A Review of Personality in Human Robot Interactions

• The Big Five, represented by OCEAN, are the most widely used personality traits and are also the most popular traits in HRI studies.Source: Introduction · pp. 3–4
• HRI personality research is fragmented and lacks a coherent framework, making it difficult to understand established knowledge and identify open questions.Source: Introduction · p. 4
• The review focuses on embodied physical-action robots because embodiment can evoke strong emotions that prompt people to project personalities onto them.Source: Introduction · p. 4
• It presents an integrated conceptual model, identifies four research thrusts, synthesizes major insights, and identifies gaps for future work.Source: Introduction · p. 4

Literature Search

• The search used Google Scholar, ACM Digital Library, Scopus, PsycINFO, and IEEE Xplore in December 2017, with three related HRI-and-personality search terms.Source: Literature Search · p. 5, Section 2.1. Study Selection Process · p. 5

Section 2.1. Study Selection Process

• The initial 220 unique articles had to mention both “robot” and “personality” and be published in English-language journals or conferences; final inclusion required an empirical study measuring personality during human interaction with an embodied physical-action robot.Source: Section 2.1. Study Selection Process · p. 5
• The review retained 83 empirical articles comprising 84 studies, excluding 15 non-empirical articles and 119 articles outside its final scope.Source: Section 2.1. Study Selection Process · p. 5

• 73.6% of included articles appeared in conferences, compared with 26.4% in journals.Source: Section 2.2. Publication Venues · p. 6
• The HRI Conference accounted for 26.3% of all included articles and 35.7% of conference publications.Source: Section 2.2. Publication Venues · p. 6
• Publication dates ranged from 2005 to 2017.Source: Section 2.2. Publication Venues · p. 6

• Figure 2.1 is titled “Publications by Year,” indicating a year-based view of publication counts.Source: Figure 2.1 · p. 6

Section 2.4. Personality Measures

• More than 89% of articles used at least one Big Five measure, and extraversion/introversion appeared in every study measuring the Big Five.The Big Five were the most widely used personality measures.Source: Section 2.4. Personality Measures · p. 6

Section 2.5. Outcome Measures

• Effectiveness measures were most common at 31.5%, followed by trust or confidence at 26.3%.Effectiveness included feelings of attachment and liking toward the robot.Source: Section 2.5. Outcome Measures · p. 6
• Distance, acceptance, robot preference, and compliance each accounted for 15.7% of reported outcomes, and most studies measured multiple outcomes.Measures of HRI quality varied.Source: Section 2.5. Outcome Measures · pp. 6–7

Section 2.6. Sample

• The average sample size was 60 participants (SD 74.77), while the median was 36 and the mode was 32.The standard deviation indicates substantial variation in sample size across studies.Source: Section 2.6. Sample · p. 7
• Average age was reported directly by 58 studies, with a mean of 31.73 years (SD 6.67); 26 studies did not directly report it.Fifteen gave an age range and 11 provided no average-age information.Source: Section 2.6. Sample · p. 7
• Among the 63 studies reporting participant gender counts, 45.8% of participants were women; 21 studies omitted those counts.Source: Section 2.6. Sample · p. 7
• Nationality was reported by 57 studies and omitted by 27; participants across the studies represented 12 countries.Several studies examined more than one population.Source: Section 2.6. Sample · p. 7
• All but one study analyzed individuals; Salam et al. (2017) was the sole group-level study.Source: Section 2.6. Sample · p. 7

Section 2.7. Study Settings

• Home settings accounted for 18 studies and caregiving for 15, compared with 7 workplace, 3 education, and 1 entertainment study.The study-setting distribution is presented in Figure 2.2.Source: Section 2.7. Study Settings · p. 7, Figure 2.2 · p. 8
• All studies were experimental lab studies, with 44 designed to mimic a real-world setting and 40 not designed to do so.Source: Section 2.7. Study Settings · p. 7

