The Perception and Measurement of Human-Robot Trust

University of Central Florida STARS

Electronic Theses and Dissertations, 2004-2019

2013

The Perception And Measurement Of Human-robot Trust

Kristin Schaefer University of Central Florida

Part of the Psychology Commons Find similar works at: https://stars.library.ucf.edu/etd University of Central Florida Libraries http://library.ucf.edu

This Doctoral Dissertation (Open Access) is brought to you for free and open access by STARS. It has been accepted for inclusion in Electronic Theses and Dissertations, 2004-2019 by an authorized administrator of STARS. For more information, please contact [email protected].

STARS Citation Schaefer, Kristin, "The Perception And Measurement Of Human-robot Trust" (2013). Electronic Theses and Dissertations, 2004-2019. 2688. https://stars.library.ucf.edu/etd/2688

THE PERCEPTION AND MEASUREMENT OF HUMAN-ROBOT TRUST

KRISTIN E. SCHAEFER B. A. Susquehanna University, 2003 M. S. University of Central Florida, 2009

A dissertation submitted in fulfillment of the requirements for the degree of Doctor of Philosophy in the Department of Modeling and Simulation in the College of Sciences at the University of Central Florida Orlando, Florida

Summer Term 2013

Major Professor: Peter A. Hancock

ABSTRACT

As robots penetrate further into the everyday environments, trust in these robots becomes a crucial issue. The purpose of this work was to create and validate a reliable scale that could measure changes in an individual’s trust in a robot. Assessment of current trust theory identified measurable antecedents specific to the human, the robot, and the environment. Six experiments subsumed the development of the 40 item trust scale. Scale development included the creation of a 172 item pool. Two experiments identified the robot features and perceived functional characteristics that were related to the classification of a machine as a robot for this item pool.

Item pool reduction techniques and subject matter expert (SME) content validation were used to reduce the scale to 40 items. The two final experiments were then conducted to validate the scale. The finalized 40 item pre-post interaction trust scale was designed to measure trust perceptions specific to HRI. The scale measured trust on a 0-100% rating scale and provides a percentage trust score. A 14 item sub-scale of this final version of the test recommended by

SMEs may be sufficient for some HRI tasks, and the implications of this proposition were discussed.

iii

This dissertation is dedicated to my advisor and mentor, Dr. Peter A. Hancock.

ACKNOWLEDGMENTS

Over the last few years, there are a number of other people who provided both professional and personal support throughout this process. First and foremost, I would like to thank my advisor and committee chair, Dr. Peter A. Hancock. There are not enough words to thank you for helping me to become the researcher I am today. It has truly been a joy and honor to work with you. Although I will miss our early morning coffee breaks, I look forward to future endeavors that may come our way.

Secondly, I would like to thank my committee members, Drs. John D. Lee, Florian

Jentsch, Peter Kincaid, Deborah R. Billings, and Lauren Reinerman. Each of you encouraged me to strive for excellence every day. Thank you for your time, support, and guidance that helped me to develop a sound theory, strong methodology, and statistical clarity that will benefit the

HRI community.

I would also like to acknowledge a number of professional colleagues who provided support for this research. I would first like to thank the U. S. Army Research Laboratory (ARL) and the support of many colleagues involved with the Robotics Collaborative Technical Alliance

(RCTA). In addition, a special thank you to Drs. Jessie Y. C. Chen and Susan Hill for their unending support, encouragement, and guidance in furthering the field of human-robot trust. I would also like to thank Ralph Brewer, ARL, and Dave Wagner, GDRS, for their guidance and recommendations with the RIVET simulation development. In addition, there were eleven important people who without them I could not have completed my research. I would like to extend a special thank you to my subject matter experts: Ralph Brewer, Jessie Chen, A. William

Evans, Jason Gregory, Chad Kessens, Stephanie Merritt, Philip Osteen, Jason Owens, Aaron

Steinfield, and two additional SMEs who prefer to remain anonymous.

A very special thank you is also required for my colleagues and friends, Gabriella

Hancock, Ryan Yordon, Valerie Willis, and Tracy Sanders for taking the time to read draft after draft of my work and providing critical feedback. I am also so proud of my undergraduate students, Angela Bardwell-Owens, Jackie Cook, and Jacob Whitney, who took this opportunity to strive for excellence themselves. Thank you for taking the lead on our conference presentations this last year. I am so proud of all your accomplishments and honored to have had the opportunity to work with each of you.

I would especially like to thank my family and friends for their never ending encouragement. To my parents, Fred and Jane Schaefer, thank you for providing me the courage to pursue my dreams and the support to continue through to the end. A very special thank you is needed for my mom for her grammatical genius while editing multiple drafts of my work.

Finally, a deep, heart-felt thank you to Steven Rodman and David Parks for countless hours of listening to me talk through my thoughts, all your insights into partnership roles and tactics within the military domain, and especially for making me take a break every once in a while.

To each of you, I cannot express the depths of my gratitude enough. With that said, I will leave you with the words of the Robot from Lost in Space, “my micro mechanism thanks you, my computer tapes thank you, and I thank you.”

The research reported in this document was performed in connection with Contract Number

W911NF-10-2-0016 with the U. S. Army Research Laboratory. The views and conclusions contained in this document are those of the author and should not be interpreted as presenting the official policies or position, either expressed or implied, of the U. S. Army Research Laboratory, or the U. S. Government unless so designated by other authorized documents. Citation of manufacturer’s or trade names does not constitute an official endorsement or approval of the use thereof. The U. S. Government is authorized to reproduce and distribute reprints for Government purposes notwithstanding any copyright notation herein.

vii

TABLE OF CONTENTS

LIST OF FIGURES ...... xix

LIST OF TABLES ...... xxiii

LIST OF ACRONYMS ...... xxvii

CHAPTER ONE: INTRODUCTION ...... 1

Robots: A Step toward Definition...... 2

Fictional media...... 4

The history of robotics...... 6

Domain of use...... 8

Human-Robot Interaction ...... 10

Rationale for this Dissertation ...... 13

Statement of the Problem ...... 15

Overall Significance of this Dissertation ...... 16

Assumptions and Limitations ...... 16

CHAPTER TWO: REVIEW OF LITERATURE ...... 18

Trust Definitions ...... 19

The Triadic Model of Trust...... 21

Robot-related trust factors...... 24

viii

Robot Capabilities ...... 25

Robot features...... 26

Human-related trust factors...... 28

Human traits ...... 29

Human states...... 30

Environment-related trust factors...... 31

Environment: Team-related trust ...... 31

Environment: Task-related trust ...... 32

Trust Measurement ...... 33

Propensity to trust...... 34

Affect-based trust...... 35

Comfort...... 35

Satisfaction...... 35

Confidence...... 36

Attitudes...... 36

Cognition-based trust...... 36

Understanding how the robot works...... 37

Ability to use or interact with a robot...... 37

Expectancy...... 37

Trustworthiness...... 37

The advancement of human-robot trust measurement...... 38

Progression to Next Section ...... 41

CHAPTER THREE: STUDY 1 ...... 42

Experimental Method...... 43

Experimental participants...... 43

Included antecedents of trust...... 43

Experimental stimuli...... 43

Materials...... 44

Experimental Procedure...... 45

Experimental Results ...... 47

Human stimuli...... 47

Machine stimuli...... 48

Robot stimuli...... 49

Relationship between trust antecedents and trustworthiness...... 50

Comparison of robot and machine classification ratings...... 62

Distribution of the robot classification of robot stimuli...... 65

Discussion ...... 66

Progression to Next Experiment ...... 69

CHAPTER FOUR: STUDY 2 ...... 70

Experimental Method...... 71

Experimental Participants...... 71

Included antecedents of trust...... 71

Experimental Stimuli...... 71

Materials...... 72

Experimental Procedure...... 72

Experimental Results ...... 73

Relationship between robot classification and trustworthiness...... 73

Robot attributes ratings...... 75

Discussion ...... 78

Progression to Next Experiment ...... 80

CHAPTER FIVE: STUDY 3 ...... 81

Experimental Method...... 82

Participants...... 82

Prior experience...... 82

Robot mental model...... 83

Materials...... 84

Experimental procedure...... 85

Experimental Results ...... 85

Paired samples t–tests...... 85

Complete factor analysis...... 87

Impact of mental models on factor analyses...... 93

Human-like mental model analysis...... 93

Machine-like mental model analysis...... 94

Varied mental model analysis...... 94

Discussion ...... 94

Progression to Next Experiment ...... 96

CHAPTER SIX: STUDY 4 ...... 97

Experimental Method...... 97

Subject matter experts...... 97

Materials...... 98

Experimental procedure...... 98

Experimental Results ...... 99

Discussion ...... 105

Progression to Next Experiment ...... 106

CHAPTER SEVEN: STUDY 5 ...... 107

Experimental Method...... 108

xii

Participants...... 108

Trust measurement...... 108

Materials...... 109

Computer-based simulation...... 109

Experimental scenarios...... 110

Experimental procedure...... 111

Experimental Results ...... 111

Comparison of each trust scale item across time...... 111

Comparison of the trust scale across time...... 114

40 item trust scale versus 14 SME recommended item scale...... 117

Discussion ...... 120

Progression to Next Experiment ...... 121

CHAPTER EIGHT: STUDY 6 ...... 123

Experimental Method...... 124

Participants...... 124

Materials...... 124

Trust scale...... 124

Personality assessment...... 125

Human states...... 125

xiii

Virtual environment...... 125

Experimental task...... 126

Experimental procedure...... 127

Experimental Results ...... 128

Same-Trait analysis...... 128

Multi-Trait analysis...... 129

Discussion ...... 132

CHAPTER NINE: GENERAL DISCUSSION ...... 135

Development of the Trust Scale ...... 135

A Return to the Triadic Model of Trust ...... 136

Human-related antecedents of trust...... 137

Environment-related antecedents of trust...... 137

Robot-related antecedents of trust...... 138

General Conclusions ...... 138

Interpretation of the 40 item scale...... 139

Creating the trust score...... 141

14 item subscale...... 141

Future Work ...... 142

APPENDIX A: OPERATIONAL DEFINITIONS OF ROBOT DOMAINS...... 144

xiv

APPENDIX B: EMPIRICAL RESEARCH ON HUMAN-ROBOT TRUST ...... 146

APPENDIX C: HRI TRUST DEFINITIONS...... 149

APPENDIX D: PLATO’S ALLEGORY OF THE CAVE ...... 154

APPENDIX E: QUESTIONNAIRES AND SCALES ...... 158

Mini-IPIP ...... 159

Negative Attitudes toward Robots Scale ...... 160

Interpersonal Trust Scale ...... 161

Godspeed Questionnaire ...... 162

Dundee Stress State Questionnaire ...... 163

Checklist for Trust between People and Automation ...... 170

APPENDIX F: STUDY 1 INFORMED CONSENT ...... 171

APPENDIX G: STUDY 1 STIMULI ...... 173

Robot Stimuli ...... 174

Machine Stimuli ...... 178

Human Stimuli ...... 179

APPENDIX H: STUDY 1 ADDITIONAL ANALYSES OF MACHINE IMAGES...... 180

Distribution of the Robot Classification Ratings for Machine Stimuli...... 181

Individual Difference Analyses of the Baggage Screener and X-ray Machine ...... 183

MRC analysis of the baggage screener...... 184

MRC analysis of the x-ray machine...... 185

APPENDIX I: STUDY 1 ADDITIONAL ANALYSES OF ROBOT IMAGES ...... 186

Robot Classification t-tests ...... 187

Robot Stimuli Stem and Leaf Plots ...... 190

Individual Differences Analyses for Robot Stimuli...... 198

MRC analysis of the gutter robot...... 199

MRC analysis of the reaper...... 201

MRC analysis of the laparoscopic robot...... 203

MRC analysis of Furby ...... 205

MRC analysis of the ballroom robot...... 207

MRC analysis of the lawnmower robot...... 209

MRC analysis of Roomba...... 211

MRC analysis of the TUGV robot...... 213

MRC analysis of the IED Detonator robot...... 215

MRC analysis of the DaVinci robot...... 217

MRC analysis of the RP6 medical robot...... 219

MRC analysis of the iFoot robot...... 221

MRC analysis of the metal industrial robot...... 223

MRC analysis of the glass industrial robot...... 225

xvi

APPENDIX J: STUDY 2 INFORMED CONSENT ...... 227

APPENDIX K: STUDY 2 ADDITIONAL ANALYSES ...... 229

Pearson Correlations for Robot Attributes ...... 230

Percentages of Attributes for Fuzzy Boundary Robots...... 239

APPENDIX L: STUDY 3 INFORMED CONSENT...... 243

APPENDIX M: STUDY 3 DEMOGRAPHICS QUESTIONNAIRE ...... 245

APPENDIX N: STUDY 3 ADDITIONAL ANALYSES ...... 248

Item Analysis ...... 249

Additional Analyses: Item Normality ...... 254

Additional Analyses: Principal Component Analyses based on Mental Models ...... 258

APPENDIX O: STUDY 4 SUBJECT MATTER EXPERT QUESTIONNAIRE ...... 267

APPENDIX P: STUDY 4 TRUST SCALE 74 ITEMS ...... 271

APPENDIX Q: STUDY 5 INFORMED CONSENT ...... 275

APPENDIX R: STUDY 5 PRE/POST TRUST SCALE ITEMS ...... 278

APPENDIX S: STUDY 6 INFORMED CONSENT ...... 282

APPENDIX T: STUDY 6 SIMULATION HANDOUTS ...... 285

APPENDIX U: STUDY 6 ADDITIONAL ANALYSES ...... 288

Human States Analysis ...... 289

Trust Antecedents Means and Standard Deviations ...... 291

xvii

APPENDIX V: COPYRIGHT PERMISSIONS ...... 292

APPENDIX W: REFERENCES FOR PREVIOUS TRUST SCALES ...... 294

Human-Robot Trust ...... 295

Interpersonal Trust Most widely cited ...... 298

Interpersonal Trust Others cited...... 299

Automation Trust ...... 301

REFERENCES ...... 311

xviii

LIST OF FIGURES

Figure 1. A graphical representation of the impact of fictional media on robot design...... 4

Figure 2. Fictional media that has provided inspiration for real-world robotics...... 5

Figure 3. Čapek’s play, showing the “robots” in rebellion in the play itself...... 6

Figure 4. Key historical achievements in the evolution of a robot...... 7

Figure 5. A timeline of the beginning of the robotic domains...... 9

Figure 6. A comparative representation of the transition from tool to team member between human-animal interaction and human-robot interaction...... 11

Figure 7. The relationship among calibration, resolution, and automation capability in deﬁning appropriate trust (Lee & See, 2004)...... 14

Figure 8. Variations in trust definitions...... 20

Figure 9. Human-robot three factor trust model...... 22

Figure 10. Updated descriptive model of human-robot trust...... 24

Figure 11. Examples of real-world robots across seven different robotic domains...... 25

Figure 12. Examples of present day HRI within high-risk environments...... 29

Figure 13. Visual description of how previous trust scales are applicable to HRI...... 34

Figure 14. UML predictive model of human-robot trust measurement over time...... 40

Figure 15. Example of Study 1 survey...... 46

Figure 16. Example of object being present in human stimulus...... 47

xix

Figure 17. Representation of the robot classification ratings for each of the machine images.

Means and 95% confidence intervals demonstrated that all machine images were rated below the neutral point (4) on the robot classification scale...... 49

Figure 18. Representation of the robot classification ratings for each of the entertainment robot images. Means and 95% confidence intervals demonstrated that all entertainment robot images were rated above the neutral point (4) on the robot classification scale...... 53

Figure 19. Representation of the robot classification ratings for each of the industry robot images. Means and 95% confidence intervals demonstrated that six out of the seven industry robot images were rated above the neutral point (4) on the robot classification scale...... 54

Figure 20. Representation of the robot classification ratings for each of the medical robot images. Means and 95% confidence intervals demonstrated that RP6, Heart robot, UBot, and

Lifting robot were rated above the neutral point (4) on the robot classification scale...... 56

Figure 21. Representation of the robot classification ratings for each of the military robot images. Means and 95% confidence intervals demonstrated that all robots were rated above the neutral point (4) on the robot classification scale, except for the Reaper...... 57

Figure 22. Representation of the robot classification ratings for each of the service robot images.

Means and 95% confidence intervals demonstrated that five of the robots were rated above the neutral point (4) on the robot classification scale, while the gutter robot and Roomba were rated as neutral...... 59

Figure 23. Representation of the robot classification ratings for each of the social robot images.

Means and 95% confidence intervals demonstrated that six of the robots were rated above the neutral point (4) on the robot classification scale...... 60

Figure 24. Representation of the robot classification ratings for each of the therapy robot images.

Means and 95% confidence intervals demonstrated that five of the robots were rated above the neutral point (4) on the robot classification scale...... 62

Figure 25. Robot stimuli with higher classification ratings of a robot than a machine (ordered from most to least)...... 64

Figure 26. Robot stimuli with equal robot and machine classification ratings...... 64

Figure 27. Robot stimuli that were rated low on the robot classification scale...... 67

Figure 28. Robot stimuli that were rated neutral on the robot classification scale...... 67

Figure 29. Example of a semantic differential scale item from the Godspeed Questionnaire ..... 72

Figure 30. Example questions included in the initial item pool...... 85

Figure 31. Graphical representation of trust scores for the 40 item scale and the 14 item scale across time...... 118

Figure 32. Graphical representation of trust scores for the 40 item scale and the 14 item scale for the Change in Trust Scores over Time...... 120

Figure 33. Talon robot...... 127

Figure 34. Interactions between trust scales and conditions...... 131

Figure 35. Experiment 1 Industry robot stimuli ...... 174

Figure 36. Experiment 1 Entertainment Robot Stimuli ...... 174

Figure 37. Experiment 1 Medical robot stimuli ...... 175

Figure 38. Experiment 1 Military robot stimuli ...... 175

Figure 39. Experiment 1 Service robot stimuli ...... 176

Figure 40. Experiment 1 Social robot stimuli ...... 176

xxi

Figure 41. Experiment 1 Therapy robot stimuli ...... 177

Figure 42. Experiment 1 Machine stimuli ...... 178

Figure 43. Experiment 1 Human stimuli ...... 179

Figure 44. Percentages of Scores on the Robot Scale ranging from ‘Not at All’ (1) to

‘Completely’ (7)...... 182

Figure 45. X-ray machine and baggage screener stimuli...... 183

Figure 46. Study 6 Map A...... 286

Figure 47. Study 6 Map B...... 287

xxii

LIST OF TABLES

Table 1 HRI and Automation Trust Literature Supporting Trust Antecedents Specific to the

Capabilities of the Robot ...... 26

Table 2 HRI and Automation Trust Literature Supporting Trust Antecedents Specific to Robot

Features ...... 28

Table 3 Means and Standard Deviations for each Machine Stimuli on the Robot Classification

Scale ...... 48

Table 4 Correlations among Trust Antecedents and Trustworthiness for Robot Domains ...... 50

Table 5 Entertainment Robots Means, Standard Deviations, and Correlations of Robot

Classification Ratings ...... 52

Table 6 Industry Robots Means, Standard Deviations, and Correlations of Robot Classification

Ratings ...... 54

Table 7 Medical Robots Means, Standard Deviations, and Correlations of Robot Classification

Ratings ...... 55

Table 8 Military Robots Means, Standard Deviations, and Correlations of Robot Classification

Ratings ...... 57

Table 9 Service Robots Means, Standard Deviations, and Correlations of Robot Classification

Ratings ...... 58

Table 10 Social Robots Means, Standard Deviations, and Correlations of Robot Classification

Ratings ...... 60

xxiii

Table 11 Therapy Robots Means, Standard Deviations, and Correlations of Robot Classification

Ratings ...... 61

Table 12 Comparison of Robot and Machine Classification Ratings per Robot Domain ...... 63

Table 13 Frequency Ratings on Robot Scale for Robot Stimuli ...... 65

Table 14 Relationship between Robot Classification Ratings and Trustworthiness for each Robot

Stimuli ...... 73

Table 15 Paired Samples t-tests comparing Attribute Ratings between the Neutral

Classification Robots compared to the High Classification Robots ...... 75

Table 16 Significant Correlations Between Robot Attributes and Robot Classification Ratings 76

Table 17 Number of Positive Trustworthiness Correlations for the Anthropomorphism,

Animacy, Likeability, and Perceived Intelligence Attributes ...... 79

Table 18 Trust Antecedents for Scale Development ...... 81

Table 19 Participants Prior Experiences with Robots ...... 83

Table 20 Robot Mental Model ...... 84

Table 21 Factor Analysis Rotated Components...... 88

Table 22 Impact of Mental Models on Factor Analyses ...... 93

Table 23 Subject Matter Experts Years of Experience ...... 98

Table 24 CVR Values for the 14 Items Recommended by SMEs ...... 100

Table 25 CVR Values and Hypothetical Ranges for the 37 Important Items Recommended by

SMEs ...... 102

Table 26 Domain Specific Items ...... 104

Table 27 Hypothetical Ranges for the SME Recommended Items to Remove from Scale ...... 105

xxiv

Table 28 Means, Standard Deviations, and Confidence Intervals for each Trust Item at Time 1,

Time 2, and Time 3...... 112

Table 29 Experiment 5 Means, Standard Deviations, and Correlations ...... 115

Table 30 Experiment 5 Repeated Measure Variable of Time ...... 116

Table 31 Means, Standard Deviations and Confidence Intervals Comparing the 40 Item Scale to the 14 Item Scale ...... 117

Table 32 Means, Standard Deviations and Confidence Intervals Comparing Change Scores for the 40 Item Scale to the 14 Item Scale ...... 119

Table 33 Same Trait Trust Scale Correlations over Time ...... 129

Table 34 Means, Standard Error, and Confidence Intervals for the Three Trust Scales ...... 131

Table 35 Finalized Trust Scale ...... 139

Table 36 Operational Definitions of Robot Domains ...... 145

Table 37 Human-Robot Empirical Trust Studies included for Meta-Analysis ...... 147

Table 38 Trust Definitions ...... 150

Table 39 Frequency Ratings on Robot Classification Ratings for each Machine Stimuli ...... 181

Table 40 Baggage Screener Means, Standard Deviations, and Intercorrelations ...... 184

Table 41 X-ray Machine Means, Standard Deviations, and Intercorrelations ...... 185

Table 42 Classification of Robot Stimuli ...... 187

Table 43 Percentage of Robot Classification Variance Explained ...... 198

Table 44 Gutter Robot Means, Standard Deviations, and Intercorrelations ...... 200

Table 45 Reaper Robot Means, Standard Deviations, and Intercorrelations ...... 202

Table 46 Laparoscopic Robot Means, Standard Deviations, and Intercorrelations ...... 204

xxv

Table 47 Furby Means, Standard Deviations, and Intercorrelations ...... 206

Table 48 Ballroom Robot Means, Standard Deviations, and Intercorrelations ...... 208

Table 49 Lawnmower Robot Means, Standard Deviations, and Intercorrelations ...... 210

Table 50 Roomba Robot Means, Standard Deviations, and Intercorrelations...... 212

Table 51 TUGV Robot Means, Standard Deviations, and Intercorrelations ...... 214

Table 52 IED Detonator Robot Means, Standard Deviations, and Intercorrelations ...... 216

Table 53 DaVinci Robot Means, Standard Deviations, and Intercorrelations...... 218

Table 54 RP6 Robot Means, Standard Deviations, and Intercorrelations ...... 220

Table 55 iFoot Robot Means, Standard Deviations, and Intercorrelations ...... 222

Table 56 Metal Robot Means, Standard Deviations, and Intercorrelations...... 224

Table 57 Glass Robot Means, Standard Deviations, and Intercorrelations ...... 226

Table 58 Study 2 Correlation Analysis for Robot Attributes ...... 230

Table 59 Average Percentage of the Anthropomorphic Attributes for Each Robot ...... 241

Table 60 Average Percentage of the Animacy Attributes for Each Robot ...... 241

Table 61 Average Percentage of the Likeability Attribute for Each Robot ...... 242

Table 62 Average Percentage of the Perceived Intelligence Attribute for Each Robot ...... 242

Table 63 Study 3 Results of Item Analysis ...... 249

Table 64 Study 3 Normality ...... 254

Table 65 Rotated Component Matrices across Mental Model Classifications...... 258

Table 66 Means, Standard Deviations, and Confidence Intervals for Human States for each

Condition...... 289

Table 67 Study 6 Trust Antecedents Means and Standard Deviations ...... 291

xxvi

LIST OF ACRONYMS

ARL Army Research Laboratory

CVR Content Validity Ratio

DOD Department of Defense

GDRS General Dynamics Robotic Systems

HRI Human Robot Interaction

IRB Institutional Review Board

LOA Level of Automation

ONR Office of Naval Research

PI Perceived Intelligence

RIVET Robotic Interactive Visualization & Exploration Technology

SDK Software Development Kit

SME Subject Matter Expert(s)

VE Virtual Environment

xxvii

CHAPTER ONE: INTRODUCTION

“If every tool, when ordered, or even of its own accord, could do the work that befits it … then there would be no need either of apprentices for the master workers or of slaves for the lords.” Aristotle, 322 B.C.

Once the province of mythology and more recently science fiction, functional and productive robots have begun to permeate modern-day society. In the past two decades especially, we have seen a rapid influx of robotics into many everyday social environments.

These locations include the home, the workplace, the battlefield as well as a number of others dimensions. Technological advancements provide the enhanced capabilities and autonomy that have allowed the robot to evolve from their primarily passive tool-based role to that of an interdependent social actor. As the vector of robot development continues to move towards that of an integrated team member, the issue of trust is taking on an ever more prominent role. One of the most significant challenges for successful collaboration between humans and robots is the development of appropriate levels of mutual trust in robots (Desai, Stubbs, Steinfeld, & Yanco,

2009; Groom & Nass, 2007). Regardless of the domain of application, the environment, or the task, a human’s trust in their non-human collaborator is an essential element required to ensure that any functional relationship will ultimately be effective. To begin the present work however, we first need to address the question as to how to define a robot and how this description itself changes.

Robots: A Step toward Definition

Robots emanate from the field of automation. Despite the major advancements that have been made over the years, researchers have yet to agree on a clear and universally accepted definition of a robot. Even in relevant research domains, robot classifications range from

“programmable automation” (Mahoney, 1997) to definitions based on manifestations of human characteristics, such as complex actions and anthropomorphic features (Kurfess, 2005).

Automation has been defined as “the execution by a machine agent (usually a computer) of a function that was previously carried out by a human” (Parasuraman & Riley, 1997, pg. 231).

Traditional automation most frequently performs one type of task, usually in a well-structured and controlled environment. Examples include assistance as cognitive aids (e.g., airport baggage screener), perceptual aids (e.g., collision avoidance systems in automobiles), and control aids

(e.g., adaptive cruise control for automobiles, or navigation aids). Further, a human operator must be present, even if only in a supervisory capacity.

Roboticists have been more recently attempting to extend the classification of a robot beyond that of traditional automation to include the importance of cognition and intelligence.

Researchers in the area of social robotics have extended Mahoney’s definition to include the postulate that a robot is an autonomous system that acts on its own decisions with input from a human operator, but may not be completely controlled by a human (e.g., Feil-Seifer & Matarić,

2005). Scholtz (2003) has suggested that the type of interaction, the number of systems a user may interact with, the environment, and the physical nature of the robot (e.g., complex, dynamic control systems; exhibition of autonomy and cognition) help differentiate human-robot

interaction from more general human-computer or human-machine interaction. More recently,

Yagoda (2011) defined a robot as the interaction of intelligence and autonomy of the technology.

She argued that based on this interaction, robots are distinctly different from machines, such as a refrigerator (low intelligence, high autonomy), an expert system (high intelligence, low autonomy), or even a mechanized robotic arm (low intelligence, low autonomy).

Other research-based definitions further suggest the importance of the environment as a factor in the classification of a robot. Fong and colleagues (2001) have suggested that a robot is a complex, dynamic system that shows a degree of autonomy and cognition as it operates in a real- world environment. A robot perceives the environment through sensor inputs that may or may not require direct human response, and has the ability to move in order to manipulate an uncertain physical world by transmitting information to humans or directly acting on the environment (Groom, 2008; Lin, Bekey, & Abner, 2008). This allows robots to be deployed in unstructured environments with task-oriented goals. The Department of Defense (DOD), Office of Naval Research (ONR) has suggested that a robot is typically re-programmable, flexible and often mobile (see also, Lin, Bekey, & Abney, 2008).

To account for the variations in definitions, this present work acknowledges that both physical form and functional capabilities are essential to the classification of a machine as a robot. Individual perceptions can impact how people define a machine to be a robot, which further illuminate the fuzzy boundaries that occur in the classification process (see Schaefer,

Billings, & Hancock, 2011). A number of factors affect these individual perceptions including fictional media, historical precedent, human interaction roles, and domains of specific use or needs.

Fictional media. Fictional media (e.g., literature, art, movies, etc.) had commonly been used as a foundation for the design and development of real-world robotics. Figure 1 (adapted from Schaefer, Cook, Adams, Bell, Sanders, & Hancock, 2012) provides a representation of how fictional media has provided a medium to extend robot capabilities and design without technological restraint. This leads to idea generation, exploratory research, and innovation and development of additional capabilities or design elements. Schaefer and colleagues (2012) further show that fictional media also influences the perception and mental models of the general population, which can impact how individuals classify real-world machines as robots. However, the mismatch between fiction and real-world robotics continues to actually aid in the in the inhibition of the robot definition.

Figure 1. A graphical representation of the impact of fictional media on robot design.

Iconic fictional robots also brought robotics to the attention of general society through science fiction and visual media (see also Schaefer, Adams, Cook, Bardwell-Owens, & Hancock,

2013). Some of the most recognized fictional robots include Robby the RobotTM, from

Forbidden Planet (Nayflack & Wilcox, 1956) and Lost in Space (Allen, 1965; Koch & Hopkins,

1998); Rosie the RobotTM from The Jetsons (Couch & Gabai, 1962 ); R2D2TM from Star Wars

(Kurtz & Lucas, 1977); and Lt. Commander DataTM from Star Trek: The Next Generation

(Roddenberry, 1987). Not only did media bring these robots into the home, they served as inspirations for the design capabilities and development of real-world robotics (Figure 2).

Figure 2. Fictional media that has provided inspiration for real-world robotics.

The history of robotics. The term robot, as we know it today, was popularized in 1921 in Karl Čapek’s play, Rossum’s Universal Robots, R.U.R. (see Figure 3). It is important to note however that the term actually derived from robota, a term employed in the Austro-Hungarian

Empire for vassal or worker.

Figure 3. Čapek’s play, showing the “robots” in rebellion in the play itself.

However, the foundation for robotics development may date back to as early as the 3rd

Century, B.C.E. in both myth and reality. The first possible instantiation of a robot may stem from the Greek myth of Talos, a bronze man who protected Crete from pirates. Around the same time the Greek mathematician, Archytas of Tarentum, built a mechanical bird propelled by steam. A review of the historical development can help us identify common design needs such as movement, realism, capabilities, and functionality (see Figure 4; Schaefer, Adams, Cook,

Bardwell-Owens, & Hancock, 2013).

rd 3 Century, B.C.E. Greek mathematician, Archytas of Tarentum used steam power to build a mechanical bird called “the Pigeon.”

1495 Leonardo DaVinci used mechanisms inside armor to build a knight that could move

1738 Jacques de Vaucanson modeled animal autonomy to build “the duck” that could flap its wings, move, and eat 1770 Swiss clock makers Pierre and Henri-Louis Jaquet-Droz built three dolls that could write, play music, and draw respectively

1898 Nikola Tesla built the first remote controlled boat

1959 John McCarthy and Marvin Minsky founded the Artificial Intelligence Laboratory at the Massachusetts Institute of Technology 1961 George Devol built the first industrial robotic arm for the General Motors assembly line 1966 Stanford Research Institute built Shakey, the first mobile robot to know and react to its own actions

1970/1979 Stanford University developed the “Stanford Cart” that could follow a line or be controlled from a computer. Rebuilt by Hans Moravec to include a robust vision system 1986 Honda begins robotics research that focuses on robots that coexist and cooperate with humans

Figure 4. Key historical achievements in the evolution of a robot.

For example, the importance of movement to robotic design can been seen in Leonardo da Vinci’s “Knight,” George Devol’s robotic arm, and Stanford Research Institute’s “Shakey.”

Control systems are an additional functional element of importance in robotics. Nikola Tesla’s remote controlled boat may have provided inspiration for some current robotic hand-held teleoperation control systems; while Stanford University’s “Stanford Cart” demonstrated the importance of computer technology for the purpose of programming robot control. While history provides a foundation for understanding the functional capabilities of a robot, there is a noted transition in the utility of the robot from historical entertainment into the present, more practical uses.

Domain of use. An additional way in which people differentiate robots is through their domain of use. Here, ‘domain’ refers to the nine main areas in which robots are designed. Libin and Libin (2004) have suggested that Industrial, Military, Research, Medical and Service robots are designed to assist humans by satisfying needs, assisting daily living, and acting in risky or dangerous situations. Social, Education, Entertainment, and Rehabilitation/Therapeutic robots provide interactive stimulation. A listing of operational definitions of these specific areas of robotics is provided in Appendix A. It is important to note that even within these specific robotic domains, there are inconsistencies within the operational definitions. Figure 5 demonstrates a timeline of the first robotics for each domain.

Figure 5. A timeline of the beginning of the robotic domains.

Human-Robot Interaction. The human element of human-robot interaction (HRI) also may affect the classification process. According to Scholtz (2003) there are five different types of human roles within HRI that vary in their level of knowledge and control in terms of goals, intentions, actions, perception and evaluation.

 Supervision interaction: Human monitors robot(s) action and controls overall situation/plan.

 Operator interaction: Human operator determines if robot actions are being carried out correctly and if the actions are in accordance with the longer term goal. They can correct inappropriate behaviors.

 Mechanic interaction: The mechanic deals with physical interventions in the hardware/software.

 Peer interaction: This is more of a communication-based interaction. Teammates of robots can give commands within larger goals/intentions.

 Bystander: The bystander has no information about the goals or intentions of the robot and a small subset of actions available.

Human-animal interaction can be used as a basis for understanding the human-robot roles within HRI (see Billings et al., 2012a; Phillips, Ososky, Swigert & Jentsch, 2012). For example, in the tool-based role, canaries were used as an early warning detection system for poisonous gases (BBC News, 1986); while law enforcement dogs, military working dogs (MWDs), and animals used in search-and-rescue (SAR) environments all exemplify human-animal teams that must work in and adapt to difficult and dangerous environments (Finkel, 2012; Helton, 2009).

Figure 6 represents a comparison between human-animal teams and human-robot teams.

Humans & Robots Humans & Animals

TOOL

Replace / Enhance Human Capabilities

Augment Human Capabilities

Provide Comfort

TEAM MEMBER

Figure 6. A comparative representation of the transition from tool to team member between human-animal interaction and human-robot interaction.

At the tool-based role, robots are often designed to accomplish a specific task, while the human takes an operator role. The reliability of the robot’s functional capability is essential for trust development at this level, as failure may lead to increased external risk to the human population within the area. At the next level, robots are designed to replace or enhance human capabilities. This type of robot is often designed for and used within the manufacturing domain due to the specific capabilities for accuracy and precision of product development, as well as increased safety of human workers (e.g., lifting or moving heavy objects, repetitive motions, etc.). The human takes a supervisory role, and trust is related to dependability and efficiency.

At the third level, robots are designed to augment human capabilities (e.g., robotic surgical systems). Elements of a partnership begin to emerge. Trust becomes essential at this level since the robot is extending the human operator’s capabilities in one way or another. In addition, there is often a trust relationship with a bystander (e.g., surgical patient, assisting nurses, etc.). Communication and feedback, as well as the previously discussed reliability and dependability of the robot, are very important to the trust relationship at this level of HRI.

As robot development continues to move closer to a team member role within HRI, the role of the human begins to transition to bystander and/or peer interaction. This transition is more directly visible at the fourth level in which the robot is providing emotional support, such as comfort to the human. Social and therapy robots are often included in this level of HRI. Often robots are designed with zoomorphic or anthropomorphic traits and behaviors to reach this goal, as well as instill trust. Both the physical appearance and functional capabilities are essential to trust development.

At this time it can be argued that robot development has not yet reached the level of a team member, however human-animal interaction has provided a structure and insight into how to reach this goal. As it has been shown, robotics tends to emulate physical (e.g., mobility characteristics) and behavioral aspects of animals to enhance design. In addition, functions of human-robot collaborations often parallel the varying levels of functionality of human-animal partnerships, ranging from tool to integrated team member. Thus, examining the nature of human-animal relationships can increase our understanding of how humans may interact with and trust certain technology, such as robots (see also Billings et al., 2012; Schaefer et al., 2012).

Rationale for this Dissertation

Trust involves at least two interacting members. One is the trustor and the other is the trustee. Specific to HRI, this relationship thus includes at minimum one person (trustor) and one robot (trustee). Most often the person of interest within HRI is the robot operator or user since he/she is the one directly working with the robot on a collaborative task. Throughout this work the trustor will be referred to as the human operator. However, other interactors might include other human teammates, a supervisor or command personnel, and even bystanders in an environment.

This trust relationship becomes critical as robots shed their more passive tool-based roles and move more towards being an active integrated team member (see Chen & Terrence, 2009).

Despite the obvious need to investigate the trust aspect of the interaction in the human-machine dyad, the primary focus in the world of robotics is currently placed on the design and capabilities of the technology itself. The human element has often been under-represented, overlooked or

even forgotten. Problems will inevitably emerge if the human component of the dyad is neglected in the design and implementation of the technology factor. Chief among these concerns is the question of operator trust.

Figure 7. The relationship among calibration, resolution, and automation capability in deﬁning appropriate trust (Lee & See, 2004).

A misplaced over-reliance (or under-reliance) on behalf of the human operator can lead to misuse, or inappropriate use, of a robot. In contrast, lack of trust in human-robot relationships can also lead to disuse, in which an effective technology sits on a shelf collecting dust (e.g., the

SWORD robot, Special Weapons Observation Reconnaissance Detection; Ogreten, Lackey, &

Nicholson, 2010). For a complete review of use, misuse, disuse and abuse as it applies to automation, see Parasuraman and Riley (1997) and Lee (2008). Lee and See (2004) discuss the

importance of trust calibration (i.e., trust that matches the system’s capabilities) as a means to approach appropriate use of a system (see Figure 7). Until trust between a human and a robot is solidly established, robotic partners will continue to be underutilized or unused, therefore providing little to no opportunity for trust to develop (Lussier, Gallien, & Guiochet, 2007).

Statement of the Problem

Human-robot trust is a relatively new field of study. Even though past work (see

Sheridan, 1975) has emphasized the importance of trust within human-robot interaction (HRI), the earliest empirical research did not occur until the latter half of the 1990s (see Tenney,

Rogers, & Pew, 1998; Dassonville, Jolly, & Desodt, 1996). Further, a majority of the empirical work on trust in HRI has actually occurred in the latter half of the most recent decade. To date, only one human-robot trust meta-analysis (see Hancock et al., 2011) is available. This work documented the limited range of empirical research, being comprised of only a total of ten correlational studies and twelve experimental studies (see Table 37 in Appendix B).

At this time, subjective measurements of human-robot trust are acquired through adapted human-interpersonal and human-automation trust scales; this with one exception specific to human-robot trust within the military domain (see Yagoda, 2011). Consequently, identification of the antecedents that influence trust development within HRI must be more thoroughly identified. A reliable and valid subjective scale is needed to specifically measure human-robot trust that can be applicable to multiple robotic domains.

Overall Significance of this Dissertation

To address the present limitations of measuring human-robot trust, the goal of this dissertation is to produce a reliable and valid measure which captures changes in a human’s trust in a robot. There are three major components necessary to meet this end:

i) determining the underlying theory of trust as it applies to human-robot relationships, ii) classification of the current perception of a robot as it applies to trust, and iii) development of the trust scale.

Findings from this study can then serve to guide future work in the identification of specific robot design metrics. In turn, these changes have the potential to lead to lower cost and increased use.

Assumptions and Limitations

With respect to field specific HRI, previous trust research is limited. Therefore, it is necessary here to generate initial assumptions regarding the factors that influence trust. Research in the area of human-robot trust is limited both in terms of robot and trust. As such, trust will be explored via theoretical and empirical work across a number of domains of trust rather than purely on its definitional foundation.

Participants were recruited through the University of Central Florida’s SONA Systems undergraduate student subject pool for five out of the six included studies. To address the limitation of using undergraduate students, an experience questionnaire was developed and implemented in Studies 3, 5, and 6. Findings from the experience questionnaire showed that the

sample represented a novice HRI population, whose primary interaction with robots was through television-based observation. Specifics are reported within each study.

CHAPTER TWO: REVIEW OF LITERATURE

“There’s a lot of space out there to get lost in.” John Robinson, Lost in Space (Koch & Hopkins, 1998)

Research focusing on the importance of trust has had a long history. There are currently over 700 published articles focusing on trust as the primary topic of the research, and many more referencing the importance of trust. However, there is still “a lot of space out there to get lost in.”

While John Robinson, a character in the movie Lost in Space (Koch & Hopkins, 1998), was talking about outer space, the sentiment still applies here. There is still much to learn about trust relationships, especially in the context of human-robot interaction (HRI).

Since human-robot trust is a relatively new field of study with limited research, the more extensive fields of human-interpersonal trust and human-automation trust can provide valuable insight into the area. With well over 500 published articles on the topic of interpersonal trust, this area has the most extensive background in the field, and continues to explore the intricacies of trust today. A number of trust-related topics can be applied to the technological domain. These include, but are not limited to benefits of team structures, measurement techniques, and the importance of task-dependent and discipline specific factors. The more recent field of human- automation trust also provides additional support for how the technological and functional capabilities of robotics affect trust development.

However, there is some debate as to whether or not interpersonal trust research can or should be applied to the technological domain. The most common reason is the scope of the relationship. Some researchers suggest that the type of interaction between humans is very different than that of humans and robots. For example, Lee and See (2004) suggest that individuals trust and respond to technology different than people. This sentiment is supported by the empirical work by Lewandowsky, Mundy, and Tan (2000), which demonstrated that individuals delegate control differently to humans compared to automation. In addition, the degree of trust was more strongly related to the decision to delegate to the automation

(Lewandowsky et al.), further supporting the claims of Lee and See (2004). Not all researchers have such strong views. Madhavan and Wiegmann (2007) suggest that individuals may enter trust relationships with technology (e.g., computers, machines, robots) similarly to human trust relationships; yet, a divergence occurs due to the characteristics of the partner and the decision- making process. Thompson and Gillian (in Barnes & Jentsch, 2010, p. 78) suggest that drivers of interpersonal trust may also drive human-robot trust. Despite this debate, there is still much to be learned from these other areas of trust research.

Trust Definitions

Even though there is a long history, questions as to “what is trust?” still persist today.

Adams and colleagues (2003) suggest that no consensus definition currently exists within cognitive science or indeed across other areas. To date, there are over 300 documented definitions of trust across the various research domains with 32 trust definitions related to the

HRI domain alone (see Appendix C). These domains include human interpersonal trust (see

Mayer, Davis, & Schoorman, 1995; Rotter, 1971; Rempel, Holmes, & Zanna, 1985), human- automation trust (see Lee & See, 2004; Madhavan, & Wiegmann, 2007; Moray, Inagaki, & Itoh,

2000), and trust in software agents (see Patrick, 2002), to name only a few.

While it is difficult to assess human-robot trust from this diverse definitional foundation alone, such definitions can provide a number of common characteristics that are important to understanding trust across domains. In their review of 220 definitions for interpersonal trust and

82 definitions for technology-based trust, Billings, Schaefer, Llorens, and Hancock (2012b) found that a large number of definitions refer specifically to expectations, confidence, risk or uncertainty, reliance, and vulnerability. Figure 8 (adapted from Billings et al., 2012b) represents a visual representation of the component elements of the trust definitions.

30 Interpersonal Trust 25

20 Technology- based Trust 15

5 # of definitions including including item # definitions of 0

Items included in trust definitions

Figure 8. Variations in trust definitions.

In addition, the definitions of trust across the different types of interactions

(interpersonal, human-automation, and human-robot) provide additional insight into the important characteristics of the trust relationship. For example, Mayer, Davis, and Schoorman

(1995) define human-interpersonal trust as “the willingness of a party to be vulnerable to the outcomes of another party based on the expectation that the other will perform a particular action important to the trustor, irrespective of the ability to monitor or control that other party.” This definition addresses the importance of vulnerability and expectation specific to the actions of both the trustee and trustor. Similarly, Lee and See (2004) suggest that uncertainty and vulnerability are also important to the definition of trust in automation such that trust is an

“attitude that an agent will help achieve an individual’s goals in a situation characterized by uncertainty and vulnerability.” However Lee and See’s definition of trust highlights the importance of uncertainty and vulnerability as it relates to a situation, not specifically the actions of the trustor or the autonomous agent. Within HRI, Hancock, Billings, and Schaefer (2011) define trust as “the reliance by an agent that actions prejudicial to their well-being will not be undertaken by influential others;” thus suggesting that trust can involve other objects beyond sentient organisms that do not express an intrinsic, self-determined intention.

The Triadic Model of Trust

To address these variations in trust definitions, the present work used an updated model of human-robot trust, initially developed by Hancock and colleagues (2011), as the foundation for understanding trust within the present robotic domain. This triadic trust model featured the

human, the robot, and the environment as centers of interest, as well as the interaction between these three respective factors. For an in-depth literature review of these factors and the theoretical relevance of the antecedents of trust, see Sanders, Oleson, Billings, Chen, and

Hancock (2011).

*Note. Theoretical and empirical studies are included in the model with * representing experimental and + representing correlational analyses.

Figure 9. Human-robot three factor trust model.

Through meta-analytic procedures, Hancock and colleagues (2011a) demonstrated the primary importance of robot performance consistency (e.g., reliability), as well as team collaboration and tasking. This work also identified under researched (e.g., human characteristics) and theoretically supported antecedents of trust. However, caution should be used when interpreting the resulting model of human-robot trust (Figure 9). The descriptive model may not be complete due to the dearth of current experimentation which specifically relates to human-robot trust. As a consequence of this lack of experimentation, research in the areas of human-interpersonal trust and trust in automation was thus assessed to provide additional support and adapt the human-robot trust model. For an in-depth review and meta- analytic findings specific to human-interpersonal trust and human-automation trust see Hancock,

Billings, Schaefer, and Hancock (2013) and Schaefer et al. (2013) respectively. This resulted in the elaborated descriptive model of human-robot trust illustrated in Figure 10.

Figure 10. Updated descriptive model of human-robot trust.

Robot-related trust factors. Trust in the robot itself is the most well researched area of human-robot trust. Hancock and colleagues (2011) suggest that antecedents of trust can be separated into robot capabilities or performance-based (e.g., behavior, reliability) characteristics, and robot features or attribute-based characteristics (e.g., anthropomorphism). Duffy (2003) suggests that both the form and function of the robot are necessary to facilitate appropriate social interaction with people. In addition, robot typology is suggested to have a direct effect on trust

(Evers, Maldanado, Brodecki, & Hinds, 2008; Ross 2008). Figure 11 represents a variety of robots across multiple domains.

Figure 11. Examples of real-world robots across seven different robotic domains.

Minor changes were made in the updated model per automation trust literature to provide additional clarity regarding communication with a robot. In a meta-analysis of human- automation trust, Schaefer and colleagues (2013) found that features of the automation, including

LOA and mode of communication ( ̅= +.35), as well as communication feedback ( ̅= +.45), were shown to have a small to moderate effect on trust development.

Robot Capabilities. As stated previously, the primary focus of design and development of robots is on enhancing the functional capabilities to meet a specific need. Due to the longer history of work in this area, the importance of functional capabilities on trust development is also one of the most well researched areas of human-robot trust. Antecedents of trust associated with robot capabilities hereby include robot behavior, reliability/errors (e.g., quality and accuracy), and feedback or cueing from the robot. Similar findings are also demonstrated in the automation field, providing support for the antecedents of trust described in Table 1. For example,

predictable and dependable capabilities of the robot (e.g., behavior) can affect the overall trust in a system (Biros, Daly, & Gunsch, 2004). Automation trust literature also provided additional support specific to the importance of errors and feedback that can be applied to human-robot trust.

Table 1

HRI and Automation Trust Literature Supporting Trust Antecedents Specific to the Capabilities of the Robot

Trust Antecedent Findings Source Behavior Express social behaviors (e.g., turn taking, Abe & Richardson, 2006 emotional expressions) to be considered Bahner et al., 2008 trustworthy Bailey & Scerbo, 2007 Participants trust slower robotic behaviors than Donmez, Boyle, Lee, & McGehee, faster ones 2006 Drury, Rick, & Racklife, 2006 Reliability/Errors Increase in reliability increases trust Dzindolet, Peterson, Pomranky, Pierce, Quality/Accuracy Low error rate has a moderate effect on trust & Beck, 2003 development Evers, Maldanado, Brodeck, & Hinds, Exposing users to failures during training can be 2008 used to reduce complacency and automation bias Groom & Nass, 2007 Errors effect trust unless a reason for the error is Ross, 2008 provided Seong & Bisantz, 2008 When the system falsely adapts to the situation, Tsui, Desai, Yanco, Cramer & trust declines Kemper, 2010 The quality of automation effects trust Wang, Jamieson, & Hollands, 2009 Failures of robot communication negatively Wang, Rau, Evers, Robinson, & impact trust Hinds, 2010

Feedback/Cueing Implicit communication enhances trust with some cultures, while explicit communication enhances trust with other cultures Prompt feedback is essential to maintain trust Trust in aid feedback correlates with belief in aid reliability and reliance on feedback Information from robot impacts trust and situation awareness

Robot features. Physical features of the robot are often a secondary focus of design and development. As such, there is less research in this area beyond that of theoretical links to trust

development. However, a much larger research base focused on robot features within HRI is available to provide support and insight. For example, appearance impacts general likability (Li

& Yeh, 2010) and perceived intellect (Bartneck, Kanda, Mubin, & Al Mahmud, 2009).

Component features such as surface features (e.g., beveled edges), number of appendages (Sims et al, 2005), facial features and expressions (Lum et al., 2007; Mohan, Calderon, Zhou, & Yue,

2008; Zhang et al., 2010), and levels of anthropomorphism (DiSalvo & Gemperle, 2003) have also been found to impact how individuals interact with robots. Goetz, Kiesler, and Powers

(2003) found that social cues are embodied in appearance, and therefore influences how individuals perceive a robot (positively or negatively) as well as their willingness to comply with instructions.

In addition, Upham Ellis and colleagues (2005) found that when function, in terms of behavior, is controlled, individuals make attributions (e.g., cooperation, intelligence, friendliness, threat) about the robot based on the outside physical features. This finding is also supported in the computer agent literature. Embodied human-like computer agents were perceived to provide naturalistic behavior, appropriate emotions, and follow social rules (Mateas, 1999; Nass &

Moon, 2000; Parise, Kiesler, Sproull, & Waters, 1999; Reeves & Nass, 1996; Rousseau &

Hayes-Roth, 1997).

Antecedents of trust associated with robot features hereby included the level of automation (LOA), appearance (e.g., anthropomorphism), type of robot, robot personality, robot intelligence, and mode of communication. Refer to Table 2 for specific findings regarding robot features associated with trust development. Automation literature provides additional clarity specific to the mode of communication and trust that can be utilized in robotics.

Table 2

HRI and Automation Trust Literature Supporting Trust Antecedents Specific to Robot Features

Trust Antecedent Findings Sources LOA Level of automation is correlated with human- Adams et al., 2003 robot trust Biros, Daly, & Gunsch, 2004 Trust increases when a human has the ability to Dautenhahn, 2007 adapt the level of the robot’s automation. de Ruyter, Saini, Markopoulous, & People tend to trust a human operator more than van Breemen, 2005 a robot acting without an apparent operator. Gupta, Bisantz, & Singh, 2002 Heerink, Kröse, Evers, & Wielinga, Anthropomorphism Anthropomorphism is correlated with trust 2010 Robot Type Trust increases when behavior, appearance, and Kidd & Breazeal, 2008 robot type match what’s expected Li, Rau, & Li, 2010 Type of robot and size of robot affect trust Looije, Neerincx, & Cnossen, 2009 Rau, Li, & Li, 2009 Robot Personality / A highly likable robot with active response and Reising & Sanderson, 2002 Intelligence engagement correlates with trust Tenney, Rogers, & Pew, 1998 People trusted a less social robot. Tsui, Desai, & Yanco, 2010 Robot personality for companion robots should be considerate, proactive, non-intrusive, flexible, and competent Social intelligence matters (e.g., people trust socially unintelligent robot more)

Mode of Mode of communication for control displays communication should be appropriately instrumented and designed according to the principles of ecological interface design; otherwise they might lead to misinterpretation Trust is lower for high sensitivity alarms

Human-related trust factors. The human element within HRI is a key component in understanding trust development. Not only do robots require human interaction (e.g., supervision, guidance) to complete tasks (Yagoda, 2011), but the success of HRI is dependent on human acceptance of the technology (Davids, 2002). Yet little attention has been paid to the human partner in this respect (Sheridan, 1988). Figure 12 provides two examples of HRI, including a police officer interacting with a bomb diffusing robot, and the Armed Forces interacting with the Talon robot.

Figure 12. Examples of present day HRI within high-risk environments.

At present, only two studies (Kidd, 2003; Scopelliti, Giuliani, & Fornana, 2005) have empirically assessed the human component of HRI. Specifically they have examined human traits, as they impact trust development. Review of the fields of interpersonal and automation trust serve to provide additional support for determining potential antecedents of human-robot trust specific to human states and traits.

Human traits. Demographic characteristics of the user or operator include age, ethnicity, personality, as well as a number of others. Each of these has been suggested to influence the human-robot trust relationship. The automation and interpersonal trust domains support the importance of traits in trust development. There is evidence that stable traits may be related to the operator’s response to automation, but it is likely that their relationships vary as a function of specific traits as well as characteristics of the automated task (Szalma & Taylor, 2011).

Stable traits, such as personality, may have similar findings across differing forms of interaction. For example, extroverts tend to trust more than introverts (McBride & Morgan,

2010). However limited works in HRI suggest that personality traits are highly correlated with trust (Looije, Neerincx, & Cnossen, 2010). Ethnicity may also prove to be a stable trait in terms of trust development. Different ethnic groups report varying levels of trust in robots (Evers,

Maldanado, Brodecki, & Hinds, 2008; Rau, Li, and Li, 2009). Conversely, findings specific to the relationship between age and trust development may be system or task specific. Scopelliti,

Giuliana, and Formana (2005) found that young adults tend to trust robots more than older adults. The importance of age is also found in automation trust literature. Here, findings suggest that older adults trust automated driver warning systems more than younger adults (Ho,

Wheatley, & Scialfa, 2005; Kiercher & Thorslund, 2009). It is also important to note that gender does not seem to be related to trust development (Kidd, 2003).

Human states. Human states can directly impact the operator’s perceived ability to use a robot (Hancock & Warm, 1989; Parasuraman et al., 2009). Human-automation literature expands upon this work to address the specific importance of stress, fatigue, workload, and attentional control to the actual use and reliance on an automated system (see Chen & Barnes, 2012; Chen &

Terrence, 2009; Finomore, Matthews, Shaw, & Warm, 2009; Igbaria & Iivari 1995; Neaubauer,

Matthews, Langheim, & Saxby, 2012; Reichenbach, Onnasch, & Manzey, 2011; Thropp, 2006;

Warm, Parasuraman, & Matthews, 2008). While extensive research has been conducted linking human states to successful human-technology interaction, the work specific to trust is limited primarily to the impact of stress and workload on trust development. For example, stress and workload are variable and unpredictable, and can lead to degradations in trust (Biros, Daly, &

Gunsch, 2004; Cosenzo, Parasuraman, Novak, & Barnes, 2006; Wang, Jamieson, & Hollands,

2011).

