2009-lepape & vatrapu-an experimental study of field dependency in altered gz environments

8/10/2019 2009-LePape & Vatrapu-An Experimental Study of Field Dependency in Altered Gz Environments

1/10

An Experimental Study of Field Dependencyin Altered Gz Environments

Marc A. Le Pape

Communication and Information SciencesUniversity of Hawaii at [email protected]

Ravi K. Vatrapu

Center for Applied ICT (CAICT)Copenhagen Business [email protected]

ABSTRACT

Failure to address extreme environments constraints at thehuman-computer interaction level may lead to the commis-sion of critical and potentially fatal errors. This experimentalstudy addresses gaps in our current theoretical understandingof the impact ofGzaccelerations and field dependency in-dependency on task performance in human-computer inter-action. It investigates the effects ofGz accelerations andfield dependency independency on human performance inthe completion of perceptual motor tasks on a personal dig-

ital assistant (PDA). We report the results of an experimen-tal study, conducted in an aerobatic aircraft under multipleGz conditions, showing that cognitive style significantlyimpacts latency and accuracy in target acquisition for per-ceptual motor tasks in altered Gz environments and pro-pose design guidelines as countermeasures. Based on theresults, we argue that developing design requirements tak-ing into account cognitive differences in extreme environ-ments will allow users to execute perceptual motor tasks ef-ficiently without unnecessarily increasing cognitive load andthe probability of critical errors.

ACM Classification Keywords

H.1.2 User\

Machine Systems: Human factors, Human in-formation processing, Software Psychology; H.5.2 User In-terfaces: Evaluation\Methodology, Theory and Methods

Author Keywords

Extreme Environments, Mobile Devices, Target Acquisition,Comparative Informatics, Perceptual Style

INTRODUCTION

When designing safety-critical systems, addressing contex-tual issues pertinent to where, how and why systems areused is of paramount importance. Extreme physical envi-ronments, where users typically operate under conditions ofrisk and stress, have a low tolerance for user error. Failure

to address extreme environments constraints at the human-computer interaction level may lead to the commission of

Permission to make digital or hard copies of all or part of this work forpersonal or classroom use is granted without fee provided that copies arenot made or distributed for profit or commercial advantage and that copiesbear this notice and the full citation on the first page. To copy otherwise, orrepublish, to post on servers or to redistribute to lists, requires prior specificpermission and/or a fee.CHI 2009, April 4-9, 2009, Boston, Massachusetts, USA.

Copyright 2009 ACM 978-1-60558-246-7/09/04...$5.00.

critical and potentially fatal errors[24]. In the age of spaceflights and high performance aircrafts, an emerging HCI chal-lenge is to create innovative interactive systems extendingcurrent design requirements from Earth ubiquitous 1Gzenvironment to altered Gz environments. Designing forextreme environments would enable users to execute percep-tual motor tasks efficiently, without unnecessarily increasingcognitive load and the probability of critical errors. How-ever, meeting such a challenge would require answering firstthe following question: How does context influence percep-

tual motor performance in human-computer interaction? Be-cause extreme environments stressors impact both humanphysiology and human psychology, the significance of un-derstanding their synergistic effects on performance is com-pelling. Indeed, such an understanding would allow us topredict far more accurately than is currently possible themanner in which extreme environments stressors affect per-ceptual motor performance outcomes. We argue that optimaluser interface design for extreme environments such as al-tered Gz extreme environments can only be derived fromsuch an understanding.

RELATED WORK

Previous work on human performance relevant to the quan-tified prediction of human performance in extreme environ-ments for cognitive and perceptual motor tasks exhibits adichotomy at the theoretic level and application level. Atthe theoretic level, the emphasis in experimental psychologyhas been on human information processing models on theone hand[18,13, 10,41], and individual differences in themanner in which information is acquired and processed inthe early stages of information processing on the other hand[44, 21]. A very brief review of theoretic models and appli-cation domains relevant to this study is presented followedby a short summary of the impact of altered Gz environ-ments on human performance.

Information Theoretic Models

Several theoretic models of information processing are wellsuited to explaining and predicting human performance interms of response to visual stimuli in low-level perceptualmotor tasks. The Keystroke-Level Model (KLM) derivedfrom Card, Moran and Newells Model Human Processor[10]has been applied to a wide range of HCI tasks, and re-mains a simple but accurate means of generating quantita-tive estimates of human performance in terms of task exe-cution time. Most appropriately, the KLM has been shown


2/10

to be a model applicable to handheld interfaces with an av-erage prediction error of 3.7% [27]. Whenever a user mustselect among alternatives in a situation in which there aretwo or more possible stimuli, Hicks-Hymans law providesa specific solution to model the decision outcome as a func-tion ofn possible alternatives and in terms of reaction timein target selection [18, 20]. This model also enables quan-tified predictions of human performance as it predicts that

reaction time increases as the logarithm of the number ofalternatives. Fitts law, in its original form and in its vari-ous incarnations, has proved imminently useful in predictingspeed-accuracy trade offs at the user interface level. How-ever, although Fitts law did provide a reliable mechanism topredict pointing time, it did not provide a direct answer to theproblem of predicting error rate. Subsequent work derivedfrom Fitts law does provide such a mechanism[45] and canbe used in conjunction with Fitts law to predict both speedand error rate in target acquisition. Most importantly for thepurpose of this study, Fitts law, and Hicks-Hymans law areboth well suited not only to the evaluation of a wide rangeof pointing and selection tasks executed with the archetypalmice and trackballs pointing devices, but also to the evalu-

ation of pointing and selection tasks executed with currenttablet-stylus devices and the increasingly prevalent touch-screen devices. Just as importantly, both laws are partic-ularly relevant in quantifying human performance in targetacquisition involving both rapid, aimed movements at lowerlevels of information processing such as low level perceptualmotor tasks on a PDA.

