See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/228368017

Altered attitudes of people toward robots: Investigation through the Negative

Attitudes toward Robots Scale

Conference Paper · January 2006

CITATIONS READS

107 1,169

4 authors, including:

Tatsuya Nomura Tomohiro Suzuki

Ryukoku University Tokyo Future University
145 PUBLICATIONS 4,318 CITATIONS 54 PUBLICATIONS 2,948 CITATIONS

SEE PROFILE SEE PROFILE

Takayuki Kanda
Kyoto University
462 PUBLICATIONS 20,249 CITATIONS

SEE PROFILE

All content following this page was uploaded by Tatsuya Nomura on 21 May 2014.

The user has requested enhancement of the downloaded file.

 Altered Attitudes of People toward Robots:
Investigation through the Negative Attitudes toward Robots Scale ∗
Tatsuya Nomura†,‡ Tomohiro Suzuki Takayuki Kanda‡ Kensuke Kato
Department of
† Graduate School of ‡
ATR Intelligent Robotics Graduate School of
Media Informatics, Sociology, and Communication Human Science,
Ryukoku University Toyo University Laboratories Osaka University
Shiga 520–2194, Japan Tokyo 112–8606, Japan Kyoto 619–0288, Japan Osaka 565–0871, Japan
nomura@rins.ryukoku.ac.jp suzukirt@h9.dion.ne.jp kanda@atr.jp DZL02550@nifty.ne.jp

Abstract interactions with AIBO. Their important suggestion is that

people in general, and children in particular, may fall prey to
It is important to measure what images, opinions, and at-
titudes people have developed toward robots and how they
accepting robotic pets without the moral responsibilities that
can be changed, from scientific and engineering perspectives. real, reciprocal companionship and cooperation involves. In
This paper reports results of social research on Japanese peo- addition, Nomura et al. (2005b) reported the results of so-
ple’s attitudes toward robots by using “the Negative Attitudes cial research on visitors to an exhibition of communication
toward Robots Scale (NARS).” They revealed that attitudes robots, called “Robovie” (Ishiguro et al. 2001), suggesting
toward robots differ depending on assumptions about robots that even in Japan, younger generations do not necessarily
such as their type and task, and there may be gender differ- like the robots more than do elder generations. These stud-
ences associated with them. Based on the results, the paper ies are focused on specific commercialized robots.
then discusses how people’s attitudes toward robots can be On the other hand, some studies examined more gen-
altered.
eral images independent of specific robots. Suzuki et al.
(2002) developed a psychological scale for measuring hu-
Introduction mans’ mental images toward robots, while Kashibuchi et
It is known that the concept of “robots” itself is very old. al. (2002) showed by using this scale that humans’ men-
However, it is only recently that they have appeared as com- tal images toward robots are positioned in the middle of a
mercialized products in daily life. This fact implies that the one–dimensional scale, where one pole corresponds to hu-
concept of robots itself or mental images of robots may be mans and another pole corresponds to just physical objects.
changed. Thus, it is important to determine what images, Woods & Dautenhahn (2005) investigated the difference in
opinions, and attitudes people have had toward robots and relations of robots’ appearances to emotions toward them
how they can be altered, from scientific and engineering between children and adults, using a questionnaire–based
perspectives such as design issues and market research on method and photographs of several robots. In addition, Bart-
robots. neck et al. (2005b; 2005a) reported the influences of cul-
Regarding the measurement of human mental images and tural differences and personal experiences with robots into
impressions toward robots, there are plenty of published attitudes toward robots by using a psychological scale mea-
studies. Shibata, Wada, & Tanie (2002; 2003; 2004) re- suring negative attitudes toward robots, “the Negative Atti-
ported international research results on people’s subjective tudes toward Robots Scale (NARS)” developed by Nomura,
evaluations of a seal–type robot they developed, called Palo, Kanda, & Suzuki (2006).
in several countries including Japan, the U.K, Sweden, Italy, This paper also reports results of social research on peo-
and Korea. Their results revealed that there were differences ple’s attitudes toward robots by using the above psychologi-
in subjective evaluations of the robot among genders and cal scale NARS. The focus of this research is on relation-
ages, and that nationality also affected the evaluation factors. ships between negative attitudes toward and assumptions
Friedman, Kahn, & Hagman (2003) investigated people’s re- about robots such as their types and tasks (Nomura et al.
lationships with robotic pets by analyzing more than 6,000 2005a). Based on the results, we discuss how people’s atti-
postings in online discussion forums about one of the most tudes toward robots can be altered.
advanced robotic pets currently on the retail market, Sony’s
robotic dog AIBO. Furthermore, Kahn et al. (2004) ex- Attitudes toward Robots
amined preschool children’s reasoning about and behavioral
Psychological Concept of Attitudes and Robots
∗
This research was supported in part by the Ministry of Internal
Affairs and Communications of Japan, and the Japan Society for
An attitude is psychologically defined as a relatively stable
the Promotion of Science, Grants–in–Aid for Scientific Research and enduring predisposition to behave or react in a certain
No. 18500207. way toward persons, objects, institutions, or issues, and the
Copyright c 2006, American Association for Artificial Intelli- source is cultural, familial, and personal (Chaplin 1991).
gence (www.aaai.org). All rights reserved. This definition of attitudes implies that they can be affected
 Table 1: The English Version of Negative Attitude toward Robots Scale and the Subscales that Each Item Is Included
Item No. Questionnaire Items Subcale
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
(∗ Reverse Item)

