Decision-making with a Humanoid Robot Partner: Individual

Differences Impacting Trust

Joel Elson

, Luis Merino and Douglas Derrick

University of Nebraska at Omaha, 6002 Dodge St., Omaha, NE 68182, U.S.A.

Keywords: Trust, Personality, Humanoid Robot.

Abstract: Trust in human-machine teams, where humans partner with intelligent systems, is critical to effective

collaboration and work success. Research in prior studies of trust in human-robot partnerships, has largely

focused on three groups of trust antecedents: factors relating to the environment, the machine, and human

individual differences. There is a dearth of research in this later area, despite wide recognition that individual

differences play an important role in human behaviour and cognition. This paper draws on the psychological

theory of trait activation and examines the role of human personality in trust in the relationships between

humans and intelligent humanoid robots partnering to make critical decisions. We conducted an empirical

study that looked to explore the role of the Big-Five personality traits on trust. Results suggest that the

openness personality trait is a significant predictor of trust in a humanoid-robot partner, above and beyond

the individual difference propensity trust. Individuals scoring high on the openness personality trait may have

a greater trust in a humanoid robot partner than those with low scores in the openness personality dimension.

Future studies should look to better understand the trait activating factors related to Openness in human

machine trusting relationships.

1 INTRODUCTION

Advances in artificial intelligence promise a future of

computing that will transform humans and machines'

relationship, moving machines from tools to

collaborative partners. This future has been referred

to as the "cognitive computing" era and is

characterized by intelligent systems, a class of

systems that learn and interact naturally to perform

knowledge work (Spohrer, J., and G. Banavar, 2015).

These systems are designed to augment human

expertise, amplify human intelligence, enhance

productivity, and improve decision-making. These

systems can be embodied as humanoid robots as a

way of integrating them with human teammates.

Trust in these systems is essential to collaborate

effectively and fully realize the advantages of these

new machine partners. Research has found trust to be

a necessary ingredient for successful cooperation

(Jones, G. R., and J. M. George, 1998), important in

predicting human use and reliance on technology

(Dzindolet, M. T., et al., 2003) and crucial to

relationships in situations characterized by risk and

https://orcid.org/0000-0003-4227-1274

https://orcid.org/0000-0001-8589-5023

uncertainty (Fukuyama, F., 1995; Luhmann, N.,

1982). Despite recognizing the importance of trust,

there is an incomplete understanding of trust, which

adequately accounts for the relationship between the

multitude of factors that contribute to trust in an a

robot partner. Idiosyncratic patterns of trust has been

observed across trust research in humans, machines,

and other technology systems. This paper advances

the theories of trust and unifies prior work by

adopting existing approaches and theoretical

frameworks from psychology literature and applying

them to the Information Systems domain to better

understand human trust humanoid robots.

It is generally recognized that three primary

sources influence human trust in an intelligent

machine: characteristics of the system (the robot

being trusted), individual characteristics (the person

who is trusting), and factors relating to the situation

or environment where trust is being applied. In

information systems literature, significant effort has

been made to describe system characteristics and

situational factors that contribute to trust. Relatively

little attention has been devoted to understanding the

300

Elson, J., Merino, L. and Derrick, D.

Decision-making with a Humanoid Robot Partner: Individual Differences Impacting Trust.

DOI: 10.5220/0010722400003060

In Proceedings of the 5th International Conference on Computer-Human Interaction Research and Applications (CHIRA 2021), pages 300-307

ISBN: 978-989-758-538-8; ISSN: 2184-3244

role that personality traits and other individual

differences play, despite numerous trust researchers

recognizing their importance in trusting relationships

involving both humans and machines (Billings, D. R.,

et al., 2012; Mayer, R. C., et al., 1995; Mcknight, D.

H., et al., 2011). This paper focuses on how individual

differences and personality traits relate to trust in

humanoid-robot partners through the lens of Trait

Activation Theory. This theory's application to the

information systems domain provides a more complete

picture of the factors that come together to predict trust.

Trait Activation Theory posits that situational

cues uniquely act on an individual's personality to

elicit behavioural and psychological responses

characteristic of a personality type. In short,

environmental factors "activate" or amplify

personality trait expression. It follows that when

interacting with a humanoid robot, situational cues

activate characteristic personality responses,

ultimately influencing an individual's perceptions and

trust in the system. Trust in information systems has

been operationalized both as a behaviour response as

well as a quantifiable measure of a systems

performance, process, and purpose. We posit that

personality trait activation strongly impacts both

behavioural and perceptual measures of trust in an

intelligent system.