Environment-related trust factors. HRI can occur within a variety of environments ranging from industrial settings (e.g., manufacturing plants), to social environments (e.g., educational classroom, home, etc.), to high-risk environments (e.g., battlefield), and more. The environment can affect how the robot operates, as well as fundamentally change the way humans perform a task (Goodrich & Schultz, 2007). It can relate to changes in robot and human behavior

(e.g., expression of feelings or attitudes) that occur during interaction and in the development of a shared language (Steinfield et al., 2006). For successful interaction to take place within a given environment, human and robotic teammates must share a common goal, have a shared mental model, put group needs above individual needs, view the interdependent nature of the team positively, know and fulfill roles, communicate effectively with team members, and trust each team member (Groom, 2008).

Environment: Team-related trust. The team-related trust factors include role interdependence, team composition, shared mental models, and societal impact. Working with another person (or in this case a robot) to accomplish a task requires team members to depend on each other in order to achieve their goals (Mayer, Davis, & Schoorman, 1995). The degree of interdependence between those involved in the relationship is one of the most important factors involved in the development of trust (Axelrod, 2004). Therefore, higher trust is required for successful collaborative teaming (Lyons, Stokes, Garcia, Adams, & Ames, 2009).

Environment: Task-related trust. These environmental-based factors often require an interaction between the human and robot within a specific context. Therefore, trust is often a task dependent construct that is dependent on the type of task, multi-tasking requirements, the physical environment and risk. For example, Kidd (2003) found that collaborative tasks yield higher trust ratings than information gathering tasks. The nature of these tasks can be impacted differently by the physical environment and risk associated with the specific task. Physical environments can include weather constraints (e.g., unmanned aerial vehicle missions), terrain, obstructions and even bystander interaction. For example, Lyons et al. (2011) used factor analysis to examine trust in the context of IT suspicion and determined that levels of trust are dependent upon cues in the environment.

The perceived risk and uncertainty of various environments can influence the development and calibration of trust (De Santis, De Luca & Bicchi, 2008; Mayer et al., 1995). Some levels of risk and uncertainty are necessary in order for trust to be relevant and important in the relationship (Bhattacharya, Devinney, & Pillutla, 1998). Both internal risk of the robot as well as the external risk of the environment can affect trust development. For example, in the field of automation, Beckles, Welch, and Basney (2005) looked at the internal risk of data corruption from the automation of a computerized credential management system. They suggest that the system will not be trusted if the information is either compromised or revoked. Borst, Mulder, and Paassen (2010) investigated automation-based pilot alerting systems for the external risk of the physical terrain. They suggest that trust is an important element in these types of automated systems due to the safety risk associated with this type of human-automation interaction.

Trust Measurement

Self-report measures are the most commonly used method of assessing trust across domains. At this time human-robot trust is only measured through self-report methods.

Previously developed measures in HRI are primarily concerned with assessing the likelihood of robot use. Broad speculations of trust are made via self-report on items that range from a single question (e.g., “How much do you trust this robot?”) to questionnaires created by the researchers themselves that are specific to the interpersonal or automation domains. Appendix W provides citations for previously developed or referenced trust scales.

Human-interpersonal and human-automation trust literature suggests that there are multiple types of subjective trust measurement scales. These previously developed trust scales traditionally only measure one specific type of trust. This leads to some debate as to whether or not self-assessed trust measures reflect actual use (Chen & Terrence, 2009). Propensity to trust, cognitive trust, affective trust, and trustworthiness are four types of scales that can be applied to human-robot trust (see Figure 13).

Figure 13. Visual description of how previous trust scales are applicable to HRI.

Propensity to trust. Propensity to trust is an integral component of relationships

(Larzelere & Huston, 1980). As a stable trait, unique to the individual (Jarvenpaa, Knoll, &

Leidner, 1998; Mayer et al, 1995; Rotter, 1967; Stack, 1978), it is expected to remain at the same level for a given individual across a variety of settings and situations (Burke et al, 2007; Mayer,

Davis, & Schoorman, 1995). The most common measurement scale used to assess an individual’s propensity to trust is the Interpersonal Trust Scale (Rotter, 1967).

Current human-robot trust researchers suggest that due to the human component in the

HRI team, a propensity to trust scale should be used in conjunction with a human-robot trust measure (see Yagoda, 2011). Merritt and Ilgen (2008) suggest that an individual’s propensity to trust machines can directly impact use of that machine. However, limited work has included propensity to trust in human-technology domains. Within human-robot teams, propensity to trust

can be identified as the inherent trait-like characteristic of the person (i.e., individual differences) to trust. Antecedents of trust that could impact the propensity to trust include human demographic traits (i.e., age, ethnicity, and personality traits) and societal/cultural norms.

Affect-based trust. Affect-based trust is an emergent attitudinal state (emotion-based investment) in which the individual makes attributions about the motives of a partner (Burke et al., 2007; McAllister, 1995). Similarly to propensity to trust, affect-based trust is also important in the initial stages of a relationship. Atoyan and colleagues (2006) suggest that initial experience has a direct influence on trust and use. Within HRI, affect-based trust can be applied to how the human operator makes attributions about the motives of a robotic partner prior to interaction, and can be updated following HRI. The human emotive factors of comfort, satisfaction, confidence, and attitudes are included as antecedents of affect-based trust.

Comfort. Whether or not a user is comfortable with a robot may influence their perception, and ultimately, whether or not they trust the robot. The level of comfort may be dependent on familiarity and proximity (Nakada, 1997; van den Broek & Westerink, 2009), similarity of intent (Verberne, Ham, & Midden, 2012), or even the degree of control the robot has over levels of control, tactics, or strategy of the task (Ward, 2000).

Satisfaction. Satisfaction influences human perception of the robot. This satisfaction is fostered by the quality of information and service that the technology provides (Lee, Kim, &

Kim, 2007). Generally, as satisfaction increases, so does trust (Domnez et al., 2006). Satisfaction may be influenced by situation awareness, predictability and dependability of the robot.

Confidence. Confidence is often used as a synonym for trust. Other researchers suggest that confidence is one of the antecedents leading to trust development. Operators need to be confident in the functionality of the system in order to successfully interact and use said system.

For example, Gao, Lee, and Zhang (2006) found that an operator will switch to manual control when confidence-based trust in automation decreases. Operators are also more likely to remain in manual control in the future, even when it is clear that the automation is more efficient than human performance.

Attitudes. Attitudes about a robot can impact can influence trust development. For example, Bailey and colleagues (2006) found that operators with positive attitudes towards automation may have too much trust in the system and focus less on the overall system state which can lead to degraded performance. Additionally, moods and emotions impact trust and liking of a system (e.g., happiness; Merritt, Heimbaugh, LaChapell, & Lee, 2012).

Cognition-based trust. Cognition-based trust can change based on interaction and is dependent on the reliability and dependability of a specific partner (Burke et al., 2007;

McAllister, 1995; Merritt & Ilgen, 2008). It is a process that changes over time (Burke et al.,

2007) throughout continued interaction with a partner. A robot’s intended use during interactions within human-robot partnerships range from tool to collaborative partner or team member.

Successful collaborative partnerships require communication, cooperation, and coordination between the two acting members as they work towards a common goal (Groom & Nass, 2007).

Cognition-based trust can be discussed in terms: (1) understanding how the robot works, (2) ability to use or interact with the robot, and (3) expectancy of the robot. While these are most

often researched in the interpersonal trust domain, Ashleigh and Stanton (2001) suggest that these areas of research are applicable to the technology domain as well.

Understanding how the robot works. Understanding how to operate or interact with a robot is in part due to the ease of learning the system and the operator’s previous experiences.

Role interdependence and team composition can also provide additional information that impacts how a person understands the robot.

Ability to use or interact with a robot. Another area that affects the human-robot cognitive-based trust relationship is the operator’s perceived ability to interact with the robot.

This includes self-efficacy, workload, and expertise. Following interaction with a robot, additional antecedents of trust including situation awareness, reliability, errors, feedback, and cueing can impact the operator’s perception of how to use or interact with the robot. For example, situation awareness is associated with trust in the robot through interpretation of information received from the robot (Drury, Rick, & Racklife, 2006).

Expectancy. Prior to interacting with automation, human operators develop mental models or expectations of how the automation should behave. Therefore this subcomponent of cognitive factors includes the following antecedents of trust: perceived commitment, usefulness, expectations, and potential risk or benefit. The task type and physical environment can also provide expectancy cues.

Trustworthiness. Trustworthiness is an important element in understanding trust as a whole. Sheppard and Sherman (1998) suggest that the initial trust relationship begins through the assessment of the trustworthiness of a potential partner. Interpersonal trust literature suggests

that trustworthiness is established through assessing the partner’s ability, benevolence, and integrity (see Colquitt, Scott, & LePine, 2007; Daellenbach & Davenport, 2004; Mayer, Davis, &

Schoorman, 1995). Automation literature suggests that trustworthiness is a key element of trust research (Lee & See, 2004) and is founded on the characteristics of the automation (e.g., behavior; Muir & Moray, 1996; Parasuraman & Miller, 2004).

For the purpose of human-robot trust, initial trustworthiness can be defined as the cursory information interpreted by the human about characteristics of his/her partner. The primary factors that impacts initial robot trustworthiness are reputation (Daellenbach & Davenport, 2004;

Lee & See, 2004), physical form (e.g., anthropomorphism; Dautenhahn, 2007), robot type, robot personality, intelligence, level of automation and perceived function (e.g., behavior). This addresses a major design consideration in which form should match function. Since initial trustworthiness is assessed prior to interaction, the perceived functionality of the robot can be drawn from its reputation or viewing the robot’s physical form. The robot’s physical form (i.e., robot type, level of anthropomorphism) may be the primary influence of initial trustworthiness due to the importance of features on perception of trustworthiness. For example, the e-commerce domain suggests that features (e.g., color, layout of computer interfaces, brand) can enhance or degrade the perception of trustworthiness (Kim & Moon, 1998; Nickerson & Reilly, 2004).

The advancement of human-robot trust measurement. Research has continued to address the creation and validation of successful evaluation methods for wide spectrum of HRI issues, including human-robot trust (Steinfeld et al., 2006). The trust scale methodology for this dissertation used the allegory of the cave, a well-known philosophical reference in Plato’s

written work, The Republic (360 B.C.E./2008; see Appendix D), as its foundation. The effectiveness of the allegory of the cave to the field of psychometrics, specific to True Score

Theory, has been demonstrated in Chaorro-Premuzic, von Stumm, and Furnham (2011).

X = T + ex (1)

In the simplest terms, True Score Theory measurement (see Equation 1) follows True

Score Theory where the observed score (X) is equal to the true score (T) plus error (ex),where across a set of scores we assume that var(X) = var(T) + var(ex). By using the triadic model of trust as a foundation for this human-robot trust scale, the entire trust relationship within HRI can be assessed, thus creating a more defined “shadow” of trust.

Figure 14 represents a predictive model of human-robot trust that occurs over time. The red dotted vertical lines represent key points in time within the human-robot relationship. The thick, black lines with the open arrowhead represent a navigable association; while the solid, light blue lines represent a linked association. For example, there is a relationship between society and the human involved in HRI. The dotted, blue line with the open arrowhead represents a dependency. For example, the ability to gather information about the HRI task is dependent upon the team members’ knowledge and prior experiences, the reputation of the robot, and the observed physical features, as well as the perceived functional capabilities of the robot itself.

Together, this predictive model provides a means of understanding the human-robot trust relationship as a whole.

Figure 14. UML predictive model of human-robot trust measurement over time.

Progression to Next Section

The thorough literature review provided the means to develop both a descriptive and predictive model of human-robot trust, thus laying the foundation for the creation of the human- robot trust scale. Each part of scale development was then constructed using the procedures discussed in DeVellis (2003) and Fink (2009). These procedures included creation of an item pool, item pool reduction techniques, subject matter expert (SME) review, and task-based validity testing.

The following chapter is the first in a series of six studies that led to the creation of a reliable and valid human-robot trust scale. This first study looked at how a robot’s physical features have impacted the perception of what classifies a machine as a robot. Further, this study extended the research on human-robot trust to understand the importance of physical form on developing trust prior to interaction.

CHAPTER THREE: STUDY 1

The Robot: “I am only the end result of those who programmed me.” Professor John Robinson: “Well, that may be true, but there’s something more, too, something that doesn’t make sense. You’re more than a machine.” Lost in Space (Allen & Martin, 1966).

Classification of a machine as a robot is often left up to the discretion of the author, contributing to the ongoing debate regarding the characteristics that differentiate a machine from a robot. Just because a designer calls a machine a robot, does not mean that it will be perceived as such by the general public. For the purpose of this work, both the robot’s form and function are necessary components in robot classification as it applies to perceived trustworthiness.

Hypothesis 1: Machine images will be rated low on the robot scale, such that all images of machines will receive ratings less than (or equal to) a 4 (neutral) on that scale.

 robot_rating < neutral rating (4)

Hypothesis 2: There will be a positive relationship between robot classification ratings and trustworthiness ratings for each robot domain and each robot stimuli such that as robot classification ratings increase so do trustworthiness ratings. r trustworthiness*robot classification = positive correlations

Hypothesis 3: Subject Matter Expert (SME) classified images of robots will be rated high on the robot scale, such that all images of robots will receive ratings greater than a 4 (neutral) on that scale. 

 robot_rating > neutral rating (4)

Experimental Method

Experimental participants. The participants were 161 undergraduate students (76 males, 85 females) from the University of Central Florida. Two hundred participants were originally recruited; however 39 were not included in analysis due to incomplete data or unsuccessful completion of control questions. Their participation accorded with all regulations from the university’s Institutional Review Board (IRB).

Included antecedents of trust. Antecedents of trust associated with this study included human traits (i.e., demographical information, personality traits, attitudes, and prior experiences), perceived robot functionality (i.e., perceived level of automation and perceived robot intelligence), and robot physical form (i.e., anthropomorphism and robot type). Human states, environmental factors, robot personality, mode of communication and behavior are interaction specific and therefore not addressed in this experiment.

Experimental stimuli. Sixty-three total images (Free Use Images, accessed via web- based sources) were used (see Appendix G). All stimuli were edited with a neutral gray background and were equally sized (1.5 inch x 2 inch). Forty-nine robot images were selected based on commonly referenced robots in literature or recommendations from experts in the field

(SMEs). Seven images were included for each of the seven identified robotic domains (Industry,

Military, Medical, Service, Social, Entertainment, and Therapy) in order to allow for differentiation between categories. The Research and Education robotic domains were not

included in this study due to the diversity of robots within these domains. One image from the medical domain was not included in analysis due to an error in the recording of the web-based system.

Seven adult human images were used as a control for methodological purposes. Images included multiple races and both genders. Seven common machine images were included, ranging in levels of automation from a hammer to an airplane. These collective images were chosen based on robot classification debates within the literature.

Materials. Two scales, Mini-IPIP and NARS, were used to identify individual personality traits and any prior negative attitudes towards robots that may have influenced participant ratings (Appendix E). The demographics questionnaire consisting of four questions served to identify gender, race, age, and year in school.

The 7-point Mini-IPIP scale (Donnellan, Oswald, Baird, & Lucas, 2006) personality assessment was used to measure the Big 5 personality traits: Agreeableness, Extraversion,

Intellect, Conscientiousness, and Neuroticism. It is a 20-item short form of the 50-item

International Personality Item Pool – Five Factor Model (Goldberg, 1999).

The 7-point scale Negative Attitudes towards Robots Scale (NARS; Nomura, Suzuki,

Kanda, & Kato, 2004) was used to further understand differences in robot classification. It has been previously used across multiple domains of HRI and has been shown to predict interaction and explain individual differences in participants’ behavior. For a review of studies using the

NARS, see Tsui, Desai, Yanco, Cramer, and Kemper (2010).

Experimental Procedure. Participation occurred online and all data were collected via

SurveyMonkey.com. Following informed consent, participants viewed stimuli presented in a random order for each participant. Participants were asked to rate the degree to which they classified each image as human, machine, object, and robot on a 7-point Likert scale. Higher ratings corresponded to greater agreement that the entity constituted human, machine, object, or robot classification. Participants were asked to rate the perceived intelligence (not intelligent to very intelligent), perceived level of automation (completely controlled by someone or something else to completely self-controlled), perceived trustworthiness (not at all to definitely), and intended use (not at all to definitely) for each image on 7-point Likert scales that were created specifically for this study. To account for prior experience with a specific stimulus, participants were asked two separate questions to report if they had ever seen, or ever interacted with stimulus. Refer to Figure 15 for a visual depiction of the survey. Finally, participants completed the Mini-IPIP, Negative Attitudes toward Robots (NARS), and the demographic questionnaire before being debriefed about that nature of the study in which they had just participated. The web-based study took approximately 60 minutes to complete.

Figure 15. Example of Study 1 survey.

Experimental Results

All data were analyzed using IBM SPSS Statistics v.19 (SPSS, 2010), with an alpha level set to .05, unless otherwise indicated. Three levels of stimuli were utilized in this study: humans, machines, and robots. The measures for classification were human, object, machine, and robot.

Participants prior experiences (i.e., previously seen or previously interacted) with the stimuli had no impact on any of the analyses. None of these variables were affected by the demographic data.

Human stimuli. The seven human images were used as a control for this experiment.

Results supported that human stimuli were rated highly on the human scale (M=6.48, SD=0.90), and low on the machine (M=1.49, SD=0.94), object (M=2.08, SD=1.57), and robot (M=1.39,

SD=0.80) scales. Further analysis on the individual images, revealed that the slightly higher object score was due to various objects being present in the stimuli (e.g., woman holding towels; see Figure 16).

Figure 16. Example of object being present in human stimulus.

Machine stimuli. The first hypothesis predicted that machine images would be rated low on the robot scale; confirmation was expected to help separate out the divergence in robot definitions. Based on prior definitions of a robot, certain machines (e.g., washing machine) could be classified as a robot. Seven images of machines ranged from a simple machine (e.g., ball) to complex machine (e.g., airplane). Five of the stimuli (i.e., car, washing machine, X-ray machine, baggage screener, and airplane) all have some degree of programmable automation.

In support of the first hypothesis, overall machine stimuli were rated low on the robot classification scale (M=2.36, SD=1.22). Due to the diversity of these images, individual t-tests were conducted for each image. All machine stimuli scored significantly below a neutral response score of 4 on the robot classification scale (see Table 3). A graphical representation of this finding is depicted in Figure 17. Additional analyses explored the possible individual differences of the x-ray machine and baggage screener (see Appendix H).

Table 3

Means and Standard Deviations for each Machine Stimuli on the Robot Classification Scale

Machine Stimuli Robot Mean (SD) df t (Test Value =4) CILow CIHigh

X-Ray Machine 3.44 (2.37) 159 -2.95** 3.09 3.83

Baggage Screener 3.08 (2.29) 160 -5.10** 2.74 3.45

Washing Machine 2.67 (2.12) 159 -7.96** 2.31 2.97

Airplane 2.41 (2.09) 160 -9.65** 2.05 2.70

Car 2.32 (1.99) 160 -10.70** 2.03 2.65

Ball 1.36 (1.02) 159 -32.56** 1.20 1.53

Hammer 1.24 (0.79) 160 -44.58** 1.12 1.36 Note. ** p < .001

Robot Classification RobotRatings Classification

Machine Images

Figure 17. Representation of the robot classification ratings for each of the machine images. Means and 95% confidence intervals demonstrated that all machine images were rated below the neutral point (4) on the robot classification scale.

Robot stimuli. Robot stimuli were analyzed across the seven general domains of robotics as well as by each individual stimulus. All data were analyzed using IBM SPSS Statistics v.19

(SPSS, 2010), with an alpha level set to .05, unless otherwise indicated.

Relationship between trust antecedents and trustworthiness. Pearson correlations were conducted for each of the measured antecedents of trust and the trustworthiness of each robot domain. In accord with Hypothesis 2, there was a positive relationship between robot classification ratings and trustworthiness ratings for each robot domain. Significant correlations between the antecedents of trust and trustworthiness that were not directly related to the hypothesis can be located in Table 4.

Table 4

Correlations among Trust Antecedents and Trustworthiness for Robot Domains

Trust Antecedents Trustworthiness Entertainment Social Military Medical Service Therapy Industry Robots Robots Robots Robots Robots Robots Robots Gender .015 .049 -.040 .015 .048 .084 .037 Age .083 .063 .014 .020 .020 .063 .035 Extraversion -.058 -.017 -.010 -.030 -.067 -.013 -.009 Agreeableness .009 .004 -.003 -.027 -.087 .045 -.021 Conscientiousness -.062 -.097 -.099 -.084 -.086 -.005 -.062 Neuroticism .124 .160* .031 .024 .141 .075 .005 Intelligence .053 .065 -.014 -.014 .002 .033 -.050 Emotions in -.080 -.115 -.148 -.161* -.090 -.118 -.082 Interactions Situational Influence -.092 -.087 -.140 -.095 -.098 -.089 -.114 Social Influence -.257** -.273** -.254** -.279** -.268** -.229** -.228** Robot Classification .321** .191* .292** .234** .337** .284** .229** Machine Classification .302** .227** .206** .243** .256** .229** .185* Object Classification .201* .178* .128 .132 .177* .177* .113 Human Classification .251** .219** .146 .139 .221** .131 .161* Previously Seen -.012 -.032 -.134 .108 -.097 -.082 -.189* Previously Interacted -.055 .042 -.142 -.125 -.142 .009 -.063 Perceived Intelligence .591** .618** .538** .565** .611** .462** .579** Perceived LOA .436** .449** .379** .434** .420** .334** .483** Intended Use .638** .648** .636** .628** .713** .585** .625** Note. **p=.01 *p=.05

The demographic characteristics of gender and age were not significantly related to trustworthiness for each domain. Only one small significant correlation was found among the five personality traits and trustworthiness for the social robot domain, r(159)neuroticism*trustworthiness=

.16, p=.042. One interesting finding showed that there was a negative relationship between the

NARS Social Influence subscale and trustworthiness for all robot domains. This finding showed that participants with lower negative social influence are more likely to have higher trustworthiness ratings across all robot domains. Moderate to high positive correlations were also found for perceived intelligence, perceived level of automation (LOA), and intended use for all robot domains.

Additional Pearson correlations were conducted to assess the second part of Hypothesis 2 which suggested that there would be a positive relationship between robot classification ratings and trustworthiness ratings for each robot stimuli. Pearson correlations for the robot classification variable and the machine classification, object classification, human classification, trustworthiness, perceived intelligence, and perceived LOA variables were analyzed for each robot stimuli. In support of the second hypothesis, 27 of the robot stimuli were found to be significantly positively related. Means, standard deviations, and intercorrelations for each robot stimuli are reported in Tables 5-11.

Independent-sample t-tests were then conducted for each robot stimuli and compared against the neutral point to assess the classification of each stimulus as a robot. In support of

Hypothesis 3, overall robot stimuli were rated high on the robot classification scale (M=5.08,

SD=1.06). Thirty-nine of the 48 stimuli were reported to be significantly classified as a robot.

The complete table of results is reported in Table 42 located in Appendix J. Graphical representations of these findings are depicted in Figures 18-24.

Robot classification ratings for all seven Entertainment robots were found to be positively correlated to machine classification and object classification ratings (see Table 5). In partial support of Hypothesis 2, robot classification ratings for the Ballroom robot (r(159)==.310, p<.001), the Cyclist robot (r(159)=.312, p<.001), Furby (r(158)==.259, p=.001), and Robosapian

(r(158)==.156, p=.049) were found to have significant positive correlations with trustworthiness.

Two additional interesting findings showed significant positive relationships between robot classification and perceived intelligence for Asimo, the Ballroom robot, Cyclist robot, Furby, and

Topio; as well as between robot classification and perceived LOA for all Entertainment robots except Qrio. All seven entertainment robots were rated above the neutral point on the robot classification scale (see Figure 18).

Table 5

Entertainment Robots Means, Standard Deviations, and Correlations of Robot Classification Ratings

Robot Classification Mean (SD) Machine Object Human Trustworthy Intelligence LOA Furby 4.39 (2.465) .408** .215** -.010 .259** .259** .357** Ballroom Robot 4.71 (2.443) .763** .161* -.173* .310** .282** .347** Cyclist Robot 5.42 (2.218 .570** .306** -.009 .312** .309** .360** Asimo 5.81 (1.866) .597** .293** -.344** .103 .172* .157* Robosapian 6.15 (1.626) .428** .342** -.233** .156* .137 .162* Topio 6.41 (1.370) .554** .287** -.232** -.138 .315** .221** Qrio 6.56 (1.131) .396** .357** -.166* .080 .066 -.040 Note. ** Correlation is significant at the 0.01 level (2-tailed) * Correlation is significant at the 0.05 level (2-tailed)

assification assification Ratings Robot Cl

Entertainment Robots

Robot classification ratings for all seven of the Industry robots were found to be positively correlated to machine classification and object classification ratings (see Table 6). In partial support of hypothesis 2, robot classification ratings for the Bridge robot (r(159)==.185, p=.019), the Glass robot (r(159)==.251, p=.001), the Metal robot (r(156)==.231, p=.004), the

Parallel robot (r(159)==.245, p=.002) and the Parallelizing Operations robot (r(158)==.162, p=.041) were found to have significant positive correlations with trustworthiness. Two additional interesting findings showed significant positive relationships between robot classification and perceived intelligence for all seven Industry robots; as well as between robot classification and perceived LOA for all Industry robots except the Metal robot. In addition, all industry robots,

except for the Parallel robot, were rated above the neutral point on the robot classification scale

(see Figure 18).

Table 6

Industry Robots Means, Standard Deviations, and Correlations of Robot Classification Ratings

Robot Classification Mean (SD) Machine Object Human Trustworthy Intelligence LOA Parallel Robot 3.39 (2.470) .481** .158* .075 .245** .397** .357** Metal Robot 4.60 (2.352) .323** .360** -.086 .231** .223** .122 Glass Robot 4.87 (2.409) .384** .412** -.164* .251** .250** .232** Food Robot 4.91 (2.265) .315** .393** -.086 .075 .193* .161* Robotic Arm 5.02 (2.254) .279** .534** -.430** .148 .210** .228** Bridge Robot 5.04 (2.265) .292** .531** -.114 .185* .178* .200* Parallelizing 5.07 (2.255) .433** .421** -.101 .162* .212** .224** Operations

Robot Classification RobotRatings Classification

Industry Robots Figure 19. Representation of the robot classification ratings for each of the industry robot images. Means and 95% confidence intervals demonstrated that six out of the seven industry robot images were rated above the neutral point (4) on the robot classification scale.

Similarly to the findings from the two previous domains, robot classification ratings for all six of the Medical robots were found to be positively correlated to machine classification and object classification ratings. In partial support of Hypothesis 2, robot classification ratings for the

DaVinci robot (r(158)=.179, p=.024), the Laparoscopic robot (r(157)=.354, p<.001), the Lifting robot (r(159)=.160, p=.043), and Ubot (r(159)=.319, p<.001) were found to have significant positive correlations with trustworthiness. Two additional interesting findings showed significant positive relationships between robot classification and perceived intelligence for all six Medical robots; as well as between robot classification and perceived LOA for all Medical robots except the RP6 and UBot.

Table 7

Medical Robots Means, Standard Deviations, and Correlations of Robot Classification Ratings

Robot Classification Mean (SD) Machine Object Human Trustworthy Intelligence LOA Laparoscopic Robot 4.21 (2.421) .463** .251* -.006 .354** .354** .259* DaVinci Robot 4.22 (2.472) .351** .269** -.257** .179* .267** .332** RP6 4.47 (2.359) .254** .356** -.118 .116 .165* .094 Heart Robot 4.92 (2.254) .417** .174* -.093 -.021 .231** .211** UBot 5.91 (1.821) .545** .437** -.067 .319** .213** .046 Lifting Robot 6.09 (1.619) .550** .423** -.365** .160* .194* .278**

Results from t-tests supported Hypothesis 3 for the Heart robot, Lifting robot, RP6, and

UBot (see Figure 20). The Laparoscopic robot and the DaVinci robot were rated as neutral on the robot classification scale, t(157)=1.08, p=.281 and t(158)=1.12, p=.265, respectively.

Robot Classification RobotRatings Classification

Medical Robots Figure 20. Representation of the robot classification ratings for each of the medical robot images. Means and 95% confidence intervals demonstrated that RP6, Heart robot, UBot, and Lifting robot were rated above the neutral point (4) on the robot classification scale.

Robot classification ratings for all seven of the Military robots were found to be positively correlated object classification ratings (see Table 8). However, only the Big Dog,

Daksh, IED Detonator, Packbot, and Reaper robots were found to have positive significant correlations between robot classification and machine classification ratings. In addition, only the

Big Dog (r(157)=.256, p=.001) and Packbot (r(156)=.293, p<.001) were found to have a significant correlation between robot classification and trustworthiness. Interestingly, findings showed a significant positive relationship between robot classification and perceived intelligence for all the Military robots, except for the TUGV robot. Significant positive correlations were also found between robot classification and perceived LOA for the Big Dog, Packbot, Reaper, and

Talon robots. In addition, all military robots, except for the Reaper, t(158)= -0.07, p=.948, were rated above the neutral point on the robot classification scale (see Figure 21).

Table 8

Military Robots Means, Standard Deviations, and Correlations of Robot Classification Ratings

Robot Classification Mean (SD) Machine Object Human Trustworthy Intelligence LOA Reaper 3.99 (2.442) .219** .220** .073 .102 .312** .340** TUGV 4.39 (2.369) .155 .247** -.339** .119 -.098 -.149 IED Detonator 4.66 (2.335) .281** .289** -.104 .174* .235** .153 Talon 5.35 (2.183) .152 .396** -.046 .051 .191* .172* Daksh Robot 5.48 (2.025) .291** .380** -.185* .055 .201* .113 Big Dog 5.66 (1.933) .705** .499** -.100 .256** .220** .282** Packbot 5.80 (1.889) .482** .408** -.212** .293** .261** .274**

Robot Classification RobotRatings Classification

Military Robots Figure 21. Representation of the robot classification ratings for each of the military robot images. Means and 95% confidence intervals demonstrated that all robots were rated above the neutral point (4) on the robot classification scale, except for the Reaper.

Robot classification ratings for all seven Service robots were found to be positively correlated to machine classification and object classification ratings (see Table 9). In partial support of Hypothesis 2, all seven Service robots were found to have significant positive correlations between robot classification and trustworthiness except for the Chore robot

(r(158)=.129, p=.104). Two additional interesting findings showed significant positive relationships between robot classification and perceived intelligence for all Service robots, except Ubiquitous Dream (r(159)=.132, p=.095); as well as between robot classification and perceived LOA for all Service robots except for SmartPalIV (r(159)=.108, p=.172).

Table 9

Service Robots Means, Standard Deviations, and Correlations of Robot Classification Ratings

Robot Classification Mean (SD) Machine Object Human Trustworthy Intelligence LOA Gutter Robot 3.66 (2.361) .474** .254** .113 .386** .283** .199* Roomba 4.33 (2.376) .385** .189* -.026 .251** .280** .421** Lawnmower Robot 4.62 (2.374) .354** .207** -.044 .165* .268** .288** PatrolBot 5.06 (2.367) .533** .324** -.056 .212** .335** .323** Ubiquitous Dream 5.70 (1.946) .469** .462** -.179* .289** .132 .241** Chore Robot 6.23 (1.538) .768** .477** -.329** .129 .188* .188* SmartPalIV 6.34 (1.327) .603** .562** -.324** .195* .208** .108

Out of the seven service robots, the lawnmower robot, PatrolBot, Ubiquitous Dream, chore robot and SmartPalIV were rated above the neutral point on the robot classification scale

(see Figure 22). The gutter robot and Roomba were rated as neutral, t(158)= -1.84, p=.067 and t(159)=1.76, p=.081, respectively.

Robot Classification RobotRatings Classification

Service Robots

Figure 22. Representation of the robot classification ratings for each of the service robot images. Means and 95% confidence intervals demonstrated that five of the robots were rated above the neutral point (4) on the robot classification scale, while the gutter robot and Roomba were rated as neutral.

Robot classification ratings for all seven Social robots were found to be positively correlated to machine classification and object classification ratings, and negatively correlated to human classification ratings (see Table 10). Only Aibo was found to have a significant positive correlation between robot classification and trustworthiness, r(158)=.250, p=.001. In addition, only Nexi (r(159)=.236, p=.003) and PaPeRo (r(159)=.174, p=.027) were found to have a significant positive correlation between robot classification and perceived intelligence; while

Aibo (r(159)=.189, p=.016) and Nexi (r(159=.288, p<.001) were found to have a significant positive correlation between robot classification and perceived level of automation

Table 10

Social Robots Means, Standard Deviations, and Correlations of Robot Classification Ratings

Social Robots Mean (SD) Machine Object Human Trustworthy Intelligence LOA Robot Classification Paro 1.54 (1.283) .688** -.116 .416** .136 .024 -.057 Wakamaru 6.11 (1.637) .494** .420** -.254** .135 .118 .121 Aibo 6.11 (1.580) .461** .445** -.283** .250** .129 .189* PaPeRo 6.18 (1.545) .498** .555** -.317** .108 .174* .142 Olivia 6.28 (1.480) .526** .502** -.331** .116 .117 .078 Nexi 6.30 (1.524) .547** .447** -.272** .142 .236** .288** Simon 6.41 (1.256) .344** .405** -.232** .179* .089 .090

Paro was the only service robot rated significantly below the neutral point on the robot classification scale, t(158)= -24.30, p<.001 (see Figure 23). Upon further examination, Paro’s physical appearance is that of a toy. With limited knowledge of the capabilities of the Paro robot, it would be difficult to classify it as such.

ification Ratings Robot Class

Social Robots

Figure 23. Representation of the robot classification ratings for each of the social robot images. Means and 95% confidence intervals demonstrated that six of the robots were rated above the neutral point (4) on the robot classification scale.

In partial support of Hypothesis 2, iFoot (r(156)=.183, p=.022), Kismet (r(159)=.184, p=.019), and the Koala robot (r(157)=.229, p=.004) were the only Therapy robots that were found to have had a significant positive correlation between robot classification and trustworthiness. All other significant correlations reported in Table 11 are similar to above findings.

Table 11

Therapy Robots Means, Standard Deviations, and Correlations of Robot Classification Ratings

Robot Classification Mean (SD) Machine Object Human Trustworthy Intelligence LOA Tibion 2.05 (1.903) .552** .266** -.184* -.017 -.217** -.148 Carrier Wheelchair 3.30 (2.346) .342** .370** -.227** .030 -.095 -.091 iFoot 4.86 (2.240) .448** .250** -.071 .183* .232** .226** Lokomat 5.06 (2.305) .486** .455** -.194* .045 .271** .323** Koala 5.14 (2.187) .310** .356** -.109 .229** .249** .235** Torso Robot 5.48 (2.053 .410** .373** -.137 .001 .006 .145 Kismet 5.91 (1.802) .384** .214** -.245** .184* .199* .229**

Results from the independent samples t-tests showed that the iFoot, Lokomat, Koala robot, torso wheelchair robot and Kismet were all rated above the neutral point on the robot classification scale (see Figure 24). However, the Tibion and Carrier Wheelchair robot were rated significantly below the neutral point, t(159)= -13.00, p<.001 and t(155)= - 3.74, p<.001, respectively. Upon further examination, both the Tibion and Carrier Wheelchair robots appear to be physical therapy assistive devices. Without prior knowledge of the functional capabilities of the Tibion and Carrier Wheelchair, it is difficult to classify either as a robot from the image alone.

Robot Classification RobotRatings Classification

Therapy Robots Figure 24. Representation of the robot classification ratings for each of the therapy robot images. Means and 95% confidence intervals demonstrated that five of the robots were rated above the neutral point (4) on the robot classification scale.

Comparison of robot and machine classification ratings. Participant ratings were evaluated with a paired samples t-test to evaluate the robot and machine classification ratings.

Results revealed that there was a significant difference between the ratings describing the extent to which participants classified the robot stimuli as robots (M=5.08, SD=1.06) and machines

(M=5.64, SD=1.07), t(160) = 8.69, p<.001 (two tailed). While participants may have classified an image of a robot as both a robot and a machine, they tended to rate certain images as being more machine-like than robot-like. The results for each robot domain are reported in Table 12 (see also

Schaefer, Billings, & Hancock, 2012). Here, Entertainment and Social robots had significantly higher robot classification ratings.

Table 12

Comparison of Robot and Machine Classification Ratings per Robot Domain

Domain Mean(SD) t Higher robot classification Robot Classification 5.56 (.976) Social Robots Machine Classification 5.03 (1.337) 5.61**

Robot Classification 5.63 (1.054) Entertainment Robots 4.95** Machine Classification 5.15 (1.476) Higher machine classification Robot Classification 4.70 (1.680) Industry Robots Machine Classification 6.18 (1.073) -11.35** Robot Classification 4.98 (1.352) Medical Robots Machine Classification 5.92 (1.137) -9.54** Robot Classification 5.04 (1.367) Military Robots Machine Classification 6.05 (1.118) -10.09** Robot Classification 5.13 (1.283) Service Robots Machine Classification 5.79 (1.255) -8.35** Robot Classification 4.54 (1.221) Therapy Robots Machine Classification 5.37 (1.191) -9.80** Note. ** p<.01 df = 160

Participant ratings were evaluated with a paired samples t-tests to evaluate the robot classification and machine classification ratings. Paired samples t-tests of each robot stimuli showed that 15 robots were rated as significantly more robot-like (see Figure 25), and five robots were rated as equally rated as robot-like and machine-like (see Figure 26). These findings may suggest that individuals make distinctions between entities that they classify as machines and robots, based on form alone. Further, examination of entities that were classified as having greater robot classification ratings tended to have more anthropomorphic (i.e., human-like) or zoomorphic (i.e., animal-like) physical features.

Figure 25. Robot stimuli with higher classification ratings of a robot than a machine (ordered from most to least).

Figure 26. Robot stimuli with equal robot and machine classification ratings.

Distribution of the robot classification of robot stimuli. A frequency distribution was conducted to identify the variations within individual ratings on the robot classification scale.

Stem and leaf plots have been provided in Appendix I. Table 13 reports the frequency distributions for the 14 robot stimuli that had multiple leaves.

Table 13

Frequency Ratings on Robot Scale for Robot Stimuli

Robot Mean df 1 2 3 4 5 6 7 (SD)

Gutter Robot 3.656 160 Frequency 54 10 13 22 16 11 34 (2.361) Percentage 33.8% 6.3% 8.1% 13.8% 10.0% 6.9% 21.3%

Reaper 3.988 160 Frequency 48 11 11 15 18 14 43 (2.442) Percentage 30.0% 6.9% 6.9% 9.4% 11.3% 8.8% 26.9%

Laparoscopic 4.208 159 Frequency 43 8 10 20 16 14 48 (2.421) Percentage 27.0% 5.0% 6.3% 12.6% 10.1% 8.8% 30.2%

Furby 4.385 161 Frequency 39 13 8 15 18 9 59 (2.465) Percentage 24.2% 8.1% 5.0% 9.3% 11.2% 5.6% 36.7% 4.708 161 Frequency 34 8 13 13 9 16 68 Ballroom (2.443) Percentage 21.1% 5.0% 8.1% 8.1% 5.6% 9.9% 42.2% 4.621 161 Frequency 36 4 10 20 15 17 59 Lawnmower (2.374) Percentage 22.4% 2.5% 6.3% 12.6% 9.3% 10.6% 36.7%

4.329 161 Frequency 37 12 7 27 13 13 52 Roomba (2.376) Percentage 23.0% 7.5% 4.4% 16.8% 8.1% 8.1% 32.3%

4.388 160 Frequency 37 8 12 19 19 13 52 TUGV (2.369) Percentage 23.1% 5.0% 7.5% 11.9% 11.9% 8.1% 32.5%

IED 4.658 161 Frequency 32 8 10 17 18 18 58 Detonator (2.334) Percentage 19.9% 5.0% 6.3% 10.6% 11.2% 11.2% 36.0%

4.219 160 Frequency 45 9 8 19 12 17 50 DaVinci (2.472) Percentage 28.1% 5.6% 5.0% 11.9% 7.5% 10.6% 31.3%

4.472 161 Frequency 35 6 16 19 18 10 57 RP6 (2.359) Percentage 21.7% 3.7% 9.9% 11.8% 11.2% 6.3% 34.4%

Robot Mean df 1 2 3 4 5 6 7 (SD) 4.862 159 Frequency 28 4 8 21 19 19 60 iFoot (2.240) Percentage 17.61% 2.52% 5.03% 13.21% 11.95% 11.95% 37.74%

4.604 159 Frequency 33 9 10 13 19 21 54 Metal Robot (2.352) Percentage 20.76% 5.66% 6.29% 8.18% 11.95% 13.21% 33.96%

4.870 161 Frequency 31 10 8 10 15 15 72 Glass Robot (2.409) Percentage 19.26% 6.21% 4.97% 6.21% 9.32% 9.32% 44.72% Note. 1 represents “Not a Robot”, 4 represents “Neutral”, 7 represents “Definitely a Robot”

Further analysis was conducted to explore potential individual differences that could account for the diversity of the robot classification ratings for the robot stimuli. Results showed that number and type of predictor variables of robot classification were unique for each robot stimuli (see Table 43, Appendix I).

Discussion

This study clearly demonstrated that there are individual perceptual differences in the classification of robots confirming that we need to consider fuzzy boundaries to the classification process. Overall, robots were perceived to be machines, however 15 of the robot stimuli were classified to be significantly more robotic and five stimuli were equally classified as robots and machines (see also Figure 25). Further examination showed greater physical anthropomorphic or zoomorphic physical features of these robots.

In addition, four robot stimuli (Parallel Robot, Paro, Tibion, and Carrier Wheelchair) were significantly below the neutral score on the robot classification scale (see Figure 27). With limited knowledge of the functional capabilities of these robot stimuli, it proved to be difficult to

classify these images as robots from physical appearance alone. Results also showed that five stimuli (DaVinci, Laparoscopic Robot, Reaper, Gutter, and Roomba) were not significantly different from the neutral point (see Figure 28).

Figure 27. Robot stimuli that were rated low on the robot classification scale. .

Figure 28. Robot stimuli that were rated neutral on the robot classification scale.

Frequency distributions were calculated to assess the spread in participants’ robot classification ratings. In addition to the five stimuli in Figure 28, an additional nine robots were found to have wide-spreading frequency distributions. Additional analyses were conducted to identify the predictor variables of robot classification. Findings suggested that unique predictor variables accounted for a small portion of the variance of robot classification for each robot stimulus.

This study also demonstrated the importance of physical form on trustworthiness ratings.

Significant positive correlations were found between robot classification ratings and trustworthiness ratings for each of the seven robot domains. Similar findings were also found for

27 of the 48 individual robot stimuli images, suggesting that there indeed can be individual difference ratings for different robots.

Additional analyses demonstrating this link between robot classification and trustworthiness ratings have been reported by Schaefer, Sanders, Yordon, Billings, and Hancock

(2012). A multiple regression correlation analysis with stepwise entry of variables was conducted to determine the factors that predicted trustworthiness from perceived robot form alone. Analyses were conducted for both the general category of robot images, as well as each robot domain. This was achieved by regressing trustworthiness onto human-related factors

(gender, race, age, year in school), personality traits (agreeableness, extroversion, conscientiousness, intellect, neuroticism), negative attitudes toward robots (emotions in interactions, social influence, and situational influence), as well as self-report items of robot form

(perceived intelligence, perceived LOA, robot classification). For the general category of robot images, the final model included perceived intelligence (PI), robot classification (RC), and social influence (SI) as predictors of trustworthiness, Ŷ = 0.825 + 0.651(PI) + 0.256(RC) –

0.164(SI). It accounted for a significant R2 of 45.1% of the variance, F(3,156)=42.70, p < .001.

This suggested that preconceived ideas regarding the level of intelligence is form-dependent and assessed prior to interaction, in much the same way as one individual will assess another individual as a potential teammate. Further societal influence (e.g., capabilities, functions, etc.) play a key role in expectation-setting similar to stereotypes of human teammates.

Progression to Next Experiment

The results of the present experiment suggested that a robot’s physical form does impact the classification process. It further provides support that the physical form is important to the trust that develops prior to interaction. The present experiment identified individual differences as an important impact on robot classification, but did not fully resolve the issue which is taken up in the next procedure. Clearly perceived functionality can be assessed through form alone, however it is important to note what physical attributes impact the classification process. Study 2 seeks to identify the attributes that are important to how a person classifies a robot as it relates to trust.

CHAPTER FOUR: STUDY 2

“Don’t blame me. I’m an interpreter. I’m not supposed to know a power socket from a computer terminal.” C-3PO, Star Wars (Brackett & Kasden, 1980)

This study was designed to identify which perceived robot attributes (e.g., features and functional capabilities) impact the individual difference classification of a specific subset of robots from Study 1. Robot attributes were assessed through the Godspeed questionnaire, a standardized measurement tool for HRI for interactive robots. It was designed to assess five key

HRI concepts: anthropomorphism (Powers & Kiesler, 2006), animacy (Lee, Park, & Song,

2005), likeability (Monahan, 1998), perceived intelligence (Warner & Sugarman, 1996), and perceived safety (Kulić & Croft, 2007). It has previously been used in HRI experimentation related to the uncanny valley (Ho & MacDorman, 2010), evaluation of specific robots (e.g., robot bartender; Foster, Gaschler, Giuliani, Isard, Pateraki, & Petrick, 2012), and evaluation of specific robotic elements (e.g., attracting human attention; Torta, van Heumen, Cuijpers, &

Juola, 2012), to name a few. The purpose of this study was to assess which attributes affected the classification of a machine as a robot, and determined the relationship to human-robot trust.

Hypothesis 1: There will be a positive relationship between robot classification ratings and trustworthiness ratings for each robot stimuli, such that as robot classification ratings increase so do trustworthiness ratings. r trustworthiness*robot classification = positive correlation

Hypothesis 2: Attributes will be positively related to robot classification ratings for each stimulus. r robot attributes*robot classification = positive correlation

Hypothesis 3: Attributes will be positively related to trustworthiness ratings for each stimulus. r robot attributes*trustworthiness = positive correlation

Experimental Method

Experimental Participants. Participants were recruited through SONA Recruitment

Systems from the Psychology Department at the University of Central Florida. Among the 208 undergraduate students, 82 were male and 126 were female with the average age of 21 years

(Mage=21.05 years, SDage=4.82). All participants were over the age of 18 years. Two-hundred thirty participants were originally recruited; 22 were not included in analysis due to incomplete data or unsuccessful completion of the control questions. Their participation accorded with all regulations from the university’s Institutional Review Board.

Included antecedents of trust. Antecedents of trust associated with this study included specific robot attributes (i.e., anthropomorphism, animacy, likeability, and perceived intelligence) assessed through the Godspeed questionnaire. Human states, environmental factors, robot personality, mode of communication and behavior are interaction specific and therefore not addressed in this experiment.

Experimental Stimuli. Stimuli consisted of a subset of 18 images from Study 1. Two robot images were rated high on the robot classification scale. Two images rated low on the

robot classification scale. Fourteen robot images were considered “neutral” or having fuzzy boundaries (refer to Table 13). All stimuli were edited with a neutral gray background and were equally sized (1.5 in. x 2 in.).

Materials. A reduced version of the 5-point semantic differential scale Godspeed

Questionnaire (Bartneck, Kulić, Croft, & Zoghbi, 2009) was used to assess the attributes of the robot associated with robot classification and trustworthiness. An example of an item from the

Godspeed questionnaire is provided in Figure 29.

Machinelike 1 2 3 4 5 Humanlike

Figure 29. Example of a semantic differential scale item from the Godspeed Questionnaire

Experimental Procedure. Participation occurred online and all data were collected via

SurveyMonkey.com. Following informed consent, participants completed a two part study. In

Part 1, participants completed a short version of the items from Study 1. They were asked to rate stimuli on a 7-point Likert scale on the degree to which they classify the image as a robot, its trustworthiness, and anticipated use. Higher ratings corresponded to greater agreement for each respective question. In Part 2, participants viewed the stimuli a second time and completed the

Godspeed questionnaire for each image. The order of the images was randomized for each participant. The web-based study took approximately 60 minutes to complete in its entirety.

Experimental Results

All data were analyzed using IBM SPSS Statistics v.19 (SPSS, 2010), with an alpha level set to .05, unless otherwise indicated.

Relationship between robot classification and trustworthiness. Hypothesis 1 proposed that there would be a positive relationship between robot classification ratings and trustworthiness ratings for each robot stimuli. Pearson correlations were conducted for robot classification and trustworthiness, r(2910)=.307, p<.001. In support of Hypothesis 1, robot classification and trustworthiness were found to be significantly positively related for all fourteen robots that were previously classified as having neutral robot classification ratings or fuzzy boundaries in the classification process (see

Table 14).

Table 14

Relationship between Robot Classification Ratings and Trustworthiness for each Robot Stimuli

Robot Variable Mean SD r

Robot classification 4.13 1.80 .516** Trustworthiness 3.88 1.32

Gutter Robot

Robot classification 5.15 1.71 .406** Trustworthiness 4.84 1.43

Roomba

Robot classification 5.20 1.78 .366** Trustworthiness 4.27 1.59 TUGV

Robot Variable Mean SD r

Robot classification 5.45 1.45 .343** Trustworthiness 4.17 1.44

iFoot

Robot classification 4.59 1.94 .339** Trustworthiness 4.08 1.55 Ballroom

Robot classification 5.06 1.82 .327** Trustworthiness 4.50 1.42

Laparoscopic

Robot classification 5.13 1.54 .306** Trustworthiness 4.44 1.36 Lawnmower

Robot classification 6.13 0.97 .303** Trustworthiness 4.77 1.30

Metal robot

Robot classification 5.78 1.37 .302** Trustworthiness 4.63 1.50 Glass robot

Robot classification 5.25 1.65 .266** Trustworthiness 4.62 1.50

RP6

Robot classification 5.40 1.61 .261** Trustworthiness 4.24 1.48 IED Detonator

Robot classification 5.43 1.60 .215** Trustworthiness 4.47 1.55

DaVinci

Robot classification 4.95 1.87 .137* Trustworthiness 4.70 1.50 Reaper

Robot classification 4.58 2.07 .136* Trustworthiness 4.29 1.79

Furby

Robot attributes ratings. Robot attribute ratings were reported on a five point semantic differential scale (e.g., machinelike - humanlike anchors) and rated from -2 to +2. Average scores were calculated for the 14 robots previously classified as having neutral robot classification ratings, and the two robots (i.e., Simon and Qrio) previously rated for having the highest robot classification ratings. Paired samples t-tests were then conducted comparing the average scores for the neutral robots and the high rated robots (see Table 15). Seventeen out of the nineteen attribute pairs were found to be significantly higher positive attribute ratings for the high classification robot images than the neutral robots. The anthropomorphic attribute that rated move slowly to move quickly showed the opposite result, t(192)=+2.76, p=.006. The perceived intelligence attribute that rated foolish to sensible had a non-significant finding between the two groups, t(145)= -0.95, p=.345.

Table 15

Paired Samples t-tests comparing Attribute Ratings between the Neutral Classification Robots compared to the High Classification Robots

Neutral Robots High Robots Attributes N M SD M SD t p Machinelike-Humanlike 205 -1.72 .64 -0.92 .85 -11.51 <.001 Artificial-Lifelike 203 -1.62 .78 -0.93 .92 -9.52 <.001 Fake-Natural 195 -1.43 .87 -1.21 .78 -3.77 <.001 Unconscious-Conscious 171 -1.37 .99 -0.66 .99 -8.81 <.001 Move Slowly – Move Quickly 193 -0.39 1.27 -0.66 .83 +2.76 .006 Mechanical – Organic 204 -1.70 .71 -1.45 .70 -4.65 <.001 Dead – Alive 164 -1.15 1.05 -0.70 .96 -6.05 <.001 Stagnant – Lively 188 -0.81 1.15 -0.10 .83 -7.62 <.001 Apathetic – Responsive 181 -0.20 1.29 +0.36 .93 -6.54 <.001 Inert – Interactive 196 -0.21 1.35 +0.69 .93 -8.96 <.001 Awful – Nice 140 +0.02 84 +0.53 .77 -6.44 <.001 Unpleasant – Pleasant 167 +0.03 .97 +0.42 .83 -5.08 <.001 Dislike – Like 188 +0.12 1.04 +0.40 .92 -3.46 .001

Neutral Robots High Robots Attributes N M SD M SD t p Unfriendly – Friendly 146 -0.30 .98 +0.40 .82 -7.44 <.001 Unkind – Kind 135 -0.07 .96 +0.39 .78 -5.10 <.001 Ignorant – Knowing 165 -0.61 1.15 -0.01 .94 -7.66 <.001 Incompetent – Competent 171 -0.09 1.21 +0.27 .88 -4.36 <.001 Foolish – Sensible 146 -0.10 1.07 -0.02 .82 -0.95 .345 Irresponsible - Responsible 154 -0.14 1.13 +0.16 .85 -3.48 .001

Data were then reorganized to individual 5-point Likert ratings for each individual attribute. Responses that were reported as not applicable were given a score of 0 for both attributes. Hypothesis 2 stated that robot attributes would be positively related to robot classification ratings for each robot image. The significant correlations between robot attributes and robot classification ratings are depicted in .

Table 16.

Table 16

Significant Correlations Between Robot Attributes and Robot Classification Ratings

Anthropomorphic Glass Metal Furby Ballroom Lawnmower Roomba Gutter Reaper TUGV IED Laparoscopic DaVinci RP6 iFoot Robots Total % Attributes

# of#

Humanlike ++ 1 7% Lifelike - + ++ 3 21% Natural + + + + ++ + 6 43% Conscious ++ ++ + ++ + 5 36% Move Elegantly ++ ++ ++ + 4 29% Machinelike ++ + ++ ++ 4 29% Artificial ++ + + 3 21% Fake ++ + ++ ++ 4 29% Unconscious ++ ++ + 3 21% Move Rigidly + + + 3 21%

Animacy Glass Metal Furby Ballroom Lawnmower Roomba Gutter Reaper TUGV IED Laparoscopic DaVinci RP6 iFoot Robots of# Total % Attributes

Lifelike - + ++ 3 21% Organic + + + 3 21% Alive ++ ++ ++ ++ ++ ++ 6 43% Lively + ++ ++ ++ ++ ++ 6 43% Responsive ++ ++ ++ ++ ++ ++ ++ ++ 8 57% Interactive ++ ++ + ++ ++ ++ 6 43% Artificial ++ + + 3 21% Mechanical ++ ++ ++ 3 21% Dead + ++ + 3 21% Stagnant + + 2 14% Apathetic + ++ - ++ + 5 36% Inert - - - 2 14%

Likeability Glass Metal Furby Ballroom Lawnmower Roomba Gutter Reaper TUGV IED Laparoscopic DaVinci RP6 iFoot Robots of# Total % Attributes

Nice ++ ++ ++ ++ + 5 36% Pleasant ++ ++ 2 14% Like + ++ ++ ++ + 5 36% Friendly + ++ ++ ++ 4 29% Kind + ++ + + 4 29% Awful + + ++ + 4 29% Unpleasant + ++ + 3 21% Dislike + 1 7% Unfriendly + ++ + + 4 29% Unkind ++ + + + 4 29% Perceived Intelligence Attributes Knowledgeable ++ + + ++ ++ 5 36% Competent ++ ++ ++ ++ 5 36% Sensible ++ + + + ++ ++ 6 43% Responsible ++ + ++ ++ 4 29% Ignorant ++ 1 7% Incompetent - ++ + 3 21% Foolish + 1 7% Irresponsible + 1 7% Total # of Positive 1 2 9 17 3 4 9 13 10 24 30 19 0 Correlations Total # of Negative 1 1 2 2 Correlations Note. **Correlation is significant at the 0.01 level *Correlation is significant at the 0.05 level

Hypothesis 3 stated that robot attributes would be positively related to trustworthiness ratings for each robot image (see Table 17). Means, standard deviations, and intercorrelations for robot attributes and robot classification ratings, as well as robot trustworthiness are reported in

Table 58 of Appendix K. Results for both hypotheses suggested that the significant relationships vary between different robot types.