by cultural backgrounds and personal experiences. More- Negative Attitudes toward Robots Scale
over, the classical psychological theory suggests that they As one of the measurement tools, in this paper we use the
can be changed based on mental congruity (Osgood & Tan- Negative Attitude toward Robots Scale (NARS) to measure
nenbaum 1955; Newcomb 1953; Heider 1958). These facts people’s attitudes toward robots through social research.
imply that attitudes toward robots can be altered by some This scale was developed to determine humans’ attitudes
factors including cultural backgrounds and personal experi- toward robots, and its internal consistency, factorial valid-
ences. ity, and test–retest reliability have been confirmed based on
Bartneck et al. (2005b; 2005a) suggested that cultures Japanese respondents (Nomura, Kanda, & Suzuki 2006). As
may influence attitudes toward robots. Moreover, Nomura, mentioned above, the research by Bartneck et al. (2005b;
Kanda, & Suzuki (2006) suggested the influence of personal 2005a) was based on this psychological scale.
experiences with robots on attitudes toward them, such as This scale consists of fourteen questionnaire items. Table
individuals’ experiences on really acting robots. However, 1 shows the English version of the NARS, which was trans-
these studies lack perspective on what types and tasks of lated using back–translation. These items are classified into
robots people assume. It is considered that attitudes toward three subscales, S1: “Negative Attitude toward Situations
robots can more directly be influenced by assumptions about of Interaction with Robots” (six items), S2: “Negative At-
robots than by cultures and personal experiences, although titude toward Social Influence of Robots” (five items), and
these assumptions can be affected by cultures and personal S3: “Negative Attitude toward Emotions in Interaction with
experiences. Thus, we should focus on assumptions about Robots” (three items). The number of grades for each item is
robots to investigate alternation of attitudes toward robots. five (1: I strongly disagree, 2: I disagree, 3: Undecided, 4: I
These assumptions can be altered by cultural situations such agree, 5: I strongly agree), and an individual’s score on each
as media, and their distribution can affect that of attitudes subscale is calculated by summing the scores of all the items
toward robots. Of course, cultural differences can produce included in the subscale, with the reverse of scores in some
differences in attitudes toward one specific type of robot items. Thus, the minimum and maximum scores are 6 and
such as humanoids, as Kaplan (2004) mentioned. Thus, we 30 in S1, 5 and 25 in S2, and 3 and 15 in S3, respectively.
can consider several types of differences in attitude toward
robots as follows: Social Research
• Differences in attitudes between different assumptions To investigate the differences in attitudes between assump-
about robots in one culture. tions about robots in one culture, we administered the fol-
• Differences in attitudes toward one specific type of robot lowing social research.
between different cultures.
Method
• Differences in assumptions about robots and their rela-
The social research was administered from November 2005
tionship to attitudes toward robots between different cul-
to March 2006. The participants were Japanese university
tures.
and special training school students. The Japanese version
As mentioned above, Bartneck et al. (2005b; 2005a) sug- of the NARS was administered during lecture time. Partici-
gested the possibility of there being cultural differences in pation by the respondents was voluntary.
attitudes toward robots. This paper deals with the first issue The face sheet used in administering this survey included
in the above list. items that asked respondents to answer which type of robots
 Table 2: The Number of Respondents Who Selected Each Robot Type and Task
Robot Task
Robot Type 1 2 3 4 5 6 7 8 9 10 11 Total
1 N 53 8 0 5 9 19 26 11 30 31 4 196
φ .255∗∗∗ .059 -.088 .000 -.024 -.072 .001 .148∗∗ -.162∗∗ -.028 -.094
2 N 12 1 1 3 0 2 9 1 21 20 3 73
φ -.012 -.047 .033 .047 -.111∗ -.137∗∗ -.014 -.047 .078 .135∗ .007
3 N 1 2 1 0 5 23 13 0 5 10 1 61
φ -.179∗∗∗ .005 .043 -.070 .060 .340∗∗∗ .101 -.077 -.143∗∗ -.006 -.049
4 N 0 0 0 1 0 2 1 0 21 1 1 27
φ -.125∗∗ -.048 -.024 .020 -.063 -.039 -.077 -.048 .367∗∗∗ -.096 -.002
5 N 0 1 1 0 2 0 1 0 0 0 0 5
φ -.052 .112 .251∗ -.018 .180∗ -.042 -.022 -.020 -.060 -.051 -.023
6 N 0 0 0 0 4 1 1 0 0 1 0 7
φ -.062 -.024 -.012 -.022 .319∗∗∗ .009 .004 -.024 -.072 -.009 -.027
7 N 2 0 0 1 0 0 1 0 9 3 6 22
φ -.053 -.043 -.021 .031 -.057 -.090 -.063 -.043 .111 ∗ -.021 .298∗∗∗
Total N 68 12 3 10 20 47 52 12 86 66 15 391
Robot Type: 1: human–size humanoids, 2: small–size humanoids, 3: acting huge objects, 4: animals,
5: stationary objects, 6: arm manipulators, 7: others
Robot Task: 1: housework, 2: office work, 3: public service such as education, 4: medical or welfare service,
5: construction or assembling tasks, 6: guard or battle,
Robot Type: 7: tasks in places hard for humans to go or hazardous locations such as the space and the deep sea,
8: the service trade, 9: communication partners or playmates, 10: amusement, 11: others
(Correlation Coefficients: φ–Coefficients, ∗ p < .05, ∗∗ p < .01, ∗∗∗ p < .001, p–Values: Results by Fisher’s Method)