Field Dependency-Independency

Jean Piaget, the noted developmental psychologist, deemedthe construct of field dependency independency to be a crit-ical factor in the development of visuoperceptual analysis[33]. As a theory of individual differences in cognition, andof individual differences in information processing modali-ties in particular, field dependency independency theory [44]is highly relevant to the optimization of human performancein the execution of cognitive and perceptual motor tasks.Field dependency independency is typically discussed as acognitive or perceptual style, that is as a heuristic that allindividuals use to process information in their environment(e.g., color, shape, structural patterns or spatial relations),and to selectively encode this information. Furthermore,such heuristics are prevalent at numerous levels of infor-mation processing, from perceptual scanning speed in colorrecognition tasks to accuracy in pattern matching such as inletter selection tasks (e.g., letter comparison), and complexproblem solving tasks [21].

Developed by Herman A. Witkin following a series of exper-iments on visual perception, the concept of field dependencyindependency evolved over time from perception of the up-right to perceptual analytical ability. In other words, theconstruct evolved from a narrow perceptual skill to a fullfledged perceptual style in its own right [44]. As a cog-nitive construct, field dependence independence refers to apersons ability to separate a stimulus from its embeddingcontext, and defines the extent to which an individuals per-ception of a focal stimulus is affected by distracting or con-

flicting contextual cues. Field independent individuals arecharacterized by an overt reliance on internal frames of ref-erences, while field dependent individuals are characterizedby an overt reliance on external frames of reference. Oper-ationally, field dependency independency measures the de-gree to which an individual has difficulty discriminating be-tween part or parts of a field, independently of a distractingor confusing contextual field.

Theoretical indices of field dependency independency in-clude tests of perception of the upright as instantiated inthe Rod and Frame Test (RFT), and tests of disambiguationas instantiated in the Embedded Figures Test (EFT) and theGroup Embedded Figures Test (GEFT). Although both teststarget the same construct, they measure two different aspectsof field dependency independency. The EFT and the GEFTreflect individual differences in cognitive and restructuringabilities arising from overt reliance either on undifferentiatedstructure or on differentiated structure in visual informationprocessing. The RFT reflects individual differences in per-ception of the upright arising from overt reliance either onexternal visual cues or internal vestibular and somatosensory

cues in information processing, in situations when vestibularcues may either conflict or conform with visual cues. Thesethree tests have been widely used as reliable tools in numer-ous studies and are considered valid indicators of the abil-ity of an individual to perceptually ignore conflicting cuesin distracting, misleading, or conflicting contexts [26, 44].Finally, Witkin [43]and his colleagues did not define fielddependency independency as a binary construct but rather asa continuum of individual differences [26] in perceptualperformance, ranging from extreme field independence toextreme field dependence. The conceptualization of field de-pendency independency as a perceptual style has been con-sistently recognized as a significant contribution to the fieldof experimental psychology, its influence has been pervasive

in several research areas, and it has unified findings reportedin many academic disciplines [22].

Safety-Critical Systems and Mission Critical Systems

The design of displays for safety-critical systems (e.g., clin-ical decision support systems [37]) and mission-critical sys-tems (e.g., aviation displays [42]) has largely been imple-mented under the desktop level paradigm. However, very lit-tle work has been done to extend design guidelines evolvedfrom our cumulative design experience within these two do-mains to mobile platforms to be used for similar purposes.Furthermore, notwithstanding the well established potentialand the increased use of mobile platforms such as PDA[s] asintegrated components of full fledged applications in a vari-ety of extreme environments ranging from altered Gzen-vironments, under sea habitats and space [28, 30], even lesshas been done to extend aforementioned design guidelines tomobile platforms for users active in such environments. Asfar as safety-critical systems are concerned, prior researchhas focused on implementing vital sign monitoring applica-tions on a PDA in the context of high-altitude [3, 32] andaltered Gz extreme environments [12]. As far as mission-critical systems are concerned, prior work investigated thearea of RFID identification in space [9]. However, there is


3/10

no indication in the literature that issues of context use, letalone individual cognitive differences, were ever given seri-ous thought in the design and evaluation of these systems.We argue that considering both the strenuous demands ex-treme environments place on human physiology and theirsevere impact on cognitive functions, due consideration tothese issues is long over due if the necessity of optimizinghuman performance while minimizing cognitive load and

the probability of critical error is ever to be enforced at theuser interface level.

Performance Degradation in Altered Gz Environments

Extreme environments can be defined as severe ecologicalsystems well outside the range of normal human survival pa-rameters, where optimal human performance predicated ona high level of physiological adaptation, technological inno-vation, and training is a sine qua non requisite for survival.Extreme environments act as physiological stressors induc-ing dysfunctional physiological activity. In particular, ex-treme environment stressors such as high acceleration loadsimpact both human physiology and cognition, sequentiallyaffecting human receptors and proprioreceptors, informationprocessing, and motor control. A significant degradation ofhuman performance ensues, particularly conspicuous in de-lays and errors in task execution [41]. Nonetheless, for theuser of a mission-critical or safety-critical system in alteredGz environments, maintaining optimal performance at alltimes remains a critical requirement. This combination offactors obviously places severe constraints on the use anddesign of mobile devices for users operating in such envi-ronments.