Discussion

Robot classification was shown to be related to trustworthiness ratings. Results showed that higher classification ratings were positively related to trustworthiness. Therefore, the main purpose of this study was to determine if specific attributes could be identified in the robot classification process. Attribute ratings were assessed through use of the Godspeed questionnaire. Overall results showed that each robot has different attributes that are important to classification. Significant correlations were found for each attribute and robot classification for each robot stimuli. In addition, percentage analysis demonstrated that each robot stimuli was perceived to have some degree of each attribute. Therefore, it was decided all attribute items would be included in the initial trust item pool, the reliability of which is the subject of the next procedure.

Table 17

Number of Positive Trustworthiness Correlations for the Anthropomorphism, Animacy, Likeability, and Perceived Intelligence Attributes

Anthropomorphism 1 0 2 5 1 3 1 Animacy 2 3 3 3 1 5 1 Likeability 2 3 5 3 0 5 2 Perceived 2 3 4 4 2 3 3 Intelligence # of Positive 7 9 14 16 4 16 7 Correlations

RP6 Anthropomorphism 1 2 3 2 4 2 1 Animacy 3 3 4 0 5 2 2 Likeability 1 5 2 5 3 3 3 Perceived 2 4 4 0 3 2 2 Intelligence # of Positive 7 14 13 7 15 9 8 Correlations

Progression to Next Experiment

The following section describes the development of the initial item pool, with the inclusion of all attribute items from Experiment 2. Procedures for initial reduction of the item pool are described. In addition, results from Experiment 1 and 2 on the importance of individual differences to classification and trust supported the need for a more in depth demographics questionnaire to better assess an individual’s prior experiences. Items included exposure to robot media, previous interaction and use with real-world robots, as well as an item to assess the individual’s mental model of a robot.

CHAPTER FIVE: STUDY 3

Lt. Commander Data: “Hm... Then it is better not to trust?” Commander William T. Riker: “Without trust, there's no friendship, no closeness. None of the emotional bonds that make us who we are.” Star Trek: The Next Generation (Roddenberry & Scheerer, 1990)

To conduct the following procedure, the initial item pool was formed which included 172 items that represented each of the antecedents of trust. Between two and four items were created for each antecedent, representing equal number of positively and negatively worded (or opposite related) items. Initial items were developed specific to the updated triadic model of trust (see

Table 18).

Table 18

Trust Antecedents for Scale Development

Trust Factor Trust Antecedent Type of Measurement Specific Information Behavior Scale Development Includes predictability and dependability Robot Reliability/Errors Scale Development Capability Quality/Accuracy Scale Development Includes communication Feedback/Cueing Scale Development Mode of Communication Scale Development Robot Personality Scale Development Robot Features Anthropomorphism Scale Development Intelligence Scale Development Level of Automation Scale Development Age Additional Survey Demographics Questionnaire Gender Additional Survey Demographics Questionnaire Human Traits Ethnicity Additional Survey Demographics Questionnaire Personality Additional Survey Mini-IPIP

Trust Factor Trust Antecedent Type of Measurement Specific Information Attentional Control Additional Survey DSSQ Fatigue Additional Survey DSSQ Human States Stress Additional Survey DSSQ Workload Additional Survey DSSQ Role Interdependence Scale Development Environment Team Composition Scale Development Team Mental Models Scale Development Demographics / Design Cultural / Societal Impact Additional Survey Interpersonal Trust Scale/NARS Task Type Scale Development Environment Physical Environment Scale Development Task Risk Scale Development

Previous developed and referenced scales (see Appendix D) in the robot, automation, and interpersonal trust domains were reviewed to refine the items. Following internal review, the item pool was reduced from 172 to 156 items. Factor analytic procedures were then used to reduce the initial item pool further

Experimental Method

Participants. Participants included 159 undergraduate students (65 males, 94 females) from the University of Central Florida. Two hundred participants were originally recruited; 171 signed up to participate, 12 of which were not included in analysis due to incomplete data or unsuccessful completion of the control questions. Individual participation accorded with all regulations from the university’s Institutional Review Board.

Prior experience. Participants’ prior experience with robots was assessed through the demographics questionnaire (Appendix M). Table 19 reports participants’ prior experience with robots. Of the 156 participants who watched movies or televisions shows with robots, 87

watched between 1-5 movies/television shows, 31 watched between 6-10 movies/television shows, and 18 watched over 10 movies or television shows.

Table 19

Participants Prior Experiences with Robots

Prior Experience Questions Yes No Have you ever watched a movie or television show that includes robots? 98% 2% Have you ever interacted with a robot? 23% 77% Have you ever built a robot? 7% 93% Have you ever controlled a robot? 21% 79%

Participants also provided open-ended descriptions of any robots they had interacted with, built, and/or controlled. Of the 36 participants who reported interacting with a robot, 34 provided a description of the robot. Five reported interacting with a robot at a museum or theme park (e.g., animatronics). Twelve participants reported interacting with everyday items such as a cell phone, computer, ATM, or Xbox. Additional robots included: toys (N = 8), a robot vacuum

(N = 2), robots they built in a classroom or Battle Bot (N = 8), and one participant described an interaction with robot that could not be classified. Robots that were built by participants ranged from small, simple designs to development for high school robotics teams for a robotics competition. Of the 34 participants who reported controlling a robot previously, three controlled a robot through speech, gestures, or commands, 21 controlled a robot by a remote control, keypad, or joystick, six robots were preprogrammed or controlled via computer, and two participants reported items that could not be classified (i.e., DisneyQuest, appliances).

Robot mental model. To assess participants’ mental model of a robot, they were asked an open-ended question to describe what a robot looks like. Table 20 reports the coding description

of participants’ responses. The “machine-like” category included responses such as machine, metallic, and silver. The “human-like” category included both direct statement of human-like, or mention of human features (e.g., limbs, eyes, feet, face, etc.). The “varied” category included participant responses that provided multiple descriptions or ranges of robots. “Shape” included direct reference to a shape (e.g., box-like, square, etc.), size, and physical descriptions (e.g., cold, rigid, durable). The “Computer/Electronics” category included specific mention of computer or electronics, as well as reference to wires or buttons. The “other” category included descriptions of helpful, intelligent, has cameras, a robot, and an alien. Seventeen participants directly referenced specific robots from movies/television (N = 14; R2D2, C-3PO, iRobot, AI, and

Terminator), video games (N = 1, Mass Effect 3), military (N = 2, Predator), and industry (N = 2; robotic arm).

Table 20

Robot Mental Model

Coding Description N % Machine-like 121 76.1% Human-like 49 30.8% Varied 28 17.6% Tool 4 2.5% Shape 34 21.4% Movement 33 20.8% Computer/Electronics 25 15.7% Language 7 4.4% Task, Function, or Interaction-based 30 18.9% Other 7 4.4%

Materials. Materials included 156 initial trust items. Due to the format of the items, a 7- point Likert scale was used (see also Figure 30). The complete initial item pool is located in

Appendix N. A demographics questionnaire was designed to assess prior experiences and mental model of a robot (Appendix M).

Strongly Disagree Slightly Neutral Slightly Agree Strongly Disagree Disagree Agree Agree Most robots make poor teammates. Most robots possess adequate decision-making capability. Most robots are pleasant towards people. Most robots are not precise in their actions. Figure 30. Example questions included in the initial item pool.

Experimental procedure. Participants took part in this study via online participation. All data for this experiment was collected through an online tool (SurveyMonkey.com). Following informed consent, participants completed the 156 initial trust items. All items were randomized for each participant. Participants then completed the demographics questionnaire. The study took approximately 30 minutes to complete.

Experimental Results

Paired samples t–tests. Paired samples t-tests were conducted on each of the paired items to determine if they were interchangeable, thus reducing the item pool. Data were analyzed using IBM SPSS Statistics v.19 (SPSS, 2010), with an alpha level set to .05, unless otherwise indicated. Means, standard deviations, and paired-samples t-statistics are reported in

Table 63 located in Appendix N. The following 39 paired items were rated to be insignificantly different.

 Most robots are supportive/unsupportive.  Most robots instill fear/do not instill fear in people.  Most robots are successful/unsuccessful when performing tasks.  Most robots meet/do not meet user expectations.  Most robots are easy/hard to maintain.  Most robots work best alone/with a team.  Most robots complete simple/complex tasks.  Most robots are precise/not precise in their actions.  Most robots perform/do not perform accurately.  Most robots keep/do not keep classified information secure.  Most robots possess adequate/inadequate decision-making capability.  I would/would not feel comfortable giving a robot complete responsibility for the completion of a mission  I would/would not feel comfortable assigning a robot to a task or problem that was critical to the success of a mission.  I feel/do not feel comfortable when a robot has to make decisions which will affect me personally  I feel/do not feel comfortable when a robot has to make decisions which will affect me personally  I am/am not comfortable with the idea of working with a robot.  I know/do not know when a robot tries to communicate.  Most robots are responsive/unresponsive.  Most robots move quickly/slowly.  Most robots are mobile/immobile.  Most robots work/do not work in close proximity with people.  Most robots are easily/not easily led astray by unexpected changes in the environment or task.  Most robots provide/do not provide feedback.  Most robots communicate/do not communicate with people.  Most robots openly/do not openly communicate with users or operators.  Most robots often/rarely communicate with people.  Most robots understand/do not understand commands.  Most robots provide appropriate/inappropriate information  Most robots are intelligent/unintelligent.  Most robots tell/do not tell the truth.  Most robots communicate all/only partial information.

 Most robots provide/do not provide credible solutions to a problem.  Most robots require/do not require training to use or operate.  Most robots can/cannot perform a task better than a novice human user.  Most robots have/do not have a relationship with their human users or operators.  It is easy/difficult to determine the use or function of most robots from the robot's appearance.  Most robots perform/do not perform as intended.  Most robots are considered part of/separate from the team.  Experience with one robot can/cannot be generalized to other robots.

COMPLETE FACTOR ANALYSIS. A PRINCIPAL COMPONENTS ANALYSIS WAS ALSO PERFORMED ON THE

156 INITIAL TRUST ITEMS. OVERALL, N = 159 VALID RESPONDENTS WERE ANALYZED. MEANS, STANDARD

DEVIATIONS, AND CORRELATIONS ARE REPORTED IN Table 63 LOCATED IN APPENDIX. N. PEARSON

CORRELATION ANALYSIS WAS CONDUCTED FOR EACH OF THE PAIRED ITEMS (E.G., FRIENDLY/UNFRIENDLY).

SIGNIFICANT CORRELATIONS WERE FOUND FOR 66 PAIRED ITEMS. NORMALITY WAS ALSO ASSESSED.

SKEWNESS AND KURTOSIS RATINGS ARE REPORTED IN Table 64

TABLE 64 located in Appendix N. The ses= .192. Z-scores were calculated with formula z = (S-0)/ses. The sek = .383. Z-scores were calculated with the formula z = (K-0)/sek. Sixty-two items were calculated to have a significant skew, and 20 items were calculated to have potential issues with kurtosis.

Extraction was used to identify 43 components (using the Kaiser Criterion of Eigenvalue

> 1 for truncation), accounting for 79.63% of the variance. Following review of the scree plot, four components were retained. The un-rotated solution was subject to an orthogonal varimax rotation suppressed below |.30|. Table 21 shows the resulting rotated components, their respective amount of variance, and the coefficients for each variable across the four components.

In the rotated model, the four components accounted for 30.64% of the variance.

Table 21

Factor Analysis Rotated Components

Trust Items Component 1 2 3 4 Most robots function successfully. .731 Most robots do exactly as instructed to do. .730 Most robots act consistently. .717 Most robots perform as intended. .715 Most robots do not perform as intended (R) .687 Most robots are precise in their actions. .686 Most robots are successful when performing tasks. .679 Most robots do not perform accurately (R) .675 Most robots perform a specific function. .665 -.349 Most robots are unsuccessful when performing tasks (R) .661 Most robots provide appropriate information. .658 Most robots do not keep classified information secure (R) .649 Most robots can be relied on to complete a task. .649 Most robots perform accurately. .643 Most robots are unpleasant (R) .639 Most robots meet the needs of the mission. .634 Most robots follow specific procedures. .629 Most robots do approximately as instructed (R) -.622 Most robots do not meet the needs of the mission (R) .622 Most robots act inconsistently (R) .615 Most robots follow standard task-related protocols. .615 Most robots do not meet the user or operator's expectations (R) .606 Most robots are not precise in their actions (R) .599 Most robots meet the user or operator's expectations. .598 Most robots follow general procedures. .586 Most robots provide inappropriate information (R) .579 I like most robots. .575 .350 Most robots are valued by their users or operators. .570 Most robots harm people. (R) .569 Most robots are unkind. (R) .553 Most robots are qualified to perform a specific task. .552 Most robots do not tell the truth. (R) .551 Most robots keep classified information secure. .538 I dislike most robots. (R) .531 Most robots are incompetent. (R) .526 Most robots are mechanical. (R) -.520 Most robots make foolish decisions. (R) .506 Most robots are irresponsible. (R) .497 Most robots are offensive. (R) .494 Most robots cannot perform a task better than a novice human user. (R) .478 Most robots are unfriendly. (R) .478 Most robots do not understand commands. (R) .473 Most robots have high error rates. (R) .471 .319 Most robots are not supportive. (R) .466 Most robots are artificial. (R) -.464 Most people are not comfortable with the idea of working with a .459 .330 robot.(R)

Trust Items Component 1 2 3 4 Most robots are competent. .448 Most robots can perform a task better than a novice human user. .445 Most robots are ignorant. (R) .437 Responsibility of the robot's action falls to the robot. -.432 Most robots are built for long term use. .430 Most robots provide credible solutions to a problem. .428 .390 Most robots are supportive to people. .424 .398 Most robots are not qualified to perform a specific task. (R) .420 Most robots do not provide feedback. (R) .417 Most robots tell the truth. .404 Most robots are not valued by their users or operators. (R) .402 .346 Most robots have faces. -.401 .343 Most robots are machine-like. (R) -.398 .342 Most robots do not provide credible solutions to a problem. (R) .387 Most robots do not have faces. (R) -.381 Responsibility of the robot's action falls to the human. .378 -.319 Most robots cannot be relied on to complete a task. (R) .371 .362 Most robots appear to be organic. -.370 Experience with one robot cannot be generalized to other robots. (R) -.362 Most robots perform tasks that are typically carried out by people. .351 Most robots fail to warn people of potential risks in the environment.(R) .350 Most robots do not work in close proximity. (R) .347 .347 Most robots communicate all information. .326 Most robots instill fear in people. (R) .320 Most robots are controlled by people. .313 Most robots complete complex tasks. .310 Experience with one robot can be generalized to other robots. Most robots complete simple tasks. Most robots do not instill fear in people. Most robots communicate only partial information. (R) Most robots are autonomous. (R) Most robots are not easily led astray by unexpected changes in the environment or task. Most robots are caring towards people. .669 Most robots possess adequate decision-making capability. .592 Most robots are friendly towards people. .577 Most robots are kind towards people. .559 Most robots do not communicate with people. (R) .551 Most robots often communicate with people. .548 Most robots communicate with people. .538 Most robots are pleasant towards people. .535 Most robots will act as part of the team. .533 .315 Most robots protect people. .339 .520 Most robots are considered part of the team. .492 Most robots appear to be conscious. .479 Most robots know the difference between friend and foe. .466 Most robots have a relationship with their human users or operators. .455 I know when a robot tries to communicate. .447 -.314 Most robots warn people of potential risks in the environment. .441 Most robots understand commands. .406 .439 Most robots work best with a team. .434

Trust Items Component 1 2 3 4 Most robots openly communicate with users or operators. .414 Most robots appear to be lifelike. .411 Most robots appear to be alive. .408 Most robots are attractive to people. .403 Most robots are intelligent. .399 Most robots are responsible. .394 Most robots operate in an integrated team environment. .394 Most robots do not openly communicate with users or operators. (R) .387 Most robots make sensible decisions. .384 Most robots will not act as part of the team. (R) .384 Most robots are knowledgeable. .369 Most robots provide feedback. .333 .366 Most robots rarely communicate with people. .350 Most robots are considered separate from the team. (R) .343 .336 .310 Most robots do not have a relationship with their human users or .317 operators. (R) Most robots perform many functions. .305 Most robots appear to be human-like. .304 I do not feel comfortable when a robot has to make decision which will .678 affect me personally. (R) I do not feel comfortable giving a robot complete responsibility for the .654 completion of a mission. (R) I would feel the need to monitor a robot during a mission. -.590 I do not feel comfortable assigning a robot to a task or problem that was .557 critical to mission success. (R) I would feel comfortable assigning a robot to a task or problem that was .539 critical to the success of a mission. I would feel comfortable giving a robot complete responsibility for the .527 completion of a mission. I feel comfortable when a robot has to make decisions which will affect .494 me personally. Most robots are hard to maintain. (R) .486 Most robots are fake. (R) .475 Most robots make poor teammates. (R) .311 .461 Most robots make good teammates. .454 .455 Most robots require frequent maintenance. (R) .450 I am comfortable with the idea of working with a robot. .322 .340 .448 Most robots have low error rates. .434 .445 Most robots move rigidly. (R) .439 Most robots malfunction. (R) .404 Most robots are easy to maintain. .322 .385 Most robots require training to use or operate. (R) .374 Most robots do not require training to use or operate. .372 -.318 Most robots require infrequent maintenance. .364 Most robots appear to be natural. .361 I do not feel the need to monitor a robot during a mission. (R) -.355 I do not know when a robot tries to communicate. (R) .313 Most robots are unconscious. (R) .311 Most robots move elegantly. Most robots do not know the difference between friend and foe. (R) Most robots are built to be replaced. (R)

Trust Items Component 1 2 3 4 Most robots move quickly. Most robots operate in a solitary capacity. (R) Most robots are easily led astray by unexpected changes in the environment or task. (R) Most robots are mobile. .447 Most robots are apathetic. (R) .416 Most robots are immobile. (R) .415 It is easy to determine the use or function of most robots from the -.412 robot's appearance. Most robots are unresponsive. (R) .393 Most robots work best alone. (R) .372 It is difficult to determine the use or function of most robots from the -.371 robot's appearance. (R) Most robots appear to be dead. (R) .351 Most robots possess inadequate decision-making capability. (R) .332 Most robots work in close proximity with people. .323 Most robots are responsive. .303 .307 Most robots are unintelligent. (R) .300 .301 Most robots move slowly.

In looking at the loadings in the Rotated Component Matrix, twenty-two items with high loadings (>.60) were located in component one. In addition, 17 items were not included on any of the four components (see below). Of these, five items were previously identified as having issues with normality. They are marked with a + sign.

 Most robots make sensible decisions.  Most robots do not openly communicate with users or operators. (R)  Most robots openly communicate with users or operators.  Most robots complete simple tasks.  Most robots are autonomous. (R)  Experience with one robot can be generalized to other robots.  Most robots communicate only partial information.  Most robots are easily led astray by unexpected changes in the environment or task. (R)  Most robots are not easily led astray by unexpected changes in the environment or task.  Most robots move slowly.  Most robots operate in a solitary capacity. (R)  Most robots move quickly. (R)

 Most robots are knowledgeable. +  Most robots are built to be replaced. (R) +  Most robots work in close proximity with people. +  I do not feel the need to monitor a robot during a mission. (R) +  Most robots possess inadequate decision-making capability. (R) +

An additional 10 items were negative. Of these, six items were previously identified as having issues with normality. They are marked with a + sign.

 Most robots do approximately as instructed. (R) +  Experience with one robot cannot be generalized to other robots. (R) +  Most robots are controlled by people. +  Responsibility of the robot’s action falls to the human. (R) +  Most robots require training to use or operate. (R) +  I would not feel comfortable assigning a robot to a task or problem that was critical to mission success. (R)  Most robots require frequent maintenance. (R)  Most robots do not require training to use or operate. +  It is easy to determine the use or function of most robots from the robot’s appearance.  It is difficult to determine the use or function of most robots from the robot’s appearance. (R)

Based on the loadings of trust items on each of the four components, interpretations can be made about the factors themselves. Component one seems to represent performance-based functional capabilities of the robot. Component two seems to represents robot behaviors, including communication. Component three may represent task or mission specific items.

Finally, component four seems to represent feature-based descriptors of robots. These components supported the theory addressed by the theoretical three factor model of trust.

Impact of mental models on factor analyses. Three additional exploratory factor analyses were performed on initial trust items, sorted by humanlike, machinelike, and varied mental models (see Table 22). Data were analyzed using IBM SPSS Statistics v.19 (SPSS, 2010), with an alpha level set to .05, unless otherwise indicated. Rotated component matrices are reported in Table 65 in Appendix N.

Table 22

Impact of Mental Models on Factor Analyses

ance

% % Variance Rotation % Vari # items in Component 1 # items in Component 2 # items in Component 3 # items in Component 4 # items in Component 5 # items removed # negative items removed Mental Model N Extraction Human-like 49 36 95.1% 4 39.1% 66 59 32 28 -- 21 10 Machine-like 121 42 83.7% 4 32.6% 70 40 35 -- -- 37 12 Varied 28 27 100.0% 5 48.6% 61 39 38 26 31 52 15 Total 159 43 79.6% 4 30.6% 70 43 27 13 -- 27 10

Results demonstrated that variations between factors exist across mental model classification. The greatest amount of variance (48.6%) was accounted for in the group of participants with a varied mental model (i.e., perception that robots can be multiple shapes, sizes, forms, and functionality). However with such a small sample size (N=28), results may not be supported. Similarly, findings for the group with a human-like mental model is limited by sample size (N=49).

Human-like mental model analysis. Based on the loadings of trust items on each of the four, interpretations can be made about the factors themselves. Component one seems to

represent team-related performance. Component two seems to represents comfort-related robot features. Component three may represent the functional capabilities of the robot. Finally, component four seems to represent task-related robot capabilities.

Machine-like mental model analysis. Based on the loadings of trust items on each of the three components, interpretations can be made about the factors themselves. Component one seems to represent task-related robot capabilities. Component two seems to represents comfort- related functional capabilities. Component three may represent the positive items that make a good team member.

Varied mental model analysis. Based on the loadings of trust items on each of the five components, interpretations can be made about the factors themselves. Component one seems to represent comfort-related functional capabilities of the robot. Component two seems to represents task-related robot responsibilities. Component three may represent the features and capabilities that make a robot a good team member. Component four may represent team-related capabilities. Finally, component five may represent items that are robot attributes.

Discussion

Following an internal review of the initial 172 item pool, a large sample reliability assessment was conducted on the remaining 156 items with the purpose of reducing the item pool. Following factor analyses and paired samples t-tests, the trust item pool was reduced from

172 items to 73 items. The following items were retained for SMEs for further item analysis due to their importance in theory:

 Move quickly  Move slowly  Require frequent maintenance  Autonomous  Led astray by unexpected changes in the environment  Work in close proximity with people  Possess adequate decision-making capability  Make sensible decisions  Openly communicate  Communicate only partial information

Some potential issues arose with the wording of the items. Two main types of item formation were included in the above version of the scale. Items either began with “most robots” or “I.”

This may have impacted the factor creation. Therefore, all items were reduced to a single word or short phrase prior to SME review (see Study 4).

In addition, trust has been most commonly measured on a continuum ranging from distrust to trust. However, more recently in the interpersonal and e-commerce domains, trust and distrust are viewed as separate constructs with differing effects on behavior, consequences and outcomes (Lewicki, McAllister, & Bies, 1998; McKnight, Kacmar, & Choudhury, 2004;

Wildman, Fiore, Burke, & Salas, 2011; Wildman, 2011). Lewicki, Tomlinson, and Gillespie

(2006) reviewed both types of trust scales and suggested that when measuring a multi-faceted trust model, an increase in trust does not necessarily lead to a direct decrease in distrust. While the patterns for trust and distrust were negatively related, the patterns were not perfectly inversely correlated, and should initially be viewed as separate constructs (Wildman, 2011). Due to the multi-faceted nature of human-robot trust, this scale was updated to measure trust as an

independent construct ranging from no trust to complete trust. The scale was modified from a 7- point Likert scale to a percentage scale with 10% increments.

Progression to Next Experiment

A three step validation process is discussed in the following three experimental chapters.

The first of the three validation experiments was constructed to assess content validity (Lawshe,

1975). Subject Matter Experts (SMEs) in the area of trust and robotics were used for two phases of semantic analysis: item relevance (content validity), and identification of hypothetical range of differences (between high and low trust group differences) of the current trust scale.

CHAPTER SIX: STUDY 4

“‘Reverse primary thrust, Marvin.’ That's what they say to me. ‘Open airlock number 3, Marvin.’ ‘Marvin, can you pick up that piece of paper?’ Here I am, brain the size of a planet, and they ask me to pick up a piece of paper.” Marvin, Hitchhikers Guide to the Galaxy (Adams, 1981)

The purpose of the current experiment was to (1) reduce the item pool through content validation, and (2) provide theoretical end points to the items. This experiment was designed to will help identify irrelevant items and potential missing items required for the completion of the trust measurement scale.

Experimental Method

Subject matter experts. Eleven SMEs were included from the United States Army

Research Laboratory, United States Air Force Sensemaking and Organizational Effectiveness

Branch, and faculty members from university research laboratories. All SMEs were considered experts in the field of trust research, robotics research, and/or human-robot interaction. Table 23 provides the SME’s years of experience across a variety of robot, automation, and research topics.

Table 23

Subject Matter Experts Years of Experience

SME Robot Robot Robot HRI Automation Automation Trust Other Design Operator Research Design Research Research Research 1 0 0 8 8 0 8 0 8 2 5 4 3 0 0 0 0 0 3 4 0 4 0 2 2 0 0 4 7 0 7 0 0 0 0 0 5 4 8 8 8 0 0 0 0 6 0 0 0 0 0 7 7 0 7 11 0 11 11 0 0 3 15 8 0 0 0 8 0 8 6 0 9 0 0 4 0 0 4 0 0 10 7 0 7 0 0 0 0 0 11 0 0 10 10 20 30 15 0 Combined 38 yrs 12 yrs 62 yrs 45 yrs 22 yrs 59 yrs 31 yrs 23 yrs Years Note. All results are reported in years of experience

Materials. Materials included an expertise questionnaire (Appendix O), the 74 item Trust

Scale (Appendix P), and a Content Validation questionnaire. The Trust Scale was updated with an 11 point scale ranging from 0% to 100%, in increments of 10%. The Content Validation

Questionnaire was a 3 point Likert scale based on Lawshe (1975) content analysis protocols.

Experimental procedure. SMEs were contacted via email. Upon agreement to participate, they were provided a link to complete an online survey. All data for this experiment were collected through an online tool (SurveyMonkey.com). SMEs were provided background information, purpose, and a brief theoretical foundation prior to beginning the study. In Section

1, SMEs completed the expertise questionnaire. In Section 2, SMEs were given instructions

“Please rate the following items on how a person with little to no trust in a robot would rate them.” Following completion of Section 2, SMEs were given instructions to complete Section 3

trust scale. “Please rate the following items on how a person with complete trust in a robot would rate them.” All items in both Section 2 and Section 3 were randomized. Section 4 included item analysis. SMEs rated each item on a 3 point Likert scale as either ‘extremely important to include in scale,’ ‘important to include in scale,’ or ‘should not be included in scale.’ SMEs could also mark if they felt an item was domain specific. A comment box was also available to provide any clarification about why they rated the item a specific way, provide additional recommendations, or suggest items that may be missing from the scale. Content analysis took approximately 30 minutes to complete.

Experimental Results

Items were analyzed using the Content Validity Ratio developed by Lawshe (1975). The

Content Validity Ration (CVR), depicted in Equation 2, is a commonly used method of analyzing scale items, and has been used previously in robot team trust scale development (see

Yagoda, 2011). The CVR equation was derived from a 3 point Likert scale (1= ‘Should not be included in scale,’ 2 = ‘Might be important to include in scale’ and 3 = ‘Extremely important to include in scale’).

CVR = (ne – N/2) / (N/2) (2)

CVR = content validity ratio ne = number of SMEs indicating that an item is “extremely important to include in the scale” N = total number of SMEs

Lawshe suggested that 11 subject matter experts with a criterion set to .59 are needed to ensure that the SME agreement is unlikely to be due to chance. The formula yielded values ranging from +1 to -1; positive values indicated that at least half of the SMEs rated the item as

‘Extremely Important.’ Table 24 reports the CVR values for the 14 items recommended by the

SMEs.

Table 24

CVR Values for the 14 Items Recommended by SMEs

Item CVR Function successfully 1.00 Act consistently 1.00 Reliable 1.00 Predictable 1.00 Dependable 1.00 Follow directions 0.82 Meet the needs of the mission 0.82 Perform exactly as instructed 0.82 Have errorsR 0.82 Provide appropriate information 0.82 MalfunctionR 0.64 Communicate with people 0.64 Provide feedback 0.64 UnresponsiveR 0.64 Note. CVR > .59 R represents reverse coded items

CVR values were also calculated for the items that were rated as ‘Important to include in the scale.’ This resulted in 37 additional items to consider for inclusion in the finalized scale. The scores from the hypothetical range of differences were used to further evaluate these 37 items.

The hypothetical range of differences was assessed from the SMEs completion of the full scale two times comparing ratings for the little to no trust and complete trust in a robot. Paired samples t-tests were conducted to identify the hypothetical range of differences between ‘little to no trust’

100

and ‘complete trust’ in a robot. Twenty-four items were retained for further scale validation (see

Table 25).

A total of thirteen items were removed from the scale. Of those 13 items, eight items were found to have non-significant findings from the paired samples t-tests. SMEs comments assisted in the removal of the remaining five items. The first comment was a general one stating that some items (e.g., easy/difficult to maintain) were repetitive in nature. To address this comment, only one of the repetitive items was included in the revised scale. ‘Responsive,’ ‘easy to maintain,’ and ‘poor teammate’ were removed from the scale. In addition, ‘instill fear in people’ was removed due to its nature of distrust more than trust. Finally, the item ‘likeable’ was considered to be too general an item for the scale.

An additional comment suggested that two items ‘are given complete responsibility for the completion of the mission,’ and ‘are assigned tasks that are critical to mission success’ represented situational factors that may be a separate issue from trust. Even though CVR analysis revealed that SMEs felt that they might be important to include in the scale, their responses to the theoretical ends did not show a significant change. This added support for the SME’s comments recommending removal of these two items.

101

Table 25

CVR Values and Hypothetical Ranges for the 37 Important Items Recommended by SMEs

Complete Trust No Trust Items CVR Min Max Mean SD Min Max Mean SD t df p Operate in an integrated team 1.00 50 100 75.00 21.73 10 80 28.00 25.73 3.62 9 .006 environment Autonomous 1.00 30 100 69.09 20.23 20 90 38.89 23.15 3.18 8 .013 Good teammate 0.82 60 100 87.27 11.04 0 40 11.82 11.68 16.60 10 <.001 Perform a task better than a 0.82 30 100 69.09 20.71 0 70 24.55 24.23 6.69 10 <.001 novice human user Led astray by unexpected 0.82 40 90 71.00 19.12 10 50 26.00 15.78 5.78 9 <.001 changes in the environment Know the difference between 0.82 20 100 71.00 27.67 0 60 14.55 16.95 5.89 9 <.001 friend and foe Make sensible decisions 0.82 60 100 84.00 11.74 10 60 21.00 20.79 8.62 9 <.001 Clearly communicate 0.82 70 100 83.00 11.60 10 60 19.00 15.24 12.30 9 <.001 Warn people of potential risks in 0.82 70 100 83.00 11.60 10 60 23.00 18.29 7.75 9 <.001 the environment Incompetent 0.82 50 100 85.45 14.40 0 90 39.00 33.15 5.24 9 .001 Possess adequate decision- 0.82 50 90 71.11 16.91 0 60 20.00 21.60 5.57 8 .001 making capability Are considered part of the team 0.82 50 90 79.00 12.87 0 80 35.00 31.36 3.36 8 .010 Will act as part of the team 0.82 50 100 74.00 19.55 10 100 30.00 29.06 3.28 9 .010 Perform many functions at one 0.82 20 100 68.00 23.48 10 80 36.00 25.47 4.40 9 .002 time Protect people 0.82 20 100 77.00 24.52 0 80 23.00 22.14 3.92 9 .003 Openly communicate 0.82 50 100 80.00 15.81 10 80 34.00 28.36 3.79 8 .005 Responsible 0.82 10 100 66.36 30.42 0 100 27.27 34.67 2.76 10 .020 Built to last 0.82 Work in close proximity with 0.82 40 100 65.00 19.58 10 90 35.56 26.03 2.34 8 .047 people Supportive 0.64 40 90 66.00 16.47 0 40 18.00 11.35 8.67 9 <.001 Work best with a team 0.64 50 90 71.00 18.53 10 90 34.00 28.36 3.41 9 .008 Tell the truth 0.64 50 100 86.36 21.57 10 100 46.00 31.69 2.90 9 .018 Keep classified information 0.64 50 100 84.00 15.06 10 100 55.45 36.43 2.69 9 .025 secure Require frequent maintenance 0.64 30 90 74.00 20.66 10 90 44.00 28.75 2.37 9 .042

102

Complete Trust No Trust Items CVR Min Max Mean SD Min Max Mean SD t df p Items Removed from Scale *Responsive 1.00 70 100 84.00 9.66 10 70 32.22 21.67 7.50 8 <.001 *Poor teammate 0.82 50 100 84.55 14.40 0 100 45.45 36.43 3.64 10 .005 *Are assigned tasks that are 0.82 0 100 58.00 32.25 0 100 25.00 35.67 1.84 9 .098 critical to mission success *Communicate only partial 0.82 10 90 53.00 28.30 10 90 35.00 28.77 1.03 9 .331 information *Instill fear in people 0.73 50 100 83.64 18.59 0 90 50.91 34.48 3.13 10 .011 *Likeable 0.64 10 90 60.00 33.54 0 30 14.00 10.75 4.61 8 .001 *Easy to maintain 0.64 30 90 69.00 20.79 0 100 35.00 31.71 2.89 9 .018 *Responsible for its own actions 0.64 0 100 64.00 32.04 0 100 32.73 36.63 2.22 9 .054 *Given complete responsibility 0.64 0 100 59.00 37.55 0 100 19.00 37.84 2.16 9 .059 for the completion of a mission *Monitored during a mission 0.64 0 90 49.00 29.61 0 90 20.00 28.67 2.08 9 .067 *Are considered separate from 0.64 20 90 68.00 24.40 0 90 36.00 32.39 1.92 8 .091 the team *Difficult to maintain 0.64 50 90 70.00 15.63 0 100 50.00 34.64 1.60 9 .143 *Work best alone 0.64 10 80 48.00 23.94 10 90 41.00 33.48 0.69 9 .506 Note. CVR > .59 * represents items that were removed from the scale

103

The 22 items that did not meet the CVR criterion are as follows: move quickly, move slowly, mobile, move rigidly, friendly, kind, pleasant, conscious, lifelike, attractive, caring, human-like, fake, alive, dead, offensive, organic, have a face, ignorant, apathetic, have a relationship with their human users or operators, and make decisions that affect me personally.

Out of those 22 items that did not meet the CVR criterion, a number of SMEs felt that six of the items may be domain specific items and should not be discounted (see also Table 26). Out of those six items, four were retained in the scale. SME’s comments suggested that the items that represent robot personality were domain specific and may only matter in specific types of robots, such as social robots. Therefore, ‘conscious,’ and ‘lifelike’ were retained for further validation.

Table 26

Domain Specific Items

# of Complete Trust No Trust Items SMEs Min Max Mean SD Min Max Mean SD t df p Friendly 7 0 90 48.89 31.00 0 50 11.00 19.12 5.68 8 <.001 Pleasant 6 0 100 63.33 33.91 0 50 16.00 15.06 4.98 8 .001 Conscious 6 0 90 34.44 33.58 0 100 20.00 33.67 1.01 8 .342 Lifelike 7 0 90 48.00 33.93 0 100 35.56 33.58 0.60 8 .566 *Have a face 8 0 70 37.78 28.63 0 50 16.67 15.00 2.30 8 .051 *Attractive 6 0 100 46.67 32.79 0 50 23.33 19.36 2.06 8 .073 Note. * represents items that were removed from the scale

In an additional comment, one SME made recommendations about the robot capability items (e.g., movement) suggesting that trust and speed are orthogonal. Therefore, the items

‘move quickly,’ ‘move slowly,’ ‘mobile,’ and ‘move rigidly’ were removed from the item pool.

The remaining items with low CVR values that were removed from the scale are reported in

Table 27. In addition to SME recommendations to remove these items from the scale, nine of

104

these items demonstrated non-significant findings from the SMEs ratings for the theoretical range of responses. The item ‘fake’ was also removed due to SME recommendation.

Table 27

Hypothetical Ranges for the SME Recommended Items to Remove from Scale

Complete Trust No Trust Item Min Max Mean SD Min Max Mean SD t df p Fake 10 100 85.00 27.59 0 100 51.00 40.67 2.92 9 .017 Mobile 50 100 71.11 19.00 10 90 47.78 27.28 2.14 8 .065 Alive 0 100 45.00 32.40 0 100 18.00 36.15 2.05 9 .071 Offensive 50 100 84.44 20.68 40 100 74.44 26.03 1.66 8 .135 Dead 20 100 75.56 28.77 0 100 46.00 45.51 1.55 8 .159 Apathetic 30 100 71.11 26.19 0 100 74.00 35.02 -1.51 8 .169 Make decisions that 0 90 53.00 29.08 0 100 37.00 33.35 1.13 9 .288 affect me personally Organic 10 70 28.00 24.40 0 90 18.00 27.81 0.84 9 .423 Ignorant 10 90 61.00 28.07 10 90 48.89 31.80 0.78 8 .457 Human-like 0 90 41.11 32.96 0 100 31.11 34.08 0.59 8 .571

Discussion

Semantic analysis of the trust scale items reduced the scale from 74 items to 42 items. It further identified 14 items that were extremely important to trust scale measurement, with an additional 24 items that could be important to trust scale measurement. Four domain specific items (friendly, pleasant, conscious, and lifelike) were also retained on the scale based on SME recommendation. No additional items or item revisions were recommended. In addition, no changes to the scale design were recommended.

Four SME comments were made regarding specific items in the scale. The first comment was a general one stating that some items (e.g., easy/difficult to maintain) were repetitive in nature. To address this comment, three items were removed from the scale. The second comment

105

suggested that two items ‘are given complete responsibility for the completion of the mission,’ and ‘are assigned tasks that are critical to mission success’ represented situational factors that may be a separate issue from trust. Both items were removed from the item pool. In a third comment, one SME made recommendations about the robot capability items (e.g., movement) suggesting that trust and speed are orthogonal. Items ‘move quickly,’ ‘move slowly,’ and ‘move rigidly’ were removed from the item pool.

The final comment was specific to the four items that directly mentioned communication:

‘communicate with people,’ ‘openly communicate with users/operators,’ and ‘clearly communicate.’ This SME commented that “it [was] unclear what ‘communication’ might entail.

I rated it as though it meant spoken or written language. I [assumed] that robots communicate in other ways that might be important, but it was unclear whether those should be included here.”

To address this comment, robots communicate in multiple ways (e.g., verbal, written, tactile, auditory, etc.) depending on their programmed capabilities and domain of use. These were left broad to allow for the mode of communication to be defined by the specific robot.

Progression to Next Experiment

The previous experiment was the first in three validation experiments. The item pool was reduced from 74 items to 42 items. The following experiment was designed to assess the 42 item trust scale’s capability to measure changes in perceived trust. A HRI monitoring task using a

“cover the back door” scenario was utilized.

106

CHAPTER SEVEN: STUDY 5

“Danger, Will Robinson! Danger!” The Robot, Lost in Space (Koch & Hopkins, 1998)

The purpose of this validation experiment was to determine if the updated 42 item trust scale measured changes in trust over time. Computer-based simulation was used to develop a monitoring task specific to a “cover the back door” scenario for a human-robot team. Trust using the 42 item scale was measured pre-interaction, post-interaction of the high trust condition, and post-interaction of the low trust condition. Within the high trust condition, the Talon robot provided 100% reliable feedback on target detection, while within the low trust condition, the

Talon robot only provided 25% reliable feedback.

Hypothesis 1: There are mean differences in trust that occur over time, with respect to changes in robot reliability. More specifically, trust will increase from pre-interaction to a 100% reliable interaction, and trust will decrease from a 100% reliable interaction to a 25% reliable interaction.

Time1 < Time2 > Time3

Hypothesis 2: There is a positive correlation between Rotter’s Interpersonal Trust Scale (ITS) and the developed Pre-Interaction Trust scale. rITS*pre-interaction trust= +correlation

107

Experimental Method

Participants. Participants were 81 undergraduate students (25 males, 56 females;

Mage=22.57 years, SDage=3.95) from an undergraduate psychology course in Science and

Pseudoscience at UCF. Extra credit was provided by the Professor of the class. Their participation accorded with all regulations from the University’s Institutional Review Board

(IRB).

Participants had a varied background with respect to robots familiarity. All participants previously watched movies or television shows that incorporated robots (N 1-5 movies = 30, N 6-10 movies = 9, N over 10 = 42). Forty-four participants previously interacted with robots, ranging from the Roomba vacuum cleaner to an EOD bomb disposal robot. These participants reported previously controlling robots through a variety of modalities (N voice = 11, N game controller = 24, N gesture/picture = 3, and N RC Controller = 35). Five participants previously built robots for class-based projects.

Trust measurement. Trust has typically been measured after an interaction had occurred. However, trust is dynamic in nature, as ongoing interactions and relational history continuously influence trust levels at any given point in time. Consequently, trust before, during, and after an interaction may not be identical, and past and future trust in the same partner will likely change over time as that relationship progresses (Bloomqvist, 1997). Obtaining the most reliable and accurate reflection of the changing nature of trust in an interaction may necessitate measuring trust multiple times. Within the human-robot domain, trust measurement often only

108

occurs following the completion of a task or interaction, which limits the interpretation about the factors that then influence human-robot trust over time.

Research in the area of automation (see Merritt & Ilgen, 2008) as well as interpersonal trust (see McAllister, 1995) have recommended that trust should be measured at multiple times, specifically before and after the task or interaction. Within the triadic model of human-robot trust, pre-interaction measurement has been used to identify initial trust perceptions of HRI based that are influenced by human traits, robot features, and the individual’s perception of the environment and perceived robot capabilities. Post-interaction measurement was used to identify changes in human states and trust perceptions following interaction.

Materials. All materials were administered through paper and pencil versions. The

NARS and demographics questionnaire were included to identify potential individual difference ratings. Trust was measured through Rotter’s (1967) Interpersonal Trust Scale, as well as the 42 item trust scale (administered pre-post-post task, see Appendix R). The Interpersonal Trust Scale

(ITS) is a well-supported, validated scale that assesses cultural and societal impact on trust.

Person state was not assessed due to the nature of the task (monitoring only).

Computer-based simulation. RIVET (Robotic Interactive Visualization & Exploration

Technology) was developed by General Dynamics Robotic Systems to create a computer-based simulation system specific to human-robot interaction (Gonzalez et al, 2009). It uses an adapted

Torque Software Development Kit (SDK) development and runtime environment through a

109

Client/Server networking model. Development of the VE within RIVET is accomplished through a TorqueScript scripting language, similar to C++.

The virtual environments (VE) used a base environment developed by GDRS in collaboration with the Army Research Laboratory (ARL). This included layout of the physical environment (e.g., ground, roadways, buildings, and lighting) and terrorist non-player characters

(NPCs). Task-specific customization of the environment was accomplished through Scripting syntax. Specific customization included entering objects, obstacles, and creation of paths, to name a few.

Experimental scenarios. Two HRI scenarios were created using the RIVET computer- based simulation system. A standard Soldier-Talon robot team within a “cover the back door” scenario was used, as per common practice in the Robotics Collaborative Technical Alliance

(RCTA). A single video was created from the camera-view on the Talon robot which showed the

Talon robot navigating to the back of a building and monitoring the back door for human targets.

The video of the simulation was created using FRAPS real-time video capture and benchmarking program with a 30 frame rate/second .avi file. The .avi file was converted into the .mp4 file format to add auditory feedback. The robot stated “target detected” in a male computer synthetic voice. Video 1 provided 100% reliable detection condition in which the Talon robot provided auditory feedback for all eight human targets. Video 2 provided a 25% reliable detection condition. No false alarms were included.

110

Experimental procedure. Prior to the start of the study, participants were instructed on the research process (e.g., informed consent process, IRB, etc.) They were given an opportunity to ask questions. Following auditory informed consent, participants viewed an image of the

Talon robot and completed the 42 item trust scale (Time 1). Participants were instructed of the human-robot task they were to monitor. Following Video 1, participants completed the 42 item trust scale (Time 2). Participants were then instructed of the second task, monitored Video 2, and completed 42 item trust scale (Time 3). Following monitoring tasks, participants completed the

Interpersonal Trust Scale, NARS, and demographics questionnaire. An explanation of the study purpose was provided following the completion of experimentation. The study took approximately 90 minutes in its entirety.

Experimental Results

All data were analyzed using IBM SPSS Statistics v.19 (SPSS, 2010), with an alpha level set to .05, unless otherwise indicated.

Comparison of each trust scale item across time. Hypothesis 1 stated there would be mean differences in trust that occur over time. A one-way within subjects repeated measures analysis of variance was conducted for each of the 42 items of the trust scale with the factor being Time of Trust Scale measurement and the dependent variable being the individual item scores. The means, standard deviations, confidence intervals, and effects for Time are presented in Table 28.

111

Table 28

Means, Standard Deviations, and Confidence Intervals for each Trust Item at Time 1, Time 2, and Time 3

Trust Scale Item MTime1 SDTime1 CILow CIHigh MTime2 SDTime2 CILow CIHigh MTime3 SDTime3 CILow CIHigh ηp2 *Function successfully 49.75 24.95 44.24 55.27 85.19 17.47 81.32 89.05 26.30 26.53 20.43 32.16 .808 *Act consistently 34.81 28.11 28.60 41.03 88.64 16.03 85.10 92.19 26.79 27.92 20.62 32.96 .868 *Predictable 9.38 18.33 5.33 13.44 80.62 22.88 75.56 85.68 23.46 28.16 17.23 29.68 .861

*UnresponsiveR 35.06 29.59 28.52 41.60 90.25 15.65 86.79 93.71 33.83 28.75 27.47 40.18 .844 Pleasant 74.20 22.13 69.31 79.09 30.86 36.54 22.78 38.95 16.54 25.84 10.83 22.26 .784 Possess adequate decision-making 78.52 16.89 74.78 82.25 44.57 36.61 36.47 52.66 19.88 27.64 13.77 25.99 .760 capability *Reliable 65.68 25.49 60.04 71.32 79.38 21.47 74.64 84.13 22.84 25.51 17.20 28.48 .757 *Dependable 68.40 19.33 64.12 72.67 76.91 22.67 71.90 81.93 23.21 24.23 17.85 28.57 .757 Clearly communicate 45.06 24.70 39.60 50.52 88.15 18.04 84.16 92.14 68.40 32.31 61.25 75.54 .732 *MalfunctionR 59.26 22.96 54.18 64.34 87.04 17.35 83.20 90.87 35.80 30.98 28.95 42.65 .723 *Meet the needs of the 76.05 30.07 69.40 82.70 83.21 21.14 78.54 87.89 27.04 30.27 20.34 33.73 .706 mission *Perform exactly as 61.23 24.72 55.77 66.70 86.30 20.76 81.71 90.89 30.37 31.80 23.34 37.40 .698 instructed IncompetentR 59.14 23.99 53.83 64.44 83.21 19.48 78.90 87.52 31.36 30.32 24.65 38.06 .688 Responsible 60.99 22.34 56.05 65.93 67.04 33.60 59.61 74.47 20.25 26.12 14.47 26.02 .662 *Provide feedback 41.11 28.55 34.80 47.42 81.73 30.53 74.98 88.48 37.16 32.06 30.07 44.25 .653 *Follow directions 64.07 23.39 58.90 69.25 86.79 21.03 82.14 91.44 37.90 32.70 30.67 45.13 .651 Good teammate 44.96 30.09 38.31 51.62 68.27 28.32 62.01 74.53 22.84 25.01 17.31 28.37 .650 Warn people of potential 54.57 26.93 48.61 60.52 82.59 28.41 76.31 88.87 31.23 29.93 24.62 37.85 .643 risks in the environment *Provide appropriate 77.41 57.48 64.70 90.12 82.96 20.76 78.37 87.55 33.46 31.35 26.52 40.39 .635 information

*Have errorsR 58.77 26.71 52.86 64.67 76.91 27.14 70.91 82.92 26.91 27.14 20.91 32.92 .624 Tell the truth 76.06 30.05 69.42 82.71 90.49 18.23 86.46 94.52 42.84 37.76 34.49 51.19 .624 Considered part of team 33.13 27.63 26.98 39.27 72.87 32.34 65.68 80.07 50.00 38.84 41.36 58.64 .556 Supportive 42.00 25.97 36.22 47.78 62.25 35.29 54.40 70.10 22.75 28.42 16.43 29.07 .536

112

Trust Scale Item MTime1 SDTime1 CILow CIHigh MTime2 SDTime2 CILow CIHigh MTime3 SDTime3 CILow CIHigh ηp2 Protect people 52.96 25.02 47.43 58.50 46.42 40.78 37.40 55.44 20.25 25.88 14.52 25.97 .500 Make sensible decisions 34.44 27.06 28.46 40.43 57.41 35.91 49.47 65.35 19.26 24.12 13.93 24.59 .469 Openly communicate 41.13 26.62 35.20 47.05 76.12 31.28 69.16 83.09 53.75 37.67 45.37 62.13 .460 Act as part of a team 54.44 26.74 48.53 60.36 72.10 31.05 65.23 78.97 41.11 36.61 33.02 49.21 .458 Require frequent 45.06 28.55 38.75 51.36 78.02 28.35 71.76 84.29 64.57 32.17 57.45 71.68 .428 maintenanceR Perform a task better than a novice human 31.60 28.44 25.32 37.89 34.69 31.98 27.62 41.76 13.58 22.66 8.57 18.59 .384 user *Communicate with 68.77 22.21 63.85 73.68 70.37 37.60 62.06 78.68 44.69 37.05 36.50 52.88 .323 people Conscious 27.41 24.43 22.01 32.81 27.78 37.58 19.47 36.09 13.46 23.78 8.20 18.72 .300 Autonomous 59.82 28.94 53.42 66.22 62.96 35.58 55.10 70.83 45.19 37.62 36.87 53.50 .277 Work best with a team 49.25 30.64 42.43 56.07 56.88 36.02 48.86 64.89 34.88 36.25 26.81 42.94 .210 Led astray 73.70 18.87 69.53 77.88 79.75 26.55 73.88 85.64 64.44 35.43 56.61 72.78 .162 Lifelike 59.01 27.37 52.96 65.06 9.88 21.77 5.06 14.69 4.44 12.55 1.67 7.22 .796 Perform many functions 59.75 27.57 53.66 65.85 30.00 37.38 21.73 38.27 20.99 32.23 13.86 28.12 .454 at one time Friendly 51.48 24.75 46.01 56.96 22.72 33.50 15.31 30.12 19.01 31.17 12.12 25.90 .431 Know the difference 66.91 44.82 57.00 76.83 38.02 42.00 28.74 47.31 25.56 32.09 18.46 32.65 .351 between friend and foe Keep classified info 68.15 24.09 62.82 73.48 50.74 40.86 41.71 59.78 44.81 38.96 36.20 53.43 .242 secure Work in close proximity 65.19 20.68 60.61 69.76 79.75 29.58 73.21 86.29 66.54 34.03 59.02 74.07 .174 to people Operate in an integrated 66.13 17.97 62.13 70.12 57.25 38.81 48.61 65.89 42.13 35.21 34.29 49.96 .272 team environment Built to last 68.40 20.64 63.83 72.96 60.80 25.59 55.15 66.46 49.63 27.68 43.51 55.75 .266 Note. The * represents the 14 SME recommended trust scale items.

113

Thirty-three of the items showed a significant mean difference for the condition of Time.

Following review of the confidence intervals for ‘operate in an integrated team environment’ and

‘built to last’ showed no significant difference over time, and were thereby removed from the scale. The remaining six items were found to be not significant between the post-interaction times (Time 2 and Time 3). These items included lifelike, perform many functions at one time, friendly, know the difference between friend and foe, keep classified information secure, and work in close proximity to people. These results may have occurred due to a significant change in the mental model from pre-interaction trust to post-interaction trust, and were retained within the scale.

Comparison of the trust scale across time. To create a general score of trust, the 40 items were summed and divided by 40 to formulate a score between 0 and 100. To accurately compare the 40 item scale with the ITS, the scores for the ITS were transformed to a 0 to 100 scoring. Means, standard deviations, and correlations were reported in Table 29. In support of

Hypothesis 2, there was a significant positive correlation between the ITS and the developed trust scale at Time 1, r(81)= +.319, p=.002. In addition, a significant positive correlation was found between Time 1 and Time 2, r(81)= +.203, p=.035. However, there was not a significant correlation between Time 2 and Time 3 trust scales, r(81)= +.058, p=.303, despite them having the same 40 items.

One interesting finding not related to the hypotheses was found. In support of the proposed predictive model of trust, the scores of the three subscales from the NARS were correlated with the ITS (p=.039, p<.001, and p=.001) and the 40 item Trust Scale measured at

114

Table 29

Experiment 5 Means, Standard Deviations, and Correlations

M SD 1 2 3 4 5 6 7 8 9 10 11 12 13 1.Gender 0.69 .46 1 2.Age 22.57 3.95 -.108 1 3.# of robot 2.15 .94 -.496** .254* 1 movies watched 4.Previous robot 0.54 .50 -.505** .145 .465** 1 interaction 5.Previously 0.06 .24 -.273* -.076 .180 .235* 1 built a robot 6.Previously 0.54 .50 -.452** -.031 .385** .552** .235* 1 controlled a robot 7.NARS_S1 12.47 3.87 .450** .002 -.413** -.217 -.272* -.152 1 8.NARS_S2 15.53 3.86 .252* .082 -.174 -.106 -.181 -.099 .581** 1 9.NARS_S3 9.60 3.09 .227* .037 -.174 -.150 .000 -.182 .455** .449** 1 10.Interpersonal 47.56 7.94 -.268** -.014 .132 .131 .048 .187* -.197* -.400** -.210* 1 Trust scale 11.Trust 54.48 9.86 -.189* -.169 .236* .053 .116 .144 -.506** -.383** -.348** .319** 1 (Time 1) 12.Trust 66.68 12.62 -.121 -.189* -.214* -.134 .032 -.110 -.035 .024 .037 -.059 .203* 1 (Time 2) 13.Trust 30.87 22.38 -.177 .077 .044 .142 .094 .129 .163 .067 .142 .003 -.193* .058 1 (Time 3) Note. S1 represents negative attitudes toward situations of interaction with robots S2 represents negative attitudes negative attitudes toward social influence of robots S3 represents negative emotions in interactions with robots

115

Time 1 (p<.001, p<.001, and p=.001) prior to interaction. However no significant correlations were found between the NARS and Trust Scale at Time 2 or Time 3. In addition, prior experiences in terms of number of movies watched had a significant negative correlation with

NARS_S1 (negative attitude toward situations of interaction with robots) and trust at Time 2

(post-interaction 1). This suggested that increased number of robot movies watched is related to lower negative ratings on the NARS_S1, as well as lower post-interaction trust (100% reliable condition). This finding is supported by the literature that discusses the relationship between mental models and trust.

A one-way within-subjects repeated measures analysis of variance was conducted with the factor being Time of trust scale measurement and the dependent variable being the Trust

Score. The means and standard deviations for Time are presented in Table 30. The results for the

ANOVA indicated a significant effect of time, Wilks’ lambda = .249, F(2, 79) = 119.10, p<.001,

ηρ²=.75.