Section 2.8. Type of Robot

• Approximately 95 robots were used across the 84 studies, spanning at least 20 types.Several studies used two to three robots.Source: Section 2.8. Type of Robot · p. 8
• Among the robots classified by form, 49 were humanoid and 15 were non-humanoid.Source: Section 2.8. Type of Robot · p. 8
• Nao was used in 18 studies, more than any other robot; iCat and PeopleBot appeared in six studies each.Meka was used in four studies and iCub in three; the remaining 15 robots appeared in one to two studies each.Source: Section 2.8. Type of Robot · p. 8

Section 2.9. Robot Interaction Control

• Wizard of Oz control appeared in 24 studies, automation autonomy in 10, pre-programmed strategy in six, and a hybrid approach in four.Wizard of Oz uses a hidden human operator; automation autonomy lets the robot respond independently; pre-programmed responses do not depend on human behavior; hybrid control combines automation and human operation.Source: Section 2.9. Robot Interaction Control · pp. 8–9
• Robot interaction control concerns robot responses to humans over multiple human responses, unlike demonstrations independent of the participant’s behavior.Source: Section 2.9. Robot Interaction Control · p. 8

Section 2.10. Thrust Area

• The review groups studies into four thrust areas: human personality, robot personality, human–robot personality similarities and differences, and factors that shape perceived robot personality.Source: Section 2.10. Thrust Area · p. 9

Section 2.11. Chapter Summary

• The review selected 83 empirical articles comprising 84 studies from an initial 220 articles.Source: Section 2.11. Chapter Summary · p. 9
• The chapter summarizes publication venues, personality measures, outcomes, samples, settings, robot types, and interaction control across the studies.Source: Section 2.11. Chapter Summary · p. 9
• The literature is organized into four thrust areas.Source: Section 2.11. Chapter Summary · p. 9

Section 3.1. Inputs

• Of 34 articles examining human personality as an independent variable, 23 investigated extraversion or introversion, the most commonly studied trait.Studies also associated extraversion with greater willingness to talk to robots and trust, while much of the literature treated its sociality effects as an assumption.Source: Section 3.1. Inputs · p. 10
• Other studied traits included openness, conscientiousness, and neuroticism; openness related to acceptance and robot ratings, while conscientiousness predicted task performance and approach preferences.Individuals high in conscientiousness performed better when reminded by a robot, whereas those low in conscientiousness were more willing to let it approach closer.Source: Section 3.1. Inputs · pp. 10–11

Section 3.2. Outcomes

• Distance and approach findings were mixed: personality sometimes predicted proximity or tolerance, but one study found no effect on approach distance.Other reported influences included proactive behavior, anxiety, prior experience with pets or robots, robot head orientation, and gender.Source: Section 3.2. Outcomes · p. 14
• Personality related to robot perceptions and attitudes in some studies, while agreeableness—but not several other traits—predicted a positive interaction evaluation.Robots were perceived as less open and agreeable than humans; demographic and technology-commitment effects were also reported, while one study found no gender effect on perceived persuasiveness.Source: Section 3.2. Outcomes · pp. 14–15
• Extraversion was associated with greater anthropomorphism and trust in some studies, but extraversion and emotional stability did not consistently predict trust development.Neuroticism and anthropomorphic thinking were negatively correlated with mental relatedness to the robot.Source: Section 3.2. Outcomes · p. 15
• Human personality traits did not affect emotion-recognition behavior, although people showed emotional reactions toward robots.Source: Section 3.2. Outcomes · p. 15
• Openness and extraversion were reported to promote robotics acceptance; studies also examined embodiment, vocabulary adoption, and learning from robots.The passage identifies embodiment and learning topics but does not report their findings.Source: Section 3.2. Outcomes · p. 15

Section 3.2. Outcomes; Section 3.3. Study Method, Sample, Context and Robot Type

• The cited studies used varied measures, including task time, attitudes toward social robots, safety assessed with the Godspeed questionnaire, and Likert-scale ratings of robot-appearance preferences.All 34 studies were laboratory experiments; reported samples ranged from 11 to 489 participants, aged 19 to 85.Source: Section 3.2. Outcomes · pp. 17–18, Section 3.3. Study Method, Sample,… · p. 19