By convention, the capitalized letterGis used to express theinertial resultant to whole body acceleration in multiples ofthe magnitude of the acceleration of gravity denotedg. Tra-ditionally, theG notation is used when referring to rapidtransitions between positive and negative G forces valuesof the acceleration vector. In particular, the notation Gzrefers to the positive or negative values of the accelerationvector along the vertical axis of the acceleration vector, i.e.,accelerations parallel to the spine of the human body.

Human physiological responses to Gz acceleration forcesare well documented in the medical literature [14, 1]. Theliterature discusses a constellation of physiological responsesto the impact ofGzacceleration forces whose symptoma-tology can be summarized as follows: (a) an inability of thenormal cardiovascular system mechanisms to provide ade-quate circulation to the brain and the eyes, resulting in com-pounded symptomatology and physiological changes; and(b) an increased normal weight hydrostatic pressure and phys-iological ventilation-perfusion gradients in the lungs, result-ing in increased pulmonary arterio-venous shunting and sub-sequent impairment of circulatory oxygenation. Associatedsymptomatology includes, but is not limited to, arterial bloodpressure fluctuations, decreased cardiac output, cardiac ar-rhythmias, visual disorders, reduction of peripheral vision(grey-out), total loss of vision (black-out), followed by lossof consciousness (G-LOC). Individuals exposed to rapidGzaccelerations experience sudden G-LOC without warning,

i.e., without attending peripheral or total loss of vision [15].Typically, G-LOC lasts for 3 to 10 seconds. After regainingconsciousness, individuals typically remain disoriented foranother 4 to 11 seconds [5]. Although generally perceived asa less dramatic response than G-LOC, smaller gravity fluc-tuations create in individuals an altered state of awarenessand a number of associated cognitive impairments referredto as Almost Loss of Consciousness (A-LOC). The effects of

A-LOC on cognitive functions were described as a discon-nection between cognition and the ability to act on it[40].Finally, and contrary to widely shared assumptions amongaircraft pilots, research shows that exposure to G forcesthrough professional training neither eliminates or dimin-ishes physiological responses to altered G environments[16], and the use of anti-Gsuits only slightly delays the on-set of G-LOC [15].

Different lines of research investigating the particular in-fluence of these factors on human performance have so farreached a consensus with regard to the dynamic interdepen-dence between extreme environments stressors, cognitionand performance. In particular, distinct lines of empirical

research have conclusively shown that (a) physiological ef-fects ofGz accelerations adversely impact performance[23, 17,14]; (b) performance in Gz environments corre-lates negatively with stress [40, 16, 35]; (c) performancecorrelates negatively with field dependency as a perceptualstyle on a wide range of sensory-motor tasks [6, 7,25]; (d)field-dependency correlates positively with stress [19, 38];and (e) field-dependency-independency is an important de-terminant in human-computer interaction [8, 2]. However,a careful review of the relevant literature suggests signifi-cant gaps in our understanding of this dynamic interdepen-dence between extreme environments stressors, individualcognitive differences, and performance. For instance, littleis known regarding interaction effects between extreme en-

vironments physical stressors and human cognitive abilities[31]. Even less is known regarding interaction effects be-tween field dependency independency as a perceptual styleand Gzaccelerations as an extreme environment physicalstressor[29,39]. Acknowledging these gaps in our under-standing of this dynamic interdependence of variables andintegrating lines of empirical research in cognitive psychol-ogy and human physiology, the experimental study reportedhere investigates how performance varies between field de-pendent and field independent groups under conditions ofalternating Gz accelerations. The primary purpose of thisresearch is to understand the nature and extent of the com-bined effects of field dependence-independence and Gzaccelerations on human perceptual motor performance.

METHOD

Design

The experimental design is a 5x2 mixed factorial design withrepeated independent measures consisting of 5 levels of thewithin-subjects factor (altered Gzaccelerations of2Gz,1Gz,+1Gz,+2Gz, and +3Gz) and 2 levels of the between-subjects factor (field dependence and field independence).It involves two independent groups of participants assignedthrough testing to the two categories of the field dependency


4/10

independency factor and randomly assigned to all five levelsof the Gz acceleration factor.

Independent variables

The independent variables are (1) altered Gz environmentand (2) field dependency independency. The alteredGzenvironment is operationalized as Gz acceleration loadscorresponding to the 5 levels of the Gzacceleration vector

coefficient used in this study. Field dependency indepen-dency is operationalized as a dichotomous variable througha block sampling strategy assigning extreme field indepen-dent and extreme field dependent participants to the two lev-els of the independent variable. Everything else being equal,any differences found in the measurements of the dependentvariables between the two levels of the field dependency in-dependency factor and across the five levels of the Gzac-celeration factor can only be attributed to either interactioneffects of perceptual style and Gz accelerations, main ef-fect of perceptual style, and/or main effect ofGz acceler-ation.

Dependent variables

The dependent variables consist of three individual measuresof human performance evaluated in terms of efficiency, ef-fectiveness, and user interaction satisfaction. Efficiency isoperationalized as latency in task execution. Effectiveness isoperationalized as accuracy in task execution. User interac-tion satisfaction is operationalized as a self-reported score onthe Questionnaire for User Interaction Satisfaction (QUIS).