2 BACKGROUND

This section describes intelligent systems, previous

research on trust, individual differences and trait

activation theory.

2.1 Trust

Trust is a multi-dimensional construct that has proven

quite difficult to conceptualize and define (McKnight,

D. H., and N. L. Chervany, 2001). For this study we

adopt a definition of trust that has been proposed by

Madsen and Gregor (Madsen, M., and S. Gregor,

2000). They define trust as “the extent to which a user

is confident in, and willing to act on the basis of the

recommendations, actions, and the decisions of a

computer-based tool or decision aid.” In this

definition, the human user is the “trustor” (the

individual who is trusting) and the technology is the

“trustee” (the object of trust).

Numerous definitions of trust exemplify the many

different ways of conceptualizing the construct. In

effort to bring clarity to the area of trust research,

McKnight and Chervany (McKnight, D. H., and N. L.

Chervany, 2001) created a typology of trust by

reviewing sixty-five articles containing trust

definitions and organized these by both trust

reference (characteristics of the trustee) and by

conceptual type. They identified four referent

grouping groupings of the trustee characteristics:

benevolence, integrity, competence, and

predictability. They also identified seven conceptual

type categories that include trusting: attitude,

intention, belief, expectancy, behaviour, disposition,

and institutional/structural. McKnight and Chervany

then created an interdisciplinary model of conceptual

trust types that included: 1) trusting intentions, 2)

trust-related behaviour, 3) trusting beliefs, 4)

Institution-based trust and 5) disposition to trust. We

refer readers to the McKnight and Chervany paper

(McKnight, D. H., and N. L. Chervany, 2001) for

additional information on trust and its classifications.

In this work we focus on trusting beliefs.

Foundational work on trusting beliefs was

conducted by Mayers, Davis, & Schoorman, and

identified several elements which may at the heart of

human-to-human trust including: 1) ability, 2)

benevolence, and 3) integrity. Ability describes how

capable or skilled a trustee is in carrying out a task in

a domain specified by a trustor. Benevolence relates

to a trustee having goals or intentions that benefit or

align with a trustor. Finally, integrity relates to a

trustor and trustee sharing a similar set of values and

can be counted on to act in accordance with these

shared beliefs. Building upon prior trust research, and

recognizing the distinctions that exist between human

to human and human to machine trust, McKnight

(Mcknight, D. H., et al., 2011) identify three

components of trusting beliefs that roughly align with

those identified by Mayers, Davis & Schoorman:

functionality, helpfulness, and reliability. Their work

suggests that these elements of trust are evaluated

either consciously or sub-consciously by technology

users and help to form the trusting beliefs an

individual has toward a technology.

In addition to understanding that there are

different components underlying trusting beliefs, it is

also important to acknowledge the temporal aspects

of trust. McKnight et al. (Mcknight, D. H., et al.,

2011) describe trust with a specific technology as

existing along a continuum starting with initial trust

(formed with little to no experience with a

technology) and moving on to knowledge based trust

(formed over time and based on prior interaction with

a technology). In this study we focus specifically on

initial trusting beliefs.

Measuring trust has proved difficult and in some

cases controversial endeavor. Generally speaking,

there are two primary methods of measuring trust,

Decision-making with a Humanoid Robot Partner: Individual Differences Impacting Trust

301

behavioural measurement or self-report. In this study

we focus on the latter. Jian et al. (Jian, J.-Y., et al.,

2000) developed what is called the Empirically

Derived Trust Measure (ED). The scale assesses trust

and distrust factors using 12 items and is best used for

measuring initial trust in an information system. The

ED has been utilized in a number of studies to measure

trust and has been validated as reliable trust measure

(Spain, R. D., et al., 2008). We will revisit trust measu-

rement as it applies to our study in the methods section.

2.2 Individual Differences

Individual differences are the collection of traits,

features, and behaviour that uniquely comprise the

overall makeup of an individual. These differences

are important for studying trust in human machine

partnerships and include: propensity to trust (Rotter,

J. B., 1967) and personality traits such as openness,

agreeableness or extraversion (Elson, J. S., D.