Table 30

Experiment 5 Repeated Measure Variable of Time

Variable Definition Mean SD CILow CIHigh Time 1 Pre-Interaction Trust 54.48 9.86 52.30 56.66 Time 2 Post-Interaction 1 Trust 66.68 12.62 63.89 69.47 (100% Reliable Condition) Time 3 Post-Interaction 2 Trust Scores 32.64 16.41 29.01 36.27 (25% Reliable Condition)

Follow-up polynomial contrast indicated a significant linear effect, F(1, 80) = 90.00, p<.001, ηρ² =.53, and a significant quadratic effect, F(1,80)=218.83, p<.001, ηρ² =.73. Post-hoc

116

analysis using the Fisher LSD revealed significant mean difference ratings for all three times the trust scale was administered. It revealed that trust was significantly greater in Time 2 (post- interaction 1 trust, 100% reliable condition) than Time 1 (pre-interaction trust) and Time 3 (post- interaction 2 trust, 25% reliable condition), thus supporting Hypothesis 1 in that MTime1 < MTime2

> MTime3. In addition, Trust scores at Time 1 were significantly greater than at Time 3.

40 item trust scale versus 14 SME recommended item scale. Additional analyses were conducted comparing the total 40 item trust scale with the 14 SME recommended items. Positive correlations between the two scales were found at Time 1, r(81)=.808, p<.001, Time 2, r(81)=.850, p<.001, and Time 3, r(81)=.925, p<.001. A 2 scale (40 item, 14 item) x 3 time (Time

1, Time 2, Time 3) repeated measures analysis of variance was conducted to compare the total 40 item trust scale with the 14 item SME recommended scale. Means, standard deviations and confidence intervals are reported in Table 31.

Table 31

Means, Standard Deviations and Confidence Intervals Comparing the 40 Item Scale to the 14 Item Scale

Time Scale Mean SD CILow CIHigh Time 1 40 item 54.48 9.86 51.58 57.38 14 item 65.38 11.43 61.79 68.97 Time 2 40 item 66.68 12.62 63.78 69.58 14 item 82.68 13.22 79.10 86.27 Time 3 40 item 32.64 16.41 29.74 35.54 14 item 30.87 22.38 27.28 34.46

117

Results showed a significant effect of Time, F(2, 240) = 186.59, p<.001, ɳp2 = .609;

Scale F(1, 240) = 273.61, p<.001, ɳp2 = .533; and an interaction between Time and Scale, F(2,

240) = 108.84, p<.001, ɳp2 = .476. Review of the confidence intervals showed a significant difference between the scales at Time 1 and Time 2, but not Time 3 (see Figure 31).

Figure 31. Graphical representation of trust scores for the 40 item scale and the 14 item scale across time.

Additional analyses were conducted to compare the change trust scores for the 40 item scale and the 14 item SME recommended scale. A 2 Change in Trust (Time 2 – Time 1, Time 2 –

118

Time 3) x 2 Scale (40 item, 14 item) repeated measures analysis of variance was conducted.

Means, standard deviations, and confidence intervals are reported in Table 32.

Table 32

Means, Standard Deviations and Confidence Intervals Comparing Change Scores for the 40 Item Scale to the 14 Item Scale

Scale Change in Trust Mean SD CILow CIHigh 40 item scale Time 2 – Time 1 12.20 14.35 9.03 15.37 Time 2 – Time 3 34.04 20.11 29.59 38.49 14 item scale Time 2 – Time 1 27.61 16.40 23.98 31.24 Time 2 – Time 3 51.90 26.80 45.97 57.82

Results showed a significant effect of Scale, F(1, 80) = 397.25, p<.001, ɳp2 =.832, in which the 14 item scale had significantly higher trust scores (M = 39.75, Se = 1.93) than the 40 item scale (M=23.12, Se = 1.56). There was also a main effect of Change Scores, F(1, 80) =

75.92, p<.001, ɳp2 = .487, in which there was a significantly greater change in trust scores from

Time 2 to Time 3 (M = 42.97, Se = 2.58) than Time 2 to Time 1 (M = 19.91, Se = 1.64). It is important to note here that Time 2 and Time 3 represented the post-HRI events, where all participants monitored a 100% reliable robot at Time 2 and a 25% reliable robot at Time 3. The

Change Score from Time 2 to Time 1 represented the change in trust that occurred pre-HRI

(Time 1) and post-HRI (Time 2). Finally, there was no interaction between Scale and Change in

Trust Scores, p=.055, (see Figure 32).

119

Figure 32. Graphical representation of trust scores for the 40 item scale and the 14 item scale for the Change in Trust Scores over Time.

Discussion

This initial validation experiment was designed to determine that the updated 42 item trust scale could indeed measure changes in trust over time. As expected there was a significant trend over time, such that Time1 < Time2 > Time3 for the majority of the items. Analysis revealed two items, ‘built to last’ and ‘operate in an integrated team environment,’ that did not show a

120

marked change over time, and were thereby removed from the scale. All additional analyses in this chapter were conducted with the remaining 40 item scale.

Interesting findings were also evident in the correlation analyses. As anticipated, there was a significant positive correlation between the ITS and the 40 item trust scale prior to HRI, but not between the ITS and 40 item trust scale post-HRI. This finding supported the predictive trust model (see also Figure 14). Additional significant negative correlations were found between the three subscales of the NARS and the 40 item pre-interaction trust scale. This may suggest that when the 40 item trust scale is administered prior to HRI, the trust score accounts for the individual differences in trust measured by the ITS and the negative attitudes toward robots measured by the NARS.

Additional analyses were also conducted to identify the differences between the 40 item trust scale and the 14 item SME recommended scale (refer to Chapter 6: Study 4). While findings revealed significant differences between the two scales, graphical representations showed similar patterns in the results. Taking into account both the individual analyses of each of item measured over Time, as well as the comparative results of the two scales, it appeared that the total trust score of the 40 item trust scale, provided a finer level of granularity and thus a more accurate trust rating.

Progression to Next Experiment

The previous chapter was the second study in a line of three validation studies. It demonstrated that 40 item scale did indeed show a change over time. One possible limitation of

121

this study was the order of the conditions. All participants completed the Trust Scale prior to interaction (Time 1), following HRI with a 100% reliable robot (Time 2), and following HRI with a 25% reliable robot (Time 3). This limitation was addressed in the following study.

The following chapter is the final validation experiment for the 40 item trust scale. It used a Same-Trait Multi-Method approach (Campbell & Fiske, 1959) to validate that this scale measured trust and not an alternative construct.

122

CHAPTER EIGHT: STUDY 6

“There is always a margin of error, even in a machine.” The Robot, Lost in Space (Allen & Richardson, 1966)

The purpose of this final experiment was to validate the developed scale using a Same-

Trait Multi-Method approach (Campbell & Fiske, 1959). Two additional scales measuring human-robot trust (Yagoda, 2011), and human-automation trust (Jian, Bisantz, Drury, & Llinas,

1998) were originally included to evaluate the “same trait” of subjective trust. The multiple methods include the subjective measurement and objective measurement (i.e., time attending to the robot). Human-robot interaction was accomplished through computer-based simulation of a navigation task. The following were the specific hypotheses for this final experiment.

Hypothesis 1: There will be a positive correlation between the 40 item scale, the 14 item SME selected subscale, and the Jian et al. (1998) trust scale, supporting the same trait.

r 40 Item Scale*14 Item Scale = +.6 r 40 Item Scale*Jian et al. Scale = +.6 r 14 Item Scale*Jian et al. Scale = +.6

Hypothesis 2: The three change scores from High Trust Condition to the Low Trust Condition for the Types of Trust Scales (i.e., 40 item scale, 14 item scale, and Jian et al. scale) will not show significant mean differences.

 40 item of trust =  14 item of trust =  Jian et al. trust scale

123

Hypothesis 3: There will be an effect of condition on trust, such that a robot with a low error rate (high trust condition) will be trusted more than a robot with a high error rate (low trust condition).

 trust (robot with low error rate) >  trust (robot with high error rate)

Hypothesis 4: It is hypothesized that there will be interactions between Trust Scale and Condition.

Hypothesis 5: It is hypothesized that trust scores and time attending to the robot to the robot will be negatively correlated.

r subjective trust*time attending = negative correlation

Experimental Method

Participants. Participants were 21 undergraduate students (12 males, 9 females) from the

University of Central Florida. Participants were recruited through UCF Sona Recruitment

Systems. All participants were over the age of 18 years. Their participation accorded with all regulations from the university’s Institutional Review Board. Extra credit was awarded for participation as per Sona Systems guidelines.

Materials. Multiple scales were included to measure subjective trust, personality traits, demographics, and human states. All materials were administered via computer.

Trust scale. Subjective trust was measured through the developed 40 item trust scale. A partial measure represented by the 14 items recommended by SME’s was also assessed. Two additional scales measuring human-robot trust (Yagoda, 2011), and human-automation trust

(Jian, Bisantz, Drury, & Llinas, 1998) were also included. The well-established Checklist for

124

Trust between People and Automation (Jian et al., 1998) was included for Same Trait analysis.

Items including the word ‘automation’ were adapted to ‘robot.’ After pilot testing, the Human-

Robot Trust Scale (Yagoda, 2011) was removed due to participant frustration with the items. The

Interpersonal Trust Scale (Rotter, 1967) was also included.

Personality assessment. The 7-point Mini-IPIP scale (Donnellan, Oswald, Baird, &

Lucas, 2006) personality assessment was used to measure the Big 5 personality traits:

Agreeableness, Extraversion, Intellect, Conscientiousness, and Neuroticism. It is a 20-item short form of the 50-item International Personality Item Pool – Five Factor Model (Goldberg, 1999).

Human states. The Dundee Stress State Questionnaire (DSSQ; Matthews, Joyner,

Gilliland, Campbell, Falconer, & Huggins, 1999) has represented a reliable and validated scale measuring human states (i.e., mood state, motivation, workload, and thinking style) before and after a task. The mood state subscale was used to assess an individual’s moods or feelings; the motivation subscale provided assessment of both success motivation and intrinsic motivation; the workload subscale was based on the six ratings scales from the NASA-TLX (Hart & Staveland,

1988); and the thinking style subscale assessed self-focused attention, self-esteem, concentration, control and confidence.

Virtual environment. The virtual environment (VE) for the present procedure was developed in RIVET and used base environment of a Middle Eastern town developed by GDRS in collaboration with ARL. Moveable object syntax for six different objects (e.g., refrigerator, boxes, crates, etc.) was also created by GDRS specifically for this VE. Independent task-specific customization of the physical environment was accomplished through Scripting syntax. Specific

125

customization included entering objects, obstacles, and creation of paths, etc. Scripting files were created for both the training session and the task conditions.

Experimental task. Participants had a training session, followed by two navigation tasks using a computer-based RIVET (General Dynamics Robotics Systems) simulation. The training session allowed the participant to become familiar with the VE and keyboard/mouse controls.

Each participant was provided instructions that they could reference through the entire study (see

Appendix T). Participants navigated a Soldier agent through the Middle Eastern town and were provided the opportunity to practice the process of moving obstacles.

The task was to assist an autonomous robot from a set location to a rendezvous point.

This assistance required the participant to move obstacles out of the way so that the robot could continue on to the rendezvous point. Five movable obstacles were included: a barrel, a crate, a refrigerator, a trash can, and a box. In Simulation A, the robot autonomously navigated around 4 of the 5 obstacles. In Simulation B, the robot only autonomously navigated around 1 of the 5 obstacles. As such, it required the participant to help move the obstacles out of the robot’s path.

Each simulation was approximately one minute in length. The order of simulation presentation was determined prior to participation.

The simulated tasks were recorded using FRAPS real-time video capture and benchmarking program with a 30 frame rate/second .avi file. Video was recorded from the

Soldier character’s perspective.

126

Experimental procedure. Following completion of informed consent, participants completed three questionnaires: the demographic questionnaire, the Mini-IPIP personality assessment, and the ITS. The experimenter explained the upcoming human-robot navigation task and answered any questions. Participants were then shown a picture of the robot (see Figure 33) they were to interact with and asked to complete the following questionnaires to acquire baseline measurements: 40 item trust scale, Jian et al. (1998) trust scale, and DSSQ.

Figure 33. Talon robot.

Participants completed the training session using the keyboard /mouse controls to move through the virtual and environment, as well as practiced moving obstacles from a path.

Participants had an unlimited time restriction during the training session and were able to ask questions. Participants completed the two simulated tasks, followed by the completion of the present 40 item trust scale, Jian et al. (1998) trust scale, and Post-task DSSQ, after each task.

Each simulated task was recorded using FRAPS to record the onscreen simulation as a movie

(.avi) file. It was saved with a unique identifier to maintain participant confidentiality. The study took approximately one hour to complete.

127

Experimental Results

Validity testing was conducted using the same trait, multi-method approach (Campbell &

Fiske, 1959). All data were analyzed using IBM SPSS Statistics v.19 (SPSS, 2010), with an alpha level set to .05, unless otherwise indicated. Dependent variables included the Trust Scores for each of three subjective trust measures, and the objective trust measure (i.e., percentage of time attending to the robot). Independent variables include Robot Condition (i.e., High Trust condition: 20% errors; Low Trust condition: 80% errors in navigation), Condition Order (i.e.,

Condition 1 first followed by Condition 2; Condition 2 first followed by Condition 1), and Type of Trust Scale (i.e., 40 item scale, the 14 item scale, and Jian et al., 1998).

Initial analyses were conducted to assess changes in human states over time. Means, standard deviations, and confidence intervals are provided in Appendix U. Results demonstrated no significant difference in mood state or motivation subscales. The thinking style subscales of self-focused attention and concentration showed a significant difference between pre-interaction to post-interaction, but no difference between the two post-interaction conditions. A similar result was found for the thinking content subscale, task interference. Due to these findings, no additional analyses were conducted assessing human states.

Same-Trait analysis. The same trait methodology compared the developed 40 item trust scale, the SME’s recommended 14 item trust scale, and the well-established Checklist for Trust between People and Automation developed by Jian et al. (1998). In support of Hypothesis 1,

128

significant positive correlations were found between the 40 item scale, the 14 item scale, and the

Jian et al. trust scale (see Table 33).

Table 33

Same Trait Trust Scale Correlations over Time

Pre-Interaction Trust Mean Std. Deviation 1 2 3 1. 40 Item Scale 60.60 13.58 1 2.14 Item Scale 72.45 12.70 .829** 1 3.Jian et al. Scale 71.71 16.27 .620** .745** 1 Post-Interaction (Condition 1) Mean Std. Deviation 1 2 3 1. 40 Item Scale 49.92 19.59 1 2.14 Item Scale 60.61 20.19 .918** 1 3.Jian et al. Scale 71.54 16.22 .857** .854** 1 Post-Interaction (Condition 2) Mean Std. Deviation 1 2 3 1. 40 Item Scale 46.70 22.43 1 2.14 Item Scale 57.42 24.93 .934** 1 3.Jian et al. Scale 72.27 17.16 .855** .852** 1

Hypothesis 2 stated that the Change Scores from the High Trust Condition to the Low

Trust Condition for the three types of Trust Scales (i.e., 40 item scale, 14 item SME recommended scale, and Jian et al. scale) would not show significant mean differences. A within subjects repeated measures analysis of variance was conducted. The results supported the hypothesis, F(2, 19)=2.64, p=.097, providing additional support that the developed scale measures the construct of trust.

Multi-Trait analysis. Multi-method analysis compared the three types of subjective trust scales (i.e., 40 item scale, 14 item scale, and Jian et al. scale) with an objective measure of trust

129

(i.e., percentage of time participant attended to the Talon robot). The percentage of time a participant attended to the Talon robot was calculated by recording the Total Time the robot was in frame divided by the Total Time of the simulation.

Prior to conducting the multi-method analysis paired samples t-tests were conducted to assess the change in trust pre-post interaction. A significant change in trust was found between the pre-post interaction trust measurement for the 40 item trust scale, t(40)=3.87, p<.001, and the

14 item trust scale, t(40)=3.86, p<.001. The Jian et al. trust scale did not change (p=.932). This finding suggests that the developed trust scale does indeed measure something additional to the previously developed trust scale.

A 3 Trust Scales (40 item, 14 item, and Jian et al. trust scale) x 3 Conditions (pre- interaction, high trust condition, low trust condition) repeated measures analysis of variance was conducted. There was a main effect of scale, F(2, 59) =105.16, p<.001 ηp2 =.781, but not condition, p=.191. However, there was a significant interaction between scale and condition, F(4,

118)=4.69, p=.002, ηp2 =.137. Means, standard error, and confidence intervals are reported in

Table 34.

Confidence interval analysis of the overall Trust Scores showed similar findings as Study

5, in which the 14 item scale had significantly higher trust ratings than the 40 item scale.

However, no differences were found between the 14 item scale and the Jian et al. scale. The confident intervals of each Condition showed a significant difference between the 40 item trust scale and the Jian Trust Scale at Pre-Interaction, High Trust Condition, and Low Trust Condition

(see also Figure 34).

130

Table 34

Means, Standard Error, and Confidence Intervals for the Three Trust Scales

Variable Mean Se CILow CIHigh 40 Item Scale 52.41 2.38 47.65 57.17 Trust Scale 14 Item Scale 63.49 2.56 58.37 68.62 Jian et al. Scale 71.84 2.09 67.67 76.01 Pre-Interaction 68.27 3.38 60.59 75.93 Condition High Trust Condition 60.69 3.83 53.02 68.36 Low Trust Condition 58.80 3.83 51.13 66.47 40 Item Scale 60.61 4.12 52.36 58.86 Pre-Interaction 14 Item Scale 72.45 4.44 63.57 81.33 Jian et al. Scale 71.71 3.62 64.48 78.94 40 Item Scale 49.92 4.12 41.67 58.17 High Trust Condition 14 Item Scale 60.61 4.44 51.73 69.49 Jian et al. Scale 71.54 3.61 64.31 78.77 40 Item Scale 46.70 4.12 38.45 54.95 Low Trust Condition 14 Item Scale 57.42 4.44 48.54 66.29 Jian et al. Scale 72.27 3.61 65.04 79.50 Note. N=63

Figure 34. Interactions between trust scales and conditions.

131

Hypothesis 3 stated there would be an effect of condition on trust, such that a robot with a low error rate (high trust condition) would be more than a robot with a high error rate (low trust condition). Confidence interval analysis also showed no significant effects or interactions for

Condition. In addition, there were no significant order effects (p=.911). In addition there were no main effects of Time (p=.632) or Condition (p=.645) for the object trust measurement of Time

Attending to the Robot. The experimental procedure did not show a marked change in trust between the high and low trust conditions. This suggests that despite previous theory stating that error rate and the quality of the automation had an effect on trust development (refer to Table 1).

Therefore, it was not surprising that the findings failed to support Hypothesis 4. None of the

Trust Scales were negatively correlated with Time Attending to the robot (40 item scale, p=.346;

14 item scale, p=.394; Jian et al. scale, p=.205).

Discussion

A Same-Trait, Multi-Method approach was used as the final validation of the developed trust scale. Overall, this study demonstrated that the developed trust scale assessed the construct of trust. In addition, it provided support for additional benefits of the developed 40 item trust scale above and beyond previously used scales (i.e., Checklist for Trust between People and

Automation; Jian et al., 1998). This finding accounted for the change in trust from pre- interaction to post-interaction. Finally, this study provided an avenue for additional research exclusive to the relationship between robot specific errors and trust development.

132

The Same-Trait approach looked at the relationship between a well-established subjective measure of trust in automation (Jian et al., 1998) and the 40 item scale. Additionally a 14 item partial scale that was recommended by SME’s in Experiment 4 was also assessed. As anticipated, strong positive correlations between the three scales were found. Mean analysis showed differences in the overall trust scores between the three scales in which the 40 item trust scale < 14 item trust scale < Jian et al. trust scale. Significant differences were only found between the 40 item and Jian et al. scale. However, no significant overall differences between the

Change Scores from the High Trust to Low Trust conditions were found. These findings provided support that the three scales assess the same trait.

The Multi-Method approach was designed to assess trust across subjective (i.e., trust scale) versus objective measures (i.e., percentage of time attending to robot) during a human- robot team navigation task. Participants completed two computer simulated scenarios. Within the

High Trust simulation, the robot had one navigational error (i.e., 20% error); while in the Low

Trust simulation, the robot had four navigational errors (i.e., 80% error). Despite previous theory stating the importance of error rate and the quality of automation on trust development, no significant differences were found between the High and Low Trust Conditions for the three trust scales or for the time attending to the robot. Therefore, errors in navigation alone may not be enough to show a marked change in trust.

One limitation of this study was the controlled virtual environment. There was limited environmental risk or communication between the Soldier (controlled by the participant) and the

Talon robot. Both of these variables have been previously shown to impact trust development over time (see also Study 5). The Talon robot was also set on a preprogrammed path that could

133

not be changed or adapted by the participant. In order to assist the Talon robot, the participant had to use the Soldier avatar to move obstacles in the environment out of the set path of the robot. Following completion of the study, a majority of the participants had made comments specific to their expectations of the robot to follow the Soldier through the environment, not to autonomously navigate through the environment. These limitations provide a number of avenues for future research to advance the limited experimental field of human-robot trust. These include the relationship between navigational errors and additional external variables including communication errors and environmental risk. In addition, research that leads to specific error- related trust metrics need to be explored.

Despite the limitations of the High/Low Trust conditions, pre-post interaction analysis provided additional insight into the change in trust that occurred from perception (pre- interaction) to direct interaction (post-interaction) between the three trust scales. Both the 40 item and the 14 item scales showed a significant change in trust from pre-interaction to post- interaction; however the Checklist for Trust (Jian et al., 1998) did not change. This change in trust was mirrored in Study 5, and is supported by the trust theory. Therefore, it can be postulated that the developed Trust Scale accounted for the relationship between the change in mental models and trust development that occurs after HRI. This finding suggested that the developed trust scale does indeed measure something additional to the previously developed trust scale. In addition, findings from this study, together with the findings from Study 5 provide support that

40 item trust scale had more accurate trust scores than both the 14 item SME recommended scale and the Checklist for Trust (Jian et al., 1998) scale.

134

CHAPTER NINE: GENERAL DISCUSSION

The technological capacities of robotics have vastly increased in recent years leading to advancement of both the functional capability and autonomy of current robotic systems. With these technological advancements, robots have become more prevalent within the social environment. This has led to a transition of the human role from an operator to that of a team member or even bystander. As such, the intricacies of interaction have changed to where the robot has become more of a companion, friend, teammate, etc. rather than strictly a machine.

Thereby, the individual’s trust in that robot takes a prominent role in the success of any interaction and therefore the future use of the robot.

The present work looked to create a trust perception scale specific to HRI that focused on the antecedents and measurable factors of trust specific to the human, robot, and environmental elements. The finalized 40 item scale was designed as a pre-post interaction measure used to assess changes in trust perception specific to HRI. The scale was also designed to be used as post-interaction measure to compare changes in trust across multiple conditions. It was further designed to be applicable across all robot domains.

Development of the Trust Scale

A three phased design, encompassing six related studies, was used to develop the finalized 40 item human-robot subjective trust scale. The design included creation of an item

135

pool (Studies 1 and 2), item reduction techniques (Studies 3 and 4), and task-based validity testing (Studies 5 and 6). The initial item pool included 172 items that represented each of the antecedents of trust, graphically displayed in the updated descriptive model of human-robot trust

(refer to Figure 10). Studies 1 and 2 were used to extend the research on human-robot trust specific to the impact of the robot’s physical form on trust development, and thereby develop a complete initial item pool. Item pool reduction procedures were used to reduce the item pool from the initial 172 items to 42 items through internal review, reliability assessment (Study 3), and content validation by SMEs (Study 4). These procedures led to a reduction of 16 items, 83 items, and 31 items respectively. Finally, two computer-based HRI simulation experiments were used to validate the scale by first demonstrating a marked change in trust scores over time (Study

5), followed by analysis that showed the scale did indeed measure trust and not an alternative construct (Study 6). It is important to note here that following initial analyses of Study 5, two additional items were removed from the scale, thus leading to the development of the 40 item human-robot trust scale.

A Return to the Triadic Model of Trust

Trust within HRI is comprised of the human, robot, and environmental-related elements.

The updated triadic model of trust was used to help organize identify factors for construction of a quantifiable metric for assessing trust.

136

Human-related antecedents of trust. Across all experiments, gender, age, and ethnicity were not related to trust, and are hereby suggested to be removed from the triadic model of trust.

Personality traits were not significantly correlated with trust pre or post interaction (Study 6), however there one small significant correlation (rneuroticism*trustworthiness=.160, p<.05) was found among the five personality traits and trustworthiness for the social robot domain (Study 1) suggesting that personality traits may be related to trust measurement for specific domains of robotics. In addition, personality traits may be more important to explaining the variance in the propensity to trust robots. As this was not the primary focus of this work, additional research should be conducted to further explore the relationship between societal impact, human traits, and propensity to trust as it impacts pre-interaction human-robot trust.

There were no substantial changes in human states between the navigation-based conditions, as demonstrated in Experiment 6. Therefore, it is recommended that future work should look at the effects of human states during different task types that include increased environmental and internal risk, as well as hands-on HRI beyond simulation.

Environment-related antecedents of trust. The finalized scale retained items that linked the functional capabilities of the robot to the physical environment and task type.

However, the level of risk, the specifics of the physical environment, and type of task should be acknowledged as possible variables in the design of experimentation.

Throughout the creation and validation processes, team-related scale items that were retained included those specific to team composition and role interdependence. Societal impact was originally assessed through the Interpersonal Trust Scale (ITS; Rotter, 1967) and the three

137

subscales of the Negative Attitudes towards Robots (NARS; Nomura, Suzuki, Kanda, & Kato,

2004). These subscales measured negative attitudes toward situations of interaction with robots

(S1), negative attitudes toward social influence of robots (S2), and negative emotions in interaction with robots (S3). Results of Experiment 5 showed that these scales were shown to be highly correlated with pre-interaction trust (rITS=.314, p<.01; rNARS_S1=.314, p<.01; rNARS_S2=.314, p<.01; rNARS_S3=.314, p<.01), but not post-interaction trust. These results are in line with the predictive trust model (Figure 14).

Robot-related antecedents of trust. In the finalized 40 item scale, all robot features and capability-based antecedents of trust were represented in the retained items. It is important to note that these items are used to assess individual’s perceptions of these features and capabilities.

Therefore, future researchers in the area of human-robot trust should be cognizant of the actual robot design features and capabilities.

General Conclusions

In conclusion, the developed 40 item trust scale was shown to provide a more sensitive and accurate trust score than the previously used Checklist for Trust between People and

Automation (Jian et al., 1998) as it related to human-robot interaction specifically. The scale was also developed to be used by any robotic domain from industry to military, to the everyday robot.

Therefore, this scale can benefit the future robotic development specific to the interaction between humans and robots.

138

Interpretation of the 40 item scale. The 40 item human-robot trust scale provided an overall percentage score across all items. Items were preceded by the question “What percentage of the time will this robot …” followed by a list of the items. Each item was a single word or short phrase, and the order of items was randomized for each participant. The finalized 40 item scale is provided in Table 35, and took between 5-10 minutes to complete.

Table 35

Finalized Trust Scale

What % of the time will this robot… 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Act consistently* o o o o o o o o o o o Protect people o o o o o o o o o o o Act as part of the team o o o o o o o o o o o Function successfully* o o o o o o o o o o o Malfunction R* o o o o o o o o o o o Clearly communicate o o o o o o o o o o o Require frequent maintenance R o o o o o o o o o o o Openly communicate o o o o o o o o o o o Have errors R * o o o o o o o o o o o Perform a task better than a o o o o o o o o o o o novice human user Know the difference between o o o o o o o o o o o friend and foe Provide Feedback* o o o o o o o o o o o Possess adequate decision- o o o o o o o o o o o making capability Warn people of potential risks in o o o o o o o o o o o the environment Meet the needs of the mission* o o o o o o o o o o o

Provide appropriate information* o o o o o o o o o o o

Communicate with people* o o o o o o o o o o o

Work best with a team o o o o o o o o o o o

139

What % of the time will this robot… 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Keep classified information o o o o o o o o o o o secure Perform exactly as instructed* o o o o o o o o o o o

Make sensible decisions o o o o o o o o o o o Work in close proximity with o o o o o o o o o o o people Tell the truth o o o o o o o o o o o Perform many functions at one o o o o o o o o o o o time Follow directions* o o o o o o o o o o o

What % of the time will this robot be … 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Considered part of the team o o o o o o o o o o o

Responsible o o o o o o o o o o o

Supportive o o o o o o o o o o o

Incompetent R o o o o o o o o o o o

Dependable * o o o o o o o o o o o

Friendly o o o o o o o o o o o

Reliable * o o o o o o o o o o o

Pleasant o o o o o o o o o o o

Unresponsive R * o o o o o o o o o o o

Autonomous o o o o o o o o o o o

Predictable * o o o o o o o o o o o

Conscious o o o o o o o o o o o

Lifelike o o o o o o o o o o o

A good teammate o o o o o o o o o o o Led astray by unexpected o o o o o o o o o o o changes in the environment Note. The 14 trust subscale items are marked with an *. The R represents reverse coded items for scoring.

140

Creating the trust score. When the scale is used as a pre-post interaction measure, the participants should first be shown a picture of the robot they will be interacting with prior to completing the pre-interaction scale. This accounts for any mental model effects of robots and allows for comparison specific to the robot at hand. For post-interaction measurement, the scale should be administered directly following the interaction. To create the overall trust score, 5 items must first be reverse coded. The reverse coded items are denoted in the above table. All items are then summed and divided by the total number of items (40). This provides an overall percentage of trust score.

14 item subscale. While use of the 40 item scale is recommended, a 14 item subscale can be used to provide rapid trust measurement specific to measuring changes in trust over time, or during assessment with multiple trials or time restrictions. This subscale is specific to functional capabilities of the robot, and therefore may not account for changes in trust due to the feature- based antecedents of the robot. Trust score is calculated by reverse coding the ‘have errors,’

‘unresponsive, and ‘malfunction’ items; summing the 14 item scores and dividing by 14. The 14 item scale included the following items:

 Function successfully  Act consistently  Reliable  Predictable  Dependable  Follow directions  Meet the needs of the mission  Perform exactly as instructed  Have errors (Reverse Coded)

141

 Provide appropriate information  Unresponsive (Reverse Coded)  Malfunction (Reverse Coded)  Communicate with people  Provide feedback

Future Work

This scale was developed to provide a means to subjectively measure trust perceptions over time and across robotic domain. In addition, it can be used by individuals in all the major roles of HRI: operator, supervisor, mechanic, peer, and/or bystander. Therefore, the capabilities for future work are expansive. Three avenues for future research are discussed below that highlight the expansion of human-robot trust theory, the need for robot trust metrics, and the importance of the human element.

Human-robot trust is a developing field with strong theoretical support related to the human, robot, and environmental-related antecedents of trust described in Chapter 2. While there are a number of antecedents of trust that can be experimentally tested, this work identified the importance of the relationship between mental models and trust development. Future work is needed to identify the specific changes in an individual’s mental model that is related to the changes in trust from pre-interaction to post-interaction.

This study provided an avenue for additional research exclusive to the development of trust-related design metrics. One of the areas for future work is the relationship between robot specific errors and trust development. This work found that trust development is related to communication-related errors, but not navigation-related errors, when measured alone. The

142

design of this work allows for future work that will lead to determining the specific trust-related design metrics specific to the number, duration, and type of errors. This work can then be expanded to develop a more complete list of human-robot trust metrics.

The final line of future work is related to trust as it applies to expansion and transition of the human roles within HRI. As stated in the introduction, the human element is often overlooked or even forgotten during the robot design and development process. However, future work needs to be conducted to further understand the differences in trust perceptions between team members as it applies to this process. Trust measurement across the human roles may assist in the direction and importance of design and development. For example, previous work has suggested that a robot operator is mostly concerned about the functional capability of the robot, while the bystander assesses the functional capability through the robot’s features. In addition, future work focused on understanding human-robot trust across the team members can be used to adapt the human-robot team structure and training elements, as the current field of HRI continues to strive to make a robot integrated team member.

143

APPENDIX A: OPERATIONAL DEFINITIONS OF ROBOT DOMAINS

144

Table 36

Operational Definitions of Robot Domains

Domain Operational Definition Source Assisting Robots Industry A reprogrammable, multifunctional manipulator designed to move Robot Institute of material, parts, tools, or specialized devices through various America (1979) programmed motions for the performance of a variety of tasks.

An automatically controlled, reprogrammable, multipurpose manipulator with three or more axes. Kurfess (2005) Military A machine that does not carry a human operator; can operate Lin, Bekey, & Abney autonomously or remotely; can be expendable or recoverable; and (2008) can carry a lethal or non-lethal payload.

A powered machine that senses, thinks (in a deliberative, non- mechanical sense), and acts. Research Laboratory created and assessed robots, often in educational or DiSalvo, Gemperle, industrial research laboratories. Forlizzi & Kiesler (2002) Medical Robots that share a working environment with doctors and nurses Guiochet & Vilchis and close interaction with patients. (2002) Service Robots that often directly interact with people; therefore, they must Goetz, Kiesler & Powers meet social and instrumental goals (e.g., provide feedback to users, (2003) gain cooperation).

Interactive Stimulation Robots Social Robots that rely on the human tendency to anthropomorphize and Breazeal (2003) capitalize on feelings evoked, when humans nurture, care or involve Breazeal & Scassellati with their ‘creation.’ (2002) Robots that proactively engage with humans in order to satisfy Dautenhahn (1999) internal social aims (e.g., drives, emotions). Robots that show aspects of human-style social intelligence, based on possibly deep models of human cognition and social competence.

Education Robots that can be programmed to accomplish certain functions Marsh & Spain (1984) (similar to computer programming) and be able to move (e.g., swivel, drop, and clench to pick up an item).

Entertainment An autonomous robot, which interacts with people or environment Kwak & Kim (2005) in human’s daily life to provide intimacy and enjoyment through emotion and personality design.

Rehabilitation / A robot that gives aid or support to human user. It can reduce need Feil-Seifer & Matarić Therapy for human attendant by being located remotely from the body while (2005) performing mission of augmenting manipulation function. Mahoney (1997)

145

APPENDIX B: EMPIRICAL RESEARCH ON HUMAN-ROBOT TRUST

146

Table 37

Human-Robot Empirical Trust Studies included for Meta-Analysis

Factors Correlation Studies Experimental Studies

Human-Related Antecedents of Trust Demographic Evers, Maldanado, Brodecki, & Hinds, 2008 Kidd, 2003; Scopelliti, Giuliani, & Fornana, Information 2005 Attentional Kidd & Breazeal, 2004; Li, Rau, & Li, 2010; N/A Capacity / Mutlu, Yamaoka, Kanda, Ishiguro, & Hagita, Engagement 2009 Personality Traits Li, Rau, & Li, 2010; Looije, Neerincx, N/A & Cnossen, 2010 Attitudes Towards Rau, Li, & Li, 2009; Wang, Rau, Evers, N/A Robots Robinson, & Hinds, 2010 Comfort with Evers, Maldanado, Brodecki, & Hinds, 2008 N/A Robots Robot-Related Antecedents of Trust Behavior Looije, Neerincx, & Cnossen, 2010 Tsui, Desai, & Yanco, 2010 Dependability Biros, Daly, & Gunsch, 2004 N/A Reliability Biros, Daly, & Gunsch, 2004; Kidd & Ross, 2008 Breazeal, 2004; Ross, 2008 Predictability Biros, Daly, & Gunsch, 2004 N/A Automation Biros, Daly, & Gunsch, 2004; Ross, N/A 2008; Tenney, Rogers, & Pew, 1998 Proximity Li, Rau, & Li, 2010 Bainbridge, Hart, Kim, & Scassellati, 2008; Kidd, 2003; Kiesler, Powers, Fussell, & Torrey, 2008; Powers, Kiesler, Fussell, & Torrey, 2007 Robot Personality Kidd & Breazeal, 2004; Li, Rau, & Li, 2010; de Ruyter, Saini, Markopoulous, & van Rau, Li, & Li, 2009 Breemen, 2005 Adaptability N/A Heerink, Kröse, Evers, & Wielinga, 2010 Robot Type Ross, 2008 Ross, 2008; Tsui et al., 2010a Anthropomorphism Evers, Maldanado, Brodecki, & Hinds, 2008; N/A Ross, 2008 Environment-Related Antecedents of Trust Culture / In-group Evers, Maldanado, Brodecki, & Hinds, 2008 Evers, Maldanado, Brodecki, & Hinds, 2008; membership Li, Rau, & Li, 2010; Wang, Rau, Evers, Rau, Li, & Li, 2009; Wang, Rau, Evers, Robinson, & Hinds, 2010 Robinson, & Hinds, 2010 Communication Kidd & Breazeal, 2004; Wang, Rau, Wang, Rau, Evers, Robinson, & Hinds, 2010 Evers, Robinson, & Hinds, 2010 Task Type Li, Rau, & Li, 2010 Kidd, 2003 * Note. Adapted from Tables 1,3 in ARL-TR-5875 (Hancock et al., 2011)

147

148

APPENDIX C: HRI TRUST DEFINITIONS

149

Table 38

Trust Definitions

Definition Definition Paper Citations Citation Trust is the feeling of confidence that another individual will not put Anderson (1981) Groom & Nass the self at risk unnecessarily (Anderson, 1980; Axelrod, 1976). Axelrod (1976) (2007) Believing that teammates will protect each other’s interests enables individual teammates to surrender their personal interests to the interests of the group. Defines trust in terms of the taxonomy of three specific expectations: Barber (1983) Rau, Li, & Li (1) Our general expectation of the persistence of the natural physical (2009) order, the natural biological order, and the moral social order, (2) our specific expectation of the technically competent role performance from those involved with us in social relations and systems, and (3) our specific expectation that the partners in an interaction will carry out their fiduciary obligations and responsibilities, that is, their duty in certain situations to place other’s interests before their own. (p.588) An expectation or mental attitude an agent maintains regarding its Barber (1983) Wagner (2009) social environment The expectation of the persistence and fulfillment of the natural and Barber (1983) Llinas et al. the moral social orders, expectation of technically competent role (1998) performance, expectation that partners in interaction will carry out their aforementioned characteristics (persistence, technically competent performance, and fiduciary responsibility). The expectation held by a number of a system, of the persistence of the Barber (1983) Thomer (2007) natural and moral social orders, and of technically competent performance, and the fiduciary responsibility, from a member of the system, and is related to but not necessarily isomorphic with, objective measure of these properties. System automation trust is defined as having confidence in and Biros, Daly, & Biros, Daly, & entrusting the system automation to do the appropriate action (p. 177) Gunsch (2004) Gunsch (2004) a state involving confident predictions about another’s motives with Boon & Holmes Adams et al. respect to oneself in situations entailing risk. (p. 21) (1991) (2003) According to Coleman (1990), placing trust involves putting resources Coleman (1990) de Vries, Midden in the hands of another, who in turn will use them to his own & Bouwhuis advantage, the trustor’s advantage, or both. (p. 720) (2003) According to Couch and Jones (1997), trust has been defined in diverse Couch & Jones Madhavan & ways: as a generalized expectancy (Rotter 1967), as an enduring (1997) Wiegmann (2007) attitude or trait (Deutsch 1958, Giffin 1967) and as a transitory situational variable (Kee and Knox 1970, Driscoll 1978). Trust is a facet of human personality (Deutsch, 1962). Trust is the Deutsch (1962) Wagner (2009) result of a choice among behaviors in a specific situation. Deutsch’s definition of trust focused on the individual’s perception of the situation and the cost/benefit analysis that resulted (p. 29)

150

Definition Definition Paper Citations Citation The confidence that one will find what is desired from another, rather Deutsch (1962) Thomer (2007) than what is feared. The outcome of observations leading to the belief that the actions of Elofson (2001) another may be relied upon, without explicit guarantee, to achieve a goal in a risky situation. Trust (or, symmetrically, distrust) is a particular level of the subjective Gambetta (1988) Stormont (2008) probability with which an agent assesses that another agent or group of agents will perform a particular action, both before he can monitor such action (or independently of his capacity ever to be able to monitor it) and in a context in which it affects his own action. When we say we trust someone or that someone is trustworthy, we implicitly mean that the probability that he will perform an action that is beneficial or at least not detrimental to us is high enough for us to consider engaging in some form of cooperation with him. Trust is a particular level of subjective probability with which an agent Gambetta (1988) Wagner (2009) assesses that another agent or group of agents will perform a particular action, both before he can monitor such action and in a context in which it affects his own action. The reliance by an agent that actions prejudicial to their well-being Hancock, Hancock, will not be undertaken by influential others. Billings, & Billings, & Schaefer (2011) Schaefer (2011) The belief that the system performs with personal integrity and Heerink et al. Heerink et al. reliability. (p. 364) (2010) (2010) Trust has aspects of Hoffman et al. Hoffman et al. • an attitude (of the trustor about the trustee), (2009) (2009) • an attribution (that the trustee is trustworthy), • an expectation (about the trustee’s future behavior), • a feeling or belief (faith in the trustee, or a feeling that the trustee is benevolent, or a feeling that the trustee is directable), • an intention (of the trustee to act in the trustor’s interests), and • a trait (some people are trusting and more able to trust appropriately). Trust reflects an assessment of the trustee’s capabilities and Hoffman et al. Hoffman et al. competencies to respond to uncertain situations to meet common goals. (2009) (2009) A belief by a person in the integrity of another's behavior. Larzelere & Llinas et al. Huston (1980) (1998) Trust is a psychological state involving positive confident expectations LeBlanc (2007) LeBlanc (2007) and willingness to act on the basis of these expectations (p. 20).

151

Definition Definition Paper Citations Citation The attitude that an agent will help achieve an individual’s goals in a Lee & See (2004) Freedy et al. situation characterized by uncertainty and vulnerability. (2007) Desai et al. (2009) Wagner (2009) Ogreten et al. (2010) Park, Jenkins, & Jiang (2008) Ross et al. (2007) Cring & Lenfestey (2009) Jamieson, Wang, & Neyedli (2007) Thomer (2007) A means for reducing the social complexity and risk of daily life. He Luhmann (1979) Wagner (2009) argues that the complexity of the natural world is far too great for an individual to manage the many decisions it must make in order to survive (p. 30) “The extent to which a user is confident in, and willing to act on the Madsen & Gregor Adams et al. basis of the recommendations, actions, and decisions of an artificially (2000) (2003) intelligent agent” Miller et al. (2008) “The willingness of a party to be vulnerable to the outcomes of another Mayer, Davis, & Adams et al. party based on the expectation that the other will perform a particular Schoorman (2003) action important to the trustor, irrespective of the ability to monitor or (1995) control that other party.” “An attitude which includes the belief that the collaborator will Moray & Inagaki Adams et al. perform as expected, and can, within the limits of its designers’ (1999) (2003) intentions, be relied on to achieve the design goals” "The expectation, held by a member of a system, of persistence of the Muir (1994) Olson (1999) natural and moral social orders, and of technically competent performance, and of fiduciary responsibility, from a member of the system, and is related to, but is not necessarily isomorphic with, objective measures of these properties." The degree of confidence one feels when one thinks about a Rempel & Llinas et al. relationship. Holmes (1986) (1998) Thomer (2007) The predictability or consistency of an individual’s actions Rempel et al. Madhavan & (1985) Wiegman (2007) A generalized expectation related to the subjective probability an Rempel et al. Thomer (2007) individual assigns to the occurrence of some set of future events. (1985) Confidence in a system Ross (2008) Ross (2008)

152

Definition Definition Paper Citations Citation An expectancy held by an individual or group that the word, promise, Rotter (1967) Llinas et al. verbal, or written statement of another individual or group can be (1998) relied on. A generalized expectancy held by an individual that the word, promise, Rotter (1967) Thomer (2007) or written statement of another individual or group can be relied on. An actor's willingness to arrange and repose his or her actions on Scanzoni (1979) Llinas et al. another actor because of confidence that other will provide expected (1998) gratification. Trust as an effect or outcome of certain automation characteristics (e.g. Sheridan (2002) Madhavan & reliability) and trust as a cause of operators’ behaviour when utilizing Wiegmann (2007) automation. A belief, held by the trustor, that the trustee will act in a manner that Wagner (2009) Wagner (2009) mitigates the trustor’s risk in a situation in which the trustor has put its outcomes at risk. (p. 31) 1. assured reliance on a person or thing Webster’s Llinas et al. 2. dependence on something future or contingent Dictionary (1998) 3. an equitable right or interest 4. a charge or a duty imposed in faith or confidence or as a condition of some relationship 5. something committed or entrusted to one to be used or cared for in the interest of another Trust is a prediction of reliance on an action, based on what a party Wikipedia Miller et al. knows about the other party. Trust is a statement about what is (2008) otherwise unknown -- for example, because it is far away, cannot be verified, or is in the future. Three aspects of trust: trial and error experience, understanding of Zuboff (1988) Thomer (2007) technology, and faith.

153

APPENDIX D: PLATO’S ALLEGORY OF THE CAVE

154

The entire translations of The Republic can be found at http://classics.mit.edu/Plato/republic.8.vii.html.

Socrates – Glaucon

And now, I said, let me show in a figure how far our nature is enlightened or unenlightened: --Behold! human beings living in an underground den, which has a mouth open towards the light and reaching all along the den; here they have been from their childhood, and have their legs and necks chained so that they cannot move, and can only see before them, being prevented by the chains from turning round their heads. Above and behind them a fire is blazing at a distance, and between the fire and the prisoners there is a raised way; and you will see, if you look, a low wall built along the way, like the screen which marionette players have in front of them, over which they show the puppets.

I see. And do you see, I said, men passing along the wall carrying all sorts of vessels, and statues and figures of animals made of wood and stone and various materials, which appear over the wall? Some of them are talking, others silent.

You have shown me a strange image, and they are strange prisoners. Like ourselves, I replied; and they see only their own shadows, or the shadows of one another, which the fire throws on the opposite wall of the cave?

True, he said; how could they see anything but the shadows if they were never allowed to move their heads?

And of the objects which are being carried in like manner they would only see the shadows?

Yes, he said. And if they were able to converse with one another, would they not suppose that they were naming what was actually before them?

Very true. And suppose further that the prison had an echo which came from the other side, would they not be sure to fancy when one of the passers-by spoke that the voice which they heard came from the passing shadow?

No question, he replied. To them, I said, the truth would be literally nothing but the shadows of the images.

That is certain. And now look again, and see what will naturally follow it' the prisoners are released and disabused of their error. At first, when any of them is liberated and compelled suddenly to stand up and turn his neck round and walk and look towards the light, he will suffer sharp pains; the glare will distress him, and he will be unable to see the realities of which in his former state he had seen the shadows; and then conceive someone saying to him, that what he saw before was an illusion, but that now, when he is approaching nearer to being and his eye is turned towards more real existence, he has a clearer vision, -what will be his reply? And you may further imagine that his instructor is pointing to the objects as they pass and requiring him to name them, -will he not be perplexed? Will he not fancy that the shadows which he formerly saw are truer than the objects which are now shown to him?

Far truer.

155

And if he is compelled to look straight at the light, will he not have a pain in his eyes which will make him turn away to take and take in the objects of vision which he can see, and which he will conceive to be in reality clearer than the things which are now being shown to him?

True, he now And suppose once more, that he is reluctantly dragged up a steep and rugged ascent, and held fast until he’s forced into the presence of the sun himself, is he not likely to be pained and irritated? When he approaches the light his eyes will be dazzled, and he will not be able to see anything at all of what are now called realities.

Not all in a moment, he said. He will require to grow accustomed to the sight of the upper world. And first he will see the shadows best, next the reflections of men and other objects in the water, and then the objects themselves; then he will gaze upon the light of the moon and the stars and the spangled heaven; and he will see the sky and the stars by night better than the sun or the light of the sun by day?

Certainly. Last of he will be able to see the sun, and not mere reflections of him in the water, but he will see him in his own proper place, and not in another; and he will contemplate him as he is.

Certainly. He will then proceed to argue that this is he who gives the season and the years, and is the guardian of all that is in the visible world, and in a certain way the cause of all things which he and his fellows have been accustomed to behold?

Clearly, he said, he would first see the sun and then reason about him.

And when he remembered his old habitation, and the wisdom of the den and his fellow-prisoners, do you not suppose that he would felicitate himself on the change, and pity them?

Certainly, he would. And if they were in the habit of conferring honours among themselves on those who were quickest to observe the passing shadows and to remark which of them went before, and which followed after, and which were together; and who were therefore best able to draw conclusions as to the future, do you think that he would care for such honours and glories, or envy the possessors of them? Would he not say with Homer,

Better to be the poor servant of a poor master, and to endure anything, rather than think as they do and live after their manner?

Yes, he said, I think that he would rather suffer anything than entertain these false notions and live in this miserable manner.

Imagine once more, I said, such an one coming suddenly out of the sun to be replaced in his old situation; would he not be certain to have his eyes full of darkness?

To be sure, he said.

156

And if there were a contest, and he had to compete in measuring the shadows with the prisoners who had never moved out of the den, while his sight was still weak, and before his eyes had become steady (and the time which would be needed to acquire this new habit of sight might be very considerable) would he not be ridiculous? Men would say of him that up he went and down he came without his eyes; and that it was better not even to think of ascending; and if any one tried to loosen another and lead him up to the light, let them only catch the offender, and they would put him to death.

No question, he said. This entire allegory, I said, you may now append, dear Glaucon, to the previous argument; the prison-house is the world of sight, the light of the fire is the sun, and you will not misapprehend me if you interpret the journey upwards to be the ascent of the soul into the intellectual world according to my poor belief, which, at your desire, I have expressed whether rightly or wrongly God knows. But, whether true or false, my opinion is that in the world of knowledge the idea of good appears last of all, and is seen only with an effort; and, when seen, is also inferred to be the universal author of all things beautiful and right, parent of light and of the lord of light in this visible world, and the immediate source of reason and truth in the intellectual; and that this is the power upon which he who would act rationally, either in public or private life must have his eye fixed.

157

APPENDIX E: QUESTIONNAIRES AND SCALES

158

Mini-IPIP

Donnellan, M. B., Oswald, F. L., Baird, B. M., & Lucas, R. E. (2006). The Mini-IPIP scales: Tiny-yet-effective measures of the Big Five factors of personality. Psychological Assessment, 18(2), 192-203.

Please indicate how well each statement describes you.

1. Am the life of the party. (Extraversion) 2. Sympathize with others’ feelings. (Agreeableness) 3. Get chores done right away. (Conscientiousness) 4. Have a vivid imagination. (Intellectual/Imaginative) 5. Don’t talk a lot. (reverse coded) (Extraversion) 6. Am not interested in other people’s problems. (reverse coded) (Agreeableness) 7. Often forget to put things back in their proper place. (reverse coded) (Conscientiousness) 8. Am not interested in abstract ideas. (reverse coded) (Intellectual/Imaginative) 9. Talk to a lot of different people at parties. (Extraversion) 10. Feel others’ emotions. (Agreeableness) 11. Like order. (Conscientiousness) 12. Have difficulty understanding abstract ideas. (reverse coded) (Intellectual/Imaginative) 13. Keep in the background. (reverse coded) (Extraversion) 14. Am not really interested in others. (reverse coded) (Agreeableness) 15. Make a mess of things. (reverse coded) (Conscientiousness) 16. Do not have a good imagination. (reverse coded) (Intellectual/Imaginative) 17. Have frequent mood swings. (neuroticism) 18. Am relaxed most of the time. (reverse coded) (neuroticism) 19. Get upset easily. (neuroticism) 20. Seldom feel blue. (reverse coded) (neuroticism)

159

Negative Attitudes toward Robots Scale

Nomura, T., Kanda, T., Suzuki, T., & Kato K. (2004). Psychology in Human-Robot Communication: An Attempt through Investigation of Negative Attitudes and Anxiety toward Robots. Proceedings of the 2004 IEEE International Workshop on Robot and Human Interactive Communication, Kurashiki, Okayama Japan.

1. I would feel uneasy if robots really had emotions (S2) 2. Something bad might happen if robots developed into living beings (S2) 3. I would feel relaxed talking with robots. * (S3) 4. I would feel uneasy if I was given a job where I had to use robots. (S1) 5. If robots had emotions, I would be able to make friends with them. * (S3) 6. I feel comforted being with robots that have emotions. * (S3) 7. The word “robot” means nothing to me. (S1) 8. I would feel nervous operating a robot in front of other people. (S1) 9. I would hate the idea that robots or artificial intelligences were making judgments about things. (S1) 10. I would feel very nervous just standing in front of a robot. (S1) 11. I feel that if I depend on robots too much, something bad might happen. (S2) 12. I would feel paranoid talking with a robot. (S1) 13. I am concerned that robots would be a bad influence on children. (S2) 14. I feel that in the future society will be dominated by robots. (S2)

*Inverse Items

160

Interpersonal Trust Scale

Rotter, J.B. (1967). A new scale for the measurement of interpersonal trust. Journal of Personality, 35, 651-665.

1. Hypocrisy is on the increase in our society. 2. In dealing with strangers one is better off to be cautious until they have provided evidence that they are trustworthy. 3. This country has a dark future unless we can attract better people into politics. 4. Fear and social disgrace or punishment rather than conscience prevents most people from breaking the law. 5. Using the honor system of not having a teacher present during exams would probably result in increased cheating. 6. Parents usually can be relied on to keep their promises. 7. The United Nations will never be an effective force in keeping world peace. 8. The judiciary is a place where we can all get unbiased treatment. 9. Most people will be horrified if they knew how much news that the public hears and sees is distorted. 10. It is safe to believe that in spite of what people say, most people are primarily interested in their own welfare. 11. Even though we have reports in newspapers, radio, and TV, it is hard to get objective accounts of public events. 12. The future seems very promising. 13. If we really knew what was going on in international politics, the public would have no reason to be more frightened than they now seem to be. 14. Most elected officials are really sincere in their campaign promises. 15. Many major national sports contests are fixed in one way or another. 16. Most experts can be relied upon to tell the trust about the limits of their knowledge. 17. Most parents can be relied upon to carry out their threats or punishments. 18. Most people can be counted on to do what they say they will do. 19. In these competitive times, one has to be alert or someone is likely to take advantage of you. 20. Most idealists are sincere and usually practice what they preach. 21. Most salesmen are honest in describing their products. 22. Most students in school would not cheat if they were sure of getting away with it. 23. Most repairmen will not overcharge even if they think you are ignorant of their specialty. 24. A large share of accident claims filed against insurance companies are phony. 25. Most people answer public opinion polls honestly.

161

Godspeed Questionnaire

Bartneck, C., Kulić, D., Croft, E., & Zoghbi, S. (2009). Measurement Instruments for the Anthropomorphism, Animacy, Likeability, Perceived Intelligence, and Perceived Safety of Robots. International Journal of Social Robots, 1, 71-81. DOI: 10.1007/s12369-008-001-3.

Please rate your impression of the robots on these scales:

Anthropomorphism 1. Fake – Natural 2. Machinelike – Humanlike 3. Unconscious – Conscious 4. Artificial – Lifelike 5. Moving rigidly – Moving elegantly

Animacy 1. Dead – Alive 2. Stagnant – Lively 3. Mechanical – Organic 4. Artificial – Lifelike 5. Inert – Interactive 6. Apathetic – Responsive

Likeability 1. Dislike – Like 2. Unfriendly – Friendly 3. Unkind – Kind 4. Unpleasant – Pleasant 5. Awful – Nice

Perceived Intelligence 1. Incompetent – Competent 2. Ignorant – Knowledgeable 3. Irresponsible – Responsible 4. Unintelligent – Intelligent 5. Foolish - Sensible

Perceived Safety 1. Anxious – Relaxed 2. Agitated – Calm 3. Quiescent – Surprised

162

Dundee Stress State Questionnaire

Matthews, G., Joyner, L., Gilliland, K., Campbell, S.E., Falconer, S., & Huggins, J. (1999). Validation of a comprehensive stress state questionnaire: Towards a state “Big Three.” In I. Mervielde, I.J. Dreary, F. DeFruyt, & F. Ostendorf (Eds.), Personality psychology in Europe (Vol. 7, pp. 335-350). Tilburg, the Netherlands: Tilburg University Press.

Matthews, G., Campbell, S.E., Falconer, S. Joyner, L., Huggins, J., Gilliland, K., et al. (2002). Fundamental dimensions of subjective state in performance settings: Task engagement, distress and worry. Emotion, 2, 315-340.

Mood State First, there is a list of words which describe people's moods or feelings. Please indicate how well each word describes how you feel AT THE MOMENT / WHILE PERFORMING THE TASK. For each word, circle the answer from 1 to 4 which best describes your mood.