Participants

A total of 18 participants participated in the study, with 9field independent and 9 field dependent participants assignedto the two levels of the field dependent independent factor. Atotal of 90 non-solicited participants were included into thesampling frame after indicating their intention to volunteerfor the experimental study and after taking a standard test forfield dependency independency, the Group Embedded Fig-ure Test (GEFT). This test, whose scale ranges from zero toeighteen, has been widely used as a reliable instrument in nu-merous studies and is considered a valid indicator of the abil-ity of an individual to perceptually ignore conflicting cuesin distracting, misleading or conflicting contexts. A typicalsampling strategy is to use a theoretical index of field depen-dency independency such as the GEFT to rank participantson the field independency dependency continuum and to usethe median score to divide them into two groups. Partici-pants whose score falls on or below the median are assignedto the field dependent group, while participants whose scorefalls above the median are assigned to the field independentgroup. However, in order to create two groups of extremefield dependent and extreme field independent participants,this study enforced block sampling as an alternate strategy.Accordingly, a sample of 18 participants was drawn out ofthe original pool of 90 participants included in the samplingframe. The field dependent experimental group consistedof 9 participants who scored between 0 and 9 of the GEFT,and the field independent experimental group consisted of 9participants who scored between 17 and 18. Due to safetyconcerns, and to control for a potential confounding variable

in the experimental design, participants screened for vestibu-lar disorders were excused from the sampling frame as theycould have experienced dizziness, vertigo, imbalance, andnausea during the experiment. Participants received no com-pensation for their participation.

Demographics

Average age of field independent participants was 28.89 years(Range:19 39, SD = 5.71). The field independent groupconsisted of 7 non-pilots and 2 pilots. Of the 9 field indepen-dent participants, 5 were male and 4 female. Average scoreon the Computerized Rod and Frame Test (CRFT) was0.35(Range: 0.20 0.54, SD = 0.11, SE = 0.03). On theGroup Embedded Figure Test (GEFT), the primary measureof field dependency independency, participants scored an av-erage of17.33(Range:17 18,SD= 0.50,S E= 0.17).

Average age of field dependent participants was 31.89 years(Range:23 64,S D = 13.28). The field dependent groupalso consisted of 7 non-pilots and 2 pilots. Of the 9 fielddependent participants, 6 were male and 3 female. Averagescore on the Computerized Rod and Frame Test (CRFT) was0.81(Range: 0.13 2.00, SD = 0.56, SE= 0.18). On theGroup Embedded Figure Test (GEFT), the primary measureof field dependency independency, participants scored an av-erage of7.22(Range:4 9, SD = 1.86, SE= 0.62).

No significant differences were observed between the exper-imental groups on age, pilot experience and gender. As ex-pected, a one-way analysis of variance with respect to CRFTscores was significant (F(1, 17) = 5.68, p = 0.03). Simi-larly, a one-way analysis of variance with respect to GEFTscores was significant (F(1, 17) = 249.05, p


5/10

Figure 1. Experimental Setup

The selection of a high performance aerobatic aircraft forthe experimental study was dictated by ecological validityconcerns.

Tasks

Participants in the experiment were asked to execute threetasks. The first task was a button selection task, the secondtask was a letter selection task, and the third task was a fielddependent independent task.

Button selection task

The first task is designed to evaluate and compare partici-pants performance in selecting a single red button randomlydisplayed within each of the four tasks configurations. Four

Figure 2. Button Selection Task

different configurations of the button selection task are pre-sented in random order to the participants. Each button se-lection task configuration on the screen displays either 4, 9,16, or 25 buttons arranged in a grid layout pattern. Figure2shows an example of a randomly generated red button tasksequence.

Participants are instructed to select the red button as quicklyand as accurately as possible. Immediately after selection,or after a three second time-out, the screen displays the nexttask to the participant. Each event of the button selectiontask times out after being displayed for three seconds.

Letter selection task

The second task is designed to evaluate and compare partic-ipants performance in identifying and selecting a single let-ter of the Roman alphabet randomly displayed within each

of the four tasks configurations. The letter to be selected

Figure 3. Letter Selection Task

is first displayed by itself in the center of the screen for a

duration of one second. Immediately thereafter, one of fourdifferent randomly selected letter grid layout screens is pre-sented to the participant. Each configuration on the screendisplays either 4, 9, 16, or 25 non-repeated letters arrangedin a grid layout pattern defining each of the four letter selec-tion tasks configurations.

Immediately after selection, or after a three or four secondstime-out, the screen displays the next task to the participant.The second task times out at three seconds for all but the 5x5task configuration. Due to floor effect observed in pilot stud-ies in the 5x5 letter selection task configuration, the time outfor this particular task configuration was increased to fourseconds. Figure3 shows an example of a randomly gener-

ated letter selection task sequence with its random combina-tions of letter and letter grid layout.

Field Dependent-Independent task

The third task, implemented around the rod and frame illu-sion (RFI), is designed to evaluate and compare participantsperformance in spatial orientation against a baseline previ-ously established on the ground.

The orientation of a vertical rod when viewed in the con-text of a tilted frame is typically misperceived as the resultof the tilt of the frame. Since the rod and frame illusion isleveraged as a test of field dependency-independency in therod and frame test (RFT) [44] this task also compares partici-pants performance against a baseline previously establishedon the ground through the administration of the RFT. One ofthe two randomly selected rod and frame events is first dis-played by itself in the center of the screen for a durationof two seconds. Immediately thereafter, a second screen ispresented to the participant askingWas the rod tilted. Theparticipant has two seconds to answer the question by select-ing either one of two buttons labeled Yes or No. Figure4shows each of the two possible field dependent indepen-dent task configurations and the associated question screen.