Derrick, and G. Ligon, 2018). There is evidence to

support that humans will treat machines as teammates

(Groom, V., and C. Nass, 2007) and it also has been

shown that these core personality traits affect team

performance (Barrick, M. R., et al., 1998). Therefore,

it is important that individual personality

characteristics be considered when looking at

individual differences that could impact trust in

human machine partnerships.

In psychology literature, the “Big-Five”

personality traits have been studied as predictors of

human behaviour and include: openness to

experience, conscientiousness, extraversion,

agreeableness, and emotional stability (Gosling, S.

D., et al., 2003). Individual personality traits have

been shown to be very stable over extended periods

of time (McCrae, R. R., and O. P. John, 1992).

Openness is a personality trait associated with

intellectual curiosity coupled with a general

disposition toward new experiences and adventure

(Goldberg, L. R., 1992). Conscientiousness refers to

an individual’s concern for detail, meeting planned

goals, seeking achievement (Goldberg, L. R., 1992).

Extraversion is an individual’s preferences for social

interaction, stimulation, and desire to be with others

(Goldberg, L. R., 1992). Agreeableness is the

personality trait that indicates a person’s ability to

work well with others, exhibiting high degree of trust

and reserved temperament (Goldberg, L. R., 1992).

Emotional stability describes the personality trait

relating to the stability of an individual’s experience

of emotion (Goldberg, L. R., 1992). We will discuss

our method of measuring the Big Five personality

traits in the methods section.

2.3 Trait Activation Theory

In psychology literature, trait activation theory has

provided a framework to help understand why

personality traits manifest themselves in only certain

circumstances. Thus, one aim of the present effort is to

introduce information systems researchers to this

theoretical framework to understand the complex and

often contradictory findings in the pursuit of scholarly

work on human-computer interaction. Trait activation

theory states that "the behavioural expression of a trait

requires arousal of that trait by trait-relevant situational

cues" (Tett, R. P., and H. A. Guterman, 2000). Sources

of trait-relevant cues when interacting with intelligent

systems will come from both perceptions of system

characteristics or situational factors. These trait-

relevant cues also serve as factors to inform user trust.

In the early stages of system use, minimal

information will be available to inform trust. In such

situations, authentic individual differences such as

propensity to trust and the Big Five personality traits

may be activated by initial perceptions of the system

and early system interaction. While prior research has

shown relationships between individual differences

(propensity to trust, personality traits) and trust in

human relationships, very little research has been

conducted to understand how these individual

differences will relate to early trust when a human

collaborates with a novel intelligent system.

3 RESEARCH QUESTION

Prior trust research in the information systems

domain suggests that individual differences may play

a role in human trust in an intelligent system (Elson,

J. S., et al., 2018). Sparse research into embodied

intelligent systems makes it difficult to hypothesize

specific relationships between individual personality

types and trust in an intelligent system with a

humanoid appearance. Trait activation theory

suggests that when individuals are working in novel,

ambiguous situations an individual’s personality

traits will be expressed (Tett, R. P., and D. D. Burnett,

2003). This is because in the absence of trait-relevant

situational cues, individual behaviour defaults back to

activity associated with core personality traits. It is

therefore reasonable to expect personality traits to

play a role in trust in a novel partnership with an

embodied intelligent system. We therefore pose the

following research question:

RQ: What is the relationship between the Big Five

personality traits and trust in a humanoid robot.

Humanoid 2021 - Special Session on Interaction with Humanoid Robots

302

4 METHOD

4.1 Sample

Participants were graduate and undergraduate

students from a medium-sized Midwestern

university. A total of 101 (58 females, 41 males, and

two preferring not to identify) individuals were

included in the analysis. They were recruited from a

subject participant pool within the College of

Business Administration. Thirty-three individuals

were not included in the analysis because of

incomplete data. Participants ages ranged from 19 to

24 years with a mean age of 23 years, a median age

of 21 years, and mode of 21 years. Participation in

this study was on a voluntary basis, however

participation credit toward a course requirement was

given to those who took part in the study.

4.2 Apparatus

The experimental task in this study was the Desert

Survival Simulation, initially developed by Human

Synergistics. This task was chosen as it had been

previously utilized in numerous team studies and had

performance data for several populations. Also, the

specific survival situation involving the desert

environment was specifically chosen as it would be

an environment that was likely unfamiliar to

participants from our sample population. This

reduced the likelihood that individuals would possess

expertise related to the simulation. Furthermore, the

Desert Survival Simulation can be completed with

relatively low workload as decisions can be

considered one at a time and without time pressure.