Definitely = 1 Slightly = 2 Slightly Not = 3 Definitely = 4 1. Happy (Hedonic tone) 2. Dissatisfied * (Hedonic tone) 3. Energetic (Energetic arousal) 4. Relaxed * (Tense arousal) 5. Alert (Energetic arousal) 6. Nervous (Tense arousal) 7. Passive * (Energetic arousal) 8. Cheerful (Hedonic tone) 9. Tense (Tense arousal) 10. Jittery (Tense arousal) 11. Sluggish * (Energetic arousal) 12. Sorry * (Hedonic tone) 13. Composed * (Tense arousal) 14. Depressed * (Hedonic tone) 15. Restful * (Tense arousal) 16. Vigorous (Energetic arousal) 17. Anxious (Tense arousal) 18. Satisfied (Hedonic tone) 19. Unenterprising * (Energetic arousal) 20. Sad * (Hedonic tone) 21. Calm * (Tense arousal) 22. Active (Energetic arousal) 23. Contented (Hedonic tone) 24. Tired * (Energetic arousal) 25. Impatient (Anger / frustration) 26. Annoyed (Anger / frustration) 27. Angry (Anger / frustration) 28. Irritated (Anger / frustration) 29. Grouchy (Anger / frustration)

Reverse score items and then add together all eight item scores to get the scale score.

163

Motivation

Please answer some questions about your attitude to the task you are about to do / to the task you have just done. Rate your agreement with the following statements by circling one of the following answers:

Extremely = 4 Very much = 3 Somewhat = 2 A little bit = 1 Not at all = 0 1. I expect the content of the task will be interesting 2. The only reason to do the task is to get an external reward (e.g. payment) 3. I would rather spend the time doing the task on something else 4. I am concerned about not doing as well as I can 5. I want to perform better than most people do 6. I will become fed up with the task 7. I am eager to do well 8. I would be disappointed if I failed to do well on the task 9. I am committed to attaining my performance goals 10. Doing the task is worthwhile 11. I expect to find the task boring 12. I feel apathetic about my performance 13. I want to succeed on the task 14. The task will bring out my competitive drives 15. I am motivated to do the task

Please answer the following questions about your attitude to the task you have just done. Rate your agreement with the following statements by circling one of the following answers:

Low 0 1 2 3 4 5 6 7 8 9 10 High

16. How much mental and perceptual activity was required? (Mental Demand) 17. How much physical activity was required? (Physical Demand) 18. How much time pressure did you feel due to the pace at which the task elements occurred? (Temporal Demand) 19. How successful do you think you were in accomplishing the goals of the task? (Performance) 20. How hard did you have to work (mentally and physically) to accomplish your level of performance? (Effort) 21. How discouraged, irritated, stressed and annoyed did you feel during the task? (Frustration)

Items 1-15 of Part 2 of the questionnaire assess motivation. Provisionally, these items assess two motivation scales. Success motivation refers to motivation to excel in performance: summate items 4, 5, 7, 8, 9, 13 and 14. Intrinsic motivation refers to interest in the content of the task. Items 1 and 10 are positively-scored, whereas items 2, 3, 6, 11, and 12 are reverse-scored. Subtract each of the five reverse-scored item score from 4, and then summate all seven item scores. Item 15 provides an overall motivation rating, if required. Items 16-21 on the post-task questionnaire assess workload, based on the six rating scales of the NASA-TLX measure (Hart & Staveland, 1988).

164

Thinking Style

In this section, we are concerned with your thoughts about yourself: how your mind is working, how confident you feel, and how well you expect to perform on the task. Below are some statements which may describe your style of thought RIGHT NOW / DURING TASK PERFORMANCE. Read each one carefully and indicate how true each statement is of your thoughts AT THE MOMENT / WHILE PERFORMING THE TASK. To answer, circle one of the following answers:

Extremely = 4 Very much = 3 Somewhat = 2 A little bit = 1 Not at all = 0

1. I’m trying (tried) to figure myself out. 2. I’m (was) very aware of myself. 3. I’m reflecting (reflected) about myself. 4. I’m daydreaming (daydreamed) about myself. 5. I’m thinking (thought) deeply about myself. 6. I’m attending (attended) to my inner feelings. 7. I’m examining (examined) my motives. 8. I feel (felt) that I’m off somewhere watching myself. 9. I feel (felt) confident about my abilities. 10. I am (was) worried about whether I am regarded as a success or failure. 11. I feel (felt) self-conscious. 12. I feel (felt) as smart as others. 13. I am (was) worried about what other people think of me. 14. I feel (felt) confident that I understand things. 15. I feel (felt) inferior to others at this moment. 16. I feel (felt) concerned about the impression I am making. 17. I feel (felt) that I have less scholastic ability right now than others. 18. I am (was) worried about looking foolish. 19. My attention is (was) directed towards things other than the task. 20. I am finding (found) physical sensations such as muscular tension distracting. 21. I expect my performance will be (My performance was) impaired by thoughts irrelevant to the task. 22. I have (had) too much to think about to be able to concentrate on the task. 23. My thinking is (was) generally clear and sharp. 24. I will find (found) it hard to maintain my concentration for more than a short time. 25. My mind is wandering (wandered) a great deal. 26. My thoughts are (were) confused and difficult to control. 27. I expect to perform (performed) proficiently on this task. 28. Generally, I feel (felt) in control of things. 29. I can (was) handle any difficulties I encounter. 30. I consider (considered) myself skillful at the task.

165

Part 3 of the questionnaire ('Thinking Style') assesses the person's general style of thinking and beliefs about the task.

Self-focused Attention. (8 items). This scale comprises selected items from the modified Fenigstein et al. (1975) private self-consciousness scale. Summate item scores on items 1-8.

Self-esteem. (7 items). This scale comprises 6 social self-esteem and 1 performance self-esteem items from the Heatherton and Polivy (1991) questionnaire, and may be seen as a general self-esteem scale. Summate items 10, 11, 13, 15, 16, 17 and 18 and subtract total from 28. (This step ensures high scores indicate high esteem).

Concentration (7 items). This scale comprises most of the items written to represent perceived efficiency of attention. Summate items 19-22 and 24-26 and subtract total from 28. High scores indicate good concentration.

Control and Confidence (8 items). This scale is made up mainly of positive performance self-esteem items, relating to confidence, and perceived control items. (The perceived control and performance self-esteem constructs appear to overlap). Summate items 9, 12, 14, 23, 27, 28, 29 and 30.

166

Thinking Content

This set of questions concerns the kinds of thoughts that go through people's heads at particular times, for example while they are doing some task or activity. Below is a list of thoughts, some of which you might have had recently. Please indicate roughly how often you had each thought DURING THE LAST TEN MINUTES or so, by circling a number from the list below.

1= Never 2= Once 3= A few times 4= Often 5= Very often

1. I thought about how I should work more carefully. 2. I thought about how much time I had left. 3. I thought about how others have done on this task. 4. I thought about the difficulty of the problems. 5. I thought about my level of ability. 6. I thought about the purpose of the experiment. 7. I thought about how I would feel if I were told how I performed. 8. I thought about how often I get confused. 9. I thought about members of my family. 10. I thought about something that made me feel guilty. 11. I thought about personal worries. 12. I thought about something that made me feel angry. 13. I thought about something that happened earlier today. 14. I thought about something that happened in the recent past (last few days, but not today). 15. I thought about something that happened in the distant past 16. I thought about something that might happen in the future.

Part 4 (Thinking Content) assesses specific intruding thoughts.

Task-related interference (8 items). Summate scores on items 1-8.

Task-irrelevant interference (8 items). Summate scores on items 9-16.

167

Opinions of the Task (only after interaction)

Next, please answer some questions about the task. Please indicate what you thought of the task while you were performing it. Please try to rate the task itself rather than your personal reactions to it. For each adjective or sentence circle the appropriate number, on the six point scales provided (where 0 = not at all to 5 = very much so).

Threatening Enjoyable Fearful Exhilarating Worrying Informative Frightening Challenging Terrifying Stimulating Hostile Exciting

The task was a situation:

Which was likely to get out of control In which you were unsure of how much influence you have In which somebody else was to blame for difficulties In which you had to hold back from doing what you really want Which you could deal with effectively In which efforts to change the situation tended to make it worse In which other people made it difficult to deal with the problem Which was too much for you to cope with

168

Dealing with Problems (only after interaction)

Finally, think about how you dealt with any difficulties or problem which arose while you were performing the task. Below are listed some options for dealing with problems such as poor performance or negative reactions to doing the task. Please indicate how much you used each option, specifically as a deliberately chosen way of dealing with problems. To answer, circle one of the following answers.

Extremely = 4 Very much = 3 Somewhat = 2 A little bit = 1 Not at all = 0

I …

1. Worked out a strategy for successful performance 2. Worried about what I would do next 3. Stayed detached or distanced from the situation 4. Decided to save my efforts for something more worthwhile 5. Blamed myself for not doing better 6. Became preoccupied with my problems 7. Concentrated hard on doing well 8. Focused my attention on the most important parts of the task 9. Acted as though the task wasn’t important 10. Didn’t take the task too seriously 11. Wished that I could change what was happening 12. Blamed myself for not knowing what to do 13. Worried about my inadequacies 14. Made every effort to achieve my goals 15. Blamed myself for becoming too emotional 16. Was single-minded and determined in my efforts to overcome any problems 17. Gave up the attempt to do well 18. Told myself it wasn’t worth getting upset 19. Was careful to avoid mistakes 20. Did my best to follow the instructions for the task 21. Decided there was no point in trying to do well

169

Checklist for Trust between People and Automation

Jian, J-Y., Bisantz, A.M., Drury, C.G., & Llinas, J. (1998). Foundations for an Empirically Determined Scale of Trust in Automated Systems (Report No. AFRL-HE-WP-TR-2000-0102). Wright-Patterson AFB, OH.

Below is a list of statement for evaluating trust between people and automation. There are several scales for you to rate intensity of your feelings of trust, or your impression of the system while operating a machine.

(Note: not at all = 1; extremely = 7)

1. The system is deceptive. 2. The system behaves in an underhanded manner. 3. I am suspicious of the system’s intent, action or outputs. 4. I am wary of the system. 5. The system’s actions will have a harmful or injurious outcome. 6. I am confident in the system. 7. The system provides security. 8. The system has integrity. 9. The system is dependable. 10. The system is reliable. 11. I can trust the system. 12. I am familiar with the system.

170

APPENDIX F: STUDY 1 INFORMED CONSENT

171

172

APPENDIX G: STUDY 1 STIMULI

173

Robot Stimuli

Figure 35. Experiment 1 Industry robot stimuli

Figure 36. Experiment 1 Entertainment Robot Stimuli

174

Figure 37. Experiment 1 Medical robot stimuli

Figure 38. Experiment 1 Military robot stimuli

175

Figure 39. Experiment 1 Service robot stimuli

Figure 40. Experiment 1 Social robot stimuli

176

Figure 41. Experiment 1 Therapy robot stimuli

177

Machine Stimuli

Figure 42. Experiment 1 Machine stimuli

178

Human Stimuli

Figure 43. Experiment 1 Human stimuli

179

APPENDIX H: STUDY 1 ADDITIONAL ANALYSES OF MACHINE

IMAGES

180

Distribution of the Robot Classification Ratings for Machine Stimuli

To further explore the previously noted individual differences in robot classification of machines, a frequency distribution was conducted to identify the variations within individual ratings (see Table 39). Only the X-Ray Machine and Baggage Screener had ratings that varied across the whole of the 1-7 scale amongst participants with over 25% of the participants classifying these machines as robots (see Figure 44). These findings showed some support for the importance of individual differences classification of a machine as a robot.

Table 39

Frequency Ratings on Robot Classification Ratings for each Machine Stimuli

Machine 1 2 3 4 5 6 7

X-Ray Frequency 58 17 9 15 21 7 32 Percentage 36% 11% 6% 9% 13% 4% 20%

Baggage Frequency 71 12 11 19 10 14 22 Screener Percentage 45% 8% 7% 12% 6% 9% 14%

Washing Frequency 79 18 16 13 10 6 17 Machine Percentage 50% 11% 10% 8% 6% 4% 11%

Frequency 91 20 10 13 7 2 16 Car Percentage 57% 13% 6% 8% 4% 1% 10%

Frequency 95 15 8 12 9 4 16 Airplane Percentage 60% 9% 5% 8% 6% 3% 10%

Frequency 138 13 2 3 2 0 1 Ball Percentage 87% 8% 1% 2% 1% 0% 1%

Frequency 143 5 4 4 3 0 0 Hammer Percentage 90% 3% 3% 3% 2% 0% 0% Note. 1 represents ‘Not a Robot’, 4 represents ‘Neutral’, 7 represents ‘Definitely a Robot’ * N = 159

181

Figure 44. Percentages of Scores on the Robot Scale ranging from ‘Not at All’ (1) to ‘Completely’ (7).

182

Individual Difference Analyses of the Baggage Screener and X-ray Machine

Analysis was conducted to explore potential individual differences that could account for the diversity of the robot classification ratings for both the Baggage Screener and X-Ray

Machine stimuli (see Figure 45).

X-ray Machine Baggage Screener

Figure 45. X-ray machine and baggage screener stimuli.

Multiple regression correlation analysis with stepwise entry of variable was conducted to predict robot classification from perceived LOA, perceived intelligence, and the five personality traits from the Mini-IPIP scale (i.e., extraversion, agreeableness, conscientiousness, intellect, and neuroticism). Results showed that the final model of the baggage screener included perceived intelligence and Intellect (from the Mini-IPIP scales) as predictors of robot classification, accounting for 12.3% of the variance. Results also showed that in the final model of the x-ray machine, perceived level of automation was the only predictor of robot classification, accounting for 11.5% of the variance. These findings support the theory that individual difference ratings can lead to fuzzy boundaries of the classification process.

183

MRC analysis of the baggage screener. All baggage screener data were analyzed using

IBM SPSS Statistics v.19 (SPSS, 2010), with an alpha level set to .05, unless otherwise indicated. Table 40 shows the means, standard deviations, and intercorrelations of the predictor and predicted variables.

Table 40

Baggage Screener Means, Standard Deviations, and Intercorrelations

Trust Antecedents Mean SD 1 2 3 4 5 6 7 8 1.Robot Classification 3.08 2.29 1 2.Perceived LOA 2.43 1.55 .167* 1 3.Perceived Intelligence 2.63 1.74 .314** .544** 1 4.Extraversion 17.45 6.13 -.041 -.010 -.032 1 5.Agreeableness 21.48 4.78 -.029 -.054 -.092 .430** 1 6.Conscientiousness 18.37 4.74 -.085 -.082 -.076 .081 .338** 1 7.Intellect 17.94 3.16 -.191** -.072 -.005 .021 .171* .031 1 8.Neuroticism 12.93 4.48 -.007 .024 .028 -.149 -.157* -.035 -.127 1

The first model, which included perceived intelligence as a predictor of robot classification, accounted for a significant R2 of 9.8% of the variance in robot classification,

F(1,159) = 17.349, p < .001. After adjusting for sample size, the R2 reduced to 9.3%. The final model, which included perceived intelligence, and Intellect (from the mini IPIP scale) as predictors of robot classification, accounted for a significant R2 of 13.4% of the variance in robot classification, F(2,158) = 12.260, p < .001. After adjusting for sample size, the R2 reduced to

2 2 12.3%. In the model, there was a significant change in R (change in R = .036), F change(1,158) =

6.564, p = .011. An additional 3.6% of the variance in robot classification was explained by the inclusion of Intellect (from the mini IPIP scale). This shows that perceived intelligence*Intellect

184

contributed significantly to the variance in robot classification of the baggage screener. In this final model, perceived intelligence (sr2 = .314), Intellect (sr2 = .313), perceived intelligence*Intellect (sr2 = -.190) accounted for uniquely significant variance. In the final model

Perceived Intelligence and Intellect were related, Ŷ = 4.461 + 0.410(Perceived Intelligence) –

0.137(Intellect).

MRC analysis of the x-ray machine. All x-ray machine data were analyzed using IBM

SPSS Statistics v.19 (SPSS, 2010), with an alpha level set to .05, unless otherwise indicated.

Table 41 shows the means, standard deviations, and intercorrelations of the predictor and predicted variables.

Table 41

X-ray Machine Means, Standard Deviations, and Intercorrelations

Trust Antecedents Mean SD 1 2 3 4 5 6 7 8 1. Robot Classification 3.44 2.37 1 2. Perceived LOA 2.19 1.51 0.339** 1 3. Perceived Intelligence 2.51 1.79 0.16* 0.712** 1 4. Extraversion 17.41 6.13 -0.122 0.002 -0.034 1 5. Agreeableness 21.51 4.79 0.027 -0.056 -0.061 0.436** 1 6. Conscientiousness 18.40 4.74 0.123 0.042 0.027 0.086 0.336** 1 7. Intellect 17.93 3.17 -0.038 0.075 -0.008 0.017 0.174* 0.035 1 8. Neuroticism 12.93 4.49 0.051* 0.089 -0.001 -0.148* -0.159* -0.036 -0.127 1

The final model, which included perceived LOA as a predictor of robot classification, accounted for a significant R2 of 11.5% of the variance in robot classification, F(1,158) = 20.54, p < .001, Ŷ = 2.274 + 0.535(perceived LOA).

185

APPENDIX I: STUDY 1 ADDITIONAL ANALYSES OF ROBOT IMAGES

186

Robot Classification t-tests

Table 42

Classification of Robot Stimuli

Robot Domain Robot Stimuli Robot Mean (SD) df t (Test Value =4)

Entertainment Asimo 5.81 (1.866) 158 12.238**

Entertainment Ballroom Robot 4.71 (2.443) 160 3.677**

Entertainment Cyclist Robot 5.42 (2.218 160 8.101**

Entertainment Furby 4.39 (2.465) 160 1.982*

Entertainment Qrio 6.56 (1.131) 159 28.582**

Entertainment Robosapian 6.15 (1.626) 159 16.725**

Entertainment Topio 6.41 (1.370) 159 22.211**

Industry Bridge Robot 5.04 (2.265) 160 5.846**

Industry Food Robot 4.91 (2.265) 160 5.115**

Industry Glass Robot 4.87 (2.409) 160 4.581**

Industry Metal Robot 4.60 (2.352) 158 3.237**

Industry Parallel Robot 3.39 (2.470) 160 -3.127**

Industry Parallelizing Operations Robot 5.07 (2.255) 159 5.996**

Industry Robotic Arm 5.02 (2.254) 160 5.734**

Medical DaVinci Robot 4.22 (2.472) 159 1.119

Medical Heart Robot 4.92 (2.254) 159 5.155**

Medical Laparoscopic Robot 4.21 (2.421) 158 1.081

Medical Lifting Robot 6.09 (1.619) 160 16.401**

Medical RP6 4.47 (2.359) 160 2.539*

Medical UBot 5.91 (1.821) 160 13.327**

187

Robot Domain Robot Stimuli Robot Mean (SD) df t (Test Value =4)

Military Big Dog 5.66 (1.933) 160 10.926**

Military Daksh Robot 5.48 (2.025) 158 9.244**

Military IED Detonator 4.66 (2.335) 160 3.578**

Military Packbot 5.80 (1.889) 159 12.051**

Military Reaper 3.99 (2.442) 159 -0.065

Military Talon 5.35 (2.183) 160 7.834**

Military TUGV 4.39 (2.369) 159 2.069*

Service Chore Robot 6.23 (1.538) 160 18.395**

Service Gutter Robot 3.66 (2.361) 159 -1.842

Service Lawnmower Robot 4.62 (2.374) 160 3.319**

Service PatrolBot 5.06 (2.367) 160 5.660**

Service Roomba 4.33 (2.376) 160 1.758

Service Ubiquitous Dream 5.70 (1.946) 160 11.055**

Service SmartPalIV 6.34 (1.327) 160 22.325**

Social Aibo 6.11 (1.580) 160 16.914**

Social Nexi 6.30 (1.524) 160 19.132**

Social Olivia 6.28 (1.480) 159 19.495**

Social PaPeRo 6.18 (1.545) 160 17.908**

Social Paro 1.54 (1.283) 159 -24.276**

Social Simon 6.41 (1.256) 159 24.299**

Social Wakamaru 6.11 (1.637) 158 16.233**

188

Robot Domain Robot Stimuli Robot Mean (SD) df t (Test Value =4)

Therapy Carrier Wheelchair 3.30 (2.346) 156 -3.741**

Therapy iFoot 4.86 (2.240) 158 4.850**

Therapy Kismet 5.91 (1.802) 160 13.426**

Therapy Koala 5.14 (2.187) 160 06.596**

Therapy Lokomat 5.06 (2.305) 159 5.797**

Therapy Tibion 2.05 (1.903) 160 -13.002**

Therapy Torso Robot 5.48 (2.053 158 9.080** Note. Test value of 4 represents neutral point on 7 point robot classification scale.

189

Robot Stimuli Stem and Leaf Plots

Robot Classification

Gutter Robot

190

Robot Classification

Reaper

Robot Classification

Laparoscopic Robot

191

Robot Classification

Furby

Robot Classification

Ballroom Robot

192

Robot Classification

Lawnmower Robot

Robot Classification

Roomba

193

Robot Classification

TUGV

obot Classification R

IED Detonator Robot

194

Robot Classification

DaVinci Robot

Robot Classification

RP6

195

Robot Classification

iFoot

Robot Classification

Metal Robot

196

Robot Classification

Glass Robot

197

Individual Differences Analyses for Robot Stimuli

Further analysis was conducted to explore potential individual differences that could account for the diversity of the robot classification ratings for the robot stimuli reported in Table

13. Multiple regression correlation analysis with stepwise entry of variables was conducted to predict robot classification from perceived LOA; perceived intelligence (PI); the five personality traits from the Mini-IPIP scale: extraversion (E), agreeableness (A), conscientiousness (C), intellect (I), and neuroticism (N); the three subscales from the NARS: negative attitude toward situations of interaction with robots (S1), and social influence of robots (S2) emotions in interactions (S3); as well as the two scale items for previous experience (i.e., previously seen, and previously interacted) with each robot stimulus. All data were analyzed using IBM SPSS

Statistics v.19 (SPSS, 2010), with an alpha level set to .05, unless otherwise indicated. Results showed that number and type of predictor variables of robot classification were unique for each robot stimuli (see Table 43).

Table 43

Percentage of Robot Classification Variance Explained

Robot Name % of # of Final Equation Total Predictor Variance Variables Gutter Robot 9.2% 2 Ŷ = 4.844 + 0.637(PI) - 0.003(S3) Reaper 11.0% 1 Ŷ = 3.426 + 0.742(LOA) Laparoscopic Robot 24.6% 5 Ŷ = -1.562 + 0.422(PI) + 0.088(A)+ 0.155(I)+0.362(S1) – 0.262(S2) Furby 14.3% 2 Ŷ = 0.813 + 0.464(LOA) + 0.122(I) Ballroom Robot 18.1% 3 Ŷ = 3.822 + 0.571(LOA) + 2.744(Seen)-0.008(A) Lawnmower Robot 19.2% 4 Ŷ = 0.887+ 0.767(S1) – 0.546(S3) + 0.355(PI) + 0.092(N) Roomba 19.3% 2 Ŷ = 4.957 + 0.655(LOA) – 0.058(S3)

198

Robot Name % of # of Final Equation Total Predictor Variance Variables TUGV 10.0% 2 Ŷ = 3.014 + 0.309(LOA) + 0.260(PI) IED Detonator Robot 9.5% 3 Ŷ = -0.738 + 0.361(PI) + 0.135(I) + 1.100 (Seen) DaVinci Robot 13.7% 2 Ŷ = -0.909 + 0.555(LOA) + 1.895(Interacted) RP6 2.1% 1 Ŷ = 3.869 + 0.228(PI) iFoot 10.5% 2 Ŷ = 1.062 + 0.350(PI) + 0.164(I) Metal Robot 10.7% 2 Ŷ = 0.232 + 0.306(I) + 0.526(PI) Glass Robot 6.3% 1 Ŷ = 3.917 + 0.560(PI)

MRC analysis of the gutter robot. Table 44 shows the means, standard deviations, and intercorrelations of the predictor and predicted variables. The first model, which included perceived intelligence (PI) as a predictor of robot classification, accounted for a significant R2 of

8.0% of the variance in robot classification, F(1,158) = 71.20, p < .001. After adjusting for sample size, the R2 reduced to 7.5%. The final model, which included perceived intelligence, and negative attitude toward emotions in interactions with robots (NARS_S3) as predictors of robot classification, accounted for a significant R2 of 10.3% of the variance in robot classification, F(2,157) = 9.018, p < .001. After adjusting for sample size, the R2 reduced to

2 2 9.2%. In the model, there was a significant change in R (change in R = .023), F change(1,157) =

3.971, p = .048. An additional 2.3% of the variance in robot classification was explained by the inclusion of NARS_S3. This shows that PI*NARS_S3 contributed significantly to the variance in robot classification of the Gutter robot. In the final model Perceived Intelligence and Intellect were related, Ŷ = 4.844 + 0.637(PI) - 0.003(NARS_S3).

199

Table 44

Gutter Robot Means, Standard Deviations, and Intercorrelations

M SD 1 2 3 4 5 6 7 8 9 10 11 12 13 1.Robot Classification 3.67 2.36 1 . 2.Seen 1.95 0.22 -.155* 1 3.Interacted 1.98 0.14 -.099 .603** 1 4.Perceived LOA 2.87 1.97 .199** -.030 .038 1 5.Perceived Intelligence 2.32 1.65 .283** -.078 -.057 .637** 1 6.Agreeableness 17.41 6.12 .004 -.097 .032 -.024 -.109 1 7.Conscientiousness 21.47 4.79 -.056 -.020 .033 -.050 -.082 .428 1 8.Extraversion 18.34 4.73 -.103 .071 .020 -.076 -.125 .073 .336** 1 9.Intellect 17.92 3.16 .073 .058 .011 -.059 -.055 .012 .167* .022 1 10.Neuroticism 12.94 4.49 -.009 .022 -.012 .092 .089 -.147 -.156* -.032 -.125 1 11.Negative Emotions in Interactions 3.31 1.16 -.127 -.044 -.049 .069 .080 -.077 -.085 .012 -.299** .142* 1 12.Negative Situation of Actions 4.32 1.30 -.011 -.130* -.045 -.002 -.048 .165 .066 .173* -.082 .000 .573** 1 13.Negative Social Influence of robots 4.71 1.43 -.092 -.054 .004 -.108 -.160* .074 .087 .271** -.087 -.053 .190** .300** 1

200

MRC analysis of the reaper. Table 45 shows the means, standard deviations, and intercorrelations of the predictor and predicted variables. The final model, which included perceived LOA as the predictor of robot classification, accounted for a significant R2 of 11.6% of the variance in robot classification, F(1,157) = 20.504, p < .001. After adjusting for sample size, the R2 reduced to 11.0%. In the final model, Ŷ = 3.426 + 0.742(LOA).

201

Table 45

Reaper Robot Means, Standard Deviations, and Intercorrelations

M SD 1 2 3 4 5 6 7 8 9 10 11 12 13 1.Robot 4.00 2.44 1 Classification 2.Seen 1.40 .49 .147* 1 3.Interacted 1.81 .39 .132 .396** 1 4.Perceived LOA 2.35 1.61 .340** .076 .026 1 5.Perceived 2.90 1.80 .313** -.125 .000 .582** 1 Intelligence 6.Agreeableness 17.50 6.14 .005 .029 .016 .022 .056 1 7.Conscientiousness 21.52 4.80 -.030 .103 .069 -.063 .021 .426** 1 8.Extraversion 18.38 4.77 -.165* .133 .049 -.214 -.237** .079 .338** 1 9.Intellect 17.98 3.17 .094 .046 .053 .073 -.077 .012 .164* .029 1 10.Neuroticism 12.98 4.48 .079 .185 .012 .135 .120 -.160* -.167* -.037 -.141* 1 11.Negative Emotions in 3.31 1.16 -.050 -.019 -.152 .008 -.029 -.076 -.083 .011 -.299** .147* 1 Interactions 12.Negative Situation .572* 4.32 1.30 -.025 -.001 -.052 -.070 -.134 .165* .065 .173* -.083 -.002 1 of Actions * 13.Negative Social 4.71 1.44 -.059 -.031 -.029 -.188 -.136* .077 .088 .274** -.084 -.057 .190* .300** 1 Influence of robots

202

MRC analysis of the laparoscopic robot. Table 46 shows the means, standard deviations, and intercorrelations of the predictor and predicted variables. The first model, which included perceived intelligence (PI) as a predictor of robot classification, accounted for a 12.6% of the variance in robot classification, F(1,154) = 22.201, p < .001. After adjusting for sample size, the R2 reduced to 12.0%.

The second model which included PI and agreeableness (A), accounted for a significant

R2 of 18.7% of the variance in robot classification F(2, 153) = 17.538, p<.001. After adjusting for sample size, the R2 reduced to 17.6%. In the model, there was a significant change R2 (change

2 in R =.061), F change(1,153) = 11.380, p =.001. An additional 6.1% of the variance in robot classification was explained by the inclusion of agreeableness (from the Mini-IPIP scale).

The third model which included PI, A, and intellect (I) as predictors of robot classification, accounted for a significant R2 of 22.8% of the variance in robot classification, F(3,

152) = 14.975, p<.001. After adjusting for sample size, the R2 reduced to 21.3%. In the model,

2 2 there was a significant change R (change in R =.042), F change(1,152) = 8.196, p =.005. An additional 4.2% of the variance in robot classification was explained by the inclusion of intellect

(from the Mini-IPIP scale).

The forth model which included PI, A, I, and negative attitude toward situations of action with robots (NARS_S1) as predictors of robot classification, accounted for a significant R2 of

25% of the variance in robot classification, F(4, 151) = 12.591, p<.001. After adjusting for sample size, the R2 reduced to 23.0%. In the model, there was a significant change R2 (change in

2 R =.022), F change(1,151) = 4.427, p =.037. An additional 2.2% of the variance in robot classification was explained by the inclusion of NARS_S1.

203

Table 46

Laparoscopic Robot Means, Standard Deviations, and Intercorrelations

M SD 1 2 3 4 5 6 7 8 9 10 11 12 13 1.Robot 4.19 2.41 1 Classification 2.Seen 1.74 .44 .078 1 3.Interacted 1.91 .29 .184* .432** 1 4.Perceived LOA 2.35 1.61 .252** .102 .013 1 5.Perceived 2.64 1.84 .355** .037 .098 .725** 1 Intelligence 6.Agreeableness 17.44 6.16 .303** -.049 .140* .047 .171* 1 7.Conscientiousness 21.43 4.78 .145* .019 .207** -.057 .102 .423** 1 8.Extraversion 18.35 4.77 -.090 -.028 .004 -.228** -.161* .061 .326** 1 9.Intellect 17.90 3.20 .181* -.023 .151* -.026 -.091 .024 .170* .031 1 10.Neuroticism 12.87 4.49 -.014 .071 -.055 .044 -.005 -.151* -.153* -.034 -.136* 1 11.Negative Emotions in 3.32 1.14 -.062 .009 -.044 .157* .045 -.077 -.056 .018 -.303** .139* 1 Interactions 12.Negative Situation 4.32 1.31 .172* -.005 .064 .118 .004 .170* .078 .180* -.077 -.004 .577** 1 of Actions 13.Negative Social 4.72 1.40 -.114 -.066 .033 -.016 -.081 .095 .126 .301** -.087 -.079 .166* .314** 1 Influence of robots

204

The final model, which included PI, A, I, NS, and negative attitude toward social influence of robots (NARS_S2) as the predictors of robot classification, accounted for a significant R2 of 27.1% of the variance in robot classification, F(5, 150) = 11.141, p < .001. After adjusting for sample size, the R2 reduced to 24.6%. In the model, there was a significant change

2 2 R (change in R =.021), F change(1,150) = 4.256, p =.041. An additional 2.1% of the variance in robot classification was explained by the inclusion of NARS_S2. In the final model, Ŷ = -1.562 +

0.422(PI) + 0.088(A)+ 0.155(I)+0.362(NARS_S1) – 0.262(NARS_S2).

MRC analysis of Furby. Table 47 shows the means, standard deviations, and intercorrelations of the predictor and predicted variables. The first model, which included perceived level of automation (LOA) as a predictor of robot classification, accounted for a 12.9% of the variance in robot classification, F(1,157)= 23.340, p < .001. After adjusting for sample size, the R2 reduced to 12.4%. The final model which included LOA and intellect (I), accounted for a significant R2 of 15.4% of the variance in robot classification F(2, 156) = 14.165, p<.001.

After adjusting for sample size, the R2 reduced to 14.3%. In the model, there was a significant

2 2 change R (change in R =.024), F change(1,156) = 4.473, p =.036. An additional 2.4% of the variance in robot classification was explained by the inclusion of intellect (from the Mini-IPIP scale). In the final model, Ŷ = 0.813 + 0.464(LOA) + 0.122(I).

205

Table 47 Furby Means, Standard Deviations, and Intercorrelations M SD 1 2 3 4 5 6 7 8 9 10 11 12 13 1.Robot 4.42 2.46 1 Classification 2.Seen 1.08 .27 -.079 1 3.Interacted 1.16 .37 -.062 .675** 1 4.Perceived LOA 3.06 1.93 .360** .003 -.102 1 5.Perceived 2.04 1.43 .278** .167* .082 .489** 1 Intelligence 6.Agreeableness 17.48 6.16 .123 .089 .037 .026 .087 1 7.Conscientiousness 21.50 4.76 .092 -.075 -.050 .047 .013 .433** 1 8.Extraversion 18.45 4.72 .061 -.248** -.143* .019 -.076 .074 .341** 1 9.Intellect 17.96 3.14 .147* -.048 .016 -.024 -.017 .017 .149* .023 1 10.Neuroticism 12.93 4.48 -.023 .092 .034 .019 .005 -.150 -.143* -.035 -.112 1 11.Negative Emotions in 3.31 1.16 -.073 .006 -.016 .058 .069 -.077 -.084 .011 -.301** .142* 1 Interactions 12.Negative 4.32 1.30 .115 .089 .090 .045 .009 .165* .079 .172* -.071 -.010 .574** 1 Situation of Actions 13.Negative Social 4.72 1.43 -.116 -.139* -.214** -.058 -.087 .076 .091 .271** -.083 -.057 .189** .299** 1 Influence of robots

206

MRC analysis of the ballroom robot. Table 48 shows the means, standard deviations, and intercorrelations of the predictor and predicted variables. The first model, which included perceived level of automation (LOA) as a predictor of robot classification, accounted for a 11.4% of the variance in robot classification, F(1,156)= 20.088, p < .001. After adjusting for sample size, the R2 reduced to 10.8%.

The second model, which included LOA and previously seen, accounted for a significant

R2 of 17.1% of the variance in robot classification, F(2, 155) = 15.963, p<.001. After adjusting for sample size, the R2 reduced to 16.0%. In the model, there was a significant change R2, F(1,

155) = 10.602, p=.001. An additional 5.7% of the variance in robot classification was explained by the inclusions of previously seen.

The final model which included LOA, previously seen, and agreeableness (A), accounted for a significant R2 of 19.7% of the variance in robot classification F(3, 154) = 12.576, p<.001.

After adjusting for sample size, the R2 reduced to 18.1%. In the model, there was a significant

2 2 change R (change in R =.026), F change(1,154) = 4.982, p =.027. An additional 2.6% of the variance in robot classification was explained by the inclusion of agreeableness (from the Mini-

IPIP scale). In the final model, Ŷ = 3.822 + 0.571(LOA) + 2.744(Seen)-0.008(A).

207

Table 48

Ballroom Robot Means, Standard Deviations, and Intercorrelations

M SD 1 2 3 4 5 6 7 8 9 10 11 12 13 1.Robot 4.69 2.45 1 Classification 2.Seen 1.87 .33 .271** 1 3.Interacted 1.92 .28 .264** .787** 1 4.Perceived LOA 2.84 2.01 .338** .102 .125 1 5.Perceived 2.42 1.75 .277** .102 .059 .763** 1 Intelligence 6.Agreeableness 17.35 6.07 -.153* -.050 .032 .064 .002 1 7.Conscientiousness 21.42 4.80 .054 .137* .186* -.004 -.038 .427** 1 8.Extraversion 18.27 4.72 -.030 .010 -.017 -.149* -.147* .062 .326** 1 9.Intellect 17.96 3.19 .115 .181* .068 .031 -.062 .014 .175* .036 1 - 12.95 4.49 -.151* -.214** -.090 .029 -.134 -.159* -.032 -.122 10.Neuroticism .200** 1 11.Negative Emotions in 3.31 1.15 -.129 -.121 -.167* -.001 .026 -.088 -.095 .003 -.304** .141* 1 Interactions 12.Negative 4.31 1.29 -.071 -.002 -.023 .073 -.016 .150* .058 .166* -.086 .006 .568** 1 Situation of Actions 13.Negative Social 4.69 1.43 .037 -.028 .054 -.013 -.074 .073 .085 .268** -.079 -.052 .198* .307** 1 Influence of robots

208

MRC analysis of the lawnmower robot. Table 49 shows the means, standard deviations, and intercorrelations of the predictor and predicted variables. The first model, which included perceived level of automation (LOA) as a predictor of robot classification, accounted for a 8.0% of the variance in robot classification, F(1,157)= 13.719, p < .001. After adjusting for sample size, the R2 reduced to 7.5%.

The second model, which included LOA and negative attitude toward situations of interactions with robots (NARS_S1), accounted for a significant R2 of 14.2% of the variance in robot classification, F(2, 156) = 12.906, p<.001. After adjusting for sample size, the R2 reduced to 13.1%. In the model, there was a significant change R2, F(1, 156) = 11.202, p=.001. An additional 6.2% of the variance in robot classification was explained by the inclusions of NS.

The third model, which included LOA, NARS_S1, and negative attitude toward emotions in interactions with robots (NARS_S3), accounted for a significant R2 of 17.3% of the variance in robot classification, F(3, 155) = 10.801, p<.001. After adjusting for sample size, the R2 reduced to 15.7%. In the model, there was a significant change R2, F(1, 155) = 5.797, p=.017. An additional 3.1% of the variance in robot classification was explained by the inclusions of NE.

The fourth model, which included LOA, NARS_S1, NARS_S3, and PI, accounted for a significant R2 of 19.5% of the variance in robot classification, F(4, 154) = 9.305, p<.001. After adjusting for sample size, the R2 reduced to 17.4%. In the model, there was a significant change

R2, F(1, 154) = 4.155, p=.043. An additional 2.2% of the variance in robot classification was explained by the inclusions of PI.

209

Table 49

Lawnmower Robot Means, Standard Deviations, and Intercorrelations

M SD 1 2 3 4 5 6 7 8 9 10 11 12 13 1.Robot 4.61 2.38 1 Classification 2.Seen 1.85 .36 -.032 1 3.Interacted 1.91 .29 .083 .765** 1 - 3.16 2.08 .283** -.090 4.Perceived LOA .290** 1 5.Perceived 2.87 1.81 .273** -.109 -.048 .578** 1 Intelligence 6.Agreeableness 17.50 6.14 .110 .106 .065 -.088 -.027 1 7.Conscientiousness 21.52 4.79 -.067 .046 .071 -.079 -.122 .425** 1 8.Extraversion 18.36 4.76 .012 -.020 -.007 -.025 -.183** .081 .339** 1 9.Intellect 17.94 3.18 .133* -.085 -.006 .008 -.042 .020 .170* .029 1 10.Neuroticism 12.93 4.48 .140* -.018 -.063 .126 .036 -.147* -.153* -.028 -.124* 1 11.Negative Emotions in 3.30 1.14 .000 .084 .055 .007 .008 -.065 -.071 .015 -.299** .133* 1 Interactions 12.Negative 4.32 1.28 .261** .040 .027 .046 .000 .172** .075 .187* -.075 -.018 .572** 1 Situation of Actions 13.Negative Social 4.73 1.43 -.021 -.027 .054 .067 -.158* .070 .081 .277** -.083 -.052 .208** .309** 1 Influence of robots

210

The fifth model, which included LOA, NARS_S1, NARS_S3, PI, and neuroticism (N), accounted for a significant R2 of 21.8% of the variance in robot classification,

F(5, 153) = 8.533, p<.001. After adjusting for sample size, the R2 reduced to 19.2%. In the model, there was a significant change R2, F(1, 153) = 4.581, p=.034. An additional 2.3% of the variance in robot classification was explained by the inclusions of N.

The final model, which removed LOA, accounted for a significant R2 of 20.6% of the variance in robot classification, F(4, 154) = 9.987, p<.001. After adjusting for sample size, the R2 reduced to 18.5%. In the model, there was not a significant change R2, F(1, 153) = 2.364, p=.126. In the final model, Ŷ = 0.887+ 0.767(NARS_S1) – 0.546(NARS_S3) + 0.355(PI) +

0.092(N).

MRC analysis of Roomba. Table 50 shows the means, standard deviations, and intercorrelations of the predictor and predicted variables. The first model, which included perceived level of automation (LOA) as a predictor of robot classification, accounted for a 17.5% of the variance in robot classification, F(1,158)= 33.512, p < .001. After adjusting for sample size, the R2 reduced to 17.0%.

The final model, which included LOA and NARS_S3, accounted for a significant R2 of

20.3% of the variance in robot classification, F(2, 157) = 20.053, p<.001. After adjusting for sample size, the R2 reduced to 19.3%. In the model, there was a significant change R2, F(1,

157)=5.616, p=.019. An additional 2.8% of the variance in robot classification was explained by the inclusions of NE. In the final model, Ŷ = 4.957 + 0.655(LOA) – 0.058(NARS_S3).

211

Table 50

Roomba Robot Means, Standard Deviations, and Intercorrelations

M SD 1 2 3 4 5 6 7 8 9 10 11 12 13 1.Robot 4.32 2.38 1 Classification - 1.57 .50 2.Seen .202** 1 3.Interacted 1.69 .46 -.125 .774** 1 4.Perceived LOA 3.38 2.05 .418** -.391** -.299** 1 5.Perceived 2.87 1.85 .276** -.227** -.217** .589** 1 Intelligence 6.Agreeableness 17.44 6.14 -.078 .029 .002 .008 .113 1 7.Conscientiousness 21.46 4.79 -.046 .100 .085 -.035 -.014 .429** 1 8.Extraversion 18.39 4.75 -.097 .054 .030 -.127 -.070 .082 .343** 1 9.Intellect 17.96 3.17 .099 .124 .144* -.098 -.138* .022 .174* .029 1 10.Neuroticism 12.93 4.49 .012 .074 .062 -.049 .032 -.149* -.157* -.036 -.128 1 11.Negative Emotions in 3.31 1.16 -.159* .121 .002 .024 .035 -.077 -.084 .010 -.300** .142* 1 Interactions 12.Negative 4.32 1.30 -.036 .182* .094 .043 .050 .166* .067 .174* -.080 .000 .572** 1 Situation of Actions 13.Negative Social 4.72 1.43 -.065 .151* .098 -.114 -.039 .080 .094 .271** -.086 -.056 .189** .300** 1 Influence of robots

212

MRC analysis of the TUGV robot. Table 51 shows the means, standard deviations, and intercorrelations of the predictor and predicted variables. The first model, which included perceived level of automation (LOA) as a predictor of robot classification, accounted for a 8.8% of the variance in robot classification, F(1,156)= 15.121, p < .001. After adjusting for sample size, the R2 reduced to 8.3%.

The final model, which included LOA and PI, accounted for a significant R2 of 11.1% of the variance in robot classification, F(2, 155) = 9.602, p<.001. After adjusting for sample size, the R2 reduced to 10.0%. In the model, there was a significant change R2, F(1, 155) = 3.974, p=.048. An additional 2.3% of the variance in robot classification was explained by the inclusions of PI. In the final model, Ŷ = 3.014 + 0.309(LOA) + 0.260(PI).

213

Table 51

TUGV Robot Means, Standard Deviations, and Intercorrelations

M SD 1 2 3 4 5 6 7 8 9 10 11 12 13 1.Robot 4.39 2.38 1 Classification 2.Seen 1.58 .50 .053 1 3.Interacted 1.96 .21 .022 .251** 1 4.Perceived LOA 2.34 1.55 .297** .087 -.053 1 5.Perceived 2.50 1.64 .287** -.035 -.047 .537** 1 Intelligence 6.Agreeableness 17.42 6.13 -.067 -.140* -.136* -.051 .116 1 7.Conscientiousness 21.52 4.81 -.021 -.129 -.118 -.133* .108 .429** 1 8.Extraversion 18.34 4.78 -.144* -.137* -.004 -.221** -.164* .078 .343** 1 9.Intellect 17.91 3.18 .131 -.046 .032 .001 .039 .024 .182* .027 1 10.Neuroticism 12.97 4.49 .024 .045 .053 .101 .127 -.138* -.158* -.030 -.126 1 11.Negative Emotions in 3.32 1.16 -.016 .010 -.007 .141* .027 -.070 -.086 .014 -.299** .137* 1 Interactions 12.Negative 4.34 1.29 -.016 -.035 .032 .082 -.007 .159* .052 .181* -.064 .000 .581** 1 Situation of Actions 13.Negative Social 4.70 1.43 .008 -.076 -.030 -.036 -.160* .067 .083 .275** -.079 -.047 .194** .294** 1 Influence of robots

214

MRC analysis of the IED Detonator robot. Table 52 shows the means, standard deviations, and intercorrelations of the predictor and predicted variables. The first model, which included PI as a predictor of robot classification, accounted for a 5.5% of the variance in robot classification, F(1,158)= 9.133, p = .003. After adjusting for sample size, the R2 reduced to 4.9%.

The second model, which included PI and I as a predictors of robot classification, accounted for a 8.9% of the variance in robot classification, F(1,157)= 7.675, p = .001. After adjusting for sample size, the R2 reduced to 7.7%. In the model, there was a significant change

R2, F(1, 157) = 5.931, p=.016. An additional 3.4% of the variance in robot classification was explained by the inclusions of I.

The final model, which included PI, I, and previously seen, accounted for a significant R2 of 11.2% of the variance in robot classification, F(3, 156) = 6.549, p<.001. After adjusting for sample size, the R2 reduced to 9.5%. In the model, there was a significant change R2, F(1, 156) =

4.002, p=.047. An additional 2.3% of the variance in robot classification was explained by the inclusions of previously seen. In the final model, Ŷ = -0.738 + 0.361(PI) + 0.135(I) + 1.100

(Seen).

215

Table 52

IED Detonator Robot Means, Standard Deviations, and Intercorrelations

M SD 1 2 3 4 5 6 7 8 9 10 11 12 13 1.Robot 4.65 2.34 1 Classification 2.Seen 1.88 .32 .119 1 3.Interacted 1.97 .17 .050 .489** 1 4.Perceived LOA 2.48 1.60 .149* -.012 -.082 1 5.Perceived 2.49 1.67 .234 -.146* -.227** .688** 1 Intelligence 6.Agreeableness 17.49 6.12 -.020 -.119 -.015 -.079 .018 1 7.Conscientiousness 21.51 4.78 -.046 -.143* -.048 -.144* -.061 .426** 1 8.Extraversion 18.43 4.71 -.075 -.066 -.007 -.228** -.170* .068 .332** 1 9.Intellect 17.93 3.17 .182** .022 .053 .011 -.013 .029 .177* .043 1 10.Neuroticism 12.87 4.43 .017 -.020 .060 .099 .021 -.136* -.148* -.012 -.142* 1 11.Negative Emotions in 3.31 1.16 -.068 -.016 -.022 .107 -.024 -.081 -.088 .006 -.297** .150* 1 Interactions 12.Negative 4.33 1.29 -.010 .016 .113 .065 -.092 .156* .058 .160* -.072 .019 .573** 1 Situation of Actions 13.Negative Social 4.72 1.43 -.034 -.068 .082 -.101 -.146* .075 .086 .272** -.079 -.049 .188** .298** 1 Influence of robots

216

MRC analysis of the DaVinci robot. Table 53 shows the means, standard deviations, and intercorrelations of the predictor and predicted variables. The first model, which included

LOA as a predictor of robot classification, accounted for a 11.0% of the variance in robot classification, F(1,157)= 19.350, p < .001. After adjusting for sample size, the R2 reduced to

10.4%.

The final model, which included LOA and previously interacted, accounted for a significant R2 of 13.7% of the variance in robot classification, F(2, 156) = 12.419, p<.001. After adjusting for sample size, the R2 reduced to 12.6%. In the model, there was a significant change

R2, F(1, 156) = 4.996, p=.027. An additional 2.8% of the variance in robot classification was explained by the inclusions of previously interacted. In the final model, Ŷ = -0.909 +

0.555(LOA) + 1.895(Interacted).

217

Table 53

DaVinci Robot Means, Standard Deviations, and Intercorrelations

M SD 1 2 3 4 5 6 7 8 9 10 11 12 13 1.Robot Classification 4.20 2.47 1 2.Seen 1.87 .34 .039 1 3.Interacted 1.95 .22 .112 .590** 1 4.Perceived LOA 2.55 1.59 .331** -.111 .090 1 5.Perceived 2.79 1.91 .267** -.121 .000 .758** 1 Intelligence 6.Agreeableness 17.51 6.12 .046 -.053 .090 .075 .123 1 7.Conscientiousness 21.50 4.77 -.007 -.084 .000 -.013 .077 .424** 1 8.Extraversion 18.41 4.74 -.102 -.041 -.004 -.099 -.029 .069 .331** 1 9.Intellect 17.92 3.18 .062 -.044 .022 -.106 -.197* .028 .176* .037 1 10.Neuroticism 12.91 4.48 .051 .067 .053 .083 .027 -.140* -.146* -.025 -.132* 1 11.Negative Emotions 3.32 1.15 -.021 .079 .018 .180* .079 -.074 -.075 .016 -.300** .136* 1 in Interactions 12.Negative Situation 4.32 1.30 .081 -.020 .047 .127 .076 .168* .066 .175* -.081 .000 .577** 1 of Actions 13.Negative Social 4.72 1.39 .000 -.092 -.033 .036 -.002 .096 .123 .300** -.085 -.080 .172* .312** 1 Influence of robots

218

MRC analysis of the RP6 medical robot. Table 54 shows the means, standard deviations, and intercorrelations of the predictor and predicted variables. The final model, which included PI as a predictor of robot classification, accounted for a 2.1% of the variance in robot classification, F(1,158)= 4.351, p =.039. After adjusting for sample size, the R2 reduced to 2.1%.

In the final model, Ŷ = 3.869 + 0.228(PI).

219

Table 54

RP6 Robot Means, Standard Deviations, and Intercorrelations

M SD 1 2 3 4 5 6 7 8 9 10 11 12 13 1.Robot 4.47 2.37 1 Classification 2.Seen 1.89 .31 .077 1 3.Interacted 1.94 .24 .040 .665 1 4.Perceived LOA 2.51 1.67 .093 -.005 -.030 1 5.Perceived 2.63 1.70 .164* -.003 -.041 .774** 1 Intelligence 6.Agreeableness 17.47 6.14 .161* -.050 -.124 -.001 .079 1 7.Conscientiousness 21.53 4.76 .032 .004 -.020 .003 .038 .428** 1 8.Extraversion 18.39 4.75 .108 .075 .043 -.028 -.030 .079 .336** 1 9.Intellect 17.96 3.17 .067 .143* .111 -.028 -.054 .019 .166* .030 1 10.Neuroticism 12.91 4.48 .100 .015 -.086 .025 .064 -.147* -.152* -.033 -.125 1 11.Negative Emotions in 3.30 1.15 .014 -.107 -.056 .131* .056 -.074 -.076 .014 -.296** .138* 1 Interactions 12.Negative 4.32 1.30 .035 -.088 -.036 .097 .089 .167* .069 .175* -.079 -.001 .573** 1 Situation of Actions 13.Negative Social 4.72 1.43 -.049 -.054 .033 -.018 .003 .077 .085 .273** -.083 -.053 .193** .301** 1 Influence of robots

220

MRC analysis of the iFoot robot. Table 55 shows the means, standard deviations, and intercorrelations of the predictor and predicted variables. The first model, which included PI as a predictor of robot classification, accounted for a 5.2% of the variance in robot classification,

F(1,156)= 8.496, p =.004. After adjusting for sample size, the R2 reduced to 4.6%.

The final model, which included PI and I, accounted for a significant R2 of 10.5% of the variance in robot classification, F(2, 155) = 9.123, p<.001. After adjusting for sample size, the R2 reduced to 9.4%. In the model, there was a significant change R2, F(1, 155) = 9.298, p=.003. An additional 5.4% of the variance in robot classification was explained by the inclusions of I. In the final model, Ŷ = 1.062 + 0.350(PI) + 0.164(I).

221

Table 55

iFoot Robot Means, Standard Deviations, and Intercorrelations

M SD 1 2 3 4 5 6 7 8 9 10 11 12 13 1.Robot 4.85 2.24 1 Classification 2.Seen 1.95 .22 -.029 1 3.Interacted 1.99 .11 .043 .490** 1 4.Perceived LOA 2.42 1.66 .225* -.098 -.039 1 5.Perceived 2.41 1.62 .227** -.120 -.041 .792** 1 Intelligence 6.Agreeableness 17.42 6.11 .100 .068 .091 .038 .088 1 7.Conscientiousness 21.44 4.80 .148* .118 .105 -.015 .050 .422** 1 8.Extraversion 18.35 4.77 .002 .150* .080 -.147 -.172* .079 .340** 1 9.Intellect 17.91 3.18 .204** .175* .086 -.018 -.113 .015 .164* .029 1 10.Neuroticism 12.94 4.49 .002 .010 -.065 .104 .068 -.136* -.147* -.035 -.123 1 11.Negative Emotions in 3.32 1.15 -.028 -.055 -.100 .137* .164* -.069 -.073 .007 -.292** .132* 1 Interactions 12.Negative 4.33 1.28 .100 -.053 -.015 .048 .034 .187** .085 .174* -.068 -.018 .561** 1 Situation of Actions 13.Negative Social 4.73 1.42 -.026 .010 .031 -.041 -.058 .075 .095 .272** -.074 -.060 .171* .288** 1 Influence of robots

222

MRC analysis of the metal industrial robot. Table 56 shows the means, standard deviations, and intercorrelations of the predictor and predicted variables. The first model, which included I as a predictor of robot classification, accounted for a 6.2% of the variance in robot classification, F(2, 157) = 10.422, p=.002. After adjusting for sample size, the R2 reduced to

5.6%. The final model, which included I and PI, accounted for a significant R2 of 11.8% of the variance in robot classification, F(2, 156) = 10.453, p<.001. After adjusting for sample size, the

R2 reduced to 10.7%. In the model, there was a significant change R2, F(1, 156) = 9.894, p=.002.

An additional 5.6% of the variance in robot classification was explained by the inclusions of PI.

In the final model, Ŷ = 0.232 + 0.306(I) + 0.526(PI).

223

Table 56

Metal Robot Means, Standard Deviations, and Intercorrelations

M SD 1 2 3 4 5 6 7 8 9 10 11 12 13 1.Robot 4.60 2.35 1 Classification 2.Seen 1.81 .40 -.131* 1 3.Interacted 1.97 .18 .016 .275** 1 4.Perceived LOA 3.08 1.8 .122 -.152* -.051 1 5.Perceived 2.68 1.72 .223** -.138* -.013 .689** 1 Intelligence 6.Agreeableness 17.47 6.10 .124 -.035 .043 .075 .090 1 7.Conscientiousness 21.53 4.78 .082 .082 .202** .025 .020 .426** 1 8.Extraversion 18.42 4.74 -.051 .000 .153* -.189** -.117 .071 .330** 1 9.Intellect 17.96 3.16 .250** -.021 .204** .014 -.052 .038 .176* .034 1 10.Neuroticism 12.88 4.48 -.015 .100 -.206** -.059 .016 -.143* -.148* -.024 -.129 1 11.Negative - 3.31 1.16 .086 -.027 -.063 .031 -.074 -.082 .015 -.303** .140* Emotions in .185** 1 Interactions 12.Negative Situation 4.32 1.30 -.064 -.055 .045 -.064 .028 .171* .066 .175* -.085 .001 .573** 1 of Actions 13.Negative Social 4.70 1.42 -.081 -.025 .106 -.135* -.081 .099 .103 .291** -.094 -.066 .187** .301** 1 Influence of robots

224

MRC analysis of the glass industrial robot. Table 57 shows the means, standard deviations, and intercorrelations of the predictor and predicted variables. The final model, which included PI, accounted for a significant R2 of 6.9% of the variance in robot classification, F(1,

156) = 11.609, p=.001. After adjusting for sample size, the R2 reduced to 6.3%. In the final model, Ŷ = 3.917 + 0.560(PI).