6/10

Unknown to the participant, the rod is invariably set exactlyto the vertical in each task event. Each event of the fielddependent independent task times out at four seconds.

Figure 4. Field Dependent-Independent Task

As explained in the Design section, all three tasks were pre-

sented to all participants at each of the five levels of the Gzfactor. Each and every specific tasks configuration as wellas its order of presentation was randomly generated beforebeing hard coded into five different task sequences. At anyparticularGzlevel, the tasks were presented to the partici-pants within the same unique tasks configurations sequenceassigned to this particularGz level. It follows that (a) ata specific level of the Gz factor all participants to the ex-periment were presented with exactly the same unique tasksequence featuring the same tasks configurations presentedin the same order, and (b) each of the five Gzlevel specifictask sequences displayed in a different order nine tasks con-figurations unique to this particular sequence. At each levelof theGz factor the entire task sequence timed out at 29

seconds.

To control for nausea at all Gz levels, participants wereto report on the PDA at the end of each task sequence theirsubjective wellness state before being allowed to continuethe experiment. The PDA start up and wellness screensare shown in Figure5.No participant to the experiment everbecame nauseated or reported feeling bad.

Figure 5. PDA Start-Up and Wellness Screen

Hypotheses

Based on available empirical evidence and a careful reviewof the literature on individual differences and Gz acceler-ation physiology, the following three hypotheses were ad-vanced:

H1 There will be a statistically significant interaction effectin performance between the Gz accelerations factor and

the field dependency field-independency factor on the buttonselection and letter selection tasks.

H2 Performance will be significantly lower at altered Gzacceleration levels than at the +1Gz acceleration level onthe button selection and letter selection tasks.

H3 Performance will be significantly lower for field depen-dent participants than field independent participants on thebutton selection and letter selection tasks.

Data Collection

For each task event, both latency and accuracy in task execu-tion were recorded. All experimental data was logged in to

a CSV file. The data output file opened at the beginning ofeach task, was written to as the time of task completion, andwas closed immediately afterward, so all data was recordedeven in the case of abnormal termination of the program.

Procedure

To ensure both familiarity with the software and the tasks,all participants were instructed on how to run the applicationand were trained on a demo task sequence before the flight.All participants were asked to execute the tasks as accuratelyand as quickly as possible before running the application onthe ground and as a last reminder just before take-off. Par-ticipants to the experiment were seated in the co-pilot seat ofa CAP10B aerobatic aircraft with the PDA strapped to their

left thigh if they were left handed, or right thigh if they wereright handed. This is a standard interactive mode for readingand writing notes for pilots and co-pilot of an aircraft.

Each participant to the experiment flew at all 5 Gz levels inthe same sequential order: +1Gz,+2Gz,+3Gz, 1Gzand2Gz. At each Gzlevel, after setting the aircraft into the ap-propriateGz flight pattern, the pilot confirmed to the partic-ipant theGz level value and instructed the participant to be-gin the tasks sequence with the simple instruction: Start!.After selecting the appropriate Gz level button on the PDAstart-up screen, the participant executed the entire tasks se-quence for this Gz level. Immediately after completing thetask sequence for a particular Gz level, the participant al-

lowed the pilot to unload theGzand resume +1Gznormalflying conditions with the simple call: All done!. The pilotthen returned for one minutes to a normal+1Gz flight pat-tern to allow the participants physiology to return to normalbefore exposure to the next Gzlevel.

RESULTS

Results are organized under the following four sections: hy-potheses testing, field dependent independent task, wellness,and user interface satisfaction.


7/10

Hypotheses Testing

Latency

A mixed factorial repeated measures multivariate analysis ofvariance of latency in task execution showed no statisticallysignificant interaction effects between the field dependencyindependency factor and the altered Gz acceleration fac-tor for latency (Roys largest root = 0.142, F(4, 13) =0.461, ns). Thus, the first hypothesis H1was not supported.

However, hypothesis H2 and H3 were supported. Analysisshowed a significant main effect of the within subjects Gzaccelerations factor on latency (H2) (Roy

s largest root =2.572, F(4, 13) = 8.358, p < 0.001, partial 2 = 0.72,observed power = 0.982. As predicted, performance onboth the button selection and letter selection tasks worsenedunder altered Gz acceleration conditions compared to the+1Gz condition. Figure 6 presents average time, in mil-liseconds, on the combined button selection and letter selec-tion tasks across the five levels of the altered Gz acceler-ation factor. The error bars display 95% CI of the means.Analysis showed a significant main effect of the between

Figure 6. Latency Across Altered Gz Levels

subject field dependency independency factor on latency (H3)(F(1, 16) = 4.644, p= 0.047, partial2 = 0.225, observedpower= 0.526). As predicted, field dependent participantstook more time than field independent participants on boththe button selection and the letter selection tasks. Figure7presents presents average time, in milliseconds, taken by thefield independent group versus the field dependent groups.

Figure 7. Latency between the Experimental Groups

Figure 8. Latency between the Experimental Groups by Task Types

Figure8 presents the results for latency across the two per-ceptual style groups for the two task types. Furthermore,results show that task complexity (2x2, 3x3, 4x4, 5x5) in-creased latency. Figure9 presents results across task com-plexity levels.

