mescaline as wellas the dissociative class of psychedelics such as PCP, ketamine and dextromethor-phan. Researchers found psilocybin significantly impaired the ability to reproduceinterval durations longer than 2.5 seconds, as well as significantly impairing syn-chronization of motor actions (taps on a computer keyboard) to regularly occurringtones, and impaired the ability to keep tempo when asked to tap on a key at aself-paced but consistent interval [45].Clinical Disorders such as Parkinson’s disease, schizophrenia, and AttentionDeficit Hyperactivity Disorder (ADHD) have also been linked to noticeable im-pairments in time perception. Neuropharmacological research indicates that theinternal clock, used to time durations in the seconds-to-minutes range, is linked todopamine function in the basal ganglia [7].9Repetitive Transcranial Magnetic Stimulation: The Right Dorsolateral Pre-frontal Cortex (RDLPFC) may be important in time perception in humans. In thepresent studies, a virtual lesion of the rDLPFC created by Repetitive TranscranialMagnetic Stimulation (RTMS) has led to underestimation of time perception forbrief intervals (lasting a few seconds) in working memory [20].Aging: Psychologists have found that the subjective perception of the passingof time tends to speed up with increasing age in humans. Aging may cause peopleto increasingly underestimate a given interval of time in older people. This fact canlikely be attributed to a variety of age-related changes in the aging brain, such asthe lowering in dopaminergic levels with older age [8].Music: The perceived duration of a time period may be influenced by proper-ties of environmental stimuli that fill the period. Findings suggest that perceptionsof duration are influenced by music in a way that contradicts conventional wis-dom (i.e., the ”time flies when you’re having fun” hypothesis). Time did not passquickly when an interval was filled with affectively positive stimulation and par-ticipants overestimated the interval. Perceived duration was longest for subjectsexposed to positively valenced (major key) music, and shortest for negatively va-lenced (atonal) music. Music pitched in a major key produced the longest averageduration estimates and the greatest disparity between actual (i.e., clock) time andperceived time. Music pitched in minor key produced a significantly shorter aver-age duration estimate. Atonal music produced the shortest and the most accurateestimations [19].Emotional States:Awe: Research has suggested that the feeling of awe has the ability to ex-pand one’s perceptions of time availability [33]. Awe can be characterized as anexperience of immense perceptual vastness that coincides with an increase in fo-cus. Consequently, it is conceivable that one’s temporal perception may slow downwhen experiencing awe [33].Fear: Research suggests that perceived time seems to slow down when a per-son sky-dives or bungee jumps [42], or when a person suddenly and unexpectedlysenses the presence of a potential predator or mate. This reported slowing in tem-poral perception may have been evolutionarily advantageous because it may haveenhanced our ability to intelligibly make quick decisions in moments that were of10critical importance to our survival [38].Depression: Depression may increase one’s ability to perceive time accurately.One study assessed this concept by asking subjects to estimate the amount of timethat passed during intervals ranging from 3 seconds to 65 seconds [21]. Resultsindicated that depressed subjects more accurately estimated the amount of timethat had passed than non-depressed patients; non-depressed subjects overestimatedthe passing of time. This difference was hypothesized to be because depressedsubjects focused less on external factors that may skew their judgment of time.Temporal Illusions:Vierordts Law: This law indicates that short time intervals tend to be over-estimated and long ones underestimated, the indifference interval being the inter-mediate length that is neither overestimated nor underestimated, usually found byexperiment to be in the region of 0.7 seconds [1].Kappa Effect: The Kappa effect (also known as perceptual time dilation) is aform of temporal illusion verified by experiment [43], wherein the temporal dura-tion between a sequence of consecutive stimuli is thought to be relatively longeror shorter than its actual elapsed time, due to the spatial/auditory/tactile separationbetween each consecutive stimuli. Brain biases perception in favor of expectation.Specifically, the results suggest that the brain automatically incorporates prior ex-pectation for speed in order to overcome spatial and temporal imprecision inherentin the sensorineural signal [12]. The kappa effect can be displayed by consideringa journey made in two parts that take an equal amount of time. Between thesetwo parts, the journey that covers more distance may appear to take longer thanthe journey covering less distance, even though they take an equal amount of time(also closely related to the Tau Effect) [18, 43].Chronostasis: A type of temporal illusion in which the first impression ofa new event or task to the brain appears to be extended in time. For example,Chronostasis temporarily occurs when fixating on a target stimulus, immediatelyfollowing a saccade (e.g., quick eye movement). It results in an overestimation inthe temporal duration for which that target stimulus (i.e., post saccadic stimulus)was perceived. This effect can extend apparent durations by up to 500 ms and isconsistent with the idea that the visual system models events prior to perception[47].11The most well-known version of this illusion is known as the stopped-clockillusion, wherein a subject’s first impression of the second-hand movement of ananalog clock, subsequent to one’s directed attention (i.e., saccade) to the clock, isthe perception of a slower-than-normal second-hand movement rate (the secondshand of the clock may seemingly temporarily freeze in place after initially lookingat it) [47]. The occurrence of chronostasis also extends beyond the visual domaininto the auditory and tactile domains.In the auditory domain, chronostasis and duration overestimation occur whenobserving auditory stimuli. One common example is a frequent occurrence whenmaking telephone calls. If, while listening to the phone’s dial tone, research sub-jects move the phone from one ear to the other, the length of time between ringsappears longer [17]. In the tactile domain, chronostasis has persisted in researchsubjects as they reach for and grasp objects. After grasping a new object, subjectsoverestimate the time in which their hand has been in contact with this object. Inother experiments, subjects turning a light on with a button were conditioned toexperience the light before the button press [28].The Oddball Effect: Humans typically overestimate the perceived durationof the initial event in a stream of identical events and unexpected oddball stim-uli seem to be perceived as longer in duration, relative to expected or frequentlypresented standard stimuli [32]. The oddball effect may serve an evolutionarilyadapted alerting function and is consistent with reports of time slowing down inthreatening situations, which is similar to the chronostasis phenomenon. The effectseems to be strongest for images that are expanding in size on the retina (imagesthat are looming or approaching the viewer), and the effect can be eradicated foroddballs that are contracting or perceived to be receding from the viewer [41].Cognitive Task (Spatial Cognition):One of the most effective means of altering perception of time is to increasespatial cognitive load [35]. Spatial ability or visuo-spatial ability is the capacityto understand, reason and remember the spatial relationships amongst objects orspace. Visual-spatial abilities are used everyday from navigation, understanding orfixing equipment, understanding or estimating distance and measurement, and psy-chomotor job performance. Spatial working memory is the ability to temporarilystore visual-spatial memories under attentional control, in order to complete a task12[9].This cognitive ability mediates individual differences in the capacity for higherlevel spatial abilities, such as mental rotation. Spatial working memory involvesstoring large amounts of short-term spatial memories in the form of a visuo-spatialsketchpad. It is used in the temporary storage and manipulation of visual-spatialinformation, such as memorizing shapes, colors, location or motion of objects inspace. It is also involved in tasks which consist of planning spatial movements,like determining one’s route through a complex building or maze.There exists a group of brain regions engaged in both time perception tasks andduring tasks requiring spatial cognitive effort. Thus, brain regions associated withworking memory and executive functions have been found to be engaged duringtime estimation tasks, and regions associated with time perception were found tobe engaged by an increase in the difficulty of non-temporal spatial cognitive tasks.The implication is that temporal perception and cognitive processes demandingcognitive control become interlinked when there is an increase in the level of cog-nitive effort demanded [3].Regarding the relationship between the neural mechanisms of time perceptionand other functions, studies on the prefrontal cortex evidence the implication ofthe same dorsolateral prefrontal cells for both cognitive timing and working mem-ory [37]. Also, the neural mechanisms of timing are recruited in a manner that ismodulated by degree of attention [30]. Therefore, there is reason to postulate thatcorrect executive functioning and cognitive control requires participation of bothfunctional, and neuroanatomical components of time perception. In fact, time per-ception and other executive components such as interference control, seem to sharea common neuroanatomical basis in early developmental stages [6].According to Block and Zakay [3], it was concluded that activation of severalcortical (supplementary motor area, insula/operculum, rDLPFC) and subcorticalregions (thalamus and striatum) during timing tasks is load-dependent. Addition-ally, activation of the dorsolateral prefrontal cortex under conditions of minimalworking memory involvement was observed. These findings support the specificinvolvement of this region in temporal processing rather than a more general in-volvement in working memory. Moreover, the overlap between regions participat-ing in both time perception and executive functions (performing spatial cognitive13tasks) could also indicate that both functions require similar cognitive abilities,such as sustained attention over time, maintaining information in working mem-ory, and making decisions and preparing motor responses. These findings are alsoconsistent with results from two previous meta-analyses carried out independentlyto explore the neuroanatomical basis of time perception and cognitive load [26, 29].During time perception tasks there is participation of various cognitive processes(such as working memory or executive functions). In a parallel manner, duringnon-temporal cognitive tasks