Section 3.4. Findings

• Extraversion was the most examined trait and was linked to greater tolerance of robot behavior, closer approach, anthropomorphism, trust-game offers, and longer conversations, though one study found no approach-distance effect.Source: Section 3.4. Findings · p. 19
• Introverted participants were more motivated by the robot to finish tasks earlier, while conscientious participants were better at being on time when reminded by a robot.Source: Section 3.4. Findings · p. 19
• Higher neuroticism was associated with standing farther from robots and longer touch latency, but also with better learning performance; higher extraversion was associated with faster touch.Source: Section 3.4. Findings · pp. 19–20
• Agreeableness was associated with positive interaction evaluation and closer movement toward robots; highly agreeable participants were also described as trusting.Source: Section 3.4. Findings · p. 20
• Higher openness predicted greater adoption of words introduced by a robot peer, while increasing openness was associated with lower extraversion and agreeableness ratings for a persuasive android.Source: Section 3.4. Findings · p. 20
• Gender findings differed by outcome, while studies of older adults reported personality-related interaction effects but no significant behavior change from tele-operation disclosure or beliefs about robot autonomy and functionality.Source: Section 3.4. Findings · p. 20

Section 3.5. Chapter Summary

• Overall, the review reports direct and indirect effects of human personality on reactions to robots, with extroverted people tending to prefer robot interaction more than introverted people.Source: Section 3.5. Chapter Summary · p. 22

Section 4.1. Inputs

• Robot traits center on extraversion and introversion and Big Five dimensions, with social ability also studied; measures vary, including IPIP and TIPI.Source: Section 4.1. Inputs · p. 23

Section 4.2. Outcomes

• Studies grouped outcomes into robot perceptions, intention to use, and interaction quality, including trustworthiness, intelligence, capability, and persuasiveness.Source: Section 4.2. Outcomes · p. 25
• People showed high acceptance of robots considered extraverted and socially intelligent; other studies examined enjoyment and perceived control.Source: Section 4.2. Outcomes · p. 25

Section 4.3. Study Method, Sample, Context

• All 25 reviewed studies used design and experimental approaches, deriving human-trait constructs, prototyping robot traits, and testing perceptions and effects.Source: Section 4.3. Study Method, Sample,… · pp. 31–32
• Studies were concentrated at home (8) and in healthcare (5), with two organizational-work studies; sample sizes ranged from 12 to 200 participants.Source: Section 4.3. Study Method, Sample,… · p. 32

Section 4.4. Findings

• Task and role affected perceived personality effects: assistant and companion jobs called for different personality aspects, and preferences were also tested across occupations.Source: Section 4.4. Findings · p. 32

Section 4.5. Chapter Summary

• Robot personality directly and indirectly influenced fun and enjoyableness; people tended to prefer extroverted robots, possibly reflecting the social nature of the tasks.Source: Section 4.5. Chapter Summary · p. 33

Section 5.1. Inputs

• Researchers usually framed human–robot personality similarity as a binary match or mismatch, most often measuring human traits and manipulating robot extraversion through speech, gaze, and gestures.Extraversion measures commonly captured energy, assertiveness, sociability, and talkativeness.Source: Section 5.1. Inputs · pp. 35, 37

Section 5.2. Outcomes

• Studies assessed perceptions of the robot, interaction quality, and willingness to interact, including enjoyment, engagement, and preference for a similar-personality robot.Positive perceptions studied included robot quality, personality, and competency.Source: Section 5.2. Outcomes · pp. 38, 40

Section 5.3. Study Method, Sample, Context & Robot Type

• Most studies used experiments; the 15 personality-match studies were all experimental, with samples of 18–86 and settings including healthcare, work, and home.Wizard-of-Oz control was most common among the personality-match studies; Nao was the most frequently used robot in that group.Source: Section 5.3. Study Method, Sample,… · pp. 40–41

Section 5.4. Findings

• Matching was often associated with positive perceptions, engagement, trust, credibility, and preference, while some studies found no significant matching effects.The authors note that people may not always notice the behavioral differences programmed to express robot personalities.Source: Section 5.4. Findings · pp. 41–42
• Matching effects depended on robot characteristics and context: participants’ preferences varied with personality type, robot role, and task.For example, introverted participants liked an opposite-personality robot more when it was a cleaner, while extraverted participants liked an extraverted tour guide more.Source: Section 5.4. Findings · pp. 41–42