Figure 9. Latency between the Experimental Groups by Task Com-plexity

Accuracy

A mixed factorial repeated measures multivariate analysisof variance of accuracy in task execution showed no statis-tically significant interaction effect between the field depen-dency independency factor and the Gzacceleration factor(Roys largest root = 0.267, F(4, 13) = 0.868, ns. Nosignificant differences were observed on accuracy for theGz accelerations within-subjects factor. Marginally sig-nificant differences in accuracy were observed between thefield dependent and field independent experimental groups

(F(1, 16) = 3.866, p= 0.067, partial2

= 0.195, observedpower = 0.456). Accuracy was higher for field dependentparticipants on the button selection task whereas field inde-pendent participants scored higher on accuracy on the letterselection task.

Field Dependent-Independent Task

Significant differences on latency for the field dependent in-dependent task were observed between the two experimentalgroups at the+1Gz level and at none of the other 4 altered


8/10

Gzlevels. Accuracy on the task was not significantly dif-ferent between the two experimental groups at any of the 5Gzlevels.

Wellness

Participants self-reported high scores on wellness across thefive levels of the altered Gz accelerations factor with nosignificant differences observed between the field dependent

and field independent experimental groups or across the lev-els of the Gzaccelerations factor. Therefore, both motionsickness and nausea can be ruled out as potential confound-ing factors in the experimental study.

User Interface Satisfaction

The QUIS questionnaire[11] was administered to collect theparticipants subjective satisfaction scores. The QUIS 7.0 in-strument also measured participants subjective satisfactionwith the instructions and the tutorial, aside from various sys-tem measures. No significant differences were observed onany of the sections of the QUIS instrument.

DISCUSSION

The experimental results support the hypothesis that perfor-mance for field dependent participants is significantly lowerthan for field independent participants on the button selec-tion and letter selection tasks. As such, it is clear that in al-teredGz environments individuals mode of informationprocessing has a significant impact on perceptual motor per-formance when interacting with the touchscreen interface ofa mobile device. This studys findings corroborate the im-pact ofGz acceleration on task performance documentedin prior research. For example, an experimental study in-vestigating the effects ofGzacceleration on cognitive per-formance showed performance degradation across individ-uals on a compensatory tracking task, a system monitoringtask, and a strategic resource management task [29]. Theresults of our study suggests that field dependent individu-als would benefit from mobile displays that provide supportfor disambiguation of objects and artifacts in a perceptualfield. The experimental results also support the hypothesisthat performance on the button selection and letter selec-tion tasks is significantly lower in alteredGzenvironmentsthan on Earths ubiquitous +1Gz environment. This find-ing corroborates prior general findings on the impact of al-tered Gz environments on human performance and firmlyestablishes their relevance at the human computer interac-tion level. The studys findings corroborate the impact ofperceptual style on task performance documented in priorresearch. For example, an experimental study investigat-ing the relationship between perceptual style and trackingability showed that among U.S. army pilots, field dependentindividuals had significantly more difficulty than filed inde-pendent individuals in tracking an object similar in color toits background [4]. The results of our study suggest thatusers of mobile devices in Gz environments would ben-efit from displays featuring an interface adaptive to alteredGz environments. The experimental results failed to sup-port hypothesized interaction effects between altered Gzenvironments and field dependency-independency. This sug-gests that the impact of altered Gz environments on hu-

man physiology and individual differences in informationprocessing operates at different levels with little cross-levelinteraction. Each of these factors influences human perfor-mance but there is no empirical evidence to indicate that theysignificantly impact each other. The results illustrate howenvironment independent and individual homogeneity de-sign assumptions can contribute to significant degradation inperformance in situated human computer interaction. They

suggest the need to develop and test innovative user-interfacedesign requirements for mobile devices aimed at mitigatingthe effects of altered Gzand individual differences on hu-man performance. They also demonstrate the need to de-velop adaptive displays addressing both the constraints al-tered Gz environment place on users, and the differencesin information processing among users. A two pronged ap-proach to the problem might indeed be more productive of asolution than either approach in isolation.

Implications for HCI

Historically, HCI has witnessed an evolution with regard tocontext of use. While HCI was originally concerned with

the desktop and desktop applications, the scope of its ap-plication has considerably widened with the emergence ofubiquitous computing. As a result, its domain of inquiryhas become increasingly more diversified. We already knowthat mobile devices break many assumptions inherent to HCIand desktop computing[36]. However, we know very littleas to how performance in HCI covaries with the constraintsimposed by the environment. As such, the HCI commu-nity might consider a proactive research agenda with regardto mobile computing in extreme environments. Indeed, theevidence is that mobile devices have a promising future inextreme environments as their increasingly ubiquitous pres-ence in such environments suggests. For instance, profes-sional users increasingly require mobile computational de-vices when working in extreme environments as diverse asdeep-sea diving [28], high altitude [3] orspace[9]. The samecould be argued with regard to the increasing propensity ofrecreational users to use mobile devices as consumer prod-ucts (e.g., GPS) in extreme recreational environments activi-ties as diverse as sky-diving, high performance flying, scuba-diving or high altitude climbing to name a few. Further-more, one of the primary goals of usability is safety [34],and as this paper pointed out extreme environments are al-ways characterized by conditions of risk and stress. As such,for users of mobile devices immersed in such environments,safety-critical issues become of paramount importance, life-critical in fact. Finally, if any claims to ecological validitycan legitimately be made, HCI needs to research these is-sues in their extreme environment context of use. The factsare that (a) mobile technology is increasingly ubiquitous inextreme environments; (b) mobile devices are a logical plat-form for users to choose in such environments; and (c) thissteady trend among users creates a host of HCI concerns thatbeg to be addressed. For the aforementioned reasons, the au-thors feel that there are urgent, tangible, and promising re-search opportunities for HCI researchers to contribute theirexpertise to, in an effort to address what we see as a newchallenge emerging on the HCI horizon.