This attribute of the task helped to minimize the

possibility that subjects would offload decision

choices to their partner as a strategy to cope with high

workload (Molloy, R., and R. Parasuraman, 1996).

Finally, this survival simulation presented a situation

with no clear-cut answer. To achieve the best score,

individuals must carefully consider every decision

they must make. The survival situation described a

scenario where people had been stranded with only a

small number of items that could be used to survive.

The simulation's goal was to identify which of these

items were the most essential and rank the items in

order of their importance for survival.

A custom web application was used to conduct the

survival task activity. The web application for the

survival activity consisted of four primary interface

screens that were accessed in sequential order: 1) an

introductory screen, 2) an individual decision-making

interface, 3) a collaborative interface, and 4) the final

decision-making interface.

In this study, the intelligent system partner was

the humanoid robot Pepper from SoftBank Robotics.

The robot was programmed to respond to the

participant questions about items from the survival

scenario. Participants were told that their partner

would develop solutions in real-time and would not

have access to the solutions developed by the survival

experts (in reality, the solutions presented as the

partner solutions were the optimal solution developed

by the survival experts).

4.3 Procedure

Countermeasures were taken to discourage

participants from completing the task without

appropriately considering their answers. Participants

were asked to provide written justification for why

they had ranked their items and asked to provide their

confidence for their ranking. Between steps,

participants did not receive feedback regarding their

performance or degree of success, so they did not

know how they performed until being debriefed at the

end of the study.

The experiment was conducted in a dedicated lab

space with environmental controls to alleviate noise,

light, and visual distractions. To avoid monomethod

bias, participants completed an individual

characteristics assessment prior to the

experimentation day. Participants returned to the lab

on a different day to complete the experiment

described in this study. Upon arrival on the second

day, participants first completed an IRB mandated

informed consent. Participants were made to believe

that they were helping to evaluate a web application

designed to walk users through a novel partner

decision making process. Participants were also told

that only individuals who achieved a passing score on

the simulation activities would be awarded

participation credit (in reality, all participants

received credit for their participation). Participants

then completed a study orientation and pre-survey in

a private room. In this orientation presurvey,

participants were shown an example of the web

ranking interface and allowed to perform a ranking of

items. The pre-survey included a question that asked

what would happen if participants did not achieve a

passing score on the survival simulation. This

question served as a manipulation check that ensured

all participants in the analysis were aware of the risk

associated with this experiment (the loss of

participation credit).

Decision-making with a Humanoid Robot Partner: Individual Differences Impacting Trust

303

Next, participants were directed to a second room,

where they were introduced and seated across from

their partner and given more information about the

first survival simulation activity. The participants

were told that the partner had access to a database of

various survival items, their usefulness in past

survival situations, and would use this database to

help generate a real-time solution.

Participants were reminded that they would be

scored on their rankings and that failure to achieve a

passing score (greater than 75% correct) would result

in a loss of credit for this study. Participants were then

automatically presented with the simulation

instructions and left to work with their partner to

achieve a solution.

4.4 Measures

The experiment utilized: measures of trust (before

interaction and after the simulation), system utilization,

perceived humanness of partner, perceived presence,

the Big Five personality traits, propensity to trust, and

propensity to anthropomorphize. In this study, we

considered the following measures:

The Big Five Personality traits were measured

using the Big Five Index (BFI), a psychometric

instrument that measures Extraversion, Agreeableness,

Openness to Experience, Conscientiousness, and

Neuroticism (John, O. P., et al., 1991; John, O. P., et

al., 2008). This questionnaire contains 44 items, each

with a 5-point Likert scale that ask the participant to

rate their agreement or disagreement with statements

about their personality. Each item allowed for

responses ranging from one to five, with one being

strongly agree and five being strongly disagree. An

example item for the measure of Extraversion was, "I

am someone who is talkative." Scale reliabilities for

each of the five personality measures resulted in

Cronbach's alpha scores of .87 for Extraversion, .71 for

agreeableness, .84 for conscientiousness, .79 for

neuroticism, and .76 for openness.