with various levels of cognitive effort, some level oftemporal processing is also needed and engaged, therefore, brain regions tradition-ally associated with working memory and executive functions. Additionally, spe-cific regions traditionally associated with time perception (such as the insula andthe putamen) would be engaged during non-temporal cognitive tasks in responseto increases in difficulty level.Other Factors Identified as Influencing Temporal Perception:Stimuli Intensity: Time durations may appear longer with greater stimulusintensity (e.g., auditory loudness or pitch) [23].Higher Change Rates: Time intervals associated with more changes may beperceived as longer than intervals with fewer changes [4].Body Temperature: The chemical clock hypothesis implies a causal link be-tween body temperature and the perception of duration [44].Immersion: Studies [14] suggest that increased immersion alters users abil-ity to reproduce a given duration whilst doing a simple task or playing a game.Auditory Stimuli: Auditory stimuli may appear to last longer than visual stimuli[22, 27].Table-E.1 (See Appendix E) will illustrate various factors discussed above, splitinto underestimation and overestimation of temporal duration columns.2.2 VR and Related Work on Temporal PerceptionAs a relatively new technical innovation, the perception of temporal duration in VRhas not been extensively researched to date. However, environmental and cognitivefactors on time perception have been explored in other contexts.In research undertaken by a group of researchers at the University of Hamburg14in Germany, results showed that duration was underestimated when people engagedin performing cognitive tasks. They also showed that the use of time symbols (suchas the sun) in an environment could alter a persons perception of time. [35] Theycreated a virtual tropical island for both stereoscopic VR headsets and a regular PCscreen with sun as its time symbol in 3 cases, which were:1. A sun with no motion2. A sun that moved in a realistic 24-hour cycle3. A sun that progressed at double the speed of the real 24-hour sunThe results showed that when a user was just sitting without performing task,the speed at which the sun moved greatly affected perception of time, and the usertended to overestimate how much time was immersed in the VR for all time symbolconditions. Once cognitive tasks were introduced, the user tended to underestimatetime, and the sun manipulation made less of an impact since the user wasnt pro-cessing the setting as vividly. This was especially true for the spatial cognition test,likely because the user needed to draw resources from the same area of the brainprocessing the spatial passage of the sun.At the University of Oxford researchers have worked on the human bodys in-ternal clock and have found evidence for a temporal oscillator underlying timeperception with some estimates of its characteristic frequency. They suggestedthat auditory pulses interact with our bodys temporal oscillator, which works likea pacemaker and can result in perturbations in time estimation. [40] -Higher fre-quencies were associated with overestimation of temporal duration while lowerfrequencies were associated with underestimation.In another study, researchers from the Institute of Neuroinformatics, in Zurich,Switzerland, found that increased immersion altered players ability to reproducea given duration while performing a simple task or playing a game in a virtualenvironment [14].At the University of Duke, School of Nursing, researchers evaluated the Effectof VR on time perception in patients receiving chemotherapy. [36] In a trial usingVR as a distraction intervention, women with breast cancer were found more likelyto experience altered time perception (underestimation) and lung cancer patientswere found less likely. They concluded that VR is a non-invasive interventionthat can make chemotherapy treatments more tolerable. Understanding factors that15predict responses to interventions can help clinicians tailor coping strategies tomeet each patient’s needs.Another study at Eindhoven University of Technology aimed to investigatehow effective VR was in manipulating and eventually training time perceptionfor children with learning and/or behavior disorders [13]. Children with attentiondeficit hyperactivity disorder (ADHD) appear to have dysfunction in time orien-tation (overestimation of passage of time). As the interconnectivity of multiplebrain regions is involved in time perception, it was hypothesized that small dys-functions in these brain regions could be causing time perception problems, andcould be improved with training. The researchers at Eindhoven presented a the-oretical and empirical framework that used a VR time simulation game for timeperception training for children with ADHD. They concluded that children withtime-perception problems could be cured in their early years by such techniques.They believe that using Game VR Environments (GVR) for children with ADHDcould have some benefits in learning time perception [13].In another study published in the European Journal of Special Needs Educa-tion, researchers examined whether deaf and hard-of-hearing children perceived atemporal sequence differently under different representational modes. [30] Theycompared the effect of Three-Dimensional (3D) VR representations on sequentialtime perception among deaf and hard-of-hearing children with pictorial, textual,spoken and signed representations. They studied 69 participants aged 410, whowere divided into two age groups: kindergarten and school age. Using differentmodes of representation, they examined the childrens ability to arrange episodes ofa script in which a temporal order existed. They included six scripts that wereadapted to the different modes of representation and thus created a sum of 30scripts. Their findings demonstrated that the VR 3D representation and the signedrepresentation enabled the most accurate perception of sequential time. The poor-est results were for the textual representation. An interesting finding was that thetwo dimensional pictorial representation scored low, indicating that this form ofrepresentation was not as easy as expected.After exploring the factors that can alter time perception and the existing stud-ies in the relevant literature, combining immersion in VR with increased cognitiveactivity appears to be a useful area to explore in terms of assessing impacts on16influencing a person’s perception of time.17Chapter 3MethodsImmersion in VR is thought to affect temporal perception [35, 36]. Additionally,spatial cognitive tasks are thought to add to the cognitive load and therefore pro-mote an underestimation of time for the user [13]. Therefore, alongside immersionin VR, a cognitive task (spatial cognition) was selected as a suitable technique tostudy in a control experiment to explore if temporal perception can be manipulatedin a computer-generated environment.The spatial task that was chosen for the experiment was navigating and solv-ing a maze. Maze navigation is a classic example of spatial cognitive tasks as itis a branch of cognitive psychology, which focuses on how people acquire knowl-edge about their spatial surroundings to determine their location and to navigatetowards the goal (maze exit in this case). This particular cognitive task was chosenbecause task difficulty and task duration could be easily adjusted by the mathe-matical design of the mazes, which added significant flexibility to the experiment.Performing spatial cognitive tasks requires resources from the same areas in thebrain as required by duration estimation, and therefore, spatial cognitive tasks cantheoretically distract the brain from accurately estimating the passage of time.In the preliminary iterations of this experiment which were conducted to estab-lish the best method and choose the time perception altering factor, the presence ofan auditory time cue (tick of a clock with a tempo of 1-tick/1.2 seconds) was used.Although the auditory time cue did seem effective, it did not look as promisingas adding cognitive tasks, hence cognitive spatial tasks were used. It appears that18immersion and cognitive load can interact with each other to affect estimation oftemporal duration.In order to test the influence of immersion in a VR experience and cognitiveload on duration estimation, an experiment was conducted in which participantsnavigated through four mazes. Their subjective perceived durations of the experi-ences were collected immediately following the experience, and were compared tothe actual recorded durations (See Appendix F).3.1 ExperimentThe four null hypotheses that we intended to test were:1) There is no significant difference in the perception of time experienced betweenparticipants using a VR maze and a non-VR maze.2) There is no significant difference in the perception of time experienced betweenparticipants solving a maze with a cognitive task and a maze with no cognitive task.3) There is no significant difference in the perception of time experienced betweenparticipants using a VR maze with a cognitive task and a VR maze with no cogni-tive task.4) There is no significant difference in the perception of time experienced betweenparticipants using a non-VR maze with a cognitive task and a non-VR maze withno cognitive task.Table 3.1 illustrates all four experimental conditions.Progression through the four maze conditions was designed to take roughly fiveminutes on average for each maze, followed by a short post-exposure survey (SeeAppendix D). Upon the completion of the last maze and questionnaire, a semi-structured interview was conducted. Afterwards, the participants were thanked forExperiment Condition VR Experience Cognitive Task1 Yes Yes2 No Yes3 Yes No4 No NoTable 3.1: Four Experiment Conditions19their time and their incentive gift cards were provided. On average, the wholeprocedure took around 40 minutes. Each participant was asked for 45 minutes oftheir time to make sure that there was enough time for all steps to be completed.3.1.1 Experimental DesignThe experiment had a retrospective design, in that the participants were not awarethat they had to judge the elapsed time until afterwards. Each trial ended as soonas the participant reached the goal, or after 600 seconds had passed. The actualpassed time was measured by the systems clock and was recorded in a data log.Maze navigation was chosen as the cognitive task for the user study. The rea-son behind this design decision was that spatial cognitive tasks share the mostresources with time tracking mechanisms in our body among all the cognitive tasktypes. Also, complicated maze navigation is considered a top demanding task inthis family of cognitive tasks. In addition, maze navigation was a suitable andcompatible task to perform both in VR and on a 2D screen.3.1.2 Ethical ReviewThe necessary ethical review was requested from the UBC Behavioral ResearchEthics Board (BREB) and approval was obtained (H17-00106) (See Appendix C)before recruitment for the experiment. The experiment was assisted by the Elec-trical and Computer Engineering Electrical and Computer Engineering (ECE) de-partment at