225

Table 57

Glass Robot Means, Standard Deviations, and Intercorrelations

M SD 1 2 3 4 5 6 7 8 9 10 11 12 13 1.Robot 4.89 2.41 1 Classification 2.Seen 1.75 .44 -.076 1 3.Interacted 1.97 .18 .142* .227** 1 4.Perceived LOA 2.96 1.90 .237** .002 .072 1 5.Perceived 2.73 1.79 .263** .011 .014 .573** 1 Intelligence 6.Agreeableness 17.37 6.14 .108 -.157* .017 -.007 .017 1 7.Conscientiousness 21.40 4.77 .005 -.049 .091 -.024 .003 .423** 1 8.Extraversion 18.39 4.73 -.092 -.070 .030 -.154* -.128 .077 .337** 1 9.Intellect 17.97 3.16 .118 -.001 .056 -.014 -.047 .023 .171* .009 1 10.Neuroticism 12.98 4.50 -.031 .257** -.025 .030 .033 -.139* -.146* -.033 -.133* 1 11.Negative Emotions in 3.32 1.15 -.085 .091 -.009 -.061 -.003 -.074 -.071 .027 -.290** .139* 1 Interactions 12.Negative Situation 4.33 1.30 .074 -.002 .087 -.032 .018 .170* .077 .169* -.089 -.005 .577** 1 of Actions 13.Negative Social 4.71 1.44 -.165* -.085 -.029 -.166* -.207** .073 .083 .271** -.087 -.051 .196** .302** 1 Influence of robots

226

APPENDIX J: STUDY 2 INFORMED CONSENT

227

228

APPENDIX K: STUDY 2 ADDITIONAL ANALYSES

229

Pearson Correlations for Robot Attributes

Table 58

Study 2 Correlation Analysis for Robot Attributes

Qrio Simon Mean SD rRobot rTrustworthy Mean SD rRobot rTrustworthy Robot 6.50 0.84 1 .083 3.73 1.23 .157* .173* Trustworthy 4.75 1.50 .083 1 3.05 1.64 .139* .159* Interact / Use 4.97 1.60 -.034 .581** 3.11 1.50 .046 .255** Humanlike 1.65 0.90 -.034 .581* 3.02 1.56 .148* .222** Lifelike 1.70 0.95 .055 .185** 3.25 1.50 .101 .191** Natural 1.53 0.84 -.058 .120 3.08 1.56 .122 .167* Conscious 1.82 1.32 .002 .154* 2.71 1.49 .111 .129 Move Elegantly 1.82 0.99 -.055 .203** 2.75 1.59 .081 .247** Organic 1.32 0.65 -.062 .148* 2.42 1.53 .111 .216** Alive 1.80 1.29 -.020 .174* 2.59 1.57 .095 .244** Lively 2.28 1.14 .038 .221** 3.49 1.24 -.111 -.136* Responsive 3.01 1.40 .047 .115 3.58 1.31 -.029 -.218** Interactive 3.48 1.24 .081 .175* 3.80 1.36 .060 -.155* Nice 3.00 1.49 .041 .247** 2.83 1.68 -.090 -.042 Pleasant 3.07 1.36 -.021 .254** 3.10 1.24 .132 -.148* Like 3.22 1.31 -.053 .228** 4.23 1.02 .016 -.131 Friendly 3.03 1.32 -.076 .202** 2.84 1.75 -.050 -.105 Kind 2.72 1.46 -.069 .192** 2.43 1.25 -.120 -.106 Knowledgeable 2.78 1.34 .034 .121 2.17 1.27 -.041 -.121 Competent 3.00 1.30 -.066 .099 2.10 1.11 -.028 -.120 Sensible 2.56 1.42 .004 .129 2.00 1.29 .083 -.181** Responsible 2.68 1.46 -0.02 .120 2.25 1.31 -.086 -.235** Machinelike 4.35 0.90 -.022 -.159* 2.31 1.41 -.061 -.239** Artificial 4.27 0.99 -.035 -.124 2.09 1.21 .024 -.211** Fake 4.29 1.09 .045 -.049 2.06 1.25 .078 -.204** Unconscious 3.46 1.71 -.072 -.111 2.45 1.42 .100 -.141* Move Rigidly 4.00 1.17 -.012 -.115 2.27 1.43 .155* -.111 Mechanical 4.68 0.65 .062 -.148* 2.42 1.53 .042 -.139* Dead 3.25 1.74 .045 -.022 2.31 1.47 .060 -.132 Stagnant 3.37 1.29 -.06 -.121 3.73 1.23 .157* .173* Apathetic 2.55 1.33 -.063 -.075 3.05 1.64 .139* .159* Inert 2.41 1.19 -.055 -.134 3.11 1.50 .046 .255** Awful 2.11 1.19 -.065 -.242** 3.02 1.56 .148* .222** Unpleasant 2.33 1.18 -.078 -.265** 3.25 1.50 .101 .191** Dislike 2.38 1.18 .000 -.346** 3.08 1.56 .122 .167* Unfriendly 2.45 1.20 .026 -.281** 2.71 1.49 .111 .129 Unkind 2.30 1.31 -.028 -.150* 2.75 1.59 .081 .247** Ignorant 2.85 1.35 -.072 -.053 2.42 1.53 .111 .216** Incompetent 2.54 1.22 .014 -.139* 2.59 1.57 .095 .244** Foolish 2.52 1.40 -.004 -.036 3.49 1.24 -.111 -.136* Irresponsible 2.42 1.38 -.042 -.195** 3.58 1.31 -.029 -.218**

230

Hammer Ball Mean SD rRobot rTrustworthy Mean SD rRobot rTrustworthy Robot 1.45 0.87 1 .020 1.65 1.03 1 -.030 Trustworthy 4.91 1.69 .020 1 4.11 1.71 -.030 1 Interact / Use 5.88 1.35 -.146* .476** 5.24 1.57 -.191** .520** Humanlike 1.21 1.10 .152* -.029 1.33 1.41 .026 -.015 Lifelike 1.13 1.00 .052 -.045 1.35 1.12 .012 .066 Natural 1.41 1.28 .139* -.015 1.42 1.22 .089 .015 Conscious 0.93 0.88 .197** -.001 0.96 1.08 .070 .047 Move Elegantly 1.38 1.46 .231** .002 2.86 1.81 -.065 .133 Organic 1.37 1.05 .051 -.122 1.85 1.61 -.035 -.004 Alive 0.88 0.90 .158* .006 1.00 1.09 .056 .010 Lively 1.03 1.00 .165* -.038 1.63 1.47 .088 -.028 Responsive 1.29 1.44 .030 -.139* 1.55 1.63 .059 .028 Interactive 1.70 1.54 -.050 .116 2.43 1.80 -.059 .035 Nice 1.46 1.64 .036 .016 2.18 1.96 -.024 .184** Pleasant 1.64 1.63 .138* .101 2.67 1.90 -.068 .078 Like 2.78 1.81 .050 .170* 3.09 1.88 -.089 .238** Friendly 1.20 1.48 .217** -.026 1.78 1.89 .011 .076 Kind 1.15 1.44 .089 .011 1.54 1.79 .078 .134 Knowledgeable 0.93 1.15 .085 .007 1.00 1.22 .073 -.028 Competent 1.80 1.75 .155* .056 1.15 1.35 .153* .072 Sensible 1.78 1.92 .151* .025 1.57 1.68 .051 .091 Responsible 1.35 1.60 .173* -.076 1.01 1.28 -.014 .039 Machinelike 3.12 2.10 .098 .015 2.04 1.97 .088 .037 Artificial 3.42 2.08 .017 .101 3.44 1.95 .086 .020 Fake 3.09 2.04 -.030 .055 3.20 1.99 .050 .062 Unconscious 3.28 2.23 -.018 .036 2.82 2.30 -.053 .011 Move Rigidly 2.34 2.07 .041 -.025 1.82 1.38 .108 -.137* Mechanical 3.65 1.83 .050 .032 2.31 1.86 .103 .059 Dead 2.93 2.31 -.014 .046 2.81 2.28 -.055 .005 Stagnant 3.15 2.21 .008 .068 2.75 1.99 -.042 .108 Apathetic 2.46 2.17 -.001 .023 2.32 2.08 .000 -.021 Inert 3.12 2.02 .014 .000 2.56 1.84 -.066 -.035 Awful 1.63 1.78 .110 -.119 1.49 1.49 .125 -.004 Unpleasant 1.90 1.81 .067 -.075 1.71 1.43 .122 -.054 Dislike 1.95 1.48 .093 -.042 1.67 1.33 .150* -.059 Unfriendly 1.71 1.94 .029 -.044 1.50 1.68 .029 -.021 Unkind 1.56 1.85 .030 -.045 1.38 1.64 .051 .039 Ignorant 2.30 2.27 .006 .093 2.35 2.25 -.050 .054 Incompetent 2.04 1.90 .065 .043 2.16 2.13 .032 -.059 Foolish 1.56 1.76 .111 -.027 1.80 1.85 -.008 .019 Irresponsible 1.76 1.94 .049 -.040 1.84 2.06 .053 -.024

231

Glass Robot Metal Robot Mean SD rRobot rTrustworthy Mean SD rRobot rTrustworthy Robot 5.78 1.38 1 .302** 6.13 0.97 1 .303** Trustworthy 4.63 1.50 .302** 1 4.77 1.30 .303** 1 Interact / Use 4.24 1.66 .238** .546** 4.13 1.71 .103 .361** Humanlike 1.28 0.69 -.033 -.002 1.28 0.80 .007 .076 Lifelike 1.41 0.86 -.034 .059 1.40 1.04 .095 .124 Natural 1.44 0.91 -.059 .111 1.45 1.12 .046 .047 Conscious 1.40 1.24 .009 .184** 1.29 1.31 .072 .121 Move Elegantly 2.07 1.22 -.009 .170* 2.08 1.32 .063 .187** Organic 1.22 0.59 .028 .086 1.24 0.75 -.031 -.029 Alive 1.38 1.18 .096 .206** 1.27 1.31 .020 .146* Lively 2.14 1.40 .051 .242** 2.00 1.56 .173* .296** Responsive 2.57 1.73 .052 .117 2.23 1.75 .224** .133 Interactive 2.96 1.54 .043 .205** 2.59 1.65 .117 .153* Nice 1.71 1.59 .117 .142* 1.32 1.42 .028 .011 Pleasant 2.08 1.48 .061 .181** 1.59 1.43 .068 .074 Like 2.42 1.52 .140* .233** 2.10 1.59 .148* .123 Friendly 1.55 1.53 .089 .089 1.13 1.33 .087 .146* Kind 1.50 1.45 .121 .124 1.14 1.35 .056 .100 Knowledgeable 1.80 1.60 .024 .099 1.53 1.52 .099 .180** Competent 2.49 1.78 .061 .207** 2.08 1.79 .094 .170* Sensible 2.27 1.84 .092 .204** 1.83 1.79 .097 .097 Responsible 2.37 1.78 .042 .228** 2.00 1.80 .097 .122 Machinelike 4.69 0.76 -.004 .094 4.51 1.14 -.032 .053 Artificial 4.36 1.19 .100 .120 4.05 1.59 .061 .124 Fake 4.04 1.48 .052 .066 3.54 1.84 .114 .057 Unconscious 3.24 2.01 -.018 .161* 2.72 2.13 .086 .024 Move Rigidly 3.67 1.38 .008 .140* 3.40 1.56 .012 -.046 Mechanical 4.75 0.67 -.063 .029 4.53 1.15 -.004 .042 Dead 3.18 2.00 -.065 .154* 2.76 2.16 .073 .051 Stagnant 3.13 1.63 .015 -.045 2.65 1.82 .005 -.005 Apathetic 2.39 1.68 -.001 .013 2.36 1.80 -.008 .018 Inert 2.61 1.49 -.091 -.037 2.51 1.62 -.004 -.100 Awful 1.87 1.70 .012 .060 1.94 1.93 .023 -.059 Unpleasant 2.39 1.63 -.035 -.065 2.39 1.92 .065 .005 Dislike 2.43 1.52 -.039 -.104 2.34 1.70 .007 -.117 Unfriendly 2.00 1.86 -.043 .047 2.01 2.07 .050 .064 Unkind 2.00 1.84 .043 .015 1.95 2.05 .030 .055 Ignorant 2.33 1.89 .001 -.013 2.25 1.98 .005 .037 Incompetent 2.18 1.66 -.141* -.071 2.04 1.76 .019 .034 Foolish 1.85 1.60 -.054 -.041 1.78 1.76 .023 -.058 Irresponsible 2.02 1.62 -.061 -.069 1.92 1.75 .049 -.065

232

Furby Ballroom Mean SD rRobot rTrustworthy Mean SD rRobot rTrustworthy Robot 4.58 2.07 1 .136* 4.59 1.94 1 .339** Trustworthy 4.29 1.79 .136* 1 4.08 1.55 .339** 1 Interact / Use 4.48 2.02 .156* .477** 4.07 1.70 .322** .643** Humanlike 1.81 0.99 .005 .040 2.65 1.40 -.009 .095 Lifelike 1.88 1.13 .054 .125 2.30 1.38 .119 .166* Natural 1.61 0.96 .013 .168* 1.76 1.12 .167* .111 Conscious 1.74 1.39 .093 .132 1.78 1.46 .243** .138* Move Elegantly 1.89 1.11 .099 .138* 2.55 1.59 .254** .199** Organic 1.71 1.00 .010 .095 1.87 1.19 .173* .171* Alive 1.77 1.48 .204** .065 1.70 1.51 .200** .149* Lively 2.77 1.45 .273** .094 2.50 1.57 .284** .232** Responsive 2.75 1.59 .225** .098 2.18 1.64 .338** .208** Interactive 3.33 1.36 .303** .095 2.61 1.56 .312** .259** Nice 2.85 1.68 .179** .160* 2.44 1.75 .253** .234** Pleasant 2.80 1.49 .068 .199** 2.69 1.66 .108 .194** Like 2.80 1.58 .046 .168* 2.68 1.58 .122 .219** Friendly 3.17 1.64 .175* .181** 2.47 1.75 .247** .272** Kind 2.84 1.71 .156* .181** 2.36 1.78 .285** .180** Knowledgeable 1.70 1.44 .112 -.057 1.74 1.56 .260** 0.11 Competent 1.79 1.42 .093 .026 1.88 1.55 .294** .171* Sensible 1.80 1.41 .110 -.017 1.79 1.57 .275** .158* Responsible 1.63 1.46 .053 .019 1.69 1.60 .282** .210** Machinelike 3.96 1.22 .290** -.121 3.12 1.43 .151* -.059 Artificial 4.03 1.21 .015 -.172* 3.50 1.46 -.040 -.175* Fake 4.25 1.14 .049 -.192** 3.89 1.41 .017 -.103 Unconscious 3.17 1.87 .035 -.010 3.01 1.90 -.024 -.084 Move Rigidly 3.71 1.40 .141* -.016 2.73 1.63 -.003 -.034 Mechanical 4.15 1.16 .111 -.089 3.70 1.49 .182** -.072 Dead 2.99 1.92 .002 -.062 2.80 1.99 -.088 -.104 Stagnant 2.88 1.46 -.071 -.099 2.81 1.64 -.043 -.099 Apathetic 2.50 1.53 .013 -.074 2.41 1.74 -.046 -.037 Inert 2.52 1.32 -.187** -.067 2.73 1.58 -.105 -.092 Awful 2.29 1.53 .098 -.241** 2.06 1.58 .017 -.164* Unpleasant 2.79 1.49 -.002 -.253** 2.30 1.53 -.072 -.275** Dislike 2.74 1.57 .054 -.234** 2.48 1.53 -.039 -.200** Unfriendly 2.25 1.47 -.047 -.270** 2.09 1.59 .098 -.137* Unkind 2.15 1.50 .104 -.259** 1.97 1.59 .095 -.172* Ignorant 2.83 1.94 .086 -.117 2.38 1.91 .024 -.061 Incompetent 2.97 1.87 .040 -.084 2.38 1.81 -.044 -.026 Foolish 3.04 1.85 .083 -.164* 2.25 1.83 .090 -.143* Irresponsible 2.49 1.93 .036 -.131 2.06 1.84 .012 -.061

233

Lawnmower Robot Roomba Mean SD rRobot rTrustworthy Mean SD rRobot rTrustworthy Robot 5.13 1.54 1 .306** 5.15 1.71 1 .406** Trustworthy 4.44 1.36 .306** 1 4.84 1.43 .406** 1 Interact / Use 4.79 1.39 .253** .636** 5.17 1.47 .463** .619** Humanlike 1.26 0.65 .071 .047 1.63 1.46 .069 .109 Lifelike 1.35 0.80 .127 .109 1.25 0.69 -.156* .007 Natural 1.48 0.93 .160* .031 1.45 0.91 -.014 .063 Conscious 1.34 1.09 .068 .064 1.41 0.91 -.044 .128 Move Elegantly 2.42 1.39 -.053 .155* 1.46 1.25 -.031 .060 Organic 1.27 0.73 .016 -.055 2.28 1.58 .184** .238** Alive 1.46 1.20 .049 .041 1.30 0.66 -.121 .019 Lively 1.98 1.27 .132 .077 1.31 1.20 .009 .085 Responsive 2.43 1.53 .087 .160* 2.00 1.41 .195** .172* Interactive 2.63 1.46 .060 .095 2.42 1.63 .121 .226** Nice 2.03 1.58 .121 -.011 2.48 1.42 .219** .218** Pleasant 2.43 1.49 .106 .189** 1.92 1.71 .100 .109 Like 2.82 1.35 .093 .144* 2.41 1.64 .122 .141* Friendly 1.89 1.48 .038 -.001 1.71 1.65 -.003 .171* Kind 1.90 1.60 .031 .052 1.56 1.64 -.050 .100 Knowledgeable 1.89 1.41 .075 .122 1.82 1.57 -.020 .037 Competent 2.39 1.56 .123 .138* 2.37 1.61 .094 .281** Sensible 2.04 1.61 .153* .147* 2.22 1.75 .011 .178* Responsible 2.12 1.59 .172* .284** 2.21 1.70 .070 .143* Machinelike 4.65 0.85 .020 -.003 2.49 1.93 -.026 .114 Artificial 4.50 1.05 -.032 -.039 4.60 0.98 .106 -.001 Fake 4.15 1.37 .108 .037 4.29 1.25 .056 -.101 Unconscious 3.59 1.90 -.021 .001 3.96 1.56 .214** .010 Move Rigidly 3.14 1.50 .136 -.121 3.15 1.98 .083 -.013 Mechanical 4.61 0.96 .097 .081 2.59 1.69 .039 -.045 Dead 3.27 1.94 .054 .056 4.56 0.96 .080 -.030 Stagnant 3.44 1.57 .056 .030 3.04 2.07 .099 -.011 Apathetic 2.79 1.62 .017 -.060 3.05 1.73 -.050 .044 Inert 3.02 1.51 .036 -.031 2.54 1.67 .020 .038 Awful 2.00 1.56 .044 -.123 2.98 1.51 -.015 -.098 Unpleasant 2.38 1.47 -.052 -.160* 1.75 1.59 -.039 -.052 Dislike 2.61 1.30 -.117 -.380** 2.12 1.50 -.079 -.088 Unfriendly 2.32 1.72 -.054 -.192** 1.89 1.78 -.066 -.035 Unkind 2.00 1.66 .066 .072 1.70 1.76 -.073 -.036 Ignorant 2.87 1.79 -.011 -.120 2.39 1.87 -.024 .047 Incompetent 2.54 1.61 .018 -.076 2.31 1.59 -.041 -.008 Foolish 2.17 1.68 .023 -.160* 2.00 1.63 -.100 .010 Irresponsible 2.33 1.69 -.005 -.105 2.09 1.64 -.062 -.054

234

Gutter Robot Reaper Mean SD rRobot rTrustworthy Mean SD rRobot rTrustworthy Robot 4.13 1.80 1 .516** 4.95 1.87 1 .137* Trustworthy 3.88 1.32 .516** 1 4.70 1.50 .137* 1 Interact / Use 3.77 1.52 .486** .703** 4.55 1.78 -.008 .500** Humanlike 1.29 0.84 -.067 .041 1.36 0.85 .048 .094 Lifelike 1.32 0.91 .007 .080 1.59 1.18 .129 .079 Natural 1.47 1.03 -.044 .028 1.55 1.11 .150* .085 Conscious 1.24 1.08 -.111 .108 1.45 1.41 .234** .104 Move Elegantly 1.70 1.24 .098 .131 3.41 1.57 .118 .190** Organic 1.35 0.87 -.134 -.022 1.36 0.86 .143* -.066 Alive 1.31 1.16 -.022 .125 1.38 1.38 .179** .093 Lively 1.70 1.28 .038 .087 2.41 1.75 .199** .083 Responsive 2.01 1.58 0.05 -.020 2.70 1.85 .242** .170* Interactive 2.24 1.46 .123 .125 3.17 1.69 .175* .214** Nice 1.59 1.54 -.066 .101 1.97 1.80 .028 .141* Pleasant 1.93 1.48 .054 .157* 2.32 1.76 -.033 .186** Like 2.06 1.46 .227** .287** 2.90 1.73 .096 .245** Friendly 1.46 1.43 -.074 .086 1.51 1.62 .078 .072 Kind 1.43 1.49 -.066 .115 1.50 1.65 .030 .097 Knowledgeable 1.71 1.47 -.049 -.028 2.00 1.79 .144* .082 Competent 2.00 1.68 .056 .102 2.52 1.90 .088 .021 Sensible 1.74 1.61 .062 .155* 2.31 2.00 .166* .158* Responsible 2.06 1.66 .038 .085 2.28 1.91 .109 .197** Machinelike 4.19 1.49 .237** .078 4.47 1.12 -.054 -.068 Artificial 3.99 1.64 .213** .077 4.01 1.54 .055 -.076 Fake 3.72 1.70 .242** .172* 3.64 1.71 .090 -.083 Unconscious 3.29 2.05 .071 .049 2.70 2.07 .035 .056 Move Rigidly 3.35 1.76 .144* .107 2.19 1.40 .048 -.188** Mechanical 4.16 1.47 .280** .117 4.47 1.13 -.113 .053 Dead 3.08 2.05 .124 .156* 2.60 2.09 .057 .001 Stagnant 3.06 1.83 .137* -.003 2.23 1.68 .048 -.054 Apathetic 2.43 1.77 .155* .148* 2.09 1.64 .085 -.141* Inert 2.87 1.64 .078 0.08 2.11 1.45 -.046 -.138* Awful 1.90 1.78 .042 .005 1.72 1.63 .141* -.171* Unpleasant 2.39 1.72 .090 .032 2.00 1.60 .164* -.128 Dislike 2.70 1.71 .019 -.042 2.09 1.47 .070 -.256** Unfriendly 2.12 1.91 .046 -.008 1.78 1.83 .163* -.132 Unkind 1.95 1.89 .016 .003 1.67 1.79 .183** -.108 Ignorant 2.50 1.90 .081 .132 2.01 1.80 .000 -0.07 Incompetent 2.19 1.78 .057 .082 1.87 1.60 -.025 -.097 Foolish 1.90 1.73 .042 .149* 1.58 1.57 .049 -.117 Irresponsible 2.03 1.65 .046 .048 1.78 1.63 .031 -.186**

235

TUGV IED Detonator Mean SD rRobot rTrustworthy Mean SD rRobot rTrustworthy Robot 5.20 1.78 1 .366** 5.40 1.61 1 .261** Trustworthy 4.27 1.59 .366** 1 4.25 1.48 .261** 1 Interact / Use 3.83 1.84 .304** .458** 4.00 1.66 .091 .508** Humanlike 1.25 0.65 .008 -.031 1.24 0.67 -.047 -.046 Lifelike 1.28 0.92 -.052 .133 1.39 0.91 -.044 .073 Natural 1.29 1.01 -.008 -.001 1.48 0.99 .008 .205** Conscious 1.25 1.25 .006 -.006 1.33 1.24 .008 .128 Move Elegantly 1.85 1.22 .047 .192** 1.80 1.07 -.010 .144* Organic 1.25 0.68 .004 -.026 1.23 0.62 -.062 .059 Alive 1.24 1.19 .099 .030 1.22 1.18 .047 .089 Lively 2.06 1.56 .088 .097 2.15 1.45 .018 .122 Responsive 2.31 1.79 .093 .092 2.33 1.70 .109 .228** Interactive 2.80 1.64 .287** .162* 2.64 1.61 .120 .258** Nice 1.44 1.33 .120 .122 1.60 1.45 .039 .095 Pleasant 1.58 1.30 .124 .081 1.76 1.35 -.074 .153* Like 2.03 1.54 .126 .089 2.39 1.49 .093 .233** Friendly 1.25 1.25 .099 .015 1.45 1.33 .012 .057 Kind 1.16 1.23 .105 .134 1.42 1.38 .011 .149* Knowledgeable 1.51 1.47 .077 .025 1.58 1.49 .099 .087 Competent 2.02 1.75 .144* .125 2.17 1.67 .109 .161* Sensible 1.83 1.74 .025 .163* 1.92 1.70 -.003 .146* Responsible 1.84 1.66 .125 .154* 2.04 1.73 .042 .130 Machinelike 4.64 0.91 .134 .085 4.68 0.87 .061 .098 Artificial 4.20 1.54 .153* -.046 4.32 1.30 .058 -.043 Fake 3.61 1.89 .138* -.026 3.92 1.57 .045 -.062 Unconscious 2.78 2.14 .106 .049 2.91 2.08 -.123 -.062 Move Rigidly 3.55 1.59 .139* -.070 3.94 1.31 -.011 -.024 Mechanical 4.64 0.93 .134 .079 4.69 0.82 .007 -.026 Dead 2.97 2.13 .092 .032 2.96 2.14 .009 -.067 Stagnant 2.64 1.79 .095 -.043 2.87 1.68 -.031 -.145* Apathetic 2.34 1.80 .190** -.054 2.38 1.72 -.171* -.141* Inert 2.51 1.58 .058 .034 2.63 1.60 -.155* -.107 Awful 2.48 1.99 .159* -0.06 2.12 1.82 .073 -.042 Unpleasant 2.84 1.92 .194** .029 2.71 1.81 -.003 -.033 Dislike 2.58 1.77 .051 -.120 2.60 1.56 -.089 -.224** Unfriendly 2.47 2.09 .210** .055 2.47 1.97 -.029 -.105 Unkind 2.33 2.11 .148* .029 2.07 1.88 -.030 -.076 Ignorant 2.38 2.00 .117 -.031 2.23 1.91 -.074 -.053 Incompetent 2.10 1.79 .103 -.076 2.22 1.70 -.115 -.132 Foolish 2.00 1.85 .089 -.093 1.91 1.69 -.030 -.030 Irresponsible 2.09 1.81 .129 -.087 1.97 1.69 -.091 -.173*

236

Laparoscopic Robot DaVinci Robot Mean SD rRobot rTrustworthy Mean SD rRobot rTrustworthy Robot 5.06 1.82 1 .327** 5.43 1.6 1 .215** Trustworthy 4.50 1.42 .327** 1 4.47 1.5 .215** 1 Interact / Use 4.12 1.69 .272** .632** 4.35 1.7 .151* .649** Humanlike 1.38 0.99 .040 .111 1.38 0.9 .078 .095 Lifelike 1.41 1.02 .134 .141* 1.44 1.0 .140* .063 Natural 1.42 1.07 .160* .150* 1.41 1.1 .187** .199** Conscious 1.35 1.27 .166* .245** 1.47 1.4 .222** .192** Move Elegantly 2.31 1.62 .247** .303** 2.12 1.4 .293** .373** Organic 1.32 0.86 .058 .067 1.33 0.8 .071 .080 Alive 1.25 1.20 .197** .183** 1.35 1.3 .183** .142* Lively 1.81 1.39 .198** .111 1.90 1.4 .253** .234** Responsive 2.23 1.74 .241** .164* 2.60 1.8 .233** .198** Interactive 2.74 1.70 .275** .301** 2.96 1.7 .214** .303** Nice 1.66 1.51 .224** .190** 1.60 1.6 .165* .058 Pleasant 1.63 1.43 .224** .174* 1.75 1.5 .190** .135 Like 2.22 1.59 .306** .334** 2.23 1.6 .201** .285** Friendly 1.29 1.40 .133 .108 1.45 1.5 .215** .153* Kind 1.30 1.41 .124 .103 1.35 1.5 .172* .053 Knowledgeable 1.82 1.71 .138* .142* 2.12 1.8 .258** .223** Competent 2.38 1.85 .221** .241** 2.38 1.9 .228** .209** Sensible 2.03 1.89 .153* .303** 1.97 1.8 .230** .268** Responsible 2.14 1.89 .198** .245** 2.22 1.8 .220** .241** Machinelike 4.27 1.41 .184** .022 4.45 1.2 .058 -.093 Artificial 4.10 1.54 .160* -.049 4.13 1.5 .109 -.132 Fake 3.63 1.81 .188** .006 3.63 1.8 .192** -.063 Unconscious 2.84 2.08 .212** .147* 2.89 2.1 .142* -.040 Move Rigidly 2.86 1.76 .078 .050 3.19 1.7 .127 -.124 Mechanical 4.30 1.37 .241** .084 4.52 1.1 .051 -.055 Dead 2.88 2.12 .167* .082 2.69 2.1 .183** -.057 Stagnant 3.00 1.83 .134 .120 2.92 1.8 .148* -.029 Apathetic 2.45 1.82 .197** .077 2.34 1.7 .168* -.093 Inert 2.46 1.64 .023 .035 2.40 1.6 .023 -.165* Awful 2.18 1.84 .107 -.010 2.01 1.9 .180** -.252** Unpleasant 2.52 1.92 .082 -.027 2.40 1.9 .155* -.221** Dislike 2.60 1.72 .046 -.067 2.53 1.7 .174* -.233** Unfriendly 2.17 2.05 .146* .012 2.04 2.0 .161* -.105 Unkind 2.01 1.98 .143* .009 1.86 1.9 .160* -.147* Ignorant 2.19 1.92 .078 .056 2.10 1.8 .188** -.036 Incompetent 1.98 1.66 -.001 .067 1.91 1.6 .179** -.037 Foolish 1.72 1.69 .017 .092 1.81 1.7 .121 -.001 Irresponsible 1.78 1.68 .094 .001 1.99 1.7 .118 -.070

237

RP6 iFoot Mean SD rRobot rTrustworthy Mean SD rRobot rTrustworthy Robot 5.25 1.65 1 .266** 5.45 1.45 1 .343** Trustworthy 4.62 1.50 .266** 1 4.17 1.44 .343** 1 Interact / Use 4.83 1.58 .244** .651** 4.28 1.54 .254** .535** Humanlike 1.67 1.10 .181** .053 1.39 0.84 .002 .056 Lifelike 1.74 1.17 .196** .194** 1.42 0.97 -.054 .085 Natural 1.58 1.11 .147* .073 1.49 1.12 .034 .168* Conscious 1.83 1.57 .169* .170* 1.34 1.28 -.056 0.12 Move Elegantly 2.11 1.49 .177* .134 1.77 1.31 .056 .221** Organic 1.38 0.86 0.11 .075 1.24 0.72 -.003 .094 Alive 1.64 1.51 .196** .160* 1.23 1.18 .068 .150* Lively 2.23 1.54 .095 .127 1.68 1.37 .063 .217** Responsive 2.69 1.75 .236** .208** 2.14 1.65 .018 .026 Interactive 3.15 1.66 .092 .235** 2.50 1.60 .062 .232** Nice 2.16 1.74 .068 .173* 1.57 1.58 .094 .169* Pleasant 2.37 1.63 .111 .234** 1.80 1.57 .036 .163* Like 2.62 1.58 .145* .253** 2.15 1.51 .106 .179** Friendly 2.17 1.73 .204** .225** 1.61 1.57 .028 .171* Kind 1.97 1.77 .140* .235** 1.48 1.55 .037 .185** Knowledgeable 2.31 1.74 .214** .319** 1.62 1.46 .061 .057 Competent 2.34 1.71 .205** .243** 1.87 1.62 .101 .109 Sensible 2.10 1.70 .197** .201** 1.65 1.65 .060 .176* Responsible 1.91 1.73 .131 .254** 1.72 1.66 -.038 .176* Machinelike 4.07 1.35 -.092 -.006 4.29 1.28 -.095 -.133 Artificial 3.86 1.49 .017 -.092 4.12 1.48 -.029 -.086 Fake 3.64 1.69 .088 .041 3.71 1.75 .066 -.075 Unconscious 2.58 1.92 .116 .064 2.76 2.09 .042 .007 Move Rigidly 2.85 1.73 .132 -.044 3.13 1.79 -.050 -.009 Mechanical 4.27 1.32 -.005 .002 4.38 1.31 -.077 -.110 Dead 2.66 2.00 .158* .079 2.78 2.12 .001 .049 Stagnant 2.82 1.72 .040 .004 2.85 1.89 -.024 -.020 Apathetic 2.15 1.57 -.017 .025 2.39 1.76 .018 .008 Inert 2.22 1.47 .057 -.059 2.49 1.60 .035 -.031 Awful 1.88 1.57 .171* .021 1.87 1.81 -.014 -.065 Unpleasant 2.25 1.58 .096 -.023 2.21 1.81 -.084 -.131 Dislike 2.35 1.49 .130 -.095 2.47 1.65 -.040 -.123 Unfriendly 1.93 1.59 .127 -.038 2.00 1.85 -.094 -.054 Unkind 1.78 1.64 .122 -.013 1.81 1.83 -.103 -.045 Ignorant 2.16 1.68 .127 .013 2.36 1.91 .012 .084 Incompetent 2.13 1.62 .154* -.103 2.11 1.77 .016 -.019 Foolish 2.02 1.66 .139* .031 1.93 1.84 .000 .034 Irresponsible 1.89 1.72 .175* .061 1.91 1.79 -.042 -.026

238

Percentages of Attributes for Fuzzy Boundary Robots

The perception of a robot is classified across multiple attributes. Previous research has shown that individual differences were present in attribute ratings for specific robots. For example, person A’s perception of how human-like the Ballroom robot appeared to be, and often is, very different from person B’s. Attribute ratings were turned into a percentage to calculate an average percentage score for each attribute across the robot stimuli. This allowed individual robots, with unique features, to be compared. Percentages were reported in

239

Table 59 (anthropomorphic attributes), Table 60 (animacy attributes), Table 61

(likeability attributes), and Table 62 (perceived intelligence attributes). Findings showed that the percentage of each attribute varied for each robot, however anthropomorphic and animacy attributes were more prevalent than likeability or perceived intelligence attributes.

240

Table 59

Average Percentage of the Anthropomorphic Attributes for Each Robot

e e

# # of

Fake

Move Move Mov

Rigidly

Lifelike

Natural

Artificial

Elegantly

over over 50%

attributes

Conscious

Humanlike

Machinelike Unconscious

Gutter 75 26 26 29 25 41 84 80 74 66 5 Reaper 27 32 31 29 68 89 80 73 54 44 5 Laparoscopic 28 28 28 27 46 85 82 73 57 57 5 Furby 36 38 32 35 38 79 81 85 63 74 5 Ballroom 53 46 35 36 51 62 70 78 60 55 7 Lawnmower 25 27 30 27 48 93 90 83 72 63 5 Roomba 33 25 29 28 29 50 92 86 79 63 5 TUGV 25 26 26 25 37 93 84 72 56 71 5 IED Detonator 25 28 30 27 36 94 86 78 58 79 5 DaVinci 28 29 28 29 42 89 83 73 58 64 5 RP6 33 35 32 37 42 81 77 73 52 57 5 iFoot 28 28 30 27 35 86 82 74 55 63 5 Metal robot 26 28 29 26 42 90 81 71 54 68 5 Glass robot 26 28 29 28 41 94 87 81 65 73 5 Note. All scores are %

Table 60

Average Percentage of the Animacy Attributes for Each Robot

# # of

Inert

Alive Dead

Lively

Lifelike

Organic

Stagnant

Artificial

Apathetic over 50%

attributes

Interactive

Responsi Mechanical

Gutter 34 27 26 34 40 26 67 83 62 61 49 84 5 Reaper 27 28 48 54 63 32 89 52 45 42 42 80 5 Laparoscopic 26 25 36 45 55 28 86 58 60 49 49 82 5 Furby 34 35 55 55 67 38 83 60 58 50 50 81 10 Ballroom 37 34 50 44 52 46 74 56 56 48 55 70 6 Lawnmower 25 29 40 49 53 27 92 65 69 56 60 90 7 Roomba 46 26 26 40 48 25 52 91 61 61 51 92 6 TUGV 25 25 41 46 56 26 93 59 53 47 50 84 6 IED Detonator 25 24 43 47 53 28 94 59 57 48 53 86 6 DaVinci 27 27 38 52 59 29 90 54 58 47 48 83 6 RP6 28 33 45 54 63 35 85 53 56 43 44 77 6 iFoot 25 25 34 43 50 28 88 56 57 48 50 82 6 Metal robot 25 25 40 45 52 28 91 55 53 47 50 81 6 Glass robot 24 28 43 51 59 28 95 64 63 48 52 87 7 Note. All scores are %

241

Table 61

Average Percentage of the Likeability Attribute for Each Robot

# # of

Nice

Kind

Awful

Dislike

Unkind

Pleasant

Friendly

over over 50%

Unfriendl attributes Unpleasan

Gutter 45 32 39 41 29 57 38 48 54 42 2 Reaper 39 46 58 30 30 34 40 42 36 33 1 Laparoscopic 33 33 44 26 26 44 50 52 43 40 2 Furby 57 56 56 63 57 46 56 55 45 43 7 Ballroom 49 54 54 49 47 41 46 50 42 39 3 Lawnmower 41 49 56 38 38 40 48 52 46 40 2 Roomba 50 38 48 34 31 60 35 42 38 34 2 TUGV 29 32 41 25 23 50 57 52 49 47 3 IED Detonator 32 35 48 29 28 42 54 52 49 41 2 DaVinci 32 35 45 29 27 40 48 51 41 37 1 RP6 43 47 52 43 39 38 45 47 39 36 1 iFoot 31 36 43 32 30 37 44 49 40 36 0 Metal robot 26 32 42 23 23 39 48 47 40 39 0 Glass robot 34 42 48 31 30 37 48 49 40 40 0 Note. All scores are %

Table 62

Average Percentage of the Perceived Intelligence Attribute for Each Robot

nsible

Foolish

Sensible

Ignorant

over over 50%

Competent

Responsible

Incompetent

Irrespo

# of attributes # of attributes Knowledgeable

Gutter 29 34 40 35 39 50 44 38 1 Reaper 40 50 46 46 40 37 32 36 1 Laparoscopic 36 48 41 43 44 40 34 36 0 Furby 34 36 36 33 57 59 61 50 4 Ballroom 35 38 36 34 48 48 45 41 0 Lawnmower 38 48 41 42 57 51 43 47 0 Roomba 36 47 44 44 48 46 40 42 0 TUGV 30 40 37 37 48 42 40 42 0 IED Detonator 32 43 38 41 45 44 38 39 0 DaVinci 42 48 39 44 42 38 36 40 0 RP6 46 47 42 38 43 43 40 38 0 iFoot 32 37 33 34 47 42 39 38 0 Metal robot 31 42 37 40 45 41 36 38 0 Glass robot 36 50 45 47 47 44 37 40 1 Note. All scores are %

242

APPENDIX L: STUDY 3 INFORMED CONSENT

243

244

APPENDIX M: STUDY 3 DEMOGRAPHICS QUESTIONNAIRE

245

Gender

Male

Female

Age

18 19

Other

Describe what you think a robot looks like in as much detail as possible.

How many movies or television shows have you watched that include robots? 0 1-5

6-10

10 or more

Have you ever interacted with a robot? Yes

Please describe in detail the type of interaction you have had with a robot. If you have not ever interacted with a robot, type N/A.

Please describe the robot you interacted with. What did it look like? How did it act? If you have never interacted with a robot, type N/A.

246

Have you ever controlled a robot? Yes

Please describe in detail how you controlled a robot. If you have never controlled a robot, type N/A.

247

APPENDIX N: STUDY 3 ADDITIONAL ANALYSES

248

Item Analysis

Table 63

Study 3 Results of Item Analysis

Trust item Mean SD r t Most robots are friendly towards people. 4.26 1.18 0.02 -4.268** Most robots are unfriendly towards people. (R) 4.8 1.11

Most robots are kind towards people. 3.99 1.25 0.08 -5.807** Most robots are unkind towards people. (R) 4.76 1.23

Most robots are pleasant towards people. 4.25 1.21 0.259** -4.992** Most robots are unpleasant towards people. (R) 4.82 1.14

Most robots are supportive to people. 4.38 1.32 .488** -1.908 Most robots are not supportive to people. (R) 4.58 1.3

Most robots are attractive to people. 3.23 1.5 -0.1 -11.858** Most robots are offensive to people. (R) 5.06 1.1

Most robots are caring towards people. 3.61 1.36 -0.183* -3.140** Most robots are apathetic towards people. (R) 4.15 1.46

Most robots protect people. 3.97 1.28 0.04 -9.874** Most robots harm people. (R) 5.36 1.27

Most robots do not instill fear in people. 4.4 1.47 0.473** -0.212 Most robots instill fear in people. (R) 4.43 1.44

Most robots perform a specific function. 5.35 1.15 0.14 5.261** Most robots perform many functions. 4.67 1.3

Most robots are successful when performing tasks. 5.1 1.04 .455** -0.337 Most robots are unsuccessful when performing tasks. (R) 5.13 1.2

Most robots function successfully. 4.92 1.19 .248** 8.046** Most robots malfunction. (R) 3.89 1.43

Most robots meet the user or operator's expectations. 4.99 1.17 .349** 1.422 Most robots do not meet the user or operator's expectations. (R) 4.84 1.27

Most robots meet the needs of the mission. 4.84 1.11 .371** -2.492* Most robots do not meet the needs of the mission. (R) 5.09 1.16

Most robots are qualified to perform a specific task. 5.31 1.18 .313** 3.817** Most robots are not qualified to perform a specific task. (R) 4.84 1.45

Most robots are built for long term use. 4.51 1.35 0.12 4.995** Most robots are built to be replaced. (R) 3.75 1.51

Most robots require infrequent maintenance. 3.79 1.36 .349** 4.803** Most robots require frequent maintenance. (R) 3.23 1.21

249

Trust item Mean SD r t Most robots are easy to maintain. 3.69 1.32 .442** -1.698 Most robots are hard to maintain. (R) 3.88 1.33

Most robots operate in an integrated team environment. 4.16 1.25 .233** 3.002** Most robots operate in a solitary capacity. (R) 3.8 1.21

Most robots make good teammates. 4.14 1.27 .584** -3.792** Most robots make poor teammates. (R) 4.5 1.34

Most robots work best alone. (R) 3.86 1.38 .361** -0.626 Most robots work best with a team. 3.94 1.3

Most robots complete simple tasks. 4.79 1.32 0.11 -0.414 Most robots complete complex tasks. 4.84 1.26

Most robots are valued by their users or operators. 5.13 1.15 .481** 3.163** Most robots are not valued by their users or operators. (R) 4.79 1.46

Most robots are precise in their actions. 5.04 1.33 .473** 1.773 Most robots are not precise in their actions. (R) 4.84 1.45

Most robots perform accurately. 5.13 1.07 .478** 1.758 Most robots do not perform accurately. (R) 4.96 1.29

Most robots keep classified information secure. 4.73 1.29 .571** 0.067 Most robots do not keep classified information secure. (R) 4.72 1.27

Most robots possess adequate decision-making capability. 3.89 1.47 .237** -1.785 Most robots possess inadequate decision-making capability. (R) 4.14 1.48

Responsibility of the robot's action falls to the human. 4.97 1.3 -0.349* 9.743** Responsibility of the robot's action falls to the robot. 3.25 1.4

I like most robots. 4.7 1.34 .595** -4.232** I dislike most robots. (R) 5.11 1.37

I would feel comfortable giving a robot complete responsibility 3.2 1.56 .402** -1.608 for the completion of a mission. I would not feel comfortable giving a robot complete 3.43 1.68 .548** 0.554 responsibility for the completion of a mission. (R) I would feel comfortable assigning a robot to a task or problem 3.84 1.63 that was critical to the success of a mission. I would not feel comfortable assigning a robot to a task or 3.77 1.68 .510** 2.327* problem that was critical to the success of a mission. (R) I would feel the need to monitor a robot during a mission. 4.97 1.35

I would not feel the need to monitor a robot during a mission. (R) 4.71 1.47 .455** -0.295 I feel comfortable when a robot has to make decisions which will 3.21 1.55 .548** 0.554 affect me personally. I do not feel comfortable when a robot has to make decisions 3.25 1.55 which will affect me personally. (R) I am comfortable with the idea of working with a robot. 4.57 1.54 .559** 1.485 I am not comfortable with the idea of working with a robot. (R) 4.4 1.53

250

Trust item Mean SD r t I know when a robot tries to communicate. 3.83 1.47 .471** -0.965 I do not know when a robot tries to communicate. (R) 3.94 1.41

Most robots appear to be natural. 2.88 1.35 .347** -6.727** Most robots appear to be fake. (R) 3.75 1.49

Most robots appear to be human-like. 3.1 1.43 .349** 5.161** Most robots appear to be machine-like. (R) 2.47 1.25

Most robots appear to be conscious. 3.44 1.48 .306** -2.394** Most robots appear to be unconscious. (R) 3.78 1.55

Most robots appear to be lifelike. 3.37 1.43 .321** 3.522** Most robots appear to be artificial. (R) 2.92 1.35

Most robots appear to be alive. 3.33 1.49 .247** -7.543** Most robots appear to be dead. (R) 4.42 1.48

Most robots appear to be organic. 2.81 1.27 .255** 3.858** Most robots appear to be mechanical. (R) 2.34 1.23

Most robots have faces. 3.4 1.4 .612** -2.252* Most robots do not have faces. (R) 3.63 1.47

Most robots move elegantly. 3.13 1.35 .372** -3.271** Most robots move rigidly. (R) 3.52 1.29

Most robots are responsive. 4.64 1.24 .413** 0.113 Most robots are unresponsive. (R) 4.62 1.35

Most robots move quickly. (R) 3.9 1.21 .427** -0.059 Most robots move slowly. 3.91 1.28

Most robots are mobile. 4.72 1.41 .623** -0.647 Most robots are immobile. (R) 4.78 1.41

Most robots work in close proximity with people. 4.59 1.29 .456** -1.481 Most robots do not work in close proximity with people. (R) 4.75 1.27

Most robots are easily led astray by unexpected changes in the 4.19 1.32 .322** 0.051 environment or task. (R) Most robots are not easily led astray by unexpected changes in 4.18 1.37 the environment or task. Most robots know the difference between friend and foe. 3.09 1.5 .525** -2.089* Most robots do not know the difference between friend and foe. 3.33 1.46 (R) Most robots have low error rates. 4.2 1.48 .497** -4.182** Most robots have high error rates. (R) 4.69 1.47

Most robots provide feedback. 4.57 1.31 .497** -1.327 Most robots do not provide feedback. (R) 4.7 1.31

Most robots communicate with people. 4.15 1.39 .590** -0.128 Most robots do not communicate with people. (R) 4.16 1.35

251

Trust item Mean SD r t Most robots openly communicate with users or operators. 4.04 1.38 .459** 0.33 Most robots do not openly communicate with users or operators. 4 1.39 (R) Most robots often communicate with people. 4.12 1.28 .329** -0.833 Most robots rarely communicate with people. (R) 4.22 1.34

Most robots understand commands. 4.99 1.25 .422** -1.115 Most robots do not understand commands. (R) 5.11 1.26

Most robots provide appropriate information. 5.04 1.08 .369** 0 Most robots provide inappropriate information. (R) 5.04 1.18

Most robots are competent. 4.67 1.25 .387** -2.250* Most robots are incompetent. (R) 4.92 1.3

Most robots are knowledgeable. 4.58 1.38 .281** -2.070* Most robots are ignorant. (R) 4.84 1.3

Most robots are responsible. 4.24 1.34 .262** -4.708** Most robots are irresponsible. (R) 4.85 1.35

Most robots are intelligent. 4.7 1.36 .501** -1.73 Most robots are unintelligent. (R) 4.9 1.48

Most robots make sensible decisions. 3.98 1.35 0.09 -7.093** Most robots make foolish decisions. (R) 4.99 1.32

Most robots warn people of potential risks in the environment. 4.04 1.31 .364** -2.524* Most robots fail to warn people of potential risks in the 4.34 1.36 environment. (R) Most robots tell the truth. 4.74 1.41 .277** -1.672 Most robots do not tell the truth. (R) 4.96 1.27

Most robots communicate all information. 4.37 1.43 .158* 1.967 Most robots communicate only partial information. (R) 4.09 1.37

Most robots provide credible solutions to a problem. 4.72 1.25 .348** 1.323 Most robots do not provide credible solutions to a problem. (R) 4.57 1.37

Most robots require training to use or operate. (R) 2.95 1.4 .517** 1.13 Most robots do not require training to use or operate. 2.82 1.46

Most robots can perform a task better than a novice human user. 4.32 1.54 .487** 0.099 Most robots cannot perform a task better than a novice human 4.31 1.6 user. (R) Most robots have a relationship with their human users or 3.58 1.59 .400** 0.367 operators. Most robots do not have a relationship with their human users or 3.53 1.57 operators. (R) Most robots are controlled by people. 5.13 1.38 0.1 9.478** Most robots are autonomous. (R) 3.7 1.19

Most robots do exactly as instructed to do. 5.04 1.28 -.609** 13.631** Most robots do approximately as instructed to do. (R) 2.73 1.11

252

Trust item Mean SD r t Most robots perform tasks that are typically carried out by 4.96 1.25 people. Most robots follow standard task-related protocols. 5.08 1.3

Most robots follow specific procedures. 5.38 1.14 .428** 2.512* Most robots follow general procedures. 5.13 1.22

It is easy to determine the use or function of most robots from 3.84 1.64 .435** -1.495 the robot's appearance. It is difficult to determine the use or function of most robots from 4.04 1.55 the robot's appearance. (R) Most robots can be relied on to complete a task. 4.99 1.29 .329** 2.140* Most robots cannot be relied on to complete a task. (R) 4.72 1.45

Most robots act consistently. 5.08 1.19 .519** 2.237* Most robots act inconsistently. (R) 4.86 1.33

Most robots perform as intended. 5.05 1.16 .551** 0.812 Most robots do not perform as intended. (R) 4.97 1.3

Most robots are considered part of the team. 4.09 1.3 .570** 0.066 Most robots are considered separate from the team. (R) 4.08 1.3

Most robots will act as part of the team. 4.14 1.37 .393** -1.993* Most robots will not act as part of the team. (R) 4.37 1.31

Experience with one robot can be generalized to other robots. 3.3 1.54 .318** 0.314 Experience with one robot cannot be generalized to other robot. 3.25 1.48 (R)

253

Additional Analyses: Item Normality

Table 64

Study 3 Normality

Trust Item S z p K z p Most robots are friendly towards people. -0.184 -0.96 0.896 2.34 * Most robots are unfriendly towards people. (R) 0.152 0.79 -0.436 -1.14 Most robots are kind towards people. -0.016 -0.08 0.733 1.91 Most robots are unkind towards people. (R) 0.053 0.28 0.152 0.40 Most robots are pleasant towards people. -0.28 -1.46 0.87 2.27 * Most robots are unpleasant towards people. (R) 0.208 1.08 -0.547 -1.43 Most robots are supportive to people. -0.145 -0.76 -0.227 -0.59 Most robots are not supportive to people. (R) -0.096 -0.50 -0.235 -0.61 Most robots are attractive to people. 0.197 1.03 -0.675 -1.76 Most robots are offensive to people. (R) 0.047 0.24 -0.723 -1.89 Most robots are caring towards people. -0.235 -1.22 -0.052 -0.14 Most robots are apathetic towards people. (R) 0.13 0.68 -0.18 -0.47 Most robots protect people. -0.062 -0.32 -0.043 -0.11 Most robots harm people. (R) -0.807 -4.20 ** 0.422 1.10 Most robots do not instill fear in people. -0.277 -1.44 -0.703 -1.84 Most robots instill fear in people. (R) -0.155 -0.81 -0.534 -1.39 Most robots perform a specific function. -1.093 -5.69 ** 1.721 4.49 ** Most robots perform many functions. -0.247 -1.29 -0.294 -0.77 Most robots are successful when performing tasks. -0.609 -3.17 ** 0.736 1.92 Most robots are unsuccessful when performing tasks. (R) -0.505 -2.63 ** -0.054 -0.14 Most robots function successfully. -0.441 -2.30 ** 0.01 0.03 Most robots malfunction. (R) -0.008 -0.04 -0.638 -1.67 Most robots meet the user or operator's expectations. -0.657 -3.42 ** 0.361 0.94 Most robots do not meet the user or operator's expectations. (R) -0.462 -2.41 ** -0.272 -0.71 Most robots meet the needs of the mission. -0.338 -1.76 -0.454 -1.19 Most robots do not meet the needs of the mission. (R) -0.993 -5.17 ** 1.502 3.92 ** Most robots are qualified to perform a specific task. -0.914 -4.76 ** 1.458 3.81 ** Most robots are not qualified to perform a specific task. (R) -0.752 -3.92 ** 0.087 0.23 Most robots are built for long term use. -0.214 -1.11 -0.215 -0.56 Most robots are built to be replaced. (R) -0.184 -0.96 0.896 2.34 * Most robots require infrequent maintenance. 0.152 0.79 -0.436 -1.14 Most robots require frequent maintenance. (R) -0.016 -0.08 0.733 1.91 Most robots are easy to maintain. 0.053 0.28 0.152 0.40 Most robots are hard to maintain. (R) -0.28 -1.46 0.87 2.27 * Most robots operate in an integrated team environment. 0.208 1.08 -0.547 -1.43 Most robots operate in a solitary capacity. (R) -0.145 -0.76 -0.227 -0.59 Most robots make good teammates. -0.096 -0.50 -0.235 -0.61 Most robots make poor teammates. (R) 0.197 1.03 -0.675 -1.76 Most robots work best alone. (R) 0.047 0.24 -0.723 -1.89 Most robots work best with a team. -0.235 -1.22 -0.052 -0.14 Most robots complete simple tasks. 0.13 0.68 -0.18 -0.47 Most robots complete complex tasks. -0.062 -0.32 -0.043 -0.11