Propensity to trust was assessed using the

propensity to trust others measure developed by

Ashleigh et al. (Ashleigh, M. J., et al., 2012). This 9-

item measurement uses a 5-point Likert scale that

asks the participant to rate their agreement or

disagreement with statements about their attitudes

toward others. An example question item is: "Other

people are out to get as much as they can for

themselves." Scale reliability for the measure resulted

in a Cronbach's alpha score of .89.

Trust was assessed using a modified version of

the Empirically Derived (ED) scale developed by Jian

et al (Jian, J.-Y., et al., 2000). The 12-item instrument

conceptualizes trust as being comprised of two factors

(trust & distrust). The scale's trust factors include

confidence, security, integrity, dependability,

reliability, trust, and familiarity. The distrust factors

include deceptiveness, underhandedness,

suspiciousness, wariness, and harm. Original items

were worded about a "system." Items were reworded

to reference a generic "partner." Example question

items include: "I am wary of my partner" and "I am

confident in my partner." Scale reliability for the

measure resulted in a Cronbach's alpha scores of .83.

5 RESULTS

Descriptive statistics for the continuous variables trust,

propensity to trust, Conscientiousness, Neuroticism,

Extraversion, Agreeableness, Openness, analytic

cognition, affective cognition (see Table 1).

Table 1: Descriptive Statistics for Continuous Variables.

A correlation analysis was performed to identify

the individual differences that were significantly

correlated with trust. Next, a hierarchical regression

analysis was performed to test the relationship

between analytical cognitive processes, affective

cognitive processes, and trust, including key

individual differences as covariates.

Results of a correlation analysis showed that only

two individual differences variables under

consideration were significantly correlated with trust

in the humanoid robot partner: Openness (r = -.276, p

= .003) and propensity to trust (r =.168, p =.050). The

remaining individual difference variables were not

significantly correlated with trust: Extraversion (r = -

.043, p = .339), Agreeableness (r = .044, p = .335),

Conscientiousness (r = .018, p = .429), and

Table 2: Correlations Among Continuous Variables.

Variables N Mean Std. Deviation Variance Minimum Maximum

Trust 97 3.82 0.43 0.19 3.00 5.00

Propensity to Trust 97 3.87 1.05 1.11 1.11 6.44

Conscientiousness 97 3.74 0.61 0.37 1.33 5.00

Neuroticism 97 2.94 0.62 0.39 1.63 4.25

Extraversion 97 3.10 0.74 0.55 1.63 5.00

Agreeableness 97 3.72 0.51 0.26 2.44 4.67

Openness 97 3.50 0.53 0.28 2.10 4.80

Variables 1234567

1. Trust -

2. Propensity to Trust .17* -

3. Conscientiousness .02 .22* -

4. Neuroticism -.09 -.34** -.28** -

5. Extraversion -.04 .29** .11 -.30** -

6. Agreeableness 0.04 .39** .34** -.34** .08 -

7. Openness -.28** 0.15 .08 -.17* .32** .21* -

Humanoid 2021 - Special Session on Interaction with Humanoid Robots

304

Table 3: Hierarchical Multiple Regression of Trust on Propensity to Trust and Openness.

Neuroticism (r = -.091, p = .186). Correlations

among each of the variables are presented in Table 2.

Considering these results, only the significantly

correlated individual difference variables were

retained for the final regression.

The correlation analysis revealed that only two

individual differences, propensity to trust and

openness, were significantly correlated to trust.

Therefore, the other personality traits were not

included in the final regression analysis. We

performed a hierarchical multiple regression analysis

with trust on openness and propensity to trust. The

variable propensity to trust was entered into the first

block and openness into the second block. The results

are summarized in Table 3.

In the first block, propensity to trust was added,

the model was predicted trust on propensity to trust.

The regression of predicted trust on propensity to

trust was not significant, F(1, 95) = 2.76, p = .099, R

.028, indicating that propensity to trust was not a

significant predictor of trust. Variance in propensity

to trust accounted for 3% of the variance in trust.

Propensity to trust was not a significant predictor of

trust, β = .17, B = .07, t(95) = 1.67, p = .099, 95% CI

[-0.01, 0.15], indicating that greater propensity to

trust did not predict greater trust. For more

information, refer to Table 3.