UBC. which, upon obtaining BREB approval, aided the participantrecruitment process by sending invitation emails to ECE graduate students.The signed consent forms were collected and it was acknowledged that all datacollected during the experiments would be confidential and would be kept in asecure location . All participant names were removed and replaced with subjectnumbers. The participants were also informed that their voice would be recordedand would be later transcribed for further analysis and kept safe and confidential atall times. Data collected from the users are completely confidential and kept in asecure area in the School of Nursing Media Room (T302) at UBC Hospital.203.2 Participant Recruitment3.2.1 Sampling PlanA convenience sample of 30 participants was recruited for the study. This samplesize was calculated using the UBC Statistics Departments online tool for samplesize/power analysis; the calculation was developed by Dr. Rollin Brant. The meanvalues we used were Mean1 = 340 ± 60 and Mean2 = 270. We set the sigma valueto 60, alpha to 0.05 and power to 80%.In addition, we used the website to make sure that the ob-tained sample size was consistent with other available tools. The desired powerelected was 80% for the test and 0.05 as the alpha. We used mean values (Mean1= 340 ± 60 and Mean2 = 270) and the Cohen’s d effect size was calculated to be0.433 (21% effect size) based on our results. The generated sample sizes were 28from UBCs tool and 27 from the clincalc online tool.In addition to all the above calculations, the effect of adding a new partici-pant gradually started to become insignificant after we gathered data from 30 par-ticipants, and therefore we stopped the user study after successfully running theexperiment with 34 participants.3.2.2 Inclusion and Exclusion CriteriaThe sample included both female and male adults.The inclusion criteria for participants were:1) Participants must have been over 19 years old by the date they signed the con-sent form;2) Participants should have been able to wear a VR head mounted display (HMD)for 5-10 minutes at a time;3) Participants should be able to perform simple cognitive tasks;4) Participants should have had no mobility problems when walking short dis-tances.The exclusion criteria were:1) Participants with hearing or visual impairments (eyeglass visual correction wasacceptable);212) Participants with a susceptibility to claustrophobia or motion sickness;3) In addition, in order to have some standardization of ability, participants had tosuccessfully pass a tutorial maze (with guiding lines on the floor) within a set timebefore starting the experiment.Participants must have successfully finished the tutorial maze in less than twominutes. If they failed to solve the tutorial maze in time, they were given anotherchance, where upon a second failure they were thanked for participating in theexperiment, received their incentives, and discontinued.3.2.3 Recruitment MethodsThe recruitment process used email invitation via the ECE department mailing lists,and flyers were placed across different departments and buildings on the UBC cam-pus. UBC professors (Dr. Little (CPSC 425 and CPSC 505), Dr. Garrett (NURS540), and Dr. Poole (CPSC 532)) invited students in their classes to participatewith direct in-class invitations before a session of their classes started. A consentform was supplied to the participants upon initial recruitment and a signed copywas obtained at the start of the experiment. The participants got the chance to useour VR equipment in lab and to play VR videogames, and they received a 10$Starbucks gift card as an incentive for their participation.3.3 Materials3.3.1 LocationThe experiment was conducted in a UBC Hospital room that had sufficient area forwalking around and exploring whilst participants were wearing HMDs.3.3.2 HardwareThe HTC Vive was used as the VR equipment due to its high graphic quality andease of movement using its hand controllers. It has motion tracking, two hand con-trollers that simulate participants hands positions in the IVE, a stereoscopic HMDand stereo headphones, and two location sensors (Figure-3.1). The resolution of22its display is 2160 1200 (1080 1200 per eye). The refresh rate is 90 Hz and itcovers approximately a 110 degree field of view. For rendering, system controland logging an Intel computer was used with a 3.6 Gigahertz (GHZ) quad Core i7processor, 16GB of main memory and an NVIDIA GeForce GTX 970 graphic cardwith 8GB of RAM.3.3.3 SoftwareIn order to develop the software for the experiences in the experiment, several mul-timedia VR and 3D mazes and a lobby environment were required. Unity3D 5.4.1engine was used to create them. Unity3D is the worlds largest gaming develop-ment platform and is widely used among various game/environment developers. Italso has one of the largest development communities, which eased the developmentprocess. Moreover, it is among the top game engines in terms of rendering qualityand optimization.The SteamVR plugin was used for integrating the HTC Vive facilities intothis study’s environment. SteamVR produces hand controllers in the game whichmirror physical location and orientation. In addition, for the VR experiences theSteamVR and the Vive tracked the physical boundaries of the real world room anddisplayed safety boundaries using blue dotted lines so the participants did not ap-proach physical walls and obstacles. Most of the code was implemented in C# andJavaScript programming languages. Some public coding libraries were also usedfor the game engine, physics engine and rendering, all of which were provided inthe basic Unity3D development environment and are free to use.Figure 3.1: HTC Vive Setup and Equipment23The materials and shaders were taken from the Unity asset store and were allfree to use. For all of the physical interactions in the environment and lighting, thestandard Unity libraries of 3D objects was used and any necessary modifications toconstruct the maze IVEs was implemented. We recorded audio voiceovers for allinstructions and descriptions.During the experiment participants walked around the VR mazes naturallywhilst wearing HMDs on their heads. They also walked through the non-VR mazesdisplayed on a multimedia PC screen using hand controllers to control their move-ment.Cyber sickness (also known as virtual reality motion sickness) symptoms inthe VR experiences can possibly result in general discomfort, headache, stomachawareness, nausea, vomiting, pallor, sweating, fatigue, drowsiness, disorientation,and apathy. In order to reduce this sickness caused by VR, a movement functional-ity was implemented allowing the user to look in a direction and then move in thatdirection by pressing the trigger button on the hand controllers.3.4 InstrumentsThe tools used to gather and analyze the data from the user studies were:1) A Demographic questionnaire2) A maze post-experiment questionnaire (completed at the end of each maze ex-perience)3) A semi-structured terminal interview (voice recording and transcript)4) Data logs of the participants’ progress through the mazes recorded during theexperiment3.4.1 Demographic QuestionnaireA brief questionnaire was created containing questions regarding participants basicdemographic information such as age and gender. It also asked whether the partici-pants had heard about this experiment and its purpose beforehand, and whether theparticipants suspected to have cyber sickness or claustrophobia. This questionnaireserved as a basic filter for excluding participants that did not meet the inclusion re-quirements. The age and gender of the participants were obtained but were not24used in this research, as most participants had similar ages; this data could be usedfor further exploration in the future (See Appendix D).3.4.2 Post-experience QuestionnairesA questionnaire was designed to be administered immediately after each maze ex-perience and contained questions regarding the color of the portal they took as adistracting question. It asked about whether they saw any moving creatures in themaze (which did not exist) and again functioned as another distracting question.The next question which was the goal of this survey was how long they thoughtthey were in the maze from the entering portal to the exit portal. The next questionwas another distraction question, which asked participants how enjoyable their ex-perience in the maze was. The final question regarded the level of interactivity ofthe maze, which also functioned as a distraction question. The purpose of thesedistraction questions was to hide the main question regarding time estimation, sothat in the next mazes the participants did not focus on the time taken to completethe maze. Most of these questions resulted in quantitative data. The participants’answers to the question asking how long they thought they were in the maze wasthe main input of the conducted analysis (See Appendix D).3.4.3 Terminal InterviewA short semi-structured individual interview script was developed to be undertakenwith each participant following the experiment. (See Appendix D). Participantswould be asked if they experienced any time distortion and if so, they would beasked to explain in which experiences it was the most present. They would also beasked what factors they thought might have resulted in the time distortion. The pur-pose of this interview was to explore the participants’ experiences in more depth.Questions were designed to provide more details of participant perceptions re-garding the passage of time in the experience, and help explain their actions andthought processes with more detail, which otherwise may have been lost. Partici-pants were informed before the interview that the main purpose of the experimentwas to assess if any time distortion had occurred for them during the trials, and thatthe interview would be recorded and transcribed for analysis. See Appendix D for25the interview questions.3.4.4 Data LogA data log was also implemented that would be automatically recorded by thecomputer during the experiment. Data was recorded to calculate the objective timetaken to complete each maze. These records included: the exact position (x, ycoordinates) of the player 40 times per second with their timestamps, the exactdirection they were looking at for each rendered frame with their timestamps, thetimestamps of each enter/exit portals in order to log the duration of each maze andthe color of the portal they took, and the time that participants spent to look atpictures on the wall of the mazes and hear their descriptions.This latter element was performed by measuring the amount of time the playersspent standing still in the areas in front of pictures that occurred at intervals in thenon-cognitive task mazes to slow their progression (see section 3.5 below). Usingthese logs, the total distance they walked was calculated, as well as their averagespeed, the cumulative time they spent looking at the pictures, and the percentageof time they spent looking at the pictures versus the total time. These gathered datawere sufficient to reproduce the whole experiment, but some of them were not usedin the analysis. These logs were dynamically