they assumed and which tasks they assumed the selected rectly related to interaction with robots (S1 and S3).
robots do. The choices for the former item were: 1: human–
size humanoids, 2: small–size humanoids, 3: acting huge Results
objects, 4: animals, 5: stationary objects, 6: arm manipu- A total of 400 people (male: 197; female: 199; unknown:
lators, and 7: others. The choices for the latter item were: 4; mean age: 21.4) participated in the research. For the 374
1: housework, 2: office work, 3: public service such as ed- samples that had no missing item in the NARS, Cronbach’s
ucation, 4: medical or welfare service, 5: construction or α denoting reliability were 0.756 for S1, 0.647 for S2, and
assembling tasks, 6: guard or battle, 7: tasks in places hard 0.735 for S3 respectively. The sample data were analyzed in
for humans to go or hazardous locations such as the space the following three ways:
and the deep sea, 8: the service trade, 9: communication
partners or playmates, 10: amusement, 11: others. These Relations between Assumptions about Robot Types and
choices were determined based on the pilot study by No- Tasks: First, we calculated how many respondents se-
mura et al. (2005a). lected each robot type and task with respect to the assump-
In addition to the above face sheet and the NARS, two tions about robots. Then, to find relations between specific
psychological scales were administered to investigate re- assumptions about types and tasks, φ–coefficients were cal-
lationships between attitudes toward robots and personal culated to reveal the extent of relationships between the as-
traits. One is the State–Trait Anxiety Inventory (STAI), sumption choices. In addition, we performed statistical tests
which is used for measuring general anxiety (Spielberger, with Fisher’s method on selection for pairs of choices to in-
Gorsuch, & Lushene 1970). The emotion of anxiety is gen- vestigate the statistical significance of these φ–coefficients,
erally classified into two categories: state and trait anxiety. based on the independence among these choices. 1
Trait anxiety is a trend of anxiety as a characteristic stable Table 2 shows the number of respondents who selected
in individuals whereas state anxiety is an anxiety transiently each robot type and task, and the correlations between robot
evoked in specific situations and changed depending on the types and tasks. Regarding assumptions about robot type,
situation and time. STAT consists of twenty items for mea- about 50% of the respondents selected “human–size hu-
suring state anxiety (STAI–S) and twenty items for measur- manoids.” The humanoid type, including small–size ones,
ing trait anxiety (STAI–T). was selected by about 70% of the respondents, while the se-
The other scale is the Report of Communication Appre- lection rate for “animals” was about 7%. The respondents
hension (PRCA–24) (Pribyl et al. 1998). PRCA–24 mea- who selected “others” tended to mention concrete names of
sures communication apprehension in four contexts: pub- 1
For example, to investigate the correlation between “small–
lic speaking, meetings, small–group discussion, and dyads. size humanoids” and “amusement,”, one 2 × 2 cross table con-
Each context corresponds to six items. In this administra- sisting of selection/no–selection of “small-size humanoids” and
tion, only six items corresponding to dyads were used to “amusement” was made, then the φ–coefficient was calculated and
investigate their relationships with the NARS subscales di- a test was done for this cross table.
 