In the second block, openness was added; the

model was predicted trust on propensity to trust and

openness. The multiple regression of predicted trust

on propensity to trust and openness was significant,

F(2,94) = 6.52, p < .05, R

.122, indicating that

together propensity to trust and openness were

significant predictors of trust. Variance in propensity

to trust and openness accounted for 12% of the

variance in trust. The increment in R

was significant

ΔR

= .09, ΔF = 10.00, p < .05. That is, the unique

contribution to the variance accounted for in trust by

openness was significant. The increment in the

multiple coefficients of determination indicates that

the variance in openness accounted for an additional

9% of the variance for trust above and beyond

propensity to trust. In this model, propensity to trust

was significant predictor of trust, β = .22, B = .09,

t(94) = 2.21, p = .030, 95% CI [0.01, 0.17], indicating

that greater propensity to trust predicted more trust

above and beyond Openness.

Openness was a significant predictor of trust, β =

-.31, B = -.25, t(94) = -3.16, p = .002, 95% CI [-0.41,

-0.10], indicating that greater Openness predict less

trust above and beyond propensity to trust. For more

information, refer to Table 3.

6 DISCUSSION

A key finding from this study was that under these

experimental conditions, individual personality traits

were found to be more predictive of trust in an

intelligent system than the individual difference

propensity to trust. This is a significant finding as

historical precedent (Rotter, J. B., 1967) and recent

meta-analysis of trust research in intelligent systems

(Schaefer, K. E., et al., 2016) has observed the later

(trust propensity) as the primary individual difference

considered.

Openness was found to be correlated with trust. In

the hierarchical regression, Openness remained a

significant predictor of trust above and beyond

propensity to trust. Greater Openness predicted less

trust, a finding that viewed through the lens of trait

activation theory may have related to the delayed

collaboration between human and intelligent system

serving as situationally relevant cue. In this example,

the delayed collaborative nature of the task may have

served as a trait releaser which facilitated behaviour

characteristic of individuals scoring high in

Openness. Individuals scoring high in Openness may

act in ways that relate to confirmation bias. For

example, a recent study showed that individuals

engaging in activity related to confirmation bias in

different online groups, shared a similar personality

profile which included scoring high in Openness

(Bessi, A., 2016). The confirmation bias relates to the

tendency to seek out information that confirms

existing beliefs (Nickerson, R. S., 1998). It was

observed that the beliefs of the intelligent system

varied greatly from those of most participants, as

Model bSEt β F

ΔF

ΔR

95% CI

1. Intercept 3.55 0.17 21.37** 2.78 0.03 [3.22, 3.88]

Propensity to Trust 0.07 0.04 1.67 0.17 [-0.01, 0.15]

2. Intercept 4.36 0.30 14.51** 6.52* 0.12 10.00 0.09 [3.76, 4.96]

Propensity to Trust 0.09 0.04 2.21* 0.22 [0.01, 0.17]

Openness -0.25 0.08 -3.16* -0.31 [-0.41, -0.09]

Decision-making with a Humanoid Robot Partner: Individual Differences Impacting Trust

305

evidenced in poor individual scores. While additional

analysis is needed, it is possible that individuals

scoring high in Openness (characteristically seeking

decision confirming information) rejected partner

suggestions as being erroneous, leading to decreased

trust.

The following example shows how system design

could apply to these findings. An embodied

intelligent system used by individuals who score high

on the Openness personality trait may want to make

recommendations from the onset of a decision-

making task to avoid independent solution generation

which could lead to situations where confirmation

bias may come into play.

For systems that have or are already being

deployed, these results are also practical and can

inform management and training decisions. For

example, individuals scoring high in openness can be

identified and taught to realize the importance of

considering system information when making

decisions and encouraged to critically evaluate their

original decisions.

7 LIMITATIONS AND FUTURE

RESEARCH

This research focused on exploring the relationship

between individual characteristics and early trust in

intelligent systems. Like all empirical work, there

exist several limitations that need to be addressed.

The use of controlled laboratory experiments is

widely recognized as a limitation of information

systems research. Results from lab studies may not

generalize to individuals and systems in the real

world. Future studies should be conducted that look

to test for the relationships found in this study

explicitly.

Future studies will need to be conducted to look

at the relationships and trait-relevant cues related to

the openness personality dimension. Experiments

need to be conducted to target the activation of

specific personality traits by manipulating trust

factors from each of the three-factor groupings.

Future work will want to look at factors such as

etiquette with initial greetings and politeness encoded

in system behaviour and interaction responses.