written on a text file and were storedon the hard disk.3.5 ProceduresThe four different maze experiences mentioned at the beginning of this chapterwere performed in varied predefined sequential orders using a Latin Square designfor each participant to prevent any possible sequencing effects and additional bias(Table-3.2). The order of trials was not randomized due to the fact that the numberof participants was not large enough to ensure randomness (34 as the number ofparticipants was not large enough in order to choose the mazes randomly and besure that all sequence of mazes would be covered). Before starting each maze, theresearcher instructed the participant which maze (portal) they should go to first.This was followed by a 5 minute VR orientation and acclimatization (taking about10 minutes in total).26Red (maze #1) Green (maze #2) White (maze #3) Yellow (maze #4)Green (maze #2) White (maze #3) Yellow (maze #4) Red (maze #1)White (maze #3) Yellow (maze #4) Red (maze #1) Green (maze #2)Yellow (maze #4) Red (maze #1) Green (maze #2) White (maze #3)Table 3.2: Latin Square DesignThe steps that the participants went through were:1) The HMD was handed to participants to put on and was made sure to be adjustedand calibrated in a way that suited them. They started the experiment by puttingon the HMD and earphones and holding the hand controller (for VR mazes). Al-ternatively for non-VR mazes, participants began the experiment by being guidedtowards their seat in front of the PC and were handed the controllers.2) Participants initially started in a small tutorial maze that had a guideline on thefloor. They heard instructions on how to move around and navigate towards theexit portal, which took them to the main lobby. If they did not pass on the firstattempt, they were given another chance to successfully finish in time. If they wereunable to do so, they did not meet the standard, and did not continue to the othermazes. Their participation ended at this point, but they still received the incentivegift card for participating.3) Upon successfully completing the tutorial maze in two minutes, the success-ful participants were teleported to the main lobby (repositioned from the exit of theactual mazes to a specific coordinates in the lobby environment). At this point,they had a chance of removing the HMD and relaxing their eyes. The main lobbyhad four colorful portals that each directed them to undertake one of the four dif-ferent mazes of the experiment. They then were directed to go through one specificportal.4) If they had started in a maze without the guidelines, they had to navigate themaze and find the exit portal that took them back to the main lobby. After complet-ing the maze, they removed the HMD and undertook a short questionnaire abouttheir experience.27Figure 3.2: Top View of the Tutorial Maze with Guiding Lines5) If they had been teleported to a maze with the guideline, they could followthe green line towards the exit portal that took them back to the main lobby. Thesemazes contained 20 pictures of interesting landmarks or universities around theworld with short audio narrations. The participants heard a 10 second descriptionof the pictures when they approved them, which were hung on the walls at specificpoints in the maze. The reason for the use of these pictures was to increase theaverage time participants took to complete the mazes with guidelines in order tomake them comparable with the other two mazes.6) The participants looped four times into the main lobby and the mazes (2 VRand 2 non-VR) until they finished all of the mazes. Moreover, one of the two VRmazes and one of the non-VR mazes were with guidelines. In brief, each partic-28Figure 3.3: Top View of the Lobby (four Portals to the four main Mazes)ipant navigated one maze with guidelines in a VR experience, one maze withoutguidelines in a VR experience, one maze with guidelines in a non-VR experience,and one maze without guidelines in a non-VR experience. The experiment wasthen complete and the participants were asked to remove the HMD and participatein a semi-structured interview answering a few questions in which their voices wererecorded.3.6 AnalysisQuantitative and qualitative data were obtained and analyzed from the experiment.These data consisted of objective measurements from timings, and logs, and sub-jective data from the participants perceptions of the experience.3.6.1 Quantitative AnalysisThe participants perceived duration for solving each maze were recorded after eachmaze experience and then compared to the actual recorded duration of the expe-29Figure 3.4: View of a Maze With Guiding Lines, Pictures and Voice Oversrience (See Appendix F). These quantitative data were collected from the surveysand timings recorded during the experiments. Inferential statistics utilizing a Lin-ear Mixed Effects (LME) model was used to compare time perceptions versus ac-tual time for both the cognitive task and immersion conditions. In this experimentthe dependent variable was time and independent variables were the two treatments(VR and Cognitive task).303.6.2 Linear Mixed Effects ModelModel and AssumptionsLinear mixed models extend simple linear models. These models allow both fixedand random effects. They are particularly beneficial when there is non indepen-dence in the data, arising from a hierarchical structure for instance. For example,students could be sampled from within classrooms, patients from within hospitals,or in our case, participants within different maze types.When dealing with hierarchical data, there are multiple approaches that we cantake. One simple approach is to aggregate. As an example, imagine 10 participantsare sampled from each maze type. Rather than using the individual participantsdata, which is not independent, we could take the average of all participants withina maze type. This aggregated data would then be independent. Although aggre-gate data analysis yields consistent and effective estimates and standard errors, itdoes not really take advantage of all the data, because participants’ data are simplyaveraged.Another approach to hierarchical data is analyzing data from one unit at a time.Again in our example, we could run four separate linear regressions, one for eachmaze type in the sample. Although this does work, there are many models and eachmodel does not take advantage of the information in data from other maze types.This can also introduce more noise in our results, such that the estimates from eachmodel are not based on sufficient amounts of data.Linear mixed models (also called multilevel models) can be thought of as atrade off between the above two alternatives. Individual regressions has many es-timates and lots of data, but is a noisy method. Aggregation is less noisy, but maylose important differences by averaging all samples within each maze type. LMEsare somewhere in between [10].Beyond simply caring about correcting standard errors for non independencein the data, there can also be important reasons to explore the difference betweeneffects within and between groups. LMEs allow us to explore and understand theseimportant effects.The core of mixed models is that they incorporate fixed and random effects.31A fixed effect is a parameter that does not vary. In contrast, random effects areparameters that are themselves random variables. In equation form:β ∼ N(µ,σ) (3.1)This equation is similar to linear regression, in which we assume the data arerandom variables, but the parameters are fixed effects. Now the data are randomvariables, and the parameters are random variables (at one level), but fixed at thehighest level (for example, we still assume some overall population mean).Independence, being the most important assumption, is one of the main reasonswe use mixed models rather than just working with linear models, and allows usto resolve non-independence in our data. However, mixed models can still violateindependence if missing important fixed or random effects.A random effect is generally something that can be expected to have a non-systematic, idiosyncratic, unpredictable, or random influence on your data [10]. Inexperiments, this is often the participant, and we normally want to generalize overthe idiosyncrasies of individual participants. Fixed effects, on the other hand, areexpected to have a systematic and predictable influence on data.The type of mixed model above (LME) is suitable for systems that containboth fixed effects and random effects. The linear mixed effect model, which isan extension of linear regression, is preferred over Analysis of Variance (ANOVA)in settings where the gathered data contains repeated measures of the same nature[35]. The collected data should describe the relationship between a response vari-able and independent variables, with coefficients that can vary with respect to oneor more grouping variables.Unlike most machine learning problems, here we are interested in using amodel of inference of correlating data, rather than a model for the prediction ofevents. The error of time estimations was distributed as a Gaussian distributionand the exploratory variables were assumed to be related linearly based on [35].Therefore, the results were analyzed using Linear Mixed Effects (LME) model us-ing the R programming language (suggested by The Statistical Opportunity forStudents (SOS) statisticians from UBCs Department of Statistics).The main difference between the LME model and linear regression is that the32linear mixed effects model treats observations obtained from the same participantas correlated, whilst a linear regression model treats them as independent. In thisexperiment, the former was more appropriate since there are likely similarities inperceived time within individuals. For example, some individuals may consistentlyunderestimate the amount of time spent in the maze, and some may consistentlyoverestimate. In order to use an LME model in R, the nlme package and the lme( )function within that package was used (See appendix A).LME in This ExperimentWhen there are multiple levels, such as participants navigating the same maze type,the variability in the outcome can be thought of as being either within group or be-tween group (no maze effect was observed from our investigations, and this hasbeen discussed in more depth in chapter 5 of this thesis). Participant level ob-servations are not independent, as within a given maze type participants act moresimilarly. Units sampled at the highest level (in our example, maze types) are in-dependent.The extra assumptions that we needed to worry about were collinearity (whichhas been checked based on [35]), influential data points (we had one, but decidedto keep all the data), and normality (which has been checked by plotting the data -see chapter 5).In this study the time estimate is the fixed effect term of the model, while in-dividuals habits and qualities represent random effects in the LME model. Thismodel is particularly useful in our setting where the individual specific effect iscorrelated with the independent variables. The linear mixed effects model is simi-lar to a linear regression model, and the goal of fitting such a model is to determinewhether or