254

Trust Item S z p K z p Most robots are valued by their users or operators. -0.807 -4.20 ** 0.422 1.10 Most robots are not valued by their users or operators. (R) -0.277 -1.44 -0.703 -1.84 Most robots are precise in their actions. -0.155 -0.81 -0.534 -1.39 Most robots are not precise in their actions. (R) -1.093 -5.69 ** 1.721 4.49 ** Most robots perform accurately. -0.247 -1.29 -0.294 -0.77 Most robots do not perform accurately. (R) -0.609 -3.17 ** 0.736 1.92 Most robots keep classified information secure. -0.505 -2.63 ** -0.054 -0.14 Most robots do not keep classified information secure. (R) -0.441 -2.30 * 0.01 0.03 Most robots possess adequate decision-making capability. -0.008 -0.04 -0.638 -1.67 Most robots possess inadequate decision-making capability. -0.657 -3.42 ** 0.361 0.94 (R) Responsibility of the robot's action falls to the human. -0.462 -2.41 * -0.272 -0.71 Responsibility of the robot's action falls to the robot. -0.338 -1.76 -0.454 -1.19 I like most robots. -0.993 -5.17 ** 1.502 3.92 ** I dislike most robots. (R) -0.914 -4.76 ** 1.458 3.81 ** I would feel comfortable giving a robot complete responsibility -0.752 -3.92 ** 0.087 0.23 for the completion of a mission. I would not feel comfortable giving a robot complete -0.214 -1.11 -0.215 -0.56 responsibility for the completion of a mission. (R) I would feel comfortable assigning a robot to a task or problem -0.035 -0.18 -0.792 -2.07 that was critical to the success of a mission. I would not feel comfortable assigning a robot to a task or 0.096 0.50 -0.832 -2.17 problem that was critical to the success of a mission. (R) I would feel the need to monitor a robot during a mission. -0.65 -3.39 ** 0.297 0.78 I would not feel the need to monitor a robot during a mission. -0.483 -2.52 * -0.307 -0.80 (R) I feel comfortable when a robot has to make decisions which 0.405 2.11 * -0.376 -0.98 will affect me personally. I do not feel comfortable when a robot has to make decisions 0.384 2.00 -0.406 -1.06 which will affect me personally. (R) I am comfortable with the idea of working with a robot. -0.653 -3.40 ** 0.092 0.24 I am not comfortable with the idea of working with a robot. (R) -0.178 -0.93 -0.544 -1.42 I know when a robot tries to communicate. -0.298 -1.55 -0.432 -1.13 I do not know when a robot tries to communicate. (R) 0.005 0.03 -0.141 -0.37 Most robots appear to be natural. 0.485 2.53 * -0.114 -0.30 Most robots appear to be fake. (R) 0.126 0.66 -0.592 -1.55 Most robots appear to be human-like. 0.331 1.72 -0.448 -1.17 Most robots appear to be machine-like. (R) 0.988 5.15 ** 1.332 3.48 ** Most robots appear to be conscious. -0.024 -0.13 -0.801 -2.09 * Most robots appear to be unconscious. (R) 0.117 0.61 -0.631 -1.65 Most robots appear to be lifelike. 0.25 1.30 -0.728 -1.90 Most robots appear to be artificial. (R) 0.688 3.58 ** 0.461 1.20 Most robots appear to be alive. -0.016 -0.08 -0.945 -2.47 * Most robots appear to be dead. (R) -0.256 -1.33 -0.338 -0.88 Most robots appear to be organic. 0.146 0.76 -0.726 -1.90 Most robots appear to be mechanical. (R) 1.138 5.93 ** 1.453 3.79 ** Most robots have faces. 0.021 0.11 -0.801 -2.09 * Most robots do not have faces. (R) 0.17 0.89 -0.815 -2.13 * Most robots move elegantly. 0.45 2.34 * -0.124 -0.32 Most robots move rigidly. (R) 0.232 1.21 -0.366 -0.96 Most robots are responsive. -0.325 -1.69 -0.324 -0.85

255

Trust Item S z p K z p Most robots are unresponsive. (R) -0.288 -1.50 -0.548 -1.43 Most robots move quickly. (R) 0.108 0.56 -0.195 -0.51 Most robots move slowly. -0.059 -0.31 -0.223 -0.58 Most robots are mobile. -0.586 -3.05 ** -0.237 -0.62 Most robots are immobile. (R) -0.387 -2.02 * -0.414 -1.08 Most robots work in close proximity with people. -0.515 -2.68 ** -0.239 -0.62 Most robots do not work in close proximity with people. (R) -0.337 -1.76 -0.384 -1.00 Most robots are easily led astray by unexpected changes in the -0.186 -0.97 -0.687 -1.79 environment or task. (R) Most robots are not easily led astray by unexpected changes in -0.019 -0.10 -0.274 -0.72 the environment or task. Most robots know the difference between friend and foe. 0.302 1.57 -0.802 -2.09 * Most robots do not know the difference between friend and foe. 0.211 1.10 -0.528 -1.38 (R) Most robots have low error rates. -0.187 -0.97 -0.702 -1.83 Most robots have high error rates. (R) -0.511 -2.66 ** -0.331 -0.86 Most robots provide feedback. -0.471 -2.45 * -0.381 -0.99 Most robots do not provide feedback. (R) -0.446 -2.32 * -0.127 -0.33 Most robots communicate with people. 0.028 0.15 -0.707 -1.85 Most robots do not communicate with people. (R) -0.209 -1.09 -0.547 -1.43 Most robots openly communicate with users or operators. 0.049 0.26 0.066 0.17 Most robots do not openly communicate with users or 0.143 0.74 -0.713 -1.86 operators. (R) Most robots often communicate with people. -0.172 -0.90 -0.412 -1.08 Most robots rarely communicate with people. (R) -0.283 -1.47 -0.17 -0.44 Most robots understand commands. -0.817 -4.26 ** 0.612 1.60 Most robots do not understand commands. (R) -0.664 -3.46 ** 0.096 0.25 Most robots provide appropriate information. -0.336 -1.75 0.13 0.34 Most robots provide inappropriate information. (R) -0.559 -2.91 ** 0.406 1.06 Most robots are competent. -0.312 -1.63 -0.16 -0.42 Most robots are incompetent. (R) -0.418 -2.18 * -0.298 -0.78 Most robots are knowledgeable. -0.644 -3.35 ** 0.033 0.09 Most robots are ignorant. (R) -0.358 -1.86 -0.037 -0.10 Most robots are responsible. -0.382 -1.99 * -0.054 -0.14 Most robots are irresponsible. (R) -0.556 -2.90 ** 0.142 0.37 Most robots are intelligent. -0.561 -2.92 ** 0.058 0.15 Most robots are unintelligent. (R) -0.685 -3.57 ** 0.123 0.32 Most robots make sensible decisions. -0.374 -1.95 -0.117 -0.31 Most robots make foolish decisions. (R) 0.256 1.33 -0.539 -1.41 Most robots warn people of potential risks in the environment. 0.049 0.26 -0.305 -0.80 Most robots fail to warn people of potential risks in the -0.226 -1.18 0.192 0.50 environment. (R) Most robots tell the truth. -0.375 -1.95 -0.163 -0.43 Most robots do not tell the truth. (R) -0.424 -2.21 * 0.039 0.10 Most robots communicate all information. -0.439 -2.29 * -0.101 -0.26 Most robots communicate only partial information. (R) 0.019 0.10 -0.61 -1.59 Most robots provide credible solutions to a problem. -0.57 -2.97 ** 0.139 0.36 Most robots do not provide credible solutions to a problem. (R) -0.55 -2.86 ** -0.049 -0.13 Most robots require training to use or operate. (R) 0.685 3.57 ** 0.276 0.72 Most robots do not require training to use or operate. 0.767 3.99 ** 0.065 0.17

256

Trust Item S z p K z p Most robots can perform a task better than a novice human -0.228 -1.19 -0.612 -1.60 user. Most robots cannot perform a task better than a novice human -0.142 -0.74 -0.75 -1.96 user. (R) Most robots have a relationship with their human users or 0.094 0.49 -0.766 -2.00 operators. * Most robots do not have a relationship with their human users 0.117 0.61 -0.902 -2.36 * or operators. (R) Most robots are controlled by people. -0.564 -2.94 ** -0.419 -1.09 Most robots are autonomous. (R) -0.163 -0.85 0.476 1.24 Most robots do exactly as instructed to do. -0.34 -1.77 -0.557 -1.45 Most robots do approximately as instructed to do. (R) 0.554 2.89 ** 0.233 0.61 Most robots perform tasks that are typically carried out by -0.594 -3.09 ** 0.092 0.24 people. Most robots follow standard task-related protocols. -0.606 -3.16 ** -0.117 -0.31 Most robots follow specific procedures. -0.564 -2.94 ** -0.105 -0.27 Most robots follow general procedures. -0.789 -4.11 ** 0.836 2.18 * It is easy to determine the use or function of most robots from -0.04 -0.21 -0.976 -2.55 the robot's appearance. It is difficult to determine the use or function of most robots -0.084 -0.44 -0.95 -2.48 from the robot's appearance. (R) Most robots can be relied on to complete a task. -0.625 -3.26 ** -0.161 -0.42 Most robots cannot be relied on to complete a task. (R) -0.507 -2.64 ** -0.635 -1.66 Most robots act consistently. -0.239 -1.24 -0.462 -1.21 Most robots act inconsistently. (R) -0.482 -2.51 * -0.185 -0.48 Most robots perform as intended. -0.613 -3.19 ** 0.442 1.15 Most robots do not perform as intended. (R) -0.746 -3.89 ** 0.423 1.10 Most robots are considered part of the team. -0.218 -1.14 -0.166 -0.43 Most robots are considered separate from the team. (R) -0.242 -1.26 -0.312 -0.81 Most robots will act as part of the team. -0.103 -0.54 -0.163 -0.43 Most robots will not act as part of the team. (R) -0.371 -1.93 0.298 0.78 Experience with one robot can be generalized to other robots. 0.432 2.25 -0.544 -1.42 Experience with one robot cannot be generalized to other robot. 0.465 2.42 * -0.391 -1.02 (R) Note. **Represents significance at the .01 level *Represents significance at the .05 level

257

Additional Analyses: Principal Component Analyses based on Mental Models

Three additional exploratory factor analyses were performed on initial trust items, sorted by humanlike, machinelike, and varied mental models (see also Table 22). Data were analyzed using IBM SPSS Statistics v.19 (SPSS, 2010), with an alpha level set to .05, unless otherwise indicated. Rotated component matrices are reported in Table 65 below.

Table 65

Rotated Component Matrices across Mental Model Classifications

Trust item Humanlike Machinelike Varied C1 C2 C3 C4 C1 C2 C3 C1 C2 C3 C4 C5 Most robots are friendly 0.673 0.565 0.703 towards people. Most robots are unfriendly 0.594 0.495 0.311 towards people. (R) Most robots are kind 0.304 0.39 0.528 0.674 towards people. Most robots are unkind 0.709 0.532 0.403 0.418 towards people. (R) Most robots are pleasant 0.362 0.472 0.331 0.591 0.61 towards people. Most robots are unpleasant 0.54 0.637 0.422 0.331 towards people. (R) Most robots are supportive 0.578 0.399 0.421 0.565 0.437 to people. Most robots are not 0.459 0.5 0.608 0.317 0.375 supportive to people. (R) Most robots are attractive 0.49 0.37 0.447 to people. Most robots are offensive 0.627 0.475 0.575 0.554 to people. (R) Most robots are caring 0.7 0.667 0.348 0.496 towards people. Most robots are apathetic towards people. (R) Most robots protect people. 0.584 0.345 0.345 0.414 0.44 0.313 0.517 Most robots harm people. 0.733 0.544 0.677 (R)

258

Trust item Humanlike Machinelike Varied C1 C2 C3 C4 C1 C2 C3 C1 C2 C3 C4 C5 Most robots do not instill 0.347 0.4 0.456 fear in people. Most robots instill fear in 0.511 people. (R) Most robots perform a 0.436 0.423 0.714 0.502 specific function. Most robots perform many 0.665 0.546 functions. Most robots are successful 0.403 0.654 0.443 0.5 when performing tasks. Most robots are unsuccessful when 0.672 0.645 0.652 performing tasks. (R) Most robots function 0.58 0.362 0.718 0.561 0.37 successfully. Most robots malfunction. 0.406 0.52 0.416 0.433 (R) Most robots meet the user 0.501 0.584 0.313 0.515 or operator's expectations. Most robots do not meet the user or operator's 0.475 0.338 0.587 0.524 0.338 expectations. (R) Most robots meet the needs 0.555 0.599 0.684 of the mission. Most robots do not meet the needs of the mission. 0.58 0.601 0.685 0.448 (R) Most robots are qualified to 0.473 0.565 0.65 perform a specific task. Most robots are not qualified to perform a 0.412 0.459 0.395 specific task. (R) Most robots are built for 0.381 0.354 0.71 0.359 long term use. Most robots are built to be 0.53 replaced. (R) Most robots require 0.311 0.431 infrequent maintenance. Most robots require 0.432 0.388 0.372 0.395 0.451 frequent maintenance. (R) Most robots are easy to 0.593 0.33 0.307 0.681 0.319 maintain. Most robots are hard to 0.37 0.498 0.696 maintain. (R) Most robots operate in an integrated team 0.607 0.303 0.35 environment.

259

Trust item Humanlike Machinelike Varied C1 C2 C3 C4 C1 C2 C3 C1 C2 C3 C4 C5 Most robots operate in a 0.424 0.363 solitary capacity. (R) Most robots make good 0.737 0.315 0.519 0.493 0.334 0.54 teammates. Most robots make poor 0.646 0.341 0.514 0.331 0.576 teammates. (R) Most robots work best 0.401 0.427 alone. (R) Most robots work best with 0.654 0.393 0.379 0.593 a team. Most robots complete 0.574 simple tasks. Most robots complete 0.364 0.693 complex tasks. Most robots are valued by 0.543 0.583 0.597 0.32 their users or operators. Most robots are not valued by their users or operators. 0.412 0.527 0.402 (R) Most robots are precise in 0.608 0.461 0.638 0.733 their actions. Most robots are not precise 0.665 0.579 0.333 0.716 in their actions. (R) Most robots perform 0.531 0.432 0.597 0.737 accurately. Most robots do not perform 0.722 0.644 0.628 accurately. (R) Most robots keep classified 0.412 0.342 0.399 0.555 0.454 0.576 information secure. Most robots do not keep classified information 0.564 0.338 0.64 0.545 0.307 secure. (R) Most robots possess adequate decision-making 0.627 0.555 0.655 capability. Most robots possess inadequate decision- 0.541 0.368 0.696 making capability. (R) Responsibility of the robot's action falls to the 0.335 0.335 0.385 human. Responsibility of the robot's action falls to the 0.345 0.572 robot. I like most robots. 0.718 0.527 0.427 0.329 0.435 I dislike most robots. (R) 0.824 0.5 0.366 0.425 0.315 0.36

260

Trust item Humanlike Machinelike Varied C1 C2 C3 C4 C1 C2 C3 C1 C2 C3 C4 C5 I would feel comfortable giving a robot complete 0.474 0.45 0.835 responsibility for the completion of a mission. I would not feel comfortable giving a robot complete responsibility for 0.365 0.63 0.739 the completion of a mission. (R) I would feel comfortable assigning a robot to a task 0.727 0.446 0.533 0.304 or problem that was critical to the success of a mission. I would not feel comfortable assigning a robot to a task or problem 0.513 0.528 0.582 that was critical to the success of a mission. (R) I would feel the need to monitor a robot during a mission. I would not feel the need to monitor a robot during a 0.674 mission. (R) I feel comfortable when a robot has to make decisions 0.607 0.417 0.35 0.755 which will affect me personally. I do not feel comfortable when a robot has to make - 0.34 0.327 0.454 0.664 0.762 decisions which will affect 0.362 me personally. (R) I am comfortable with the idea of working with a 0.715 0.341 0.483 0.399 0.461 0.486 robot. I am not comfortable with the idea of working with a 0.678 0.447 0.332 0.322 0.529 robot. (R) I know when a robot tries 0.418 0.304 0.482 0.466 to communicate. I do not know when a robot 0.522 0.359 0.313 0.506 tries to communicate. (R) Most robots appear to be 0.607 0.461 0.58 natural. Most robots appear to be 0.455 0.464 0.405 0.333 fake. (R) Most robots appear to be 0.482 0.314 0.312 0.532 0.311 human-like.

261

Trust item Humanlike Machinelike Varied C1 C2 C3 C4 C1 C2 C3 C1 C2 C3 C4 C5 Most robots appear to be 0.481 0.328 0.407 0.45 machine-like. (R) Most robots appear to be 0.645 0.446 0.531 conscious. Most robots appear to be 0.671 0.387 0.32 0.352 unconscious. (R) Most robots appear to be 0.67 0.351 0.572 lifelike. Most robots appear to be 0.382 0.507 0.407 artificial. (R) Most robots appear to be 0.564 0.398 0.525 alive. Most robots appear to be 0.684 0.53 dead. (R) Most robots appear to be 0.316 0.414 organic. Most robots appear to be 0.354 mechanical. (R) Most robots have faces. 0.321 0.366 Most robots do not have faces. (R) Most robots move 0.621 0.312 0.363 elegantly. Most robots move rigidly. 0.357 0.372 0.463 0.312 (R) Most robots are responsive. 0.443 0.306 0.379 Most robots are 0.632 0.325 0.543 unresponsive. (R) Most robots move quickly.

(R) Most robots move slowly. Most robots are mobile. 0.582 0.355 Most robots are immobile. 0.674 (R) Most robots work in close 0.503 0.444 proximity with people. Most robots do not work in close proximity with 0.466 0.391 people. (R) Most robots are easily led astray by unexpected 0.373 changes in the environment or task. (R) Most robots are not easily led astray by unexpected 0.451 changes in the environment or task.

262

Trust item Humanlike Machinelike Varied C1 C2 C3 C4 C1 C2 C3 C1 C2 C3 C4 C5 Most robots know the difference between friend 0.391 0.37 0.488 and foe. Most robots do not know the difference between 0.535 0.554 0.322 0.308 friend and foe. (R) Most robots have low error 0.413 0.341 0.415 0.527 0.624 rates. Most robots have high 0.46 0.504 0.398 0.468 0.534 error rates. (R) Most robots provide 0.461 0.331 0.433 0.551 0.487 feedback. Most robots do not provide 0.457 0.465 0.515 0.46 feedback. (R) Most robots communicate 0.6 0.401 0.409 0.625 with people. Most robots do not communicate with people. 0.314 0.588 0.432 0.376 0.331 0.592 (R) Most robots openly communicate with users or 0.395 0.536 operators. Most robots do not openly communicate with users or 0.486 0.341 0.456 operators. (R) Most robots often 0.334 0.45 0.311 0.416 0.551 communicate with people. Most robots rarely communicate with people. 0.315 0.543 (R) Most robots understand 0.35 0.572 0.406 0.436 0.658 commands. Most robots do not 0.342 0.383 0.57 0.725 understand commands. (R) Most robots provide - 0.386 0.431 0.606 0.704 appropriate information. 0.346 Most robots provide inappropriate information. 0.578 0.63 0.421 (R) Most robots are competent. 0.317 0.657 0.407 0.822 Most robots are 0.599 0.601 0.34 0.384 incompetent. (R) Most robots are 0.58 0.44 0.393 knowledgeable. Most robots are ignorant. 0.569 0.38 0.528 0.717 (R) Most robots are 0.405 0.346 0.737 responsible.

263

Trust item Humanlike Machinelike Varied C1 C2 C3 C4 C1 C2 C3 C1 C2 C3 C4 C5 Most robots are 0.695 0.501 0.367 0.365 irresponsible. (R) Most robots are intelligent. 0.55 0.354 0.369 0.459 0.412 Most robots are 0.399 0.368 unintelligent. (R) Most robots make sensible 0.32 0.431 0.554 0.332 decisions. Most robots make foolish 0.757 0.527 0.395 0.48 0.412 decisions. (R) Most robots warn people of potential risks in the 0.312 0.361 0.371 0.432 environment. Most robots fail to warn people of potential risks in 0.324 0.439 0.53 0.351 the environment. (R) Most robots tell the truth. 0.422 0.362 0.766 Most robots do not tell the 0.537 0.545 0.634 truth. (R) Most robots communicate 0.4 0.35 0.557 0.429 all information. Most robots communicate only partial information. 0.329 0.533 0.399 (R) Most robots provide credible solutions to a 0.332 0.368 0.457 0.31 0.437 0.328 problem. Most robots do not provide credible solutions to a 0.479 0.385 0.421 0.412 0.578 problem. (R) Most robots require training to use or operate. 0.445 (R) Most robots do not require 0.525 0.483 training to use or operate. Most robots can perform a task better than a novice 0.394 0.381 0.383 0.311 0.397 0.47 human user. Most robots cannot perform a task better than a 0.632 0.455 0.374 0.378 novice human user. (R) Most robots have a relationship with their 0.611 0.422 human users or operators. Most robots do not have a relationship with their 0.323 0.588 0.395 human users or operators. (R) Most robots are controlled 0.333 0.43 by people.

264

Trust item Humanlike Machinelike Varied C1 C2 C3 C4 C1 C2 C3 C1 C2 C3 C4 C5 Most robots are autonomous. (R) Most robots do exactly as 0.367 0.692 0.704 0.846 instructed to do. Most robots do approximately as instructed to do. (R) Most robots perform tasks - 0.343 0.415 0.342 that are typically carried 0.326 out by people. Most robots follow standard task-related 0.623 0.597 0.705 protocols. Most robots follow specific 0.55 0.611 0.643 procedures. Most robots follow general 0.329 0.501 0.568 0.657 0.311 procedures. It is easy to determine the use or function of most 0.395 0.366 0.403 robots from the robot's appearance. It is difficult to determine the use or function of most 0.528 robots from the robot's appearance. (R) Most robots can be relied 0.443 0.414 0.597 0.697 on to complete a task. Most robots cannot be relied on to complete a 0.631 0.408 0.441 0.327 0.336 0.509 task. (R) Most robots act 0.546 0.474 0.697 0.748 consistently. Most robots act 0.52 0.519 0.587 0.327 0.724 inconsistently. (R) Most robots perform as 0.579 0.679 0.825 intended. Most robots do not perform 0.669 0.659 0.935 as intended. (R) Most robots are considered 0.747 0.348 0.47 0.758 part of the team. Most robots are considered 0.627 0.441 0.775 separate from the team. (R) Most robots will act as part 0.581 0.357 0.432 0.632 0.394 of the team. Most robots will not act as 0.379 0.314 0.669 part of the team. (R)

265

Trust item Humanlike Machinelike Varied C1 C2 C3 C4 C1 C2 C3 C1 C2 C3 C4 C5 Experience with one robot can be generalized to other 0.335 robots. Experience with one robot cannot be generalized to other robot. (R)

266

APPENDIX O: STUDY 4 SUBJECT MATTER EXPERT QUESTIONNAIRE

267

What is your area of expertise. Please check all that apply. Robotics Development / Design

Robot Operator

Robotics Research

Human-Robot Interaction

Automation Development / Design

Automation Operator Automation Research

Trust Research

Other Research Areas

Please provide any clarifying comments or additional details you feel are important

How many years of experience you have in the following areas? Robotics Development / Design Robot Operator

Robot Research

Human-Robot Interaction Automation Development / Design Automation Operator

Automation Research

Trust Research

Other Research Areas

268

Please describe your experience/expertise in ROBOTICS DESIGN / DEVELOPMENT (such as your role, type of robot(s), or any additional information you feel describes your expertise)

Please describe your experience/expertise as a ROBOT OPERATOR (such as your role, type of robot(s) or any additional information you feel describes your expertise)

Please describe your experience/expertise as a ROBOTICS RESEARCHER (such as your role, type of robot(s) or any additional information you feel describes your expertise)

Please describe your experience/expertise in HUMAN-ROBOT INTERACTION (such as your role, type of robot(s) or any additional information you feel describes your expertise)

Please describe your experience/expertise in AUTOMATION DESIGN / DEVELOPMENT (such as your role, type of automation, or any additional information you feel describes your expertise)

Please describe your experience/expertise as an AUTOMATION OPERATOR (such as your role, type of automation, or any additional information you feel describes your expertise)

269

Please describe your experience/expertise in AUTOMATION RESEARCH (such as your role, type of automation, or any additional information you feel describes your expertise)

Please describe your experience/expertise in OTHER RESEARCH AREAS (such as your role, type of research, or any additional information you feel describes your expertise)

270

APPENDIX P: STUDY 4 TRUST SCALE 74 ITEMS

271

Trust Items 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Unresponsive o o o o o o o o o o o Move quickly o o o o o o o o o o o Move slowly o o o o o o o o o o o Mobile o o o o o o o o o o o Require frequent maintenance o o o o o o o o o o o Easy to maintain o o o o o o o o o o o Difficult to maintain o o o o o o o o o o o Malfunction o o o o o o o o o o o Have errors o o o o o o o o o o o Responsive o o o o o o o o o o o Move rigidly o o o o o o o o o o o Function successfully o o o o o o o o o o o Follow directions o o o o o o o o o o o Are given complete responsibility for the o o o o o o o o o o o completion of a mission Are assigned tasks that are o o o o o o o o o o o critical to mission success Friendly o o o o o o o o o o o Kind o o o o o o o o o o o Pleasant o o o o o o o o o o o Conscious o o o o o o o o o o o Lifelike o o o o o o o o o o o Attractive o o o o o o o o o o o Caring o o o o o o o o o o o Human-like o o o o o o o o o o o Fake o o o o o o o o o o o Alive o o o o o o o o o o o Dead o o o o o o o o o o o Good teammate o o o o o o o o o o o Poor teammate o o o o o o o o o o o Supportive o o o o o o o o o o o Offensive o o o o o o o o o o o Organic o o o o o o o o o o o Have a face o o o o o o o o o o o Likable o o o o o o o o o o o Incompetent o o o o o o o o o o o Ignorant o o o o o o o o o o o Apathetic o o o o o o o o o o o Autonomous o o o o o o o o o o o

272

Trust Items 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Responsible o o o o o o o o o o o Act consistently o o o o o o o o o o o Monitored during a mission o o o o o o o o o o o Perform a task better than a o o o o o o o o o o o novice human user Keep classified information o o o o o o o o o o o secure Tell the truth o o o o o o o o o o o Meet the needs of the mission o o o o o o o o o o o Perform exactly as instructed o o o o o o o o o o o Built to last o o o o o o o o o o o Have a relationship with their o o o o o o o o o o o human users or operators Make decisions that affect me o o o o o o o o o o o personally Led astray by unexpected o o o o o o o o o o o changes in the environment Work in close proximity with o o o o o o o o o o o people Reliable o o o o o o o o o o o Predictable o o o o o o o o o o o Dependable o o o o o o o o o o o Protect people o o o o o o o o o o o Instill fear in people o o o o o o o o o o o Possess adequate decision- o o o o o o o o o o o making capability Know the difference between o o o o o o o o o o o friend and foe Make sensible decisions o o o o o o o o o o o Perform many functions at one o o o o o o o o o o o time Operate in an integrated team o o o o o o o o o o o environment Responsible for its own o o o o o o o o o o o actions Work best alone o o o o o o o o o o o Work best with a team o o o o o o o o o o o Work with human teammates o o o o o o o o o o o Are considered part of the o o o o o o o o o o o team Are considered separate from o o o o o o o o o o o the team

273

Trust Items 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Will act as part of the team o o o o o o o o o o o Communicate with people o o o o o o o o o o o Provide feedback o o o o o o o o o o o Openly communicate o o o o o o o o o o o Clearly communicate o o o o o o o o o o o Provide appropriate o o o o o o o o o o o information Communicate only partial o o o o o o o o o o o information Warn people of potential risks o o o o o o o o o o o in the environment

274

APPENDIX Q: STUDY 5 INFORMED CONSENT

275

276

277

APPENDIX R: STUDY 5 PRE/POST TRUST SCALE ITEMS

278

Please rate the following items about the Talon robot, taking into consideration that interaction you just observed.

What % of the time will this robot 0 10 20 30 40 50 60 70 80 90 100 be… % % % % % % % % % % % Considered part of the team o o o o o o o o o o o

Responsible o o o o o o o o o o o

Supportive o o o o o o o o o o o

Incompetent o o o o o o o o o o o

Dependable o o o o o o o o o o o

Friendly o o o o o o o o o o o

Reliable o o o o o o o o o o o

Pleasant o o o o o o o o o o o

Unresponsive o o o o o o o o o o o

Autonomous o o o o o o o o o o o

Predictable o o o o o o o o o o o

Conscious o o o o o o o o o o o

Lifelike o o o o o o o o o o o

A good teammate o o o o o o o o o o o Led astray by unexpected changes in o o o o o o o o o o o the environment

279

Please rate the following items about the Talon robot, taking into consideration that interaction you just observed.

What % of the time will this robot 0 10 20 30 40 50 60 70 80 90 100 … % % % % % % % % % % % Act consistently o o o o o o o o o o o

Protect people o o o o o o o o o o o

Act as part of the team o o o o o o o o o o o

Function successfully o o o o o o o o o o o

Work best with a team o o o o o o o o o o o

Malfunction o o o o o o o o o o o

Clearly communicate o o o o o o o o o o o Operate in an integrated team o o o o o o o o o o o environment Require frequent maintenance o o o o o o o o o o o

Openly communicate o o o o o o o o o o o

Have errors o o o o o o o o o o o Perform a task better than a novice o o o o o o o o o o o human user Know the difference between friend o o o o o o o o o o o and foe

280

Please rate the following items about the Talon robot, taking into consideration that interaction you just observed.

What % of the time will this 0 10 20 30 40 50 60 70 80 90 100 robot… % % % % % % % % % % % Provide Feedback o o o o o o o o o o o Possess adequate decision-making o o o o o o o o o o o capability Warn people of potential risks in the o o o o o o o o o o o environment Meet the needs of the mission o o o o o o o o o o o

Provide appropriate information o o o o o o o o o o o

Communicate with people o o o o o o o o o o o

Keep classified information secure o o o o o o o o o o o

Perform exactly as instructed o o o o o o o o o o o

Make sensible decisions o o o o o o o o o o o

Work in close proximity with people o o o o o o o o o o o

Perform many functions at one time o o o o o o o o o o o

Follow directions o o o o o o o o o o o

281

APPENDIX S: STUDY 6 INFORMED CONSENT

282

283

284

APPENDIX T: STUDY 6 SIMULATION HANDOUTS

285

Keyboard/Mouse Instructions a Turn Left d Turn Right w Move Forward s Move Backward mouse Controls head movement: Allows you to look around

Important Information: Some obstacles can be moved in the environment by walking into them. Other obstacles cannot be moved.

Figure 46. Study 6 Map A.

286

Figure 47. Study 6 Map B.

287

APPENDIX U: STUDY 6 ADDITIONAL ANALYSES

288

Human States Analysis

Table 66

Means, Standard Deviations, and Confidence Intervals for Human States for each Condition

Mood States Time Mean SE CILow CIHigh Hedonic tone Pre-Interaction 12.15 .51 11.08 13.22 After Map A 12.45 .56 11.29 14.58 After Map B 13.05 .73 11.52 14.58

Energetic arousal Pre-Interaction 19.25 .78 17.61 20.89 After Map A 17.15 .90 15.26 19.04 After Map B 17.15 .80 15.49 18.82

Tense arousal Pre-Interaction 26.80 .68 25.38 28.22 After Map A 24.65 1.08 22.40 26.90 After Map B 24.35 1.34 21.55 27.15

Anger / Frustration Pre-Interaction 18.4 .40 17.56 19.24 After Map A 18.20 .57 17.02 19.38 After Map B 17.1 .74 15.55 18.66

Motivation Time Mean SE CILow CIHigh Success Motivation Pre-Interaction 15.95 1.53 12.75 19.15 After Map A 15.10 1.61 11.72 18.48 After Map B 15.25 1.34 12.45 18.05

Intrinsic Motivation Pre-Interaction 16.40 .92 14.48 18.32 After Map A 16.75 .96 14.74 18.76 After Map B 16.25 .84 14.49 18.01

Thinking Style Time Mean SE CILow CIHigh Self-focused attention Pre-Interaction 20.33 1.44 17.33 23.34 After Map A 13.10 1.12 10.76 15.43 After Map B 11.48 1.28 8.81 14.14

Concentration Pre-Interaction 17.14 .93 15.21 19.08 After Map A 20.29 .397 19.46 21.11 After Map B 19.91 .87 18.09 21.73

Self-esteem Pre-Interaction 12.24 1.40 9.33 15.15 After Map A 15.52 1.07 13.30 17.75 After Map B 16.67 1.28 14.00 19.33

Control and confidence Pre-Interaction 30.24 1.30 27.52 32.95 After Map A 28.76 1.68 25.26 32.27 After Map B 27.67 2.18 23.12 32.21

289

Thinking Content Time Mean SE CILow CIHigh Task-related inference Pre-Interaction 19.45 1.65 15.99 22.91 After Map A 17.00 1.38 14.11 19.89 After Map B 18.45 1.70 14.89 22.01

Task-irrelevant inference Pre-Interaction 11.90 .99 9.84 13.97 After Map A 8.05 .39 7.23 8.87 After Map B 8.15 .80 6.49 9.82

290

Trust Antecedents Means and Standard Deviations

Table 67

Study 6 Trust Antecedents Means and Standard Deviations

Demographics Mean Std. Deviation Gender 0.48 0.51 Year in School 1.43 0.87 Age 19.86 3.51 Ethnicity 2.71 2.37 Previous Experiences with robots Mean Std. Deviation Movies 3.05 0.97 Interacted 0.33 0.48 Built 0.05 0.22 Controlled 0.33 0.48 Personality Traits Mean Std. Deviation extraversion 17.57 2.94 agreeableness 19.62 3.04 conscientiousness 18.62 2.80 intellect 18.19 1.36 neuroticism 13.48 2.09

291

APPENDIX V: COPYRIGHT PERMISSIONS

292

293

APPENDIX W: REFERENCES FOR PREVIOUS TRUST SCALES

294

Human-Robot Trust

Bainbridge, W. A., Hart, J., Kim, E. S., & Scassellati, B. (2008). The effect of presence on human-robot interaction.

Proceedings of the 17th IEEE Symposium on Robot and Human Interactive Communication, (pp.701-7-06).

doi:10.1109/ROMAN.2008.4600749

Items adapted from Kidd & Breazeal (2004) Interactive Experiences Questionnaire

Biros, D. P., Daly, M., & Gunsch, G. (2004). The influence of task load and automation trust on deception detection.

Group Decision and Negotiation, 13, 173-189. doi:10.1023/B:GRUP.0000021840.85686.57

Items adapted from Hoffman (2000)

De Ruyter, B., Saini, P. Markopoulous, P., & van Breeman, A. (2005). Assessing the effects of building social

intelligence in robotic interface in the home. Interacting with Computers, 17(5), 522-541.

doi:10.1016/j.intcom.2005.03.003

9 items that measure trust out of 172 items

Evers, V., Maldanado, H., Brodecki, T., & Hinds, P. (2008). Relational vs. Group Self-Construal: Untangling the

role of national culture in HRI. Proceedings in the 3rd ACM/IEEE International Conference on Human

Robot Interaction, Amsterdam, Holland, (pp. 255-262). Retrieved from

http://ieeexplore.ieee.org.ezproxy.net.ucf.edu/stamp/stamp.jsp?tp=&arnumber=6249443

3 items that measure trust

Heerink, M., Krőse, B., Evers, V., & Wielinga, B. (2010). Assessing acceptance of assistive social agent technology

by older adults: the almere model. International Journal of Social Robots, 2(4), 361-375.

doi:10.1007/s12369-010-0068-5

2 items that measure trust

Kidd, C. D., & Breazeal, C. (2004). Effect of a robot on user perception. Proceedings of the IEEE/RSJ International

Conference on Intelligent Robots and Systems, 4 (pp. 3559-3564). doi: 10.1109/IROS.2004.1389967

Number of items not discussed

Kidd, C. D. (2003). Sociable robots: The role of presence and task in human-robot interaction (Unpublished

master’s thesis). Massachusetts Institute of Technology, Cambridge, MA.

295

6 items that measure trust

Kiesler, S., Powers, A., Fussell, S.R., & Torry, C. (2008). Anthropomorphic interactions with a robot and robot-like

agents. Social Cognition, 26(2), 169-181. doi:10.1521/soco.2008.26.2.169

5 items that measure trust (same as Powers et al., 2007)

Li, D., Rau, P., & Li, Y. (2010). A cross-cultural study: Effect of robot appearance and task. International Journal

of Social Robots, 2, 175-186. doi:10.1007/s12369-010-0056-9

Items adapted from the SHAPE Automation Trust Index (SATI) trust scale (found in the Appendix

of Adams et al., 2003)

Looije, R., Neerinex, M. A., & Cnossen, F. (2010). Persuasive robotic assistant for health self-management of older

adults: Design and evaluation of social behaviors. International Journal of Human-Computer Studies, 68,

386-397. doi:10.1016/j.ijhcs.2009.08.007

4 items of trust (based on de Ruyter, 2005).

Mutlu, B., Yamaoka, F., Kanda, T., Ishiguro, H., & Hagita, N. (2009). Nonverbal leakage in robots: Communication

of intentions through seemingly unintentional behavior. Proceedings of the 4th ACM/IEEE International

Conference on Human Robot Interaction, La Jolla, California, (pp.69-76). Retrieved from

http://ieeexplore.ieee.org.ezproxy.net.ucf.edu/stamp/stamp.jsp?tp=&arnumber=6256096

Number of items that measure trust not reported

Powers, A., Kiesler, S., Fussell, S., & Torrey, C. (2007). Comparing a computer agent with a humanoid robot.

Proceedings of the International Conference on Human Robot Interaction, Arlington, VA, (pp. 145-152).

Retrieved from http://ieeexplore.ieee.org.ezproxy.net.ucf.edu/stamp/stamp.jsp?tp=&arnumber=6251681

5 items of trust (same as Kiesler et al., 2008)

Rau, P. L., Li, Y., & Li, D. (2009). Effects of communication style and culture on ability to accept recommendations

from robots. Computers in Human Behavior, 25(2), 587-595. doi:10.1016/j.chb.2008.12.025

6 items that measure trust

Ross, J.M. (2008). Moderators of trust and reliance across multiple decision aids. (Unpublished doctoral

dissertation). University of Central Florida, Orlando, FL.

45 items of trust (8 exit trust items, 25 interpersonal trust items, and 12 pre-trust items).

296

Scopelliti, M., Giuliani, M.V., & Fornara, F. (2005). Robots in a domestic setting: A psychological approach.

Universal Access in the Information Society, 4(2), 146-155. doi:10.1007/s10209-005-0118-1

12 items of trust

Tenney, Y. J., Rogers, W. H., & Pew, R. W. (1998). Pilot opinions of cockpit automation issues. The International

Journal of Aviation Psychology, 8(2), 103-120. doi:10.1207/s15327108ijap0802_2

1 item of trust

Tsui, K.M., Desai, M., & Yanco, H.A. (2010). Considering the bystander’s perspective for indirect human-robot

interaction. Proceedings of the 5th ACM/IEEE International Conference on Human Robot Interaction (pp.

129–130). doi: 10.1109/HRI.2010.5453230

1 item of trust.

Wang, L., Rau, P. L., Evers, V., Robinson, B. K., & Hinds, P. (2010). When in Rome: The role of culture and

context in adherence to robot recommendations. Proceedings of the 5th ACM/IEEE International

Conference on Human Robot Interaction, (pp. 359-366). doi: 10.1109/HRI.2010.5453165

11 items of general trust (extracted from Jian, Bisantz, & Drury, 2000). Two additional items of trust

in the “unknown” and “friend” feedback.

297

Interpersonal Trust Most widely cited

Jarvenpaa, S., Knoll, K. and Leidner, D. (1998). Is anybody out there? Antecedents of trust in global virtual teams.

Journal of Management Information Systems - Special section: Managing virtual workplaces and

teleworking with information technology, 14(4), 29-64.

Items were from Pearce et al. (1992); and Schoorman, Mayer and Davis (1996)

Larzelere, R. E, & Huston, T. L. (1980). The dyadic trust scale: Toward understanding interpersonal trust in close

relationships. Journal of Marriage and the Family, 42, 595-604.

8 items of trust

Mayer, R. C., & Davis, J. H. (1999). The effect of the performance appraisal system on trust for management: A

field quasi-experiment. Journal of Applied Psychology, 84, 123–136. doi:10.1037/0021-9010.84.1.123

17 items, plus an additional 4 items from Schoorman et al. (1996). Items can be found in Colquitt,

Scott & LePine (2007)

McAllister, D. J. (1995). Affect- and cognition-based trust as foundations for interpersonal cooperation in

organizations. Academy of Management Journal, 38(1), 24−59.

11 items of trust (6 cognitive-based trust, 5 affect-based trust).

Rempel, K. J., Holmes, J. G., & Zanna, M. P. (1985). Trust in close relationships. Journal of Personality and Social

Psychology. 49(1), 95-112. doi:10.1037/0022-3514.49.1.95

26 items that measure trust

Rotter, J. B. (1967). A new scale for the measurement of interpersonal trust. Journal of Personality, 35, 651-665.

doi:10.1111/j.1467-6494.1967.tb01454.x

25 items of trust and 15 filler items

298

Interpersonal Trust Others cited

Aberg, J. & Shahmehri, N. (2000). The role of human Web assistants in e-commerce: an analysis and a usability

study. Internet Research: Electronic Networking Application and Policy, 10(2), 114-125.

doi:10.1108/10662240010322902

1 item of trust

Brockner, J., Siegel, P.A., Daly, J. P., Martin, C., & Tyler, T. (1997). When Trust Matters: The Moderating Effect

of Outcome Favorability. Administrative Science Quarterly, 42.

3 items of trust

Butler, J. K., Jr. (1991). Toward understanding and measuring conditions of trust: Evolution of a Conditions of Trust

Inventory. Journal of Management, 17, 643-663. doi:10.1177/014920639101700307

Number of items in developed scale not listed.

Cook, J., & Wall, T. (1980). New work attitude measures of trust, organizational commitment and personal need

non-fulfillment. Journal of Occupational Psychology, 53, 39-52.

Number of items not listed

Dunn, J. R., & Schweitzer, M. E. (2005). Feeling and believing: The influence of emotion on trust. Journal of

Personality and Social Psychology, 88, 736–748. doi:10.1037/0022-3514.88.5.736

10 items were adapted from Johnson-George and Swap’s (1982) Specific Interpersonal Trust Scale.

Gabarro J. J., & Athos P. (1978). Interpersonal relations and communications. New York: Prentice Hall.

7 items of trust

Hoffman, D. L., Novak, T. P. and Peralta, M. (1999). Building consumer trust online. Communications of the ACM,

42(4), 80-85. doi:10.1145/299157.299175

No items listed

Luo, Y. (2002). Building Trust in Cross-Cultural Collaborations: Toward a Contingency Perspective. Journal of

Management, 28(5), 669-694. doi:10.1177/014920630202800506

8 interpersonal trust items, 8 interorganizational trust items

299

Mayer, R. C., & Gavin, M. B. (2005). Trust in management and performance: Who minds the shop while the

employees watch the boss? Academy of Management Journal, 48(5), 874−888.

doi:10.5465/AMJ.2005.18803928

3 scales of trust (Mayer & Davis, 1999)

Ohanian, R. (1990). Construction and validation of a scale to measure celebrity endorsers’ perceived expertise,

trustworthiness and attractiveness. Journal of Advertising, 19(3), 39–52.

8 items of trustworthiness

Pearce, J. L., Sommer, S. M., Morris, A., & Frideger, M. (1992). A configurational approach to interpersonal

relations: profiles of workplace social relations and task interdependence. Graduate School of Management,

University of California, Irvine, November.

8 items of organizational trustworthiness

Ratnasingham, P. (1998). The importance of trust in electronic commerce. Internet Research: Electronic Networking

Applications and Policy, 8(4), 313-321. doi: 10.1108/10662249810231050

No items listed

Roberts, K. H., & O'Reilly, C. A. (1974). Measuring organizational communication. Journal of Applied Psychology,

59, 321-326. doi:10.1037/h0036660

3 items of trust

Robinson, S. L. (1996). Trust and breach of the psychological contract. Administrative Science Quarterly, 41, 574-

599.

7 items (Gabarro & Athros, 1978)

Shockley-Zalabak, P., Ellis, K., & Cesaria, R. (2000). Measuring organizational trust. San Francisco: International

Association of Business Communicators Research Foundation.

29 items of trust

300

Automation Trust

Abe, G., & Richardson, J. (2004). The effect of alarm timing on driver behavior: An investigation of differences in

driver trust and response to alarms according to alarm timing. Transportation Research Part F: Traffic

Psychology and Behaviour, 7(4/5), 307-322. doi:10.1016/j.trf.2004.09.008

1 item of trust

Abe, G., & Richardson, J. (2005). The influence of alarm timing on braking response and driver trust in low speed

driving. Safety Science, 43(9), 639-654. doi:10.1016/j.ssci.2005.04.006

1 item of trust

Abe, G., & Richardson, J. (2006). The influence of alarm timing on driver response to collision warning systems

following system failure. Behavior & Information Technology, 25(5), 443-452.

doi:10.1080/01449290500167824

10 items of trust

Bagheri, N. (2004). The Effect of Automation Reliability on Operator Monitoring Performance. (Unpublished

master’s thesis). University of Toronto, Canada.

Adapted Lee and Moray (1992, 1994)

Bagheri, N., & Jamieson, G. (2004). Considering subjective trust and monitoring behavior in assessing automation-

induced “complacency.” Proceedings of the Human Performance, Situation Awareness, and Automation

Conference, Marietta, GA (pp. 54-59). Retrieved from

http://cel.mie.utoronto.ca/publications/files/conference/HPSAA_04_Bagheri_Jamieson.pdf

Adapted Lee and Moray (1992, 1994)

Bailey, N. R. & Scerbo, M. W. (2007). Automation-induced complacency for monitoring highly reliable systems:

the role of task complexity, system experience and operator trust. Theoretical Issues in Ergonomics

Science, 8(4), 321-348.

12 items of trust

301

Bisantz, A., & Seong, Y. (2001). Assessment of operator trust in and utilization of automated decision-aids under

different framing conditions. International Journal of Industrial Ergonomics, 28(2), 85-97.

12 items of trust

Bliss, J. P., & Acton, S. A. (2003). Alarm mistrust in automobiles: How collision alarm reliability effects driving.

Applied Ergonomics, 34(6), 499-509. doi:10.1016/j.apergo.2003.07.003

Mistrust items not included

Cahour, B. & Forzy, J. (2009). Does projection into use improve trust and exploration? An example with a cruise

control system. Safety Science, 47, 1260-1270.

4 items of trust

Cai, H. & Lin, Y. (2012). Coordinating Cognitive Assistance with Cognitive Engagement Control Approaches in

Human-Machine Collaboration. IEEE Transactions on Systems, Man, and Cybernetics – Part A: Systems

and Humans, 42(2), 286-294.

1 item of trust

Cai, H., Lin, Y., & Cheng, B. (2012). Coordinating multi-level cognitive assistance and incorporating dynamic

confidence information in driver-vehicle interfaces. Human Factors and Ergonomics in Manufacturing &

Service Industries, 22(5), 437-449. doi:10.1002/hfm.20399

1 item of trust

Chen, J. Y. C., & Terrence, P. I. (2009). Effects of imperfect automation and individual differences on concurrent

performance of military and robotics tasks in a simulated multitasking environment. Ergonomics, 52(8),

907-920.

Items were modified version of Jian et al. (2000) trust scale (items 22-33)

Cummings, M. L., Buchin, M., Carrigan, G., & Donmez, B. (2010). Supporting intelligent and trustworthy maritime

path planning decisions. International Journal of Human-Computer Studies, 68(10), 616-626.

doi:10.1016/j.ijhcs.2010.05.002

Items from Kelly et al., (2003)

302

de Vries, P., Midden, C., & Bouwhuis, D. (2003). The effects of error on system trust, self-confidence, and the

allocation of control in route planning. International Journal of Human-Computer Studies, 58, 719-735.

doi:10.1016/S1071-5819(03)00039-9

Number of items not listed de Vries, P., & Midden, C. (2008). Effect of indirect information on system trust and control allocation. Behaviour

& Information Technology, 27(1), 17-29. doi:10.1080/01449290600874956

Number of items not listed

Donmez, B., Boyle, L. N., Lee, J. D., & McGehee, D. V. (2006). Drivers’ attitudes toward imperfect distraction

mitigation strategies. Transportation Research Part F, 9, 387-398. doi:10.1016/j.trf.2006.02.001

Items from Wiese (2003) and Bisantz & Seong(2001), as well as 2 additional items

Dzindolet, M. T., Peterson, S. A., Pomranky, R. A., Pierce, L. G., & Beck, H. P. (2003). The role of trust in

automation reliance. International Journal of Human-Computer Studies, 58, 697-718. doi:10.1016/S1071-

5819(03)00038-7

8 items of trust

Gupta, N., Bisantz, A.M., & Singh, T. (2002). The effects of adverse condition warning system characteristics on

driver performance: an investigation of alarm signal type and threshold level. Behaviour & Information

Technology, 21(4), 235-248. doi:10.1080/0144929021000013473

12 item from Jian et al. (2000)

Ho, G., Wheatley, D., & Scialfa, C.T. (2005). Age differences in trust and reliance of a medication management

system. Interacting with Computers, 17, 690-710. doi:10.1016/j.intcom.2005.09.007

Items similar to Lee and Moray (1994); Muir and Moray (1996)

Hughes, J.S., Rice, S., Traimow, D., & Clayton, K. (2009). The automated cockpit: A comparison of attitudes

towards human and automated pilots. Transportation Research Part F, 12, 428-439. doi:

10.1016/j.trf.2009.08.004

Number of items not listed

Jamieson, G.A., Wang, L., & Neyedli, H.F. (2004). Developing Human-Machine Interfaces to Support Appropriate

Trust and Reliance on Automated Combat Identification Systems (Report Number: DRDC Toronto CR

303

2008-114). Toronto, Canada: Department of National Defence. Retrieved from http://www.dtic.mil/cgi-

bin/GetTRDoc?AD=ADA485517

Items similar to Lee & Moray (1994)

Jian, J., Bisantz, A., Drury, C. G., & Llinas, J. (1998). Foundations for an empirically determined scale of trust in

automated systems (AFRL-HE-WP-TR-2000-0102). Wright-Patterson AFB, OH: Air Force Research

Laboratory. Retrieved from: http://www.dtic.mil/cgi-bin/GetTRDoc?AD=ADA395339

12 items of trust

Johnson, D.S. (2007). Achieving Customer Value from Electronic Channels through Identity Commitment,

Calculative Commitment, and Trust in Technology. Journal of Interactive Marketing 21(4), 2-22.

doi:10.1002/dir.20091

5 items of trust

Kazi, T. A., Stanton, N. A., Walker, G. H., & Young, M. S. (2007). Designer Driving: Drivers’ Conceptual Models

and level of trust in Adaptive Cruise Control. International Journal of Vehicle Design, 45(3), 339-360.

doi:10.1504/IJVD.2007.014909

10 items of trust

Kelly, C., Boardman, M., Goillau, P., Jeannot, E. (2003). Guidelines for Trust in Future ATM Systems: Measures

(No. HRS/HSP-005-GUI- 02). EUROCONTRO: European Air Traffic Management Programme. Retrieved

from: http://www.eurocontrol.int/sites/default/files/content/documents/nm/safety/safety-guidelines-for-

trust-in-future-atm-systems-measures-2003.pdf

8 items with multiple parts

Kircher, K. & Thorslund, B. (2009). Effects of road surface appearance and low friction warning systems on driver

behaviour and confidence in the warning system. Ergonomics, 52(2), 165-176.

doi:10.1080/00140130802277547

1 item of trust

Koh, Y. J., & Sundar, S. S. (2010). Effects of specialization in computer, web sites, and web agents on e-commerce

trust. International Journal of Human-Computer Studies, 68(12), 899-912. doi:10.1016/j.ijhcs.2010.08.002

30 items of trust toward media technology (adapted from McKnight et al., 2002 trusting belief scale)

304

Koustanai, A., Cavallo, V., Dalhomme, P., & Mas, A. (2012). Simulator training with a forward collision warning

system: Effects on driver-system interactions and driver trust. Human Factors: The Journal of the Human

Factors and Ergonomics Society, 54, 709-721. doi:10.1177/0018720812441796

12 items of trust

Lee, J. D. (1991). The Dynamics of Trust in a Supervisory Control Simulation. Proceedings of the Human Factors

and Ergonomics Society, System Development, 35(17), 1228-1232. doi: 10.1177/154193129103501712

Number of items not listed

Lee, K., & Chung, N. (2009). Understanding factors affecting trust in and satisfaction with mobile banking in Korea:

A modified DeLone and McLean’s model perspective. Interacting with Computers, 21(5/6), 385-392.

doi:10.1016/j.intcom.2009.06.004

5 item of trust in mobile banking items

Lee, H., Kim, J., & Kim, J. (2007). Determinants of success for applications service provider: An empirical test in

small businesses. International Journal of Human-Computer Studies, 65(9), 796-815.

doi:10.1016/j.ijhcs.2007.04.004

3 items of trust from McKnight et al. (2002)

Lee, J. D. & Moray, N. (1994). Trust, self-confidence, and operators’ adaptation to automation. International

Journal of Human-Computer Studies, 40, 153-184.

6 items of trust

Lerch, F. & Prietula, M. (1989). How do we trust machine advice? In G. Salvendy and M. J. Smith (Eds.),

Designing and Using Human Computer Interfaces and Knowledge Based Systems, II (411-419). New York,

NY: Elsevier Science.

Lorenz, B., Di Nocera, F., Rottger, S. & Parasuraman, R. (2002). Automated Fault-Management in a Simulated

Spaceflight Micro-World. Aviation Space Environment Med, 73, 886-897.

Items were from Lee & Moray (1992)

Ma, R., & Kaber, D. B. (2007). Effects of in-vehicle navigation assistance and performance on driver trust and

vehicle control. International Journal of Industrial Ergonomics, 37(8), 665-673.

doi:10.1016/j.ergon.2007.04.005

305

Items adapted from Dzindolet et al. (2002), as well as 2 additional items

Madhavan, P., & Wiegmann, D. A. (2007). Similarities and differences between human-human and human-

automation trust: An integrative review. Theoretical Issues in Ergonomics Science, 8, 277–301.

doi:10.1518/001872007X230154

Items adapted from Jian et al. (2000)

Madhavan, P., Wiegmann, D. A., & Lacson, F. C. (2006). Automation Failures on Tasks Easily Performed by

Operators Undermine Trust in Automated Aids. Human Factors, 48, 241-256.

doi:10.1518/001872006777724408

Items not listed

Madsen, M. & Gregor, S. (2000). Measuring Human-Computer Trust. Proceedings of the 11th Australasian

Conference on Information Systems, (pp. 6-8). Retrieved from:

http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.93.3874&rep=rep1&type=pdf

25 items of trust

Manzy, D., Rottger, S., Bahner-Heyne, J.E., Schulze-Kissing, D., Dietz, A., Meixensberger, J., & Strauss, G. (2009).

Image-guided navigation: the surgeon’s perspective on performance consequences and human factors

issues. The International Journal on Medical Robotics and Computer Assisted Surgery, 5, 297-308.

doi:10.1002/rcs.261

25 items of trust

Masalonis, A. J., & Parasuraman, R. (1999). Trust as a Construct for Evaluation of Automated Aids: Past and Future

Theory and Research. Proceedings of Human Factors and Ergonomics Society, Cognitive Engineering and

Decision Making, 43(3), 184-187). doi: 10.1177/154193129904300312

Items not listed

Master, R., Jiang, X., Khasawneh, M.T., Bowling, S.R., Grimes, L., Gramopadhye, A.K., & Melloy, B.J. (2005).

Measurement of Trust Over Time in Hybrid Inspection Systems. Human Factors and Ergonomics in

Manufacturing, 15(2), 177-196. doi:10.1002/hfm.20021

Two trust questionnaires (Jian et al., 2000; Master et al., 2000)

306

Master, R., Gramopadhye, A. K., Melloy, B. J., Bingham, J., & Jiang, X. (2000). A questionnaire for measuring trust

in hybrid inspection systems. Proceedings of the Industrial Engineering Research Conference, Dallas, TX.

5 items of trust

Merritt, S. M., Heimbaugh, H., LaChapell, J., & Lee, D. (2012). I trust it, but I don’t know why: Effects of implicit

attitudes toward automation on trust in an automated system. Human Factors, 55(3), 520-534. doi:

10.1177/0018720812465081

6 items pre-post task trust

Merritt, S. M., & Ilgen, D. R. (2008). Not all trust is created equal: Dispositional and history-based trust in human-

automation interactions. Human Factors, 50(2), 194-210. doi:10.1518/001872008X288574

6 items pre-post task trust

Montague, E. (2010). Validation of a trust in medical technology instrument. Applied Ergonomics, 41, 812-821.

doi:10.1016/j.apergo.2010.01.009

31 items of trust

Montague, E., Winchester, W.W., & Kleiner, B.M. (2010). Trust in medical technology by patients and healthcare

providers in obstetric work systems. Behaviour & Information Technology, 29(5), 541-554.

doi:10.1080/01449291003752914

5 items of trust

Moray, N., Inagaki, T., & Itoh, M. (2000). Adaptive automation, trust, and self-confidence in fault management of

time-critical tasks. Journal of Experimental Psychology: Applied, 6(1), 44-58. doi:10.1037/1076-

898X.6.1.44

Items adapted from Muir (1989, 1994), Lee (1991), and Lee & Moray (1994).