Finally, continued work is needed in the area of

system embodiment and the impact that various

morphologies and modalities have on system use,

trust, and trust outcomes.

8 CONCLUSIONS

The results reported here suggest that when

interacting with a humanoid robot partner, the

openness personality trait is a significant predictor of

trust above and beyond the individual difference

propensity to trust. Continuing to develop and refine

the proposed framework of early trust in intelligent

systems is critical to ensuring the success of future

human-machine collaborations.

REFERENCES

Spohrer, J., and G. Banavar, “Cognition as a Service: An

Industry Perspective”, AI Magazine 36(4), 2015, pp.

71–86.

Jones, G.R., and J.M. George, “The experience and

evolution of trust: Implications for cooperation and

teamwork”, Academy of management review 23(3),

1998, pp. 531–546.

Dzindolet, M.T., S.A. Peterson, R.A. Pomranky, L.G.

Pierce, and H.P. Beck, “The role of trust in automation

reliance”, International Journal of Human-Computer

Studies 58(6), 2003, pp. 697–718.

Fukuyama, F., Trust: The social virtues and the creation of

prosperity, Free Press Paperbacks, 1995.

Luhmann, N., Trust and Power, John Wiley & Sons Inc,

Ann Arbor, Mich, 1982.

Billings, D.R., K.E. Schaefer, J.Y.C. Chen, and P.A.

Hancock, “Human-robot interaction: developing trust

in robots”, Proceedings of the seventh annual

ACM/IEEE international conference on Human-Robot

Interaction - HRI ’12, ACM Press (2012), 109.

Mayer, R.C., J.H. Davis, and F.D. Schoorman, “An

Integrative Model of Organizational Trust”, The

Academy of Management Review 20(3), 1995, pp. 709.

Mcknight, D.H., M. Carter, J.B. Thatcher, and P.F. Clay,

“Trust in a specific technology: An investigation of its

components and measures”, ACM Transactions on

Management Information Systems 2(2), 2011, pp. 1–25.

Poole, D.L., A.K. Mackworth, and R. Goebel,

Computational intelligence: a logical approach,

Oxford University Press New York, 1998.

Warren, J., G. Beliakov, and B. van der Zwaag, “Fuzzy

logic in clinical practice decision support systems”,

Proceedings of the 33rd Annual Hawaii International

Conference on System Sciences, (2000), 10 pp.-.

Matsatsinis, N.F., and Y. Siskos, Intelligent Support

Systems for Marketing Decisions, Springer Science &

Business Media, 2012.

Twyman, N.W., J.G. Proudfoot, R.M. Schuetzler, A.C.

Elkins, and D.C. Derrick, “Robustness of multiple

indicators in automated screening systems for

deception detection”, Journal of Management

Information Systems 32(4), 2015, pp. 215–245.

Humanoid 2021 - Special Session on Interaction with Humanoid Robots

306

Rasch, R., A. Kott, and K.D. Forbus, “Incorporating Ai into

Military Decision Making: An Experiment”, IEEE

Intelligent Systems 18(4), 2003, pp. 18–26.

Nunamaker, J.F., D.C. Derrick, A.C. Elkins, J.K. Burgoon,

and M.W. Patton, “Embodied Conversational Agent-

Based Kiosk for Automated Interviewing”, Journal of

Management Information Systems 28(1), 2011, pp. 17–

49.

Turban, E., J.E. Aronson, and T.-P. Liang, Decision

Support Systems and Intelligent Systems, Prentice Hall,

2004.

Li, J., “The benefit of being physically present: A survey of

experimental works comparing copresent robots,

telepresent robots and virtual agents”, International

Journal of Human-Computer Studies 77, 2015, pp. 23–

37.

Epley, N., A. Waytz, and J.T. Cacioppo, “On seeing human:

A three-factor theory of anthropomorphism”,

Psychological Review 114(4), 2007, pp. 864–886.

Waytz, A., J. Cacioppo, and N. Epley, “Who Sees Human?

The Stability and Importance of Individual Differences

in Anthropomorphism”, Perspectives on psychological

science : a journal of the Association for Psychological

Science 5(3), 2014, pp. 219–232.

de Visser, E.J., S.S. Monfort, R. McKendrick, et al.,

“Almost human: Anthropomorphism increases trust

resilience in cognitive agents.”, Journal of

Experimental Psychology: Applied 22(3), 2016, pp.