not independent variables (e.g. VR, no VR, cognitive spatial task andno cognitive spatial task) had a significant effect on perceived time (i.e the differ-ence between perceived time and actual time recorded). The estimated coefficientsof either model give the estimated magnitude of the effect of each variable andwhether or not they may be statistically significant. A linear mixed effects modelwith interaction was used for statistical analysis. By adding an interaction term,the model became more flexible and was able to detect differences in how the re-33sponse was affected by different combinations of VR/no-VR and cognitive spatialtask/no-cognitive spatial task experiment settings.The interaction term estimates the difference of the VR effect in a setting withthe solution. If the interaction term in the model is found to be statistically sig-nificant, then the effect of both of those factors would be the total of the effect ofbeing in VR, the effect of having a solution, and the effect of the interaction term.This model should only be used if the interaction term is found to be significant;otherwise the additive model described above is sufficient.3.6.3 Qualitative AnalysisFor analyzing the qualitative data from the semi-structured interview at the endof the experiment, the recorded script of the interview was subjected to a simplecontent analysis. Transcripts of the interview were read and reread to identify sim-ilar emerging thematic items. These were coded and synthesized into significantthematic elements in order to shed light on the participants perceptions about theirexperiences in the mazes. The main goal of our thematic analysis was to pinpointthemes and patterns in the interview scripts that can help us answer our researchquestions.A six phase coding and reviewing of the themes were performed, throughwhich the repeating patterns and themes were identified and coded in order to helpunderstand the outcomes in more depth.34Chapter 4ResultsThirty-four participants (15/34 females and 19/34 males) participated in the ex-periment (zero drop outs, no participant left the study due to cyber sickness, oneparticipant suspected that she was feeling minor cyber sickness but wanted to con-tinue the experiment, and two participants had prior experience using VR). Mostof the participants were UBC graduate students and faculty members. Objectivequantitative and qualitative data, as well as some subjective data were obtained.The results were analyzed using the Linear Mixed Effects Model, as mentioned insection 3.6.1 of this thesis. Two different fitting functions were used and compared,one with the VR-Cognitive task interaction term and one without the the interac-tion term. The interaction term was found not significant (p = 0.1511) (Degree ofFreedom (DF) = 99, STD. Error = 0.112) and therefore we used the first model thatdid not include the interaction term. The interaction term estimates the differenceof the VR effect in a setting with the solution. Here in this chapter, the followingresults are reported.4.1 Quantitative Data4.1.1 Descriptive Univariate StatisticsFigure-4.1 shows the error distribution of 136 valid observations (and zero outliers)obtained through all trials that were analyzed. The histogram plots the frequency of35each absolute-error bucket. The vertical red line passing zero on the horizontal axisindicates an absolute accurate estimation. A negative error (on the left side of thered line) represents underestimation of time while a positive error (on the right sideof the red line) corresponds to overestimation. The mean of all 136 data points is-4.3% (4.3% underestimation of time), and the median is -7%. In terms of absolutevalues, the mean error is found to be -13 seconds (median is -25 seconds). In termsof data distribution, this is a normal graph, which indicates that the time estimationerror’s distribution is Gaussian. One last observation from the error distributionhistogram is that overall, more people have underestimated than overestimated thedurations, while the overestimations are larger in value.4.1.2 Inferential StatisticsLinear Mixed Effects Model: The model was implemented using R program-ming language (See Appendix A) and formatted in a long table with 34 groupsand 136 observations (each row represents one of the four observations for eachparticipant). The first four rows of the table are shown below (Table-4.1) as anexample to indicate the first participant’s data. See Appendix F for the full table ofall participant data.In this table, perceived time represents participants responses to the questionon post-experiment questionnaires which asks, How long were you in the mazefrom the entrance portal to the exit portal?. The actual time represents the amountof time participants actually were in the maze, which was automatically loggedand stored by the application. Ratio is defined as perceived time over actual time.VR column indicates whether the experiment was a VR experience (1) or a screen-based non-VR experience (0). The solution column specifies whether the maze hadguidelines (1) or did not have guidelines (0). The maze column shows which mazethe experiment was conducted in. Finally, the last column shows the ratio of thedifference between perceived and actual times over actual time. All durations arein seconds.Table-4.2 shows the achieved results of the linear mixed effects model using Rfor the various experimental conditions. A comparison of All VR and All No-VRwill show us the effect of immersion within a VR experience on time perception,36Figure 4.1: Error Distribution of all Trials (The Horizontal Axis Shows theAbsolute Error in Seconds, Vertical Axis Shows the Participant Count)while comparing all the mazes without guidelines (Cog. Task) with all the mazeswith guidelines will indicate the effect of increased spatial cognitive load on timeperception. Finally, comparing the combination of VR and Cog. Task with control(no VR no Cog. Task) will indicate the effect of combining the two. In this table,VR maze refers to the runs in the experiment in which the participant navigated themaze using the HTC Vice headset. A No-VR maze refers to the maze navigatedby the participant on the 2D computer screen (without the VIVE headset). A Cog.Task maze is one that did not have guidelines on the floor and where the participanthad to use spatial cognition in order to navigate the maze. A No-Cog maze refers toruns in which there were guidelines on the floor which the participant could followand reach the maze exit. The analysis of these achieved results and the relateddiscussion around their significance is throughly discussed in the fifth chapter ofthis thesis.37Participant VR No-Task Maze Perceived Actual Ratio Diff/ID Time (s) Time (s) Actual1 1 1 R 180 351 0.51 49%(U)1 1 0 G 300 572 0.52 48%(U)1 0 1 W 180 383 0.47 43%(U)1 0 0 Y 240 415 0.58 42%(U)Table 4.1: Results from the four runs of the experiment for participant 1Experiment # Under est. (U) P F R DF Std.Condition Obs. Over est. (O) Val. Val. sqr Err.All VR 68 16.10% (U) ≤0.01 26.32 0.28 99 0.09All No-VR 68 7.50% (O) ≤0.01 31.14 0.32 99 0.07All Cog. Task 68 6.45% (U) 0.36 22.68 0.26 99 0.08All No-Cog.Task 68 1.65% (U) 0.36 28.54 0.30 99 0.07VR and 34 22.18% (U) ≤0.01 21.44 0.32 99 0.15Cog. TaskNo-VR and 34 4.98% (O) ≤0.01 15.79 0.33 99 0.10No-Cog. TaskTable 4.2: Results of the R Code for Various Experiment Conditions (ABolded P-value indicates Significant Effect)Figure-4.2 illustrates the actual time estimate means, as well as perceived timeestimate means for all the experiment conditions. The Error in the graph is thedifference between actual and perceived mean values. A negative error representsan underestimation, while a positive error is an indicator of overestimation.Table-4.3 and Table-4.4 indicate the demographic questionnaire results aboutage and gender distribution of the participants. The non-uniform bucketing of agebrackets in Table 4.3 is due to the non-linear effect of aging on time perception,based on [8].Demographic Age Age Age Age Age(19-24) (25-30) (31-40) (41-50) (50+)Portion of Participants (%) 41.18% 47.06% 8.82% 2.94% 0%Table 4.3: Age Distribution of Participants38Again, the four null hypotheses that we intended to test were:1) There is no significant difference in the perception of time experienced betweenparticipants using a VR maze and a non-VR maze.2) There is no significant difference in the perception of time experienced betweenparticipants solving a maze with a cognitive task and a maze with no cognitive task.3) There is no significant difference in the perception of time experienced betweenparticipants using a VR maze with a cognitive task and a VR maze with no cogni-tive task.4) There is no significant difference in the perception of time experienced betweenparticipants using a non-VR maze with a cognitive task and and a non-VR mazewith no cognitive task.The obtained results provide sufficient evidence to reject hypothesis 1 and 3, whilethey do not provide sufficient evidence to reject hypothesis 2 and 4.4.2 Qualitative Data4.2.1 Demographic QuestionnairesNone of the participants had heard about the details of the experiment before par-ticipating in the experiment. None of the participants were suspected to suffer fromcyber-sickness or motion-sickness (four of the participants were not sure whetherthey suffered from cyber-sickness or motion-sickness). None of the participantswere suspected to have claustrophobia (one of the participants was not sure whetherthey suffered from claustrophobia).Demographic Female Male OtherPortion of Participants (%) 44.12% (15/34) 55.88% (19/34) 0% (0/34)Table 4.4: Gender Distribution of Participants39Figure 4.2: Actual and Perceived Time on Various Experiment Conditions4.2.2 Post-Experience Questionnaire DataOnly time estimates that participants reported for their experience (See AppendixD) were used from the post-experiment questionnaires. The other results fromdistraction questions were discarded.4.2.3 Terminal Interview DataAfter analyzing the terminal interview recorded scripts, it was found that frustra-tion level, VR factors (factors related to VR and its effect on time estimation),cognitive factors (factors related to the effect of undertaking a cognitive task ontime estimation), maze design, maneuver system, and engagement level are someof the pinpointed themes.Transcripts of the post experience terminal interviews were subjected to a con-ventional content analysis (no themes or keywords were predefined; rather, theywere generated through the analysis). The transcripts were read and reread a fewtimes until recurring themes were identified and categorized into groups. Thegroups were given names using the terminologies used in this thesis. The main40themes that were easily identifiable were the following: most participants believedthat the no-VR version of the experiment was not a satisfying experience and wasexhausting. Participant 6 mentioned “Time was really slow and I was bored whenI was finishing the maze using screen”. Another common theme was that someof the participants believed that the speed their avatar character was moving wasslow and made them overestimate the passage of time. As