Section 5.5. Chapter Summary

• Personality matching generally improved interaction quality, positive robot perceptions, and robot preference, but results were not consistent across studies.Reported perceptions included enjoyment, empathy, intelligence, social attraction, credibility, trust, perceived performance, and compliance.Source: Section 5.4. Findings · p. 41, Section 5.5. Chapter Summary · p. 42

Section 6.1. Inputs

• Robot behavior was the most common way to manipulate perceived robot personality, while appearance, role, embodiment, and human personality were also studied.Behavior manipulations included physical cues such as gestures and gaze, and communicative cues such as voice, language, and responsiveness.Source: Section 6.1. Inputs · p. 43

Section 6.2. Outcomes

• Studies measured perceived robot personality with Big Five and alternative scales, alongside attitudes, acceptance, behavioral intentions, and physical behavior such as interpersonal distance.Attitudes included usefulness, ease of use, control, enjoyment, performance, attachment, satisfaction, and animacy.Source: Section 6.2. Outcomes · pp. 45, 47, 48

Section 6.3 Study Methods, Samples, Contexts and Robot Type

• Most studies used questionnaires around experimental conditions; samples were often small and young, while settings, countries, and robots varied.Sixty-one percent of studies had fewer than 50 participants; the literature included healthcare and organizational settings, and Wizard-of-Oz and autonomous control methods.Source: Section 6.3 Study Methods, Samples,… · p. 49

Section 6.4. Findings

• User traits and robot features shaped perceptions: extraverts tended to rate mechanoid robots more positively, while humanoid appearance and role-congruent personalities were often favored.Extraverts and introverts both showed greater compliance with a personality-matching robot; findings also linked responsiveness with sociability and role-congruent personality with willingness to interact.Source: Section 6.4. Findings · pp. 50–51

Section 6.5. Chapter Summary

• The literature’s most common approach to changing perceived robot personality was altering robot behavior, including physical movement and communicative cues.Physical cues included gestures, movement patterns, facial expressions, and gaze; communicative cues included audio style, written and linguistic style, voice gender and speed, and responsiveness.Source: Section 6.5. Chapter Summary · pp. 51–52

Section 7.1. Major Findings

• The review’s four major findings concern favorable responses by extraverts, cues that evoke extraverted robot personalities, moderated responses to extravert-type robots, and responses to similar or different personalities.These are presented as the review’s most consistent and generalizable findings.Source: Section 7.1. Major Findings · p. 53

Section 7.2. Critique of the Major Findings

• Extraversion may be easier to display through robot gestures, speech, and movement, while short experimental interactions may make subtler traits less salient.The review notes that expressing traits such as openness may require technological approaches not commonly used by social-science researchers.Source: Section 7.2. Critique of the Major… · p. 53
• Task context may moderate the importance of robot extraversion, and the influence of moderators on human personality effects remains little understood.Extraversion was less important for a security robot than a health-care robot; the review suggests task sociality may help explain the difference.Source: Section 7.2. Critique of the Major… · pp. 53–54
• Because robots do not always know a human’s personality, the effects of human and robot personalities still need study apart from personality similarity.The review describes research on similar versus different personalities as small but growing.Source: Section 7.2. Critique of the Major… · p. 54

Section 7.3. A Way Forward

• Future research should examine context and task type, extend beyond lab experiments through field, longitudinal, and qualitative work, and consider traits beyond the Big Five.The review notes that no study examined context, health-care studies were lacking, and qualitative approaches had rarely been primary methods.Source: Section 7.3. A Way Forward · pp. 54–55

Section 7.4. Chapter Summary

• The chapter concludes that human or robot extraversion is positively associated with HRI, while personality-similarity effects are less clear.It highlights more contextual and field research and attention to non-Big-Five personalities as opportunities.Source: Section 7.4. Chapter Summary · p. 55