9/10

FUTURE WORK

The long term purpose of this line of research is to enableusers of mission-critical and safety-critical systems in al-tered Gz environments to execute low level interactiontasks quickly and efficiently, without unnecessarily increas-ing cognitive load and the probability of critical errors. Aparticularly promising application domain specifically requir-ing individuals to be mobile while using small, lightweight

computing devices, lies within spaces zero Gzenvironment.As a case in point, to allow space crews to perform mission-critical tasks, both the ESA and NASA are increasingly re-lying on mobile computing platforms, and particularly onPDA[s]. They are also relying increasingly on these mobileplatforms to perform safety-critical tasks such as monitor-ing crew members vital signs. To extend this research tozero Gz environments, the authors have initiated prelimi-nary negotiations with NASA to establish modalities of useof NASA Ames C9-B parabolic aircraft which would sim-ulate a near zeroGz environment for future experiments.

CONCLUSION

This experimental study analyzed the main effects and inter-

action effects of altered Gz accelerations and individualdifferences in perceptual style on performance. The studyshowed that both Gz accelerations and perceptual stylesignificantly impact human performance. The findings couldhelp predict in an accurate manner how altered Gz envi-ronments and individual differences in information process-ing impact human performance. The findings could also pro-vide guidelines for the design of interactive mobile comput-ing platforms used in altered Gzenvironments. Additionalexperimental evidence would help in the development of atheory of human-computer interface design in extreme envi-ronments.

ACKNOWLEDGMENTS

The authors would like to acknowledge Kim Binsted, DanielSuthers, Scott Robertson, and Curtis Ikehara for insightfuladvice and encouragement in investigating HCI in extremeenvironments; Pat Gilbert for re-writing the experimentalstudy software; Jeff Bagust for providing the CRAF soft-ware. The second author was supported by NSF awards #0093505 to Daniel Suthers and # 0535036 to Scott Robert-son.

REFERENCES

1. W. B. Albery. Acceleration in other axes affects +Gztolerance: Dynamic centrifuge simulation of agileflight.Aviation, Space, and Environmental Medicine,

75(1):16, 2004.

2. L. Andrea, Z. Shumin, and B. William. Manual andcognitive benefits of two-handed input: anexperimental study.ACM Trans. Comput.-Hum.Interact., 5(4):326359, 1998.

3. P. B. Angood, R. Satava, C. Doarn, and R. Merrel.Telemedicine at the top of the world: the 1998 and1999 Everest extreme expeditions.TelemedicineJournal and e-Health, 6(3):31525, 2000.

4. Atchley:Paul. Perceptual style and trackingperformance.International Journal of AviationPsychology, 1(3):255260, 1991.

5. P. Benni, J. Li, B. Chen, J. Cammarota, and D. Amory.NIRS monitoring of pilots subjected to +Gzacceleration and G-induced loss of consciousness(G-LOC).Advances in experimental medicine andbiology, 530:3719, 2003.

6. M. Bloomberg. Field independence-dependence andsusceptibility to distraction.Perceptual and MotorSkills, 20:805813, 1965.

7. M. Bloomberg. Differences between field independentand field dependent persons on the stroop color-wordtest.Journal of Clinical Psychology, 25(1):4545,1969.

8. J. A. Bohan, D. A. Boehm-Davis, and R. Marshall. Therole of individual differences in choice of strategy andperformance in a computer-based task. Xbnrl, GeorgeMason University, Fairfax, VA., June 1995.

9. P. Brown, P. Hartmann, A. Schellhase, A. Powers,T. Brown, C. Hartmann, and D. Gaines. 2E-3 Assettracking on the international space station using globalSAW tag RFID technology. In Ultrasonics Symposium,2007. IEEE, pages 7275, 2007.

10. S. K. Card, T. P. Moran, and A. Newell.The psychologyof human-computer interaction. Erlbaum, 1984.

11. J. P. Chin, V. A. Diehl, and K. L. Norman. Developmentof an instrument measuring user satisfaction of thehuman-computer interface. In ACM, editor,SIGCHI88, pages 213218, New York, 1988. ACM.

12. M. Di Rienzo, G. Parati, F. Rizzo, P. Meriggi,G. Merati, A. Faini, and P. Castiglioni. Heart ratemonitoring and control in altered gravity conditions. InEngineering in Medicine and Biology Society (EMBS)2007. 29th Annual International Conference of theIEEE, pages 6681 6684, 2007.

13. P. M. Fitts. The information capacity of the humanmotor system in controlling the amplitude ofmovement. Journal of Experimental Psychology,47(6):381391, 1954.

14. K. Fong and S. Fan. An overview of the physiologicaleffects of sustained high+Gz forces on human being.Annals Academy of Medicine Singapore, pages 94103,

1997.

15. D. Gaffie, A. Guillaume, and P. Quandieu. Modelingand numerical simulation of the cerebral blood flowunder acceleration.The Physiologist, 36(1 Suppl.),1993.