participant 19 men-tioned,“also movement speed very much affects your sense of how time passesbecause if you move faster time seems to be faster and if you are moving slowertime seems slower”.The last observation, which was clearly reflected in the results, was that thoseparticipants who thought the experiment was more fun and more interactive under-estimated the passage of time. Participant 13 indicates “There were the signs or thepictures on the wall and the fact that when you got close to them there was a voiceor history it was interactive. It made it a little bit more fun and interactive, and alsodistractive, which I think might contribute to perception of time.” This relationbetween how fun and interactive the experiment was and the perceived passage oftime needs further analysis, which was not the focus of this thesis and can be astarting point for any future work of this research.41Chapter 5DiscussionThe experiment revealed a significant (16.1% underestimation, p ≤0.01, DF = 99,STD. Error = 0.09) underestimation of passage of time in the experiment runs con-ducted in VR compared to the control. Moreover, the effect of cognitive spatialtasks on time underestimation was not within the significance boundary (6.45%underestimation, p = 0.36, DF = 99, STD. Error = 0.08) and was below the 0.05significance level. The most significant underestimation occurred when the exper-iment was in VR and when a cognitive spatial task was present (22.18% underes-timation, p ≤0.01, DF = 99, STD. Error = 0.15). This reveals that being immersedin an experience results in a significant shortening of perceived duration and theeffect is amplified when coupled with undertaking a cognitive spatial task. More-over, our findings illustrate that immersion within VR is a more effective methodfor shortening perceived duration in virtual environments.Maze DesignThere was no significant maze effect observed in the results. None of the fourdesigned mazes (Red, Green, Yellow, White) had any significant effect on theachieved results by participants. Moreover, the similar mean values for all VRand all non-VR mazes show that VR didn’t effect the mean time taken to solve amaze, and is a sign of good design as it makes the achieved statistics of both casesvery comparable. Figure-5.1 shows the achieved mean time to solve the mazes inall four mazes.42Figure 5.1: Actual, Perceived and Mean Error for all Four MazesIt is worth mentioning that the effect of cognitive tasks on perception of timehas been found significant in other studies (Schatzschneider et. al., 2016). Thismay be due to relatively small sample sizes of 34 or less, or the pleasing experi-ence of simple screen based non-immersive versions of the experiment. It is worthmentioning that cognitive tasks had a much more significant effect on time under-estimation when the experiment was conducted in VR. In fact, the underestimationbecomes significant when the cognitive task is performed while being immersed ina VR experience (22.18% underestimation, p ≤0.01, DF = 99, STD. Error = 0.15),which is the most significant achieved underestimation. Finally, an interesting ob-servation was that adding a cognitive spatial task to the experience significantlyincreases time estimation errors (both underestimation and overestimation). Thisshows that undertaking the task of navigating mazes significantly affected partici-pants’ time estimation abilities, but our findings do not support that the task onlyresulted in an underestimation of passage of time. The most accurate time esti-43mates were related to the trials in which there was no cognitive task present andwhere participants could follow the guiding lines.Effects of VR and Cognitive Spatial TaskThe combination of immersion in VR and performing a cognitive spatial task re-sulted in the largest underestimation of time. The effect of immersion on underesti-mation was the most frequent theme of the terminal interviews. 71% of participantsindicated that immersion affected their temporal perception and resulted in an un-derestimation of time, while 6% of participants believed that immersion resultedin an overestimation of time. Our findings show that solving the mazes on a 2Dscreen resulted in a 7.5% overestimation of time (p ≤0.01, DF = 99, STD. Error =0.07).Second to being immersed in a virtual experience, participants believed thatthe difficulty level of the maze was affected temporal perception. Indeed, difficultylevel of the maze required a higher cognitive load level, and more than 56% of theparticipants indicated that they underestimated time in the mazes that did not haveguidelines. However, our results from the post-experience questionnaires revealthat how “fun” and “difficult” the mazes were for participants had no significanteffect on their judgment of passage of time and did not result in any significantunderestimation of time.InteractivityAnother interesting finding from the questionnaires is that those participants whofound the mazes more interactive, (based on their mean interactivity scores fromthe questionnaires) significantly underestimated the passage of time compared tothose who did not find the mazes interactive (this was the initial finding from go-ing over the data briefly, but further investigation is necessary before making anystrong claims). This also shows that interactivity, which can also result in furtherimmersion, can be an effective way of shortening the duration of an experience.Further statistical analysis was not done on this matter as this was not the focus ofthis thesis.44Demographic FactorsParticipants’ gender was found not to be significant regarding time estimation,however, participants’ age was found to be significant. Among the age bracketsdefined on our demographic questionnaire, it was found that participants of age 19to 30 significantly underestimated the passage of time, more than those who werein the 31 to 40 and 41 to 50 brackets. We suspect this may have been caused by howmuch participants were familiar with VR technologies and how experienced theywere with playing video games. No further analysis was done on the matter of ageand VR technology familiarity. This is a good area to begin further investigationswith regard to this study.Effect of FrustrationThe frustration level (a term that was assigned to a group of very similar commentsgiven by participants during the interviews) of the experience was believed to af-fect the temporal perception with third most significance. Generally, this term wasassigned to those who either directly used the word ”frustrated”, or gave feedbackabout their frustration created by movement difficulties on the 2D screen and thespeed of the player avatar. Around 21% of the participants believed that the morefrustrating mazes resulted in overestimation of time. Another interesting point re-lated to frustration level was that 9% of the participants believed that the movementspeed was too slow for the player avatar, which resulted in their overestimation oftime.One factor that might have contributed to frustration levels was the implemen-tation of the 2D controller. The challenge in this implementation was that we didnot want to add a new variable to our user study by implementing a separate methodof moving and interacting specifically for the mazes that were navigated on a 2Dscreen. Therefore, we utilized the HTC Vive controller’s touchpads to control theavatar both in VR and in 2D. However, those touchpads have been designed forVR movement and interactions, and might not have the best user experience for2D movements on a PC compared to a more traditional controller that has beendesigned for 2D screens specifically. The finalized design decision was to not addthe new variable due to the scope of the research. For any further work on this45research it seems essential to evaluate the effect of controllers on the mazes solvedon a 2D traditional screen.Other factors were also discussed during the interviews, such as the backgroundaudio, how fun the mazes were, learning about the pictures, or even getting lost inthe beautiful sky. Again, due to the focus of this thesis, no further statistical anal-ysis was done on these factors. Further investigation is necessary before makingany claims with regard to the effect of frustration on time perception.Form of the Cognitive TaskIt is also necessary to pick a cognitive task that suits the purpose of the virtual en-vironment. Tasks that involve substantial and significant spatial cognition in theirperformance are (ordered based on cognition level): 1) Way-finding as part of nav-igation, 2) Acquiring and using spatial knowledge from direct experience, 3) Usingspatially iconic symbolic representations, 4) Using spatial language, 5) Imaginingplaces and reasoning with mental models, 6) Location allocation [25]. Way-findingwas chosen for the maze solving task, as navigation involves one of the highest lev-els of spatial cognition and hence, was a suitable task for this research. However,some of the participants lost their sense of navigation in the VR mazes and useda brute force approach to solving the mazes. This reduced the cognitive load ofthe experiment. In a way, the cognitive task that was picked (maze navigation),although generally a very engaging cognitive spatial task, in the VR domain it mayresult in losing sense of direction and contribute to less cognitive load. Other spa-tial tasks, such as playing an real time, action shooter video game might be a moresuitable task for determining underestimation in VR.It is important to note that due to the way we have defined percentage error,100% overestimation can be achieved if a participant’s perceived duration doublesthe actual one, while 100% underestimation occurs only if the participant believedzero seconds had passed during the experiment, which is not possible. Moreover,there is no upper bound on overestimation, while the upper bound for underestima-tion is 100%. This was one of our reasons for using absolute errors in figure-4.1to check error distribution rather than a percentage error. The absolute error dis-tribution is normal and is the reason why we chose the linear mixed effects model46to analyze our gathered data. Figure 5.2 shows the distribution of perceived timeover actual time ratio. In this graph a ratio of one indicates an accurate, perfectestimation of time. Ratios smaller than one indicate underestimation of time andratios larger than one indicate overestimation of time.5.1 LimitationsThere are a few limitations in this experiment that need to be addressed. Thesample size was initially calculated to be 28 by a UBC-based sample size calculatorand 27 by another online power calculator, however, we recruited 34 participants.This is still a very small sample size for a multiple intervention study. With suchsmall samples the risk that each observation could have a considerable impact onthe results is large. Although the observed effect size matches the expected initialeffect size, recruiting more participants can be a point of improvement for anyfurther investigations on this study.Another limitation was the sampling bias. Although the ordered sequence ofmazes in the experiment followed a Latin Square