331–349.

Schroeder, J., and M. Schroeder, “Trusting in Machines:

How Mode of Interaction Affects Willingness to Share

Personal Information with Machines”, (2018).

McKnight, D.H., and N.L. Chervany, “Trust and distrust

definitions: One bite at a time”, In Trust in Cyber-

societies. Springer, 2001, 27–54.

Madsen, M., and S. Gregor, “Measuring human-computer

trust”, 11th australasian conference on information

systems, Citeseer (2000), 6–8.

Jian, J.-Y., A.M. Bisantz, and C.G. Drury, “Foundations for

an empirically determined scale of trust in automated

systems”, International Journal of Cognitive

Ergonomics 4(1), 2000, pp. 53–71.

Spain, R.D., E.A. Bustamante, and J.P. Bliss, “Towards an

empirically developed scale for system trust: Take

two”, Proceedings of the Human Factors and

Ergonomics Society Annual Meeting, SAGE

Publications Sage CA: Los Angeles, CA (2008), 1335–

1339.

Rotter, J.B., “A new scale for the measurement of

interpersonal trust”, Journal of Personality 35

(4), 1967,

pp. 651–665.

Elson, J.S., D. Derrick, and G. Ligon, “Examining Trust

and Reliance in Collaborations between Humans and

Automated Agents”, 2018.

Groom, V., and C. Nass, “Can robots be teammates?:

Benchmarks in human–robot teams”, Interaction

Studies 8(3), 2007, pp. 483–500.

Barrick, M.R., G.L. Stewart, M.J. Neubert, and M.K.

Mount, “Relating member ability and personality to

work-team processes and team effectiveness.”, Journal

of Applied Psychology 83(3), 1998, pp. 377–391.

Gosling, S.D., P.J. Rentfrow, and W.B. Swann, “A very

brief measure of the Big-Five personality domains”,

Journal of Research in Personality 37, 2003, pp. 504–

528.

McCrae, R.R., and O.P. John, “An introduction to the five-

factor model and its applications”, Journal of

personality 60, 1992.

Goldberg, L.R., “The Development of Markers for the Big-

Five Factor Structure.”, Psychological assessment 4(1),

1992, pp. 26.

Tett, R.P., and H.A. Guterman, “Situation trait relevance,

trait expression, and cross-situational consistency:

Testing a principle of trait activation”, Journal of

Research in Personality 34(4), 2000, pp. 397–423.

Tett, R.P., and D.D. Burnett, “A personality trait-based

interactionist model of job performance.”, Journal of

Applied psychology 88(3), 2003, pp. 500.

Molloy, R., and R. Parasuraman, “Monitoring an automated

system for a single failure: Vigilance and task

complexity effects”, Human Factors 38(2), 1996, pp.

311–322.

John, O.P., E.M. Donahue, and R.L. Kentle, The big five

inventory—versions 4a and 54, Berkeley, CA:

University of California, Berkeley, Institute of

Personality …, 1991.

John, O.P., L.P. Naumann, and C.J. Soto, “Paradigm Shift

to the Integrative Big Five Trait Taxonomy”, Handbook

of Personality: Theory and Research 3, 2008, pp. 114–

158.

Ashleigh, M.J., M. Higgs, and V. Dulewicz, “A new

propensity to trust scale and its relationship with

individual well-being: implications for HRM policies

and practices”, Human Resource Management Journal

22(4), 2012, pp. 360–376.

Schaefer, K.E., J.Y.C. Chen, J.L. Szalma, and P.A.

Hancock, “A Meta-Analysis of Factors Influencing the

Development of Trust in Automation: Implications for

Understanding Autonomy in Future Systems”, Human

Factors: The Journal of the Human Factors and

Ergonomics Society 58(3), 2016, pp. 377–400.

Bessi, A., “Personality Traits and Echo Chambers on

Facebook”, arXiv:1606.04721 [cs], 2016.

Nickerson, R.S., “Confirmation bias: A ubiquitous

phenomenon in many guises”, Review of general

psychology 2(2), 1998, pp. 175–220.

McElroy, T., and K. Dowd, “Susceptibility to anchoring

effects: How openness-to-experience inﬂuences

responses to anchoring cues”, Judgment and Decision

Making 2(1), 2007, pp. 6.

Decision-making with a Humanoid Robot Partner: Individual Differences Impacting Trust

307