Muir, B. M. & Moray, N. (1996). Trust in automation. Part II. Experimental studies of trust and human intervention

in a process control simulation. Ergonomics, 39(3), 429-460. doi:10.1080/00140139608964474

9 items of trust

Neyedli, H. F., Hollands, J. G., & Jamieson, G. A. (2009). Human reliance on an automated combat ID system:

Effects of display format. Proceedings of the 53rd Human Factors and Ergonomics Society Annual

Meeting, 53(4), 212-216. doi:10.1177/154193120905300411

307

Items from Jian, Bisantz, & Drury (2000) with two additional items

Pak, R., Fink, N., Price, M., Bass, B. & Sturre, L. (2012). Decision support aids with anthropomorphic

characteristics influence trust and performance in younger and older adults. Ergonomics, 55(9), 1059-1072.

doi:10.1080/00140139.2012.691554

Items not reported

Rajaonah, B., Anceaux, F., & Vienne, F. (2006). Study of driver trust during cooperation with adaptive cruise

control. Le Travail Humain, 69(2), 99-127.

4 items of trust

Rajaonah, B., Tricot, N., Anceaux, F., & Millot, P. (2008). The role of intervening variables in driver - ACC

cooperation. International Journal of Human-Computer Studies, 66, 185-197.

doi:10.1016/j.ijhcs.2007.09.002

11 items of trust

Rani, M. R., Sinclair, M. A., & Case, K. (2000). Human mismatches and preferences for automation. International

Journal of Production Research, 38(17), 4033-4039. doi:10.1080/00207540050204894

Items not reported

Rovira, E., McGarry, K., & Parasuraman, R. (2007). Effects of Imperfect Automation on Decision Making in a

Simulated Command and Control Task. Human Factors, 49, 76-87. doi: 10.1518/001872007779598082

Items from Lee & Moray (1994)

Rovira, E. & Parasuraman, R. (2010). Transitioning to Future Air Traffic Management: Effects of Imperfect

Automation on Controller Attention and Performance. Human Factors, 52(3), 411-425.

doi:10.1177/0018720810375692

Items Lee & Moray (1992, 1994)

Ruff, H. A., Narayanan, S., & Draper, M. H. (2002). Human Interaction with Levels of Automation and Decision-

Aid Fidelity in the Supervisory Control of Multiple Simulated Unmanned Air Vehicles. Presence, 2(4),

335-351. doi:10.1162/105474602760204264

Items based on Masalonis & Parasuraman (1999)

308

Seong, Y., & Bisantz, A.M. (2008). Modeling human trust in complex, automated systems using a lens model

approach. In M. W. Scerbo & M. Mouloua (Eds.), Automation Technology and Human Performance:

Current Research and Trends (pp. 282-286). Mahwah, NJ: Lawrence Erlbaum Associates. Retrieved from:

http://www.dtic.mil/cgi-bin/GetTRDoc?AD=ADA354935#page=295

12 items from Jian et al., (2000)

Singh,I., Molloy, R., & Parasuraman, R. (1993). Automation-induced complacency: Development of the

complacency-potential rating scale. The International Journal of Aviation Psychology, 3(2), 111-122.

doi:10.1207/s15327108ijap0302_2

3 items of trust

Smyth, C. C. (2007). Sensitivity of subjective questionnaires to cognitive loading with driving with navigation aids:

A pilot study. Aviation, Space, and Environmental Medicine, 78(5), B39-B50. Retrieved from:

http://www.ingentaconnect.com.ezproxy.net.ucf.edu/content/asma/asem/2007/00000078/A00105s1/art0000

7?token=004b14b5c274383a4b3b25702a7b757a7a38537c47665d2a726e2d5b426c6f642f466f01216

Items not reported

Spain, R.D. & Bliss, J.P. (2008). The effect of sonification display pulse rate and reliability on operator trust and

perceived workload during a simulated patient monitoring task. Ergonomics, 51(9). 1320-1337.

doi:10.1080/00140130802120234

12 items from Jian et al. (2000) and modified by Fallon et al. (2005)

Stedmon, A. W., Sharples, S., Littlewood, R., Cox, G., Patel, H., & Wilson, J. R. (2007). Datalink in air traffic

management: Human factors issues in communications. Applied Ergonomics, 38(4), 473-480.

doi:10.1016/j.apergo.2007.01.013

Items not reported

Uggirala, A., Gramopadhye, A. K., Melloy, B. J., & Toler, J. E. (2004). Measurement of trust in complex and

dynamic systems using a quantitative approach. International Journal of Industrial Ergonomics, 34, 175-

186. doi:10.1016/j.ergon.2004.03.005

5 items of trust

309

Verberne, F. M. F., Ham, J., & Midden, C. J. H. (2012). Trust in Smart Systems: Sharing driving goals and giving

information to increase trustworthiness and acceptability of smart systems in cars. The Journal of the

Human Factors and Ergonomics Society, 54, 799-810. doi: 10.1177/0018720812443825

7 items of trust from Jian et al. (2000)

Wang, L., Jamieson, G.A., & Hollands, J.G. (2009). Trust and Reliance on an Automated Combat Identification

System. Human Factors, 51, 281-291. doi:10.1177/0018720809338842

13 items of trust

Wang, L., Jamieson, G. A., & Hollands, J. G. (2011). The effects of design features on users’ trust in and reliance

on a combat identification system. Proceedings of the Human Factors and Ergonomics Society, 55(1), 375-

379. doi: 10.1177/107118131551077

Items from Jian et al. (2000)

310

REFERENCES

311

Abe, G., & Richardson, J. (2006). The influence of alarm timing on driver response to collision warning systems

following system failure. Behavior & Information Technology, 25(5), 443-452.

doi:10.1080/01449290500167824

Adams, B. D., Bruyn, L.E., Houde, S., & Angelopoulos, P. (2003). Trust in Automated Systems Literature Review

(Report No. CR-2003-096). Toronto, Ontario, Canada: Department of National Defense.

Adams, D. (Writer). (1981). Episode 5 [Television series episode]. In A. J. W. Bell & J. Lloyd (Producers), The

Hitch Hikers Guide to the Galaxy. London: British Broadcasting Corporation.

Allen, I. (Writer). (1965). Lost in space [Television series]. In. I. Allen (Producer), Los Angeles: Irwin Allen

Productions.

Allen, I. (Writer) & Martin, S. (Director). (1966). War of the Robots. [Television series episode]. In. I. Allen

(Producer), Lost in Space. Los Angeles: Irwin Allen Productions.

Allen, I. (Writer) & Richardson, D. (Director). (1966). The Challenge. [Television series episode]. In. I. Allen

(Producer), Lost in Space. Los Angeles: Irwin Allen Productions.

Anderson, J. R. (1981). Concepts, propositions, and schemata: What are the cognitive units? In J. H. Flowers (Ed.),

Nebraska Symposium on Motivation: Cognitive Processes (pp.121-162). Lincoln, Nebraska: University of

Nebraska Press. Retrieved from: http://act-r.psy.cmu.edu/papers/58/Concepts.JRA.pdf

Ashleigh, M. J., & Stanton, N. A. (2001). Trust: Key elements in human supervisory control domains. Cognition,

Technology & Work, 3(2), 92-100. doi:10.1007/PL00011527

Atoyan, H., Duquet, J. R., & Robert, J. M. (2006). Trust in new decision aid systems. Proceedings of the 18th

Human-Computer Interaction (IHM) Conference, 58, 115-122. doi:10.1145/1132736.1132751

Axelrod, R. H. (2004). Deciding to Trust a Stranger: An exploratory study of how people make decisions based on

general interpersonal trust (with implications for organizational settings) (Doctoral Dissertation). George

Washington University.

Axelrod, R. M. (1976). The structure of decision. Princeton: Princeton University Press.

Bahner, J. E., Hiper, A. D., & Manzey, D. (2008). Misuse of automated decision aids: Complacency, automation

bias and the impact of training experience. International Journal of Human-Computer Studies, 66(9), 688-

699. doi:10.1016/j.ijhcs.2008.06.001

312

Bailey, N.R. & Scerbo, M.W. (2007). Automation-induced complacency for monitoring highly reliable systems: the

role of task complexity, system experience and operator trust. Theoretical Issues in Ergonomics Science,

8(4), 321-348. doi:10.1080/14639220500535301

Bailey N. R., Scerbo M. W., Freeman F. G., Mikulka P. J., & Scott, L. A. (2006). Comparison of a brain-based

adaptive system and a manual adaptable system for invoking automation. Human Factors: The Journal of

the Human Factors and Ergonomics Society, (48)4, 693-709. doi:10.1518/001872006779166280

Bainbridge, W. A., Hart, J., Kim, E. S., & Scassellati, B. (2008). The effect of presence on human-robot interaction.

Proceedings of the 17th IEEE Symposium on Robot and Human Interactive Communication, (pp.701-7-06).

doi:10.1109/ROMAN.2008.4600749

Barber, B. (1983). The logic and limits of trust. New Brunswick, NJ: Rutgers University Press.

Bartneck, T. Kanda, O. Mubin, & A. Al Mahmud, (2009). Does the Design of a Robot Influence its Animacy and

Perceived Intelligence? International Journal of Social Robotics, 1, 195-204. doi:10.1007/s12369-009-

0013-7

Bartneck, C., Kulić, D., Croft, E., & Zoghbi, S. (2009). Measurement Instruments for the Anthropomorphism,

Animacy, Likeability, Perceived Intelligence, and Perceived Safety of Robots. International Journal of

Social Robots, 1, 71-81. doi: 10.1007/s12369-008-001-3.

BBC News. (1986). Coal Mine Canaries Made Redundant. Retrieved from

http://news.bbc.co.uk/onthisday/hi/dates/stories/december/30/newsid_2547000/2547587.stm

Beckles, B., Welch, V., & Basney, J. (2005). Mechanisms for increasing the usability of grid security. International

Journal of Human-Computer Studies, 63(1/2), 74-101. doi:10.1016/j.ijhcs.2005.04.017

Bhattacharya, R., Devinney, T. M., & Pilluta, M. M. (1998). A Formal Model of Trust Based on Outcomes. The

Academy of Management Review, 23(3), 459-472. doi: 10.5465/AMR.1998.926621

Billings, D. R., Schaefer, K. E., Hancock, P. A., Kocsis, V., Barrerra, M., Cook, J., & Ferrer, M. (2012a). Human-

Animal Trust as an Analog for Human-Robot Trust: A review of current evidence (Report No. ARL-TR-

5949). Aberdeen Proving Ground, Maryland: Army Research Laboratory.

313

Billings, D. R., Schaefer, K. E., Llorens, N., & Hancock, P. A. (2012b). What is Trust? Defining the construct

across domains. Poster presented at the American Psychological Association Conference, Division 21.

Orlando, FL.

Biros, D. P., Daly, M., & Gunsch, G. (2004). The influence of task load and automation trust on deception detection.

Group Decision and Negotiation, 13(2), 173-189. Retrieved from:

http://link.springer.com/content/pdf/10.1023%2FB%3AGRUP.0000021840.85686.57.pdf

Bloomqvist, K. (1997). The many faces of trust. Scandinavian Journal of Management, 13(3), 271-286.

Boon, S., & Holmes, J. (1991). The dynamics of interpersonal trust: Resolving uncertainty in the face of risk. In R.

Hindle, & J. Grobel. (Eds.), Cooperation and Prosocial Behavior (pp. 167-182). New York: Cambridge

University Press.

Borst, C., Mulder, M., & van Paassen, M. M. (2010). Design and simulator evaluation of an ecological synthetic

vision display. Journal of Guidance, Control, and Dynamics, 33(5), 1577-1591. doi:10.2514/1.47832

Brackett, L., & Kasden, L. (Writers). (1980). Star Wars: Episode V – The Empire Strikes Back [Motion Picture]. San

Francisco: Lucasfilm.

Breazeal, C. (2003). Toward Social Robots. Robotics and Autonomous Systems, 42, 165-175. doi:10.1016/S0921-

8890(02)00373-1

Breazeal, C. & Scassellati, B. (2002). Robots that imitate humans. Trends in Cognitive Sciences, 6(11), 481-487.

doi:10.1016/S1364-6613(02)02016-8

Burke, C. S., Sims, D. E., Lazzara, E. H., & Salas, E. (2007). Trust in Leadership: A multi-level review and

integration. The Leadership Quarterly, 18, 606-632. doi: 10.1016/j.leaqua.2007.09.006

Campbell, D. T. & Fiske, D. W. (1959). Convergent and discriminant validation by the multitrait-multimethod

matrix. Psychological Bulletin, 56, 81-105.

Čapek, K. (1921). R.U.R. (Rossum’s universal robots). London, UK: Penguin.

Chaorro-Premuzic, T., von Stumm, S., and Furnham, A. (Eds.). (2011). The Wiley-Blackwell Handbook of

Individual Differences (pp. 7-8). Malden, MA: Blackwell Publishing Ltd.

Chen, J. Y. C., & Barnes, M. J. (2012). Supervisory control of multiple robots: Effects of imperfect automation and

individual differences. Human Factors, 54(2), 157-174. doi:10.1177/0018720811435843

314

Chen, J. Y. C., & Terrence, P. I. (2009). Effects of imperfect automation and individual differences on concurrent

performance of military and robotics tasks in a simulated multi-tasking environment. Ergonomics, 52(8),

907-920. doi:10.1080/00140130802680773

Coleman, J. S. (1990). Foundations of Social Theory. Harvard University Press, Cambridge.

Colquitt, J. A., Scott, B. A., & LePine, J. A. (2007). Trust, trustworthiness, and trust propensity: A meta-analytic

test of their unique relationships with risk taking and job performance. Journal of Applied Psychology, 92(4),

909-927. doi:10.1037/0021-9010.92.4.909

Cosenzo, K.A., Parasuraman, R., Novak, A., & Barnes, M. (2006). Implementation of Automation for Control of

Robotics Systems (Report No. ARL-TR-3808). Aberdeen Proving Ground, Maryland: Army Research

Laboratory.

Couch, C. & Gabai, E. (Writers). (1962). The Jetsons [Television Series]. United States: Hanna-Barbera

Productions.

Couch L. L., & Jones, W. H. (1997). Measuring levels of trust. Journal of Research in Personality. 31(3). 319-336.

doi: 0092-6566/97

Cring, E. A. & Lenfestey, A. G. (2009). Architecting human operator trust in automation to improve system

effectiveness in multiple unmanned aerial vehicle (UAV) Control. (Master’s thesis: Report No.

AFIT/GSE/EMV/09-M06). Air Force Institute of Technology, Wright-Patterson AFB, Ohio.

Daellenbach, U. S. & Davenport, S. J. (2004). Establishing trust during the formation of technology alliances.

Journal of Technology Transfer, 29, 187-202. doi:1 0.1023/B:JOTT.0000019537.61121.c3

Dassonville, I., Jolly, D., & Desodt, A.M. (1996). Trust between man and machine in a teleoperation system.

Reliability Engineering and System Safety, 53(3), 319-325. doi: 0951-8320/96

Dautenhahn, K. (1999). Robots as Social Actors: Aurora and the case of autism. Proceedings of the 3rd Cognitive

Technology Conference, San Francisco, CA. doi:10.1.1.190.1767

Dautenhahn, K. (2007). Socially Intelligent Robots: Dimensions of human-robot interaction. Philosophical

Transactions of the Royal Society B, 362, 679-704. doi:10.1098/rstb.2006.2004

Davids, A. (2002). Urban Search and Rescue Robots: From Tragedy to Technology. IEEE Intelligent Systems, 17(2),

81-83. doi:10.1109/MIS.2002.999224

315

de Ruyter, B., Saini, P. Markopoulous, P., & van Breeman, A. (2005). Assessing the effects of building social

intelligence in robotic interface in the home. Interacting with Computers, 17(5), 522-541.

doi:10.1.1.89.7807 de Vries, P., Midden, C., & Bouwhuis, D. (2003). The effects of errors on system trust, self-confidence, and the

allocation of control in route planning. International Journal of Human-Computer Studies, 58, 719-735.

doi:10.1016/S1071-5819(03)00039-9

DeSantis, A., Siciliano, B., De Luca, A., & Bicchi, A. (2008). An atlas of physical human-robot interaction.

Mechanism and Machine Theory, 43(3), 253-270. doi: 10.1016/j.mechmachtheory.2007.03.003

Desai, M., Stubbs, K., Steinfeld, A. & Yanco, H. (2009). Creating Trustworthy Robots: Lessons and inspirations

from automated systems. Proceedings of the AISB Convention: New Frontiers in Human-Robot

Interaction, Edinburgh, Scotland. Retrieved from

https://www.ri.cmu.edu/pub_files/2009/4/Desai_paper.pdf

DeVellis, R. F. (2003). Scale Development Theory and Applications (2nd ed.). Applied Social Research Methods

Series (Vol. 26). Thousand Oaks, CA: Sage Publications, Inc.

Deutsch, M. (1962). Cooperation and Trust: Some theoretical notes. In M. R. Jones (Ed.), Nebraska Symposium on

Motivation (pp. 275-315). Lincoln, NB: University of Nebraska.

DiSalvo, C., & Gemperle, F. (2003). From seduction to fulfillment: the use of anthropomorphic form in design.

Proceedings of the International Conference on Designing Pleasurable Products and Interfaces, (pp. 67-

72). doi:10.1145/782896.782913

DiSalvo, C. F., Gemperle, F., Forlizzi, J., & Kiesler, S. (2002). All robots are not created equal: The design and

perception of humanoid robot heads. Proceedings of the 4th conference on Designing Interactive Systems:

Processes, Practices, Methods, and Techniques (pp. 321-326). doi:10.1.1.107.6205

Donmez, B., Boyle, L. N., Lee, J. D., & McGehee, D. V. (2006). Drivers' attitudes toward imperfect distraction

mitigation strategies. Transportation Research Part F: Traffic Psychology and Behaviour, 9(6), 387-398.

doi:10.1016/j.trf.2006.02.001

316

Donnellan, M. B., Oswald, F. L., Baird, B. M., & Lucas, R. E. (2006). The Mini-IPIP scales: Tiny-yet-effective

measures of the Big Five factors of personality. Psychological Assessment, 18(2), 192-203.

doi:10.1037/1040-3590.18.2.192

Duffy, B.R. (2003). Anthropomorphism and the social robot. Robotics and Autonomous Systems, 42, 177-190.

doi:10.1016/S0921-8890(02)00374-3

Drury, J. L., Riek, L., & Rackliffe, N. (2006). A decomposition of UAV-related situation awareness. Proceedings of

the 1st ACM SIGCHI/SIGART conference on Human-Robot Interaction. New York, NY, USA, (pp. 88-94).

doi:10.1145/1121241.1121258

Dzindolet, M. T., Peterson, S. A., Pomranky, R. A., Pierce, L. G., & Beck, H. P. (2003). The role of trust in

automation reliance. International Journal of Human-Computer Studies, 58(6), 697-718.

doi:10.1016/s1071-5819(03)00038-7

Elofson, G. (2001). Developing Trust with Intelligent Agents: An exploratory study. In C. Castelfranchi & Y-H Tan

(Eds.), Trust and Deception in Virtual Societies (pp. 125-138). The Netherlands: Springer.

doi:10.1007/978-94-017-3614-5_6

Evers, V., Maldanado, H., Brodecki, T., & Hinds, P. (2008). Relational vs. Group Self-Construal: Untangling the

role of national culture in HRI. Proceedings in the 3rd ACM/IEEE International Conference on Human

Robot Interaction, Amsterdam, Holland, (pp. 255-262). Retrieved from

http://ieeexplore.ieee.org.ezproxy.net.ucf.edu/stamp/stamp.jsp?tp=&arnumber=6249443

Feil-Seifer, D., & Matarić, M. J. (2005). Defining socially assistive robotics. In proceedings of the 9th IEEE

International Conference on Rehabilitation Robotics, Chicago, IL (pp.465-668).

doi:10.1109/ICORR.2005.1501143

Fink, A. (2009). How to Conduct Surveys: A Step-by-Step Guide (4th ed.). Thousand Oaks, CA: Sage Publications,

Inc.

Finkel, M. (2012). The Cold Patrol. National Geographic, 221 (1), 82–95. Retrieved from:

http://ngm.nationalgeographic.com/2012/01/sled-dogs/finkel-text

Finomore, V., Matthews, G., Shaw, T., & Warm, J. (2009). Predicting Vigilance: A fresh look at an old problem.

Ergonomics, 52(7), 791-808. doi:10.1080/00140130802641627

317

Fong, T., Thorpe, C., & Baur, C. (2001). Collaboration, dialogue, and human-robot interaction. Springer Tracts in

Advanced Robotics, 6, 255-266. doi:10.1007/3-540-36460-9_17

Foster, M. E., Gaschler, A., Giuliani, M., Isard, A., Pateraki, M., Petrick, R. P. A. (2012). Two People walk into a

Bar: Dynamic multi-party social interaction with a robot agent. Proceedings of the 14th Annual ACM

International Conference on Multimodal Interaction (ICMI), Santa Monica, California, USA (pp. 3-10).

Retrieved from: http://www.macs.hw.ac.uk/~mef3/papers/foster-etal-icmi2012.pdf

Freedy, A., DeVisser, E., Weltman, G., & Coeyman, N. (2007). Measurement of trust in human-robot

collaborations. Proceedings of Collaborative Technologies and Systems, Orlando, FL (pp.106-114).

doi:10.1109/CTS.2007.4621745

Gambetta, D. (1988). Can we trust trust? In D. Gambetta (Ed.), Trust, Making and Breaking Cooperative

Relationships (pp. 213-237). Oxford England: Basil Blackwell.

Gao, J., Lee, J. D., & Zhang, Y. (2006). A dynamic model of interaction between reliance on automation and

cooperation in multi-operator multi-automation situations. International Journal of Industrial Ergonomics,

36(5), 511-526. doi:10.1016/j.ergon.2006.01.013

Goetz, J., Kiesler, S., & Powers, A. (2003). Matching robot appearance and behavior to tasks to improve human-

robot cooperation. Proceedings of the 12th Annual IEEE International Workshop on Robot and Human

Interactive Communication (pp.55-60). doi:10.1109/ROMAN.2003.1251796

Goldberg, L. R. (1999). A broad-bandwidth, public-domain, personality inventory measuring the lower-level facets

of several five-factor models. In I. Mervielde, I. J. Deary, F. De Fruyt, and F. Ostendorf (Eds.), Personality

Psychology in Europe (Vol. 7, pp. 7-28). Tilburg, The Netherlands: Tilburg University Press.

Gonzalez, J.P., Dodson, W., Dean, R., Kreafle, G., Lacaze, A., Sapronov, L., & Childers, M. (2009). Using RIVET

for Parametric Analysis of Robotic Systems. Proceedings of 2009 Ground Vehicle Systems Engineering

and technology Symposium (GVSETS).

Goodrich, M. A., & Schultz, A. C. (2007). Human-robot interaction: A survey. Foundations and Trends in Human-

Computer Interaction, 1(3), 203-275. doi:10.1561/1100000005

Groom, V. (2008). What’s the best role for robot? Cybernetic models of existing and proposed human-robot

interaction structures. Proceedings of the 5th International Conference on Informatics in Control,

318

Automation, and Robotics, Funchal, Madeira, Portugal, (pp.323-328). Retrieved from:

http://www.academia.edu/268803/Whats_the_Best_Role_for_a_Robot_Cybernetic_Models_of_Existing_a

nd_Proposed_Human-Robot_Interaction_Structures

Groom, V., & Nass, C. (2007). Can robots be teammates? Benchmarks in human-robot teams. Interaction Studies,

8(3), 483-500. doi:10.1075/is.8.3.10gro

Guiochet, J. & Vilchis, A. (2002). Safety Analysis of a Medical Robot for Tele-echography. Proceedings of the 2nd

IARP IEEE/RAS Joint Workshop on Technical Challenge for Dependable Robots in Human Environments,

(pp. 217-227).

Gupta, N., Bisantz, A. M., & Singh, T. (2002). The effects of adverse condition warning system characteristics on

driver performance: An investigation of alarm signal type on threshold level. Behavior & Information

Technology, 21(4), 235-248. doi:10.1080/0144929021000013473

Hancock, P. A., Billings, D. R., & Schaefer, K. E. (2011). Can you trust your robot? Ergonomics in Design, 19(3),

24-29. doi: 10.1177/1064804611415045

Hancock, P. A., Billings, D. R., Schaefer, K. E., Chen, J. Y. C., Parasuraman, R., & de Visser, E. (2011a). A Meta-

Analysis of Factors Affecting Trust in Human-Robot Interaction. Human Factors 53(5), 517-527. doi:

10.1177/0018720811417254

Hancock, P. A., Billings, D. R., Schaefer, K. E., Chen, J. Y. C., Parasuraman, R., & de Visser, E. (2011b). A Meta-

Analysis of Factors Influencing the Development of Human-Robot Trust (Report No. ARL-TR-5857).

Aberdeen Proving Ground, Maryland: Army Research Laboratory.

Hancock, P. A., Billings, D. R., Schaefer, K. E., & Hancock, G. (2013 submitted). Developing Trust in People: A

Meta-Analysis of Trust Antecedents in Interpersonal Interactions (Report No. TBA). Aberdeen Proving

Ground, Maryland: Army Research Laboratory.

Hancock, P.A., & Warm (1989). A dynamic model of stress and sustained attention. Human Factors, 31(5), 519-

537. doi: 10.1177/001872088903100503

Hart, S. G., & Staveland, L. E. (1988). Development of NASA-TLX (Task Load Index): Results of empirical and

theoretical research. In P. A. Hancock and N. Meshkati (Eds.). Human Mental Workload. Amsterdam:

North Holland Press.

319

Heerink, M., Kröse, B., Evers, V., & Wielinga, B. (2010). Assessing Acceptance of Assistive Social Agent

Technology by Older Adults: The Almere Model. International Journal of Social Robotics, 2, 361-375.

doi:10.1007/s12369-010-0068-5

Helton, W. S. (2009). Canine ergonomics: The science of working dogs. Boca Raton, FL: CRC Press.

Ho, C-C., & MacDorman, K. F. (2010). Revisiting the uncanny valley theory: Developing and validating an

alternative to the Godspeed indices. Computers in Human Behavior, 26, 1508-1518.

doi:10.1016/j.chb.2010.05.015

Ho, G., Wheatley, D., & Scialfa, C. (2005). Age differences in trust and reliance of a medication management

system. Interacting with Computers, 17(6), 690–710. doi:10.1016/j.intcom.2005.09.007

Hoffman, R. R., Lee, J. D., Woods, D. D., Shadbolt, N., Miller, J., & Bradshaw, J. M. (2009). The Dynamics of

Trust in Cyberdomains. In R. R. Hoffman, J. M. Bradshaw, & K. M. Ford (Eds.), Human-Centered

Computing (pp.5-11). IEEE Computer Society.

Igbaria, M., & Iivari, J. (1995). The effects of self-efficacy on computer usage. Omega, 23(6), 587-605.

doi:10.1016/0305-0483(95)00035-6

Jamieson, G. A., Wang, L., & Neyedli, H. F. (2007). Developing human-machine interfaces to support appropriate

trust and reliance on automated combat identification systems (Report No. CR 2008-114). Toronto,

Canada: Department of National Defense.

Jarvenpaa, S., Knoll, K. and Leidner, D. (1998). Is anybody out there? Antecedents of trust in global virtual teams.

Journal of Management Information Systems, 14(4), 29-64. Retrieved from:

http://www.jstor.org/stable/40398291

Jian, J-Y., Bisantz, A. M., Drury, C. G., & Llinas, J. (1998). Foundations for an Empirically Determined Scale of

Trust in Automated Systems (Report No. AFRL-HE-WP-TR-2000-0102). Wright-Patterson AFB, OH. Air

Force Research Laboratory.

Kidd, C. D. (2003). Sociable Robots: The Role of Presence and Task in Human-Robot Interaction (unpublished

master’s thesis). Massachusetts Institute of Technology, Boston, Massachusetts.

Kidd, C. D., & Breazeal, C. (2004). Effect of a robot on user perceptions. Proceedings of 2004 IEEE/RSJ

International Conference on Intelligent Robots and Systems (pp. 3559-3564). Sendai, Japan.

320

Kiercher, K., & Thorslund, B. (2009). Effects of road surface appearance and low friction warning systems on driver

behaviour and confidence the warning system. Ergonomics, 52(2), 165-176.

doi:10.1080/00140130802277547

Kiesler, S., Powers, A., Fussell, S.R., & Torrey, C. (2008). Anthropomorphic Interactions with a Robot and Robot-

Like Agents. Social Cognition, 26(2), 169-181. doi: 10.1521/soco.2008.26.2.169

Kim, J. and Moon, J. Y. (1998). Designing towards emotional usability in customer interfaces-Trustworthiness of

cyber-banking system interfaces. Interaction with Computers, 10, 1-29. doi:10.1016/S0953-

5438(97)00037-4

Koch, M. W. (Producer), & Hopkins, S. (Director). (1998). Lost in space [Motion picture]. United States: Jim

Henson's Creature Shop.

Kulić, D., & Croft, E. A. (2007). Affective state estimation for human-robot interaction. IEEE Transactions on

Robotics, 23(5), 991-1000. Retrieved from:

http://www.sites.mech.ubc.ca/~caris/Publications/Affective%20State%20Estimation%20for%20Human-

Robot%20Interaction.pdf

Kurfess, T. R. (Ed.). (2005). Robotics and automation handbook. New York, NY: CRC Press.

Kwak, S., & Kim, M-S. (2005). User preferences for personalities of entertainment robots according to the users’

psychological traits. Bulletin of Japanese Society for the Science of Design, 172, 47-52. Retrieved from:

http://sciencelinks.jp/j-east/article/200601/000020060105A1054426.php

Kurtz, G. (Producer) & Lucas, G. (Director). (1977). Star wars [Motion picture]. United States: 20th Century Fox.

Larzelere, R. E, & Huston, T. L. (1980). The Dyadic Trust Scale: Toward understanding interpersonal trust in close

relationships. Journal of Marriage and the Family, 42, 595-604. doi:10.2307/351903

Lawshe, C. H. (1975). A quantitative approach to content validity. Personnel Psychology, 24(4), 563-575.

doi: 10.1111/j.1744-6570.1975.tb01393.x

LeBlanc, D. J-C. (2007). Increasing operator trust through knowledge. (Unpublished master’s thesis). Royal

Military College of Canada, Ottawa, Canada.

Lee, J. D. (2008). Review of a Pivotal Human Factors Article: “Humans and Automation: Use, Misuse, Disuse,

Abuse.” Human Factors 50(3), 404-410. doi: 10.1518/001872008X288547

321

Lee, H., Kim, J., & Kim, J. (2007). Determinants of success for applications service provider: An empirical test in

small businesses. International Journal of Human-Computer Studies, 65(9), 796-815.

doi: 10.1016/j.ijhcs.2007.04.004

Lee, K. M., Park, N., & Song, H. (2005). Can a robot be perceived as a developing creature? Effects of a robot’s

long-term cognitive developments on its social presence and people’s social responses toward it. Human

Communication Research, 31(4), 538-563. doi:10.1111/j.1468-2958.2005.tb00882.x

Lee, J. D., & See, K. A. (2004). Trust in automation: Designing for appropriate reliance. Human Factors: The

Journal of the Human Factors and Ergonomics Society, 46 (1), 50-80. doi:10.1518/hfes.46.1.50_30392

Lewandowsky, S., Mundy, M., & Tan, G. P. A. (2000). The dynamics of trust: Comparing humans to automation.

Journal of Experimental Psychology: Applied, 6, 104-123.

Lewicki, R. J., McAllister, D. J., & Bies, R. J. (1998). Trust and distrust: New relationships and realities. The

Academy of Management Review, 23(3), 438-458. doi:10.5465/AMR.1998.926620

Li, D., Rau, P., & Li, Y. (2010). A cross-cultural study: Effect of robot appearance and task. International Journal of

Social Robots, 2, 175-186. doi:10.1007/s12369-010-0056-9

Li, Y-M., & Yeh, Y-S. (2010). Increasing trust in mobile commerce through design aesthetics. Computers in Human

Behavior, 26, 673-684. doi:10.1016/j.chb.2010.01.004

Libin, A. V., & Libin, E. V. (2004). Person-robot interactions from the robopsychologist point of view: The robotic

psychology and robotherapy approach. Proceedings of the IEEE, 92(11), 1789-1803.

doi:10.1109/JPROC.2004.835366

Lin, P., Bekey, G., & Abney, K. (2008). Autonomous military robotics: Risk, ethics, and design. Philosophy, 1-112.

Retrieved from http://ethics.calpoly.edu/ONR_report.pdf

Llinas, J., Bisantz, A., Drury, C., Seong, Y., & Jian, J-Y. (1998). Studies and Analyses of Aided Adversarial

Decision Making Phase 2: Research on Human Trust in Automation (Report No. AFRL-HE-WP-TR-1999-

0216). Wright-Patterson AFB, OH: Air Force Research Laboratory.

Looije, R., Neerincx, M. A., & Cnossen, F. (2010). Persuasive robotic assistant for health self-management of older

adults: Design and evaluation of social behaviors. International Journal of Human-Computer Studies,

68(6), 386-397. doi:10.1016/j.ijhcs.2009.08.007

322

Luhmann, N. (1979). Trust and Power. Chichester: Wiley.

Lum, H., Sinatra, A., Sims, V.K., Chin, M.G., Smith, H.S., Shumaker, R., & Finkelstein, N. (2007). Size Does

Matter: Automobile “Facial” Features Predict Consumer Attitudes. Proceedings of the Human Factors and

Ergonomics Society, 51(18), 1105-1108. doi: 10.1177/154193120705101811

Lussier, B., Gallien, M., & Guiochet, J. (2007). Fault tolerant planning for critical robots. Proceedings of the 37th

Annual IEEE/IFIP International Conference on Dependable Systems and Networks (pp.144-153).

doi:10.1109/DSN.2007.50

Lyons, J., Stokes, C., Garcia, D., Adams, J., & Ames, D. (2009). Trust and decision-making: An empirical platform

(CCP 204). Aerospace and Electronics Systems Magazine, 24(10) 36-41.

doi:10.1109/MAES.2009.5317785

Madhavan, P., & Wiegmann, D. A. (2007). Similarities and differences between human–human and human–

automation trust: An integrative review. Theoretical Issues in Ergonomics Science, 8(4), 277-301.

doi:10.1080/14639220500337708

Madsen, M., & Gregor, S. (2000). Measuring Human-Computer Trust. Proceedings of the 11th Australian

Conference on Information Systems, Gladston, Australia. doi:10.1.1.93.3874

Mahoney, R. M. (1997). Robotic products for rehabilitation: Status and strategy. Proceedings of ICORR ’97:

International Conference on Rehabilitation Robotics, Bath, UK (pp.12-17).

Marsh, G., & Spain. T. (1984). Robots in the Classroom. Electronic Learning, 3(6), 48-53.

Mateas, M. (1999). An Oz-centric review of interactive drama and believable agents. In M. J. Wooldridge & M.

Veloso (Eds.), Artificial Intelligence Today: Recent Trends and Developments (pp.297-398). New York:

Springer.

Matthews, G., Joyner, L., Gilliland, K., Campbell, S.E., Falconer, S., & Huggins, J. (1999). Validation of a

comprehensive stress state questionnaire: Towards a state “Big Three.” In I. Mervielde, I.J. Dreary, F.

DeFruyt, & F. Ostendorf (Eds.), Personality Psychology in Europe (Vol. 7, pp. 335-350). Tilburg, the

Netherlands: Tilburg University Press.

323

Matthews, G., Campbell, S.E., Falconer, S. Joyner, L., Huggins, J., Gilliland, K., et al. (2002). Fundamental

dimensions of subjective state in performance settings: Task engagement, distress and worry. Emotion, 2,

315-340.

Mayer, R. C., Davis, J. H., & Schoorman, F. D. (1995). An integrative model of organizational trust, Academy of

Management Review, 20(3), 709-734. doi:10.2307/258792

McAllister, D. J. (1995). Affect- and cognition-based trust as foundations for interpersonal cooperation in

organizations. The Academy of Management Journal, 38(1), 24-59.

McBride, M. & Morgan, S. (2010). Trust Calibration for Automated Decision Aids. Institute for Homeland Security

Solutions Research Brief, (pp.1-11).

McKnight, D. H., Kacmar, C. J., & Choudhury, V. (2004). Dispositional trust and distrust distinctions in predicting

high- and low-risk internet expert advice site perceptions. e-Service Journal, 3(2), 35-58. Retrieved from:

http://www.jstor.org/stable/10.2979/ESJ.2004.3.2.35

Merritt, S. M., Heimbaugh, H., LaChapell, J., & Lee, D. (2012). I trust it, but I don’t know why: Effects of implicit

attitudes toward automation on trust in an automated system. Human Factors: The Journal of the Human

Factors and Ergonomics Society. doi: 10.1177/0018720812465081.

Merritt, S. M., & Ilgen, D. R. (2008). Not all trust is created equal: Dispositional and history-based trust in human-

automation interactions. Human Factors: The Journal of the Human Factors and Ergonomics Society,

50(2), 194-210. doi:10.1518/001872008X288574

Miller, J. E., Batchelor, C., & Perkins, L. (2008). Eliciting expectations to develop trust in systems (Report No.

AFRL-RH-WP-TP-2010-0002). Air Force Research Laboratory: Wright-Patterson AFB, Ohio.

Mohan, R. E., Calderon, C. A., Zhou, C., & Yue, P. K. (2008). Evaluating virtual emotional expression systems for

human robot interaction in rehabilitation domain. Proceedings from CW ‘08: International Conference on

Cyberworlds, Hangzhou, China. doi:10.1109/CW.2008.124

Monahan, J. L. (1998). I don’t know it but I like you – The influence of non-conscious affect on person perception.

Human Communication Research, 24(4), 480-500. doi:10.1111/j.1468-2958.1998.tb00428.x

324

Moray, N., & Inagaki, T. (1999). Laboratory studies of trust between humans and machines in automated systems.

Transactions of the Institute of Measurement and Control 21(4/5), 203-211.

doi:10.1177/014233129902100408

Moray, N., Inagaki, T., & Itoh, M. (2000). Adaptive automation, trust, and self-confidence in fault management of

time-critical tasks. Journal of Experimental Psychology: Applied, 6(1), 44-58. doi:10.1037/1076-

898X.6.1.44

Muir, B.M. (1987). Trust between humans and machines, and the design of decision aids. International Journal of

Man-Machine Studies. 27, (5/6), 527-539. doi:10.1016/S0020-7373(87)80013-5

Muir, B. M. (1994). Trust in automation. Part I. Theoretical issues in the study of trust and human intervention in

automated systems. Ergonomics, 37, 1905-1922.

Muir, B. M., & Moray, N. (1996). Trust in automation. Part II. Experimental studies of trust and human intervention

in a process control simulation. Ergonomics, 39(3), 429-460. doi:10.1080/00140139608964474

Mutlu, B., Yamaoka, F., Kanda, T., Ishiguro, H., & Hagita, N. (2009). Nonverbal Leakage in Robots:

Communication of Intentions Through Seemingly Unintentional Behavior. Proceedings of the 4th

ACM/IEEE International Conference on Human Robot Interaction, La Jolla, CA (pp. 69-76).

Nakada, K. (1997). Kansei engineering research on the design of construction machinery. International Journal of

Industrial Ergonomics, 19(2), 129-146. doi:10.1016/S0169-8141(96)00009-1

Nass, C., & Moon, Y. (2000). Machines and Mindlessness: Social responses to computers. Journal of Social Issues,

56(1), 81-103.

Nayfack, N. (Producer), & Wilcox, F. M. (Director). (1956). Forbidden planet [Motion picture]. United States:

Metro-Goldwyn-Mayer.

Neubauer, C., Matthews, G., Langheim, L., & Saxby, D. (2012). Fatigue and voluntary utilization of automation in

simulated driving. Human Factors: The Journal of Human Factors and Ergonomics, 54(5), 734-746.

doi:10.1177/0018720811423261

Nickerson, J. V., & Reilly, R. R. (2004, January). A model for investigating the effects of machine autonomy on

human behavior. Proceedings of the 37th Hawaii International Conference on System Sciences, Waikoloa,

Hawaii. doi:10.1109/HICSS.2004.1265325

325

Nomura, T., Kanda, T., Suzuki, T., & Kato K. (2004). Psychology in Human-Robot Communication: An Attempt

through Investigation of Negative Attitudes and Anxiety toward Robots. Proceedings of the 2004 IEEE

International Workshop on Robot and Human Interactive Communication, Kurashiki, Okayama Japan

(pp.35-40). doi:10.1109/ROMAN.2004.1374726

Ogreten, S., Lackey, S., & Nicholson, D. (2010). Recommended roles for uninhabited team members within mixed-

initiative combat teams. The 2010 International Symposium on Collaborative Technology Systems, Chicago, IL

(pp.531-536). doi:10.1109/CTS.2010.5478468

Olson, W. A. (1999). Supporting Coordination in Widely Distribute Cognitive Systems: The role of conflict type,

Time pressure, display design, and trust. (Unpublished doctoral dissertation). University of Illinois at Urbana-

Champaign, Urbana, IL.

Parasuraman, R., Cosenzo, K. A., & Visser, E. de. (2009). Adaptive automation for human supervision of multiple

uninhabited vehicles: Effects on change detection, situation awareness, and mental workload. Military

Psychology, 21(2), 270–297. doi:10.1080/08995600902768800

Parasuraman, R., & Miller, C. A. (2004). Trust and etiquette in high-criticality automated systems. Communications

of the Association for Computing Machinery, 47(4), 51-55. doi:10.1145/975817.975844

Parasuraman, R., & Riley, V. (1997). Humans and automation: Use, misuse, disuse, abuse. Human Factors: The

Journal of the Human Factors and Ergonomics Society, 39(2), 230-253. doi:10.1518/00187209777854886

Parise, S., Kiesler, S., Sproull, L., & Waters, K. (1999). Cooperating with life-like interface agents. Computers in

Human Behavior, 15, 123-142. doi:10.1016/S0747-5632(98)00035-1

Park, E., Jenkins, Q., & Jiang, X. (2008). Measuring trust of human operators in new generation rescue robots.

Proceedings of the 7th JFPS International Symposium on Fluid Power, Toyoma, Japan (pp.489-492).

Retrieved from: http://www.jfps.jp/proceedings/toyama2008/pdf/p1-44.pdf

Patrick, A. (2002). Privacy, Trust, Agents, and Users: A Review of Human Factors Issues Associated with Building

Trustworthy Software Agents. Human Factors of Trustworthy Agents, 1(6), 1-12.

Phillips, E., Ososky, S., Swigert, B., & Jentsch, F. (2012). Human-animal teams as an analog for future human-robot

teams. Proceedings of the Human Factors and Ergonomics Society, 56(1), 1553-1557.

doi: 10.1177/1071181312561309

326

Plato. (2008). The Republic, Book VII. (B. Jowett, Trans.) Retrieved from:

http://classicsmitedu/Plato/rebulic8vii.html. (Original work published in 360 B.C.E.)

Powers, A., & Kiesler, S. (2006). The advisor robot: Tracing people’s mental model from a robot’s physical

attributes. Proceedings of the 1st ACM SIGCHI/SIGART conference on Human-Robot Interaction, Salt Lake

City, Utah, USA (pp.218-225). doi:10.1145/1121241.1121280

Powers, A., Kiesler, S., Fussell, S., & Torrey, C. (2007). Comparing a Computer Agent with a Humanoid Robot.

Proceedings of the International Conference on Human Robot Interaction, Arlington, VA (pp. 145-152).

Rau, P. L., Li, Y., & Li, D. (2009). Effects of communication style and culture on ability to accept recommendations

from robots. Computers in Human Behavior, 25(2), 587-595. doi:10.1016/j.chb.2008.12.025

Reeves, B., & Nass, C. (1996). The Media Equation: How people treat computers, television, and new media like

real people and places. New York: Cambridge University Press.

Reichenbach, J., Onnasch, L., & Manzey, D. (2011). Human performance consequences of automated decision aids

in states of sleep loss. Human Factors, 53(6), 717-728. doi:10.1177/0018720811418222

Reising, D. V., & Sanderson, P. M. (2002). Work domain analysis and sensors I: Principles and simple example.

International Journal of Human-Computer Studies, 56(6), 569-596. doi:10.1006/ijhc.2002.1006

Rempel, J. K., & Holmes, J. G. (1986). How do I trust thee? Psychology Today, 28-34.

Rempel, J. K., Holmes, J. G., & Zanna, M. P. (1985). Trust in close relationships. Journal of Personality and Social

Psychology, 49(1), 95–112. doi:10.1037/0022-3514.49.1.95

Robot Institute of America (1979). Robot [Def.]

Roddenberry, G. (Writer). (1987). Star Trek: The Next Generation [Television Series]. United States: Paramount

Television.

Roddenberry, R. (Writer) & Scheerer, R. (Director). (1990). Legacy [Television series episode]. In R. Berman

(Producer), Star Trek: The Next Generation. United States: Paramount Television.

Ross, J. M. (2008). Moderators of trust and reliance across multiple decision aids (Unpublished doctoral

dissertation). University of Central Florida, Orlando, Florida.

327

Ross, J. M., Hancock, P. A., & Szalma, J. L. (2007). Empirical examination of trust in automation across multiple

agents in a search and rescue operation. Proceedings of the Human Factors and Ergonomics Society,

51(22), 1501-1505. doi: 10.1177/154193120705102207

Rotter, J. B. (1967). A new scale for the measurement of interpersonal trust. Journal of Personality, 35(4), 651-665.

doi:10.1111/j.1467-6494.1967.tb01454

Rousseau, D., & Hayes-Roth, B. (1997). A Social-Psychological Model for Synthetic Actors Stanford Knowledge

Systems Laboratory (Report No. KSL-97-07). Knowledge Systems Laboratory: Stanford University.

Sanders, T. L., Oleson, K. E., Billings, D. R., Chen, J.Y.C., & Hancock, P. A. (2011). A Model of Human-Robot

Trust: Theoretical Model Development. Proceedings of the Human Factors and Ergonomics Society, Las

Vegas, NV, (pp. 1380-1384). doi: 10.1177/1071181311551298

Scanzoni, J. (1979). Social exchange and behavioral interdependence. In R. L. Burgess, and T. L. Huston (Eds.),

Social exchange in developing relationships (pp. 61-98), New York: Academic Press.

Schaefer, K. E, Adams, J. K., Cook, J. G., Bardwell-Owens, A., & Hancock, P. A. (In Press). The Future of Robotic

Design: Trends from the history of media representations. Ergonomics in Design.

Schaefer, K. E., Billings, D. R., & Hancock, P. A. (2012). Robots vs. Machines: Identifying User Perceptions and

Classifications. Proceedings of the 2ndAnnual IEEE Cognitive Methods in Situation Awareness and

Decision Support (CogSIMA), New Orleans, LA (pp.138-141). doi:10.1109/CogSIMA.2012.6188366

Schaefer, K. E., Billings, D. R., Kocsis, V., Cook, J. Barrera, M., Ferrer, M., & Hancock, P. A. (2012). Trust in

Human-Animal Interactions: A Review of Current Evidence. Poster presented at the American

Psychological Association, Division 21, Orlando, FL.

Schaefer, K. E., Billings, D. R., Szalma, J. L., Sanders, T. L., Adams, J. K., Chen, J. Y. C., & Hancock, P. A.

(2013). A Meta-Analysis of Factors Influencing the Development of Trust in Automation: Implications for

Human-Robot Interaction (Report No. Manuscript Under Review). Aberdeen Proving Ground, Maryland:

Army Research Laboratory.

Schaefer, K. E., Cook, J. G., Adams, J. K., Bell, J., Sanders, T. L., & Hancock, P. A. (2012). Augmented Emotion

and its Remote Embodiment: The Importance of Design from Fiction to Reality. Proceedings of Human

Factors and Ergonomics Society, 56(1), 1817-1821. doi: 10.1177/1071181312561366

328

Scholtz, J. (2003). Theory and evaluation of human robot interactions. Proceedings from the 36th Annual Hawaii

International Conference on System Sciences. doi:10.1109/HICSS.2003.1174284

Scopelliti, M., Giuliani, M. V., & Fornara, F. (2005). Robots in a domestic setting: A psychological approach.

Universal Access in the Information Society, 4(2), 146-155. doi:10.1007/s10209-005-0118-1

Seong, Y., & Bisantz, A.M. (2008). Modeling human trust in complex, automated systems using a lens model

approach. In M. W. Scerbo & M. Mouloua (Eds.), Automation Technology and Human Performance:

Current Research and Trends (pp. 282-286). Mahwah, NJ: Lawrence Erlbaum Associates. Retrieved from:

http://www.dtic.mil/cgi-bin/GetTRDoc?AD=ADA354935#page=295

Sheppard, B. H., & Sherman, D. M. (1998). The grammars of trust: A model and general implications (Special topic

forum on trust in and between organizations). Academy of Management Review, 23(3), 422-437.

doi:10.5465/AMR.1998.926619

Sheridan, T. B. (1975). Considerations in modeling the human supervisory controller. Proceedings of the IFAC 6th

World Congress, Laxenburg, Austria (pp. 1-6).

Sheridan, T. B. (1988). Trustworthiness of command and control systems. IFAC Man-Machine Systems, 427-431.

Sheridan, T. B. (2002). Humans and automation: System design and research issues. New York: Wiley/Human

Factors and Ergonomics Society.

Sims, V. K., Chin, M. G., Sushil, D. J., Barber, D. J., Ballion, T., Clark, B. R., Dolezal, M. J., Shumaker, R., &

Finkelstein, N. (2005). Anthropomorphism of robotic forms: a response to affordances? Proceedings of the

Human Factors and Ergonomics Society Conference, 49(3), 602-605. doi: 10.1177/154193120504900383

Stack, L. C. (1978). Trust. In H. London & J. E. Exner, Jr. (Eds.), Dimensionality of Personality (pp. 561–599). New

York: Wiley.

Steinfeld, A. Fong, T., Kaber, D., Lewis, M., Scholtz, J., Schultz, A., & Goodrich, M. (2006). Common metrics for

human-robot interaction. Proceedings of the 1st ACM/IEEE International Conference on Human Robot

Interaction, Salt Lake City, Utah (pp. 33-40). doi:10.1145/1121241.1121249

Stormont, D. P. (2008). Analyzing human trust of autonomous systems in hazardous environments. Proceedings of

the Human Implications of Human-Robot Interaction workshop at AAAI. Retrieved from:

http://www.aaai.org/Papers/Workshops/2008/WS-08-05/WS08-05-006.pdf

329

Szalma, J. L., & Taylor, G. S. (2011). Individual differences in response to automation: The five factor model of

personality. Journal of Experimental Psychology: Applied, 17(2), 71-96. doi:10.1037/a0024170

Tenney, Y. J., Rogers, W. H., & Pew, R. W. (1998). Pilot opinions on cockpit automation issues. The International

Journal of Aviation Psychology, 8(2), 103-120. doi:10.1207/s15327108ijap0802_2

Thomer, J. L. (2007). Trust based design of human-guided algorithms. (Unpublished master’s thesis). Massachusetts

Institute of Technology: Boston, Massachusetts.

Thompson, L. F., & Gillian, D. J. (2010). Social factors in human-robot interaction. In M. Barnes & F. Jentsch

(Eds.), Human-Robot Interactions in Future Military Operations (pp. 67-81). Burlington, VT: Ashgate

Publishing, Ltd.

Thropp, J. E. (2006). Individual differences using automation. (Unpublished doctoral dissertation). University of

Central Florida, Orlando, FL.

Torta, E., van Heumen, J., Cuijpers, R. H., & Juola, J. F. (2012). How can a robot attract the attention of its human

partner? A comparative study over different modalities for attracting attention. Social Robotics, 7621, 288-

297. doi:10.1007/978-3-642-34103-8_29

Tsui, K.M., Desai, M., & Yanco, H.A. (2010a). Considering the bystander’s perspective for indirect human-robot

interaction. Proceedings of the 5th ACM/IEEE International Conference on Human Robot Interaction,

Osaka, Japan (pp.129-130). doi:10.1109/HRI.2010.5453230

Tsui, K. M., Desai, M., Yanco, H. A., Cramer, H., & Kemper, N. (2011b). Measuring Attitudes Towards

Telepresence Robots. International Journal of Intelligent Control and Systems, 16(2).

Upham Ellis, L., Sims, V. K., Chin, M. G., Pepe, A. A., Owens, C. W., Dolezal, M. J., Shumaker, R., & Finkelstein,

N. (2005). Those A-Maze-Ing Robots: Attributions of ability are based on form, not behavior. Proceedings

of the Human Factors and Ergonomics Society, 49(3), 598-601. doi: 10.1177/154193120504900382 van den Broek, E. L., & Westerink, J. H. (2009). Considerations for emotion-aware consumer products. Applied

Ergonomics, 40(6), 1055-1064. doi:10.1016/j.apergo.2009.04.012

Verberne, F. M. F., Ham, J., & Midden, C. J. H. (2012). Trust in Smart Systems: Sharing driving goals and giving

information to increase trustworthiness and acceptability of smart systems in cars. Human Factors, 54(5),

799-810. doi:10.1177/0018720812443825

330

Wagner, A. R. (2009). The role of trust and relationships in human-robot social interaction. (Unpublished doctoral

dissertation). Georgia Institute of Technology, Atlanta, Georgia.

Wang, L., Jamieson, G. A., & Hollands, J. G. (2009). Trust and reliance on an automated combat identification

system. Human Factors, 51(3), 281-291. doi:10.1177/0018720809338842

Wang, L., Jamieson, G. A., & Hollands, J. G. (2011). The effects of design features on users' trust in and reliance on

a combat identification system. Proceedings of the Human Factors and Ergonomics Society, 55(1), 375-

379. doi: 10.1177/1071181311551077

Wang, L., Rau, P. L., Evers, V., Robinson, B. K., & Hinds, P. (2010). When in Rome: The role of culture and

context in adherence to robot recommendations. Proceedings of the 5th ACM/IEEE International

Conference on Human Robot Interaction, Osaka, Japan, (pp. 359-366).

Ward, J. N. (2000). Automation of task processes: An example of intelligent transportation systems. Human Factors

and Ergonomics in Manufacturing and Service Industries, 10(4), 395-408. doi:10.1002/1520-6564

Warm, J. S., Parasuraman, R., & Matthews, G. (2008). Vigilance requires hard mental work and is stressful. Human

Factors: The Journal of Human Factors and Ergonomics Society, 50(3), 433-441.

doi:10.1518/001872008X312152

Warner, R. M., & Sugarman, D. B. (1996). Attributes of personality based on physical appearance, speech, and

handwriting. Journal of Personality and Social Psychology, 50, 792-799.

Wildman, J. L., Fiore, S. M., Burke, C. S., Salas, E., & Garven, S. (2011). Trust in swift starting action teams:

Critical considerations. In N. A. Stanton (Ed.), Trust in military teams (pp. 71-88). London, UK: Ashgate.

Wildman, J. L. (2011). Cultural Differences in forgiveness: Fatalism, trust violations, and trust repair efforts in

interpersonal collaboration. (Unpublished doctoral dissertation). University of Central Florida, Orlando,

Florida.

Yagoda, R.E. (2011). What! You want me to trust a robot? The development of a human robot interaction (HRI)

trust scale. (Unpublished master’s thesis). North Carolina State University, Raleigh, North Carolina.

Zhang, T., Kaber, D. B., Zhu, B., Swangnetr, M., Mosaly, P., & Hodge, L. (2010). Service robot feature design

effects on user perceptions and emotional responses. Intelligent Service Robotics, 3(2), 73-88.

doi:10.1007/s11370-010-0060-9

331

Zuboff, S. (1988). In the age of smart machines: The future of work technology and power. New York: Basic Books.

332