style to avoid any possible se-quencing effects, the convenience sample of participants were self-selecting andthe majority had an interest in VR technologies and applications, which can beconsidered a sampling bias. This may affect the reliability and generalizability ofthe findings. Including randomly chosen participants can potentially increase thevalidity of the study.In terms of technical limitations, participants used the HTC Vive headset whichis wired to a computer. Occasionally participants became tangled with the wires,which may have affected their sense of immersion. At the beginning of each trialthe experimenter informed the participants of the risk of the wire and instructedthem how to pass their feet above the wire while rotating to avoid such situations.A good solution for any future work on this research is to use a hanger from theceiling to pass the wire through, so that the headset is connected from above, avoid-ing the tangling issue.Moreover, the physical boundaries of the room and the participants’ worriesabout collisions may have resulted in a lower sense of immersion, which may haveaffected our results. One of the reasons that we picked HTC Vive as our head47Figure 5.2: Histogram of Perceived Over Actual Ratio Distributionmounted display was the fact that Vive tracks the borders and boundaries of theroom and any furniture inside it, and plots these boundaries as light blue dottedlines displayed to the user. However, it was obvious from participant body languagethat they were still concerned about colliding with the walls.48Chapter 6ConclusionIn this thesis, the effects of being immersed in a virtual environment and undertak-ing a cognitive spatial task on time perception were explored. An experiment wasdesigned and presented in which participants perception of time under the influenceof immersion and cognitive load was analyzed.The results achieved indicate that immersion within a VR experience can sig-nificantly affect time perception and results in a significant underestimation of thepassage of time. Moreover, our results show that performing a spatial cognitivetask (maze navigation) also leads to an underestimation of time, but the impact isnot significant. The highest underestimation occurred when both immersion withina VR experience and a cognitive task were present. On the other hand, the largestoverestimation of time occurred when immersion within VR was not present butcognitive task was present.To address our main research questions explicitly, it was found that immer-sion within a virtual experience results in a significant underestimation of time(16.10% underestimation). When combined with undertaking a cognitive spatialtask, it resulted in a 22.18% underestimation of time among 136 observations from34 participants. Undertaking a cognitive spatial task did not result in significantunderestimation of time on its own, however, it did result in significant underesti-mation of time when performed in a VR experience.In order to build VR experiences that maximize underestimation of time, thecombination of increased immersion with cognitive spatial tasks is effective. It49is also possible to benefit from factors such as Zeitgebers [29], which are previ-ously shown to be effective (table-E.1 shows effective factors from other publishedstudies). Our qualitative results also indicate that a more interactive and engaginginterface can also contribute to the effectiveness of the environment and can resultin significant underestimations of time by participants.To summarize the contributions of this thesis:1) It was found that being immersed within a virtual experience will significantlyaffect perception of time and can lead to an underestimation of the passage of time.2) It was found that undertaking a cognitive spatial task did not result in significantunderestimation of time on its own.3) The most underestimation occurred when both immersion within a VR experi-ence and a cognitive spatial task were present.4) Although undertaking a cognitive spatial task did not result in significant under-estimation of time on its own, when the task was performed in a VR environment,it significantly affected users perception of time and resulted in underestimation oftime.5) It was found that factors such as frustration level can contribute to overestima-tion of time by users. Moreover, participants’ age also had a significant effect ontime estimations. A maze’s interactivity level also affected participant time estima-tions and contributed to underestimations of time.6) It was found that participant gender did not have any significant effect on timeestimations. It was also found that how ”fun” or ”difficult” the navigation experi-ence was also did not result in any significant effect.6.1 Implications and Future WorkThe relationship between how fun and interactive the experiment was and the per-ceived passage of time needs further analysis, as this was not the focus of this thesisand can be a starting point for future work.Adding other factors from the metadata, such as tech-savviness, into the linearmixed effects model and determining whether such parameters can also have a sig-nificant effect on time perception could also be another interesting area for further50investigations.Plotting the gathered data using other visualization techniques and methodsmight also give us new insights. A good example of this for any further inves-tigation on this research is to plot data in a way that possible clusters form. Ifany cluster existed, further investigation on the cause would be essential. Thesepossible clusters can give us further insight on other factors in our metadata, suchas tech-savviness or age, and can perhaps explain some of the achieved results inmore depth.As previously stated in the discussion chapter, choosing another cognitive spa-tial task can also be an interesting area of exploration for further work.To conclude, the combination of immersion within VR and performing a cog-nitive spatial task seems like a very practical, and yet non-expensive, solution formany virtual environments in which underestimation of time has value.51Bibliography[1] L. G. Allan. The perception of time. Perception & Psychophysics, 26(5):340–354, 1979. → page 11[2] H. K. Andersen and R. Grush. A brief history of time-consciousness:historical precursors to james and husserl. Journal of the History ofPhilosophy, 47(2):277–307, 2009. → page 7[3] R. A. Block and D. Zakay. Prospective and retrospective durationjudgments: A meta-analytic review. Psychonomic bulletin & review, 4(2):184–197, 1997. → page 13[4] S. W. Brown. Time, change, and motion: The effects of stimulus movementon temporal perception. Perception & Psychophysics, 57:105–116, 1995. →page 14[5] G. Bruder and F. Steinicke. Threefolded motion perception duringimmersive walkthroughs. In Proceedings of the 20th ACM symposium onvirtual reality software and technology, pages 177–185. ACM, 2014. →page 8[6] C. Constantinidis, G. V. Williams, and P. S. Goldman-Rakic. A role forinhibition in shaping the temporal flow of information in prefrontal cortex.Nature neuroscience, 5(2):175, 2002. → page 13[7] D. B. Davalos, M. A. Kisley, and R. G. Ross. Deficits in auditory and visualtemporal perception in schizophrenia. Cognitive Neuropsychiatry, 7(4):273–282, 2002. → page 9[8] J.-C. Dreher, A. Meyer-Lindenberg, P. Kohn, and K. F. Berman. Age-relatedchanges in midbrain dopaminergic regulation of the human reward system.Proceedings of the National Academy of Sciences, 105(39):15106–15111,2008. → pages 10, 3852[9] S. Eden. The effect of 3d virtual reality on sequential time perception amongdeaf and hard-of-hearing children. European Journal of Special NeedsEducation, 23(4):349–363, 2008. → page 13[10] J. Fox. Applied regression analysis and generalized linear models. SagePublications, 2015. → pages 31, 32[11] B. Garrett, T. Taverner, W. Masinde, D. Gromala, C. Shaw, and M. Negraeff.A rapid evidence assessment of immersive virtual reality as an adjuncttherapy in acute pain management in clinical practice. The Clinical journalof pain, 30(12):1089–1098, 2014. → pages 8, 9[12] D. Goldreich. A bayesian perceptual model replicates the cutaneous rabbitand other tactile spatiotemporal illusions. PloS one, 2(3):e333, 2007. →page 11[13] P. Gongsook. Time simulator in virtual reality for children with attentiondeficit hyperactivity disorder. In International Conference on EntertainmentComputing, pages 490–493. Springer, 2012. → pages 16, 18[14] K. Ha¨gni, K. Eng, M.-C. Hepp-Reymond, L. Holper, B. Keisker,E. Siekierka, and D. C. Kiper. The effect of task and ownership on timeestimation in virtual environments. Barcelona, Spain, 2007. → pages2, 14, 15[15] J. Heron, C. Aaen-Stockdale, J. Hotchkiss, N. W. Roach, P. V. McGraw, andD. Whitaker. Duration channels mediate human time perception. In Proc. R.Soc. B, volume 279, pages 690–698. The Royal Society, 2012. → page 7[16] R. E. Hicks, G. W. Miller, and M. Kinsbourne. Prospective and retrospectivejudgments of time as a function of amount of information processed. TheAmerican journal of psychology, pages 719–730, 1976. → page 1[17] I. Hodinott-Hill, K. V. Thilo, A. Cowey, and V. Walsh. Auditorychronostasis: hanging on the telephone. Current Biology, 12(20):1779–1781, 2002. → page 12[18] B. Jones and Y. L. Huang. Space-time dependencies in psychophysicaljudgment of extent and duration: Algebraic models of the tau and kappaeffects. Psychological bulletin, 91(1):128, 1982. → page 11[19] J. J. Kellaris and R. J. Kent. The influence of music on consumers’ temporalperceptions: does time fly when you’re having fun? Journal of consumerpsychology, 1(4):365–376, 1992. → page 1053[20] G. Koch, M. Oliveri, S. Torriero, and C. Caltagirone. Underestimation oftime perception after repetitive transcranial magnetic stimulation.Neurology, 60(11):1844–1846, 2003. → page 10[21] D. E. Kornbrot, R. M. Msetfi, and M. J. Grimwood. Time perception anddepressive realism: judgment type, psychophysical functions and bias.PLOS one, 8(8):e71585, 2013. → page 11[22] P. A. Lewis and R. C. Miall. Distinct systems for automatic and cognitivelycontrolled time measurement: evidence from neuroimaging. Currentopinion in neurobiology, 13(2):250–255, 2003. → page 14[23] W. J. Matthews, N. Stewart, and J. H. Wearden. Stimulus intensity and theperception of duration. Journal of Experimental Psychology: Humanperception and performance, 37(1):303, 2011. → page 14[24] G. K. McMillan, D. M. Considine, et al. Process/industrial instruments andcontrols handbook, volume 7. McGraw Hill, 1999. → page 7[25] D. R. Montello and M. Raubal. Functions and applications of spatialcogniticdn˜. 