16. S. Galvagno, T. Massa, and S. Price. Acceleration riskin student fighter pilots: preliminary analysis of amanagement program.Aviation, Space, andEnvironmental Medicine, 75(12):10771080, 2004.


10/10

17. K. Gillingham. High-G stress and orientational stress:physiologic effects of aerial maneuvering.Aviation,Space & Environmental Medicine, 59, 1988.

18. W. E. Hick. On the rate of gain of information.TheQuarterly Journal of Experimental Psychology,4(1):1126, 1952.

19. E. Hill and K. Eigenbaum. Altering field dependence

through stress.Perceptual and Motor Skills, 23(3),1966.

20. R. Hyman. Stimulus information as a determinant ofreaction time.Journal of Experimental Psychology,45:188196., 1953.

21. M. Kozhevnikov. Cognitive styles in the context ofmodern psychology: Toward an integrated frameworkof cognitive style.Psychological Bulletin,133(3):464481, 2007.

22. R. M. Kurtz. A conceptual investigation of Witkinsnotion of perceptual style.Mind, 78(312):522533,1969.

23. S. Leverett and R. Burton. Physiological effect of high,sustained+Gz forces on man. Life sciences and spaceresearch, 17:17185, 1979.

24. N. Leveson. Software safety: Why, what, and how.ACM Computing Surveys, 18(2), 1986.

25. G. Long, R. Ambler, and G. F.E.. Relationship betweenperceptual style and reactivity to motion. AppliedPsychology, 60(5):599605, 1975.

26. G. M. Long. Field dependencyinterdependency. areview of the literature. Monograph 19, NavalAerospace Medical Research Lab, Pensacola,FL., June21 1972.

27. L. Luo and B. John. Predicting task execution time onhandheld devices using the keystroke-level model,2005.

28. K. Montgomery, C. Mundt, G. Thonier, A. Tellier,U. Udoh, V. Barker, R. Ricks, L. Giovangrandi,P. Davies, Y. Cagle, J. Swain, J. Hines, and G. Kovacs.Lifeguard: A personal physiological monitor forextreme environments. In26th Annual InternationalConference of the IEEE Engineering in Medicine andBiology Society, 2004.

29. J. G. Morrison, E. Forster, E. M. Hitchcock, C. A.Barba, T. P. Santarelli, and M. W. Scerbo. Cumulativeeffects of+Gz on cognitive performance.In:Proceedings of the Human Factors and ErgonomicsSociety 38th Annual Meeting, 1:4650, 1994.

30. K. Mundt, C.W.and Montgomery. A multiparameterwearable physiologic monitoring system for space andterrestrial applications. In Information Technology inBiomedicine, IEEE Transactions on, volume 9, pages382 391, 2005.

31. NASA. Man-systems integration standards.NASA-STD-3000-Man-Systems Integration Standards(MSIS)Revision B, July 1995 Section 5.9.2 CombinedEnvironmental Effects Design Considerations, 1, 1995.

32. J. Onopa, M. Le Pape, G. Thonier, S. Saiki,K. Montgomery, and L. Burgess. High altitude researchHawaii. In13th International Hypoxia Symposium,pages 451451, Banff, Alberta, Canada., 2003.

33. J. Piaget and B. Inhelder.The childs conception ofspace. W. W. Norton, New York, 1967.

34. J. Preece, Y. Rogers, and H. Sharp.Interaction Design:Beyond Human-Computer Interaction. J. Wiley andSons, New York, NY, 2002.

35. C. A. Rickards and D. G. Newman. G-induced visualand cognitive disturbances in a survey of 65 operationalfighter pilots.Aviation, Space, and EnvironmentalMedicine, 76(5):496500, 2005.

36. T. Rodden, K. Chervest, N. Davies, and A. Dix.Exploiting context in HCI design for mobile systems.

InWorkshop on Human Computer Interaction withMobile Devices, 1998.

37. P. M. Sanderson, M. O. Watson, and W. J. Russell.Advanced patient monitoring displays: Tools forcontinuous informing.Anesthesia & Analgesia,101(1):161168, 2005.

38. V. Sarris, E. Heineken, and H. Peters. Effects of stresson field dependence.Perceptual and motor skills,43(1):1212, 1976.

39. M. W. Scerbo. The effects of positive acceleration oncognitive performance: Big G - little know.In:Proceedings of the Human Factors and ErgonomicsSociety 39th Annual Meeting

, 1:3539, 1995.40. B. Shender, E. Forster, L. Hrebien, H. Ryoo, and

J. Cammarota. Acceleration-induced near-loss ofconsciousness: The A-LOC syndrome.Aviation,Space, and Environmental Medicine, 74(10):10218,2003.

41. C. D. Wickens.Engineering psychology and humanperformance. Prentice Hall, Upper Saddle River, NJ,3rd edition, 2000.

42. C. D. Wickens. Attention and aviation display layout:Research and modeling. Technical Report ContractNASA NAG 2-1535, NASA Ames Research Center,2005.

43. H. Witkin. The perception of the upright.ScientificAmerican, 200(2):5156, 1959.

44. H. A. Witkin and D. R. Goodenough.Cognitive Styles,Essence and Origins: Field Dependence and FieldIndependence. International Universities Press, NewYork, 1981.

45. J. Wobbrock, E. Cutrell, S. Harada, and S. MacKenzie.An error model for pointing based on Fitts law, 2008.

2009-lepape & vatrapu-an experimental study of field dependency in altered gz environments

Documents