2013. → page 46[26] S. Neufang, G. R. Fink, B. Herpertz-Dahlmann, K. Willmes, and K. Konrad.Developmental changes in neural activation and psychophysiologicalinteraction patterns of brain regions associated with interference control andtime perception. Neuroimage, 43(2):399–409, 2008. → page 14[27] T. A. Niendam, A. R. Laird, K. L. Ray, Y. M. Dean, D. C. Glahn, and C. S.Carter. Meta-analytic evidence for a superordinate cognitive control networksubserving diverse executive functions. Cognitive, Affective, & BehavioralNeuroscience, 12(2):241–268, 2012. → page 14[28] R. Nijhawan and B. Khurana. Space and time in perception and action.Cambridge University Press, 2010. → page 12[29] A. C. Nobre and J. O’Reilly. Time is of the essence. Trends in cognitivesciences, 8(9):387–389, 2004. → page 14[30] J. Radua, N. O. Del Pozo, J. Go´mez, F. Guillen-Grima, and F. Ortun˜o.Meta-analysis of functional neuroimaging studies indicates that an increaseof cognitive difficulty during executive tasks engages brain regionsassociated with time perception. Neuropsychologia, 58:14–22, 2014. →page 1354[31] S. M. Rao, A. R. Mayer, and D. L. Harrington. The evolution of brainactivation during temporal processing. Nature neuroscience, 4(3):317, 2001.→ page 7[32] D. Rose and J. Summers. Duration illusions in a train of visual stimuli.Perception, 24(10):1177–1187, 1995. → page 12[33] M. Rudd, K. D. Vohs, and J. Aaker. Awe expands peoples perception oftime, alters decision making, and enhances well-being. Psychologicalscience, 23(10):1130–1136, 2012. → page 10[34] T. Sanders and P. Cairns. Time perception, immersion and music invideogames. In Proceedings of the 24th BCS interaction specialist groupconference, pages 160–167. British Computer Society, 2010. → page 2[35] C. Schatzschneider, G. Bruder, and F. Steinicke. Who turned the clock?effects of manipulated zeitgebers, cognitive load and immersion on timeestimation. IEEE transactions on visualization and computer graphics, 22(4):1387–1395, 2016. → pages 12, 15, 18, 32, 33[36] S. M. Schneider, C. K. Kisby, and E. P. Flint. Effect of virtual reality on timeperception in patients receiving chemotherapy. Supportive Care in Cancer,19(4):555–564, 2011. → pages 15, 18[37] J. T. Shelton, E. M. Elliott, B. Hill, M. R. Calamia, and W. D. Gouvier. Acomparison of laboratory and clinical working memory tests and theirprediction of fluid intelligence. Intelligence, 37(3):283–293, 2009. → page13[38] C. Stetson, M. P. Fiesta, and D. M. Eagleman. Does time really slow downduring a frightening event? PloS one, 2(12):e1295, 2007. → page 11[39] M. Treisman. Temporal discrimination and the indifference interval:Implications for a model of the” internal clock”. PsychologicalMonographs: General and Applied, 77(13):1, 1963. → page 1[40] M. Treisman, A. Faulkner, P. L. Naish, and D. Brogan. The internal clock:Evidence for a temporal oscillator underlying time perception with someestimates of its characteristic frequency. Perception, 19(6):705–742, 1990.→ page 15[41] P. U. Tse, J. Intriligator, J. Rivest, and P. Cavanagh. Attention and thesubjective expansion of time. Perception & psychophysics, 66(7):1171–1189, 2004. → page 1255[42] V. Van Wassenhove, M. Wittmann, A. Craig, and M. P. Paulus.Psychological and neural mechanisms of subjective time dilation. Frontiersin neuroscience, 5:56, 2011. → page 10[43] Y. Wada, T. Masuda, and K. Noguchi. Temporal illusion called’kappaeffect’in event perception. Perception ECVP abstract, 34:0–0, 2005. →page 11[44] J. H. Wearden and I. S. Penton-Voak. Feeling the heat: Body temperatureand the rate of subjective time, revisited. The Quarterly Journal ofExperimental Psychology, 48(2):129–141, 1995. → page 14[45] M. Wittmann, O. Carter, F. Hasler, B. R. Cahn, U. Grimberg, P. Spring,D. Hell, H. Flohr, and F. X. Vollenweider. Effects of psilocybin on timeperception and temporal control of behaviour in humans. Journal ofPsychopharmacology, 21(1):50–64, 2007. → page 9[46] M. Wittmann, D. S. Leland, J. Churan, and M. P. Paulus. Impaired timeperception and motor timing in stimulant-dependent subjects. Drug &Alcohol Dependence, 90(2):183–192, 2007. → page 9[47] K. Yarrow, P. Haggard, R. Heal, P. Brown, and J. C. Rothwell. Illusoryperceptions of space and time preserve cross-saccadic perceptual continuity.Nature, 414(6861):302, 2001. → pages 11, 1256Appendix AThis appendix shows the R code used in order to analyze the obtained data. This isthe code for the Linear Mixed Effects ModelR Code:install.packages(nlme)library(nlme)dat <−read.csv(C : /Users/Derek/Desktop/data f ile.csv,header = T )f it1 <−lme(Di f f erence VR+Solution,random= 1|ID,data= dat)summary(fit1)summary(fit1)f it2 < −lme(Di f f erence VR ∗ Solution,random = 1|ID,data = dat) sum-mary(fit2)57Appendix BThe next two pages (Figure B.1 and B.2) show the top view of all the four mainmazes as well as participant’s view of the tutorial maze.Figure B.1 shows two mazes with green guidelines and two mazes without theguidelines. In this picture it is obvious that all four mazes have identical sizes (andalso identical difficulties).Figure B.2 gives a closer look at the design of the maze environment and showshow exactly participants perceived the maze environments.58Figure B.1: Top View of All 4 Identical Main Mazes (2 with guiding linesand pictures, 2 without lines and pictures)59Figure B.2: Players View of a Maze60Appendix CFigures D.1, D.2, and D.3 show the Demographic Questionnaire, the Post-ExposureVR Experience Questionnaire, and the Post Experience Semi-structured InterviewQuestions that participants received during the user study.Each participant completed:1) The Demographic Questionnaire at the beginning of the user study (1 time)2) The Post-Exposure VR Experience Questionnaire at the end of each of the VRmazes (4 times)3) The Post Experience Semi-structured Interview Questions at the end of the userstudy (1 time)61Figure C.1: Demographic Questionnaire62Figure C.2: Post Exposure VR Experience Questionnaire63Figure C.3: Post Exposure Semi-Structured Interview Questions64Appendix DTable E.1 (next page) shows identified factors that can possibly affect one’s per-ception of time. The effects have been divided into two columns (Overestimationand Underestimation).65Overestimation UnderestimationPsychoactive Drugs Psychoactive DrugsMusic (Positively Valenced) Music (Negatively Valenced, Casino Ambient Sound)Emotional States (Awe, Fear, ) Clinical DisordersTelescoping Effect AgingVierordts Law Vierordts LawKappa Effect Kappa EffectChronostasis Cognitive Tasks (especially Spatial cog. Tasks)The Oddball Effect Repetitive Transcranial Magnetic StimulationHigh Stimulus Intensity Low Stimulus IntensityHigh Change Rates in Scenes Low Change Rates in ScenesBody Temperature Body TemperatureInterrupted Scenes Uninterrupted ScenesVisual Time Symbols (Zeitgebers) Visual Time Symbols (Zeitgebers)Performing temporal tasks while Performing non-temporal tasks/in immersive environments playing games while experiencing immersionHigh Frequency Auditory Time Cues Low Frequency Auditory Time CuesTable D.1: Factors That Effect Perception of Time66Appendix ETable F.1 (next 5 pages) shows all the recorded data. The columns show the par-ticipant ID, whether the experience was VR or non-VR, whether it had a cognitivespatial task or not, which maze it was, the perceived time (from the Post-experiencequestionnaire), the logged actual time taken to solve each maze, the ratio of per-ceived time over actual time, and finally the underestimation/overestimation per-centage.67User Study DataParticipant VR No-Task Maze Perceived Actual Ratio Diff/ID Time Time Actual1 1 1 R 180 351 0.51 49%(U)1 1 0 G 300 572 0.52 48%(U)1 0 1 W 180 383 0.47 43%(U)1 0 0 Y 240 415 0.58 42%(U)2 1 1 R 360 327 1.10 -10%(O)2 1 0 Y 510 532 0.96 4%(U)2 0 1 W 210 201 1.04 4%(O)2 0 0 G 300 247 1.21 21%(O)3 1 1 W 180 141 1.28 28%(O)3 1 0 Y 420 826 0.51 49%(U)3 0 1 R 120 125 0.96 4%(U)3 0 0 G 180 208 0.87 13%(U)4 1 1 W 300 300 1.47 47%(O)4 1 0 G 420 216 1.94 94%(O)4 0 1 R 900 475 1.89 89%(O)4 0 0 Y 600 362 1.66 66%(O)5 1 0 Y 180 162 1.11 11%(O)5 1 1 R 180 137 1.31 31%(O)5 0 0 G 420 469 0.90 10%(U)5 0 1 W 240 176 1.36 36%(O)6 1 0 Y 135 317 0.43 57%(U)6 1 1 W 60 156 0.38 62%(U)6 0 0 G 120 239 0.50 50%(U)6 0 1 R 90 112 0.80 20%(U)7 1 0 G 600 321 1.87 87%(O)7 1 1 R 240 149 1.61 61%(O)7 0 0 Y 240 159 1.51 51%(O)7 0 1 W 240 111 1.08 8%(O)8 1 0 G 300 322 0.93 7%(U)8 1 1 W 45 172 0.26 74%(U)8 0 0 Y 30 189 0.15 85%(U)8 0 1 R 45 124 0.36 64%(U)68Participant VR No-Task Maze Perceived Actual Ratio Diff/ID Time Time Actual9 0 1 R 180 240 0.75 25%(U)9 0 0 Y 300 344 0.87 13%(U)9 1 1 W 180 189 0.95 5%(U)9 1 0 G 420 408 1.02 2%(O)10 0 1 R 90 171 0.53 47%(U)10 0 0 G 150 217 0.69 31%(U)10 1 1 W 120 344 0.35 65%(U)10 1 0 Y 150 324 0.46 54%(U)11 0 1 W 240 154 1.56 56%(O)11 0 0 Y 600 287 2.09 109%(O)11 1 1 R 180 121 1.49 49%(O)11 1 0 G 300 337 0.89 11%(U)12 0 1 W 180 229 0.79 21%(U)12 0 0 G 360 327 1.10 10%(O)12 1 1 R 240 202 1.19 19%(O)12 1 0 Y 180 191 0.94 6%(U)13 0 0 Y 225 416 0.54 46%(U)13 0 1 R 230 380 0.60 40%(U)13 1 0 G 130 316 0.41 59%(U)13 1 1 W 105 179 0.59 41%(U)14 0 0 Y 240 204 1.18 18%(U)14 0 1 W 240 344 0.70 30%(U)14 1 0 G 360 291 1.24 24%(O)14 1 1 R 420 379 1.10 10%(O)15 0 0 G 300 495 0.61 39%(U)15 0 1 R 240 391 0.61 39%(U)15 1 0 Y 180 200 0.90 10%(U)15 1 1 W 240 344 0.70 30%(U)16 0 0 G 193 441 0.44 56%(U)16 0 1 W 120 377 0.32 68%(U)16 1 0 Y 90 285 0.32 68%(U)16 1 1 R 210 517 0.41 59%(O)69Participant VR No-Task Maze Perceived Actual Ratio Diff/ID Time Time Actual17 1 1 R 420 289 1.45 45%(O)17 1 0 Y 600 838 0.72 28%(U)17 0 1 W 230 231 0.99 1%(U)17 0 0 G 340 383 1.89 11%(O)18 1 1 R 480 337 1.42 42%(O)18 1 0 G 300 342 0.88 12%(U)18 0 1 W 300 372 0.81 19%(U)18 0 0 Y 180 355 0.51 49%(U)19 1 1 W 300 341 0.88 12%(U)19 1 0 Y 270 315 0.86 14%(U)19 0 1 R 450 410 1.09 9%(O)19 0 0 G 780 850 0.92 8%(U)20 1 1 W 210 366 0.57 43%(U)20 1 0 G 110 230 0.48 52%(U)20 0 1 R 390 554 0.70 30%(U)20 0 0 Y 405 231 1.75 75%(O)21 1 0 Y 300 166 1.81 81%(O)21 1 1 R 480 207 2.32 32%(O)21 0 0 G 900 365 2.47 47%(O)21 0 1 W 600 162 3.70 270%(O)22 1 0 Y 87 194 0.45 55%(U)22 1 1 W 135 173 0.78 22%(U)22 0 0 G 300 276 1.09 9%(O)22 0 1 R 195 159 1.22 22%(O)23 1 0 G 165 322 0.51 49%(U)23 1 1 R 345 437 0.79 21%(U)23 0 0 Y 600 592 1.01 1%(O)23 0 1 W 420 484 0.87 23%(U)24 1 0 G 420 365 1.15 15%(O)24 1 1 W 600 372 1.61 61%(O)24 0 0 Y 300 149 2.01 101%(O)24 0 1 R 600 456 1.32 32%(O)70Participant VR No-Task Maze Perceived Actual Ratio Diff/ID Time Time Actual25 0 1 R 390 338 1.15 15%(O)25 0 0 Y 330 232 1.42 42%(O)25 1 1 W 205 376 0.54 56%(U)25 1 0 G 180 389 0.46 54%(U)26 0 1 R 265 165 1.60 60%(O)26 0 0 G 340 245 1.39 39%(O)26 1 1 W 95 135 0.70 30%(U)26 1 0 Y 130 363 0.36 64%(U)27 0 1 W 240 167 1.43 43%(O)27 0 0 Y 360 228 1.58 58%(U)27 1 1 R 120 141 0.85 15%(U)27 1 0 G 135 214 0.63 37%(U)28 0 1 W 270 189 1.43 43%(O)28 0 0 G 340 266 1.28 28%(O)28 1 1 R 90 132 0.68 32%(U)28 1 0 Y 130 203 0.64 36%(U)29 0 0 Y 460 326 1.41 41%(O)29 0 1 R 265 146 1.82 82%(O)29 1 0 G 185 382 0.48 52%(U)29 1 1 W 150 246 0.61 39%(U)30 0 0 Y 300 244 1.23 23%(O)30 0 1 W 420 313 1.34 34%(O)30 1 0 G 480 355 1.35 35%(O)30 1 1 R 480 398 1.21 21%(O)31 0 0 G 210 274 0.77 23%(U)31 0 1 R 150 246 0.61 39%(U)31 1 0 Y 135 226 0.60 40%(U)31 1 1 W 150 236 0.64 36%(U)32 0 0 G 660 329 2.00 100%(O)32 0 1 W 480 382 1.26 26%(O)32 1 0 Y 210 265 0.80 20%(U)32 1 1 R 240 350 0.69 31%(U)71Participant VR No-Task Maze Perceived Actual Ratio Diff/ID Time Time Actual33 1 1 R 220 281 0.78 22%(U)33 1 0 Y 210 301 0.70 30%(U)33 0 1 W 420 331 1.27 27%(O)33 0 0 G 525 475 1.10 10%(O)34 1 1 R 240 288 0.83 17%(U)34 1 0 G 180 349 0.52 48%(U)34 0 1 W 332 277 1.20 20%(O)34 0 0 Y 450 363 1.24 24%(O)Table E.1: User Study Data72


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items