 ,*
,-)
 
,-(


 ,'

+
  *
)


(
 '
        . / . /./ . / . /

   
   ! "#$%$& , ( 0 )2134*35

S1 S2 S3

Figure 1: Means and Standard Deviations of NARS Subscale Scores based on Gender and Robot Type Subgroups (Type 1:
Male N = 92, Female N = 89, 2: Male N = 27, Female N = 46, 3: Male N = 34, Female N = 21, 4: Male N = 13,
Female N = 13, 5+6+7: Male N = 16, Female N = 17)

: 7 N L bI_
bI^
:6 NGK
b9]
8 7 M L a

896 M-K `

_
7 L
^

6 K ]
;#<=;#<=;#<=;#<=;#<=;#<=;#< O#P=O#P=O#P=O#P=O#P=O#P=O2P c#d=c#d=c#d=c#d=c#d=c#d=c#d

8 7 > ? @ 896BADC EGFIH J M L Q R S M-KUTWV XY[Z \ b e ` f g b-]UhDi jGkIl m

S1 S2 S3

Figure 2: Means and Standard Deviations of NARS Subscale Scores based on Gender and Robot Task Subgroups (Task 1: Male
N = 30, Female N = 33, 5: Male N = 12, Female N = 5, 6: Male N = 27, Female N = 16, 7: Male N = 28, Female
N = 19, 9: Male N = 22, Female N = 59, 10: Male N = 28, Female N = 34, Others (2+3+4+8+11): Male N = 31, Female
N = 18)

some robots appearing in the media, such as “Doraemon” manoids” and “the service trade,” between “small–size hu-
and “Asimo” in their written answers. manoids” and “amusement,”, between “stationary objects”
Regarding assumptions about robot tasks, there was no and “public service such as education,” between “station-
bias of respondents toward a specific task; the selection rates ary objects” and “construction or assembling tasks,” and
for “housework,” “communication partner or playmates,” “others” and “communication partners or playmates.” In
and “amusement” varied from 16% and 22%. The selec- addition, there was a low level of negative correlation be-
tion rates for “guard or battle,” and “tasks in places hard tween “human–size humanoids” and “communication part-
for humans to go or hazardous locations” were about 12– ners or playmates,” between “small–size humanoids” and
13%. Few respondents selected “public service such as ed- “construction or assembling tasks”, between “small-size hu-
ucation.” The respondents who selected “others” tended not manoids” and “guard or battle,” between “huge acting ob-
to mention concrete tasks in their writtten answers. jects” and “housework,” between “huge acting objects” and
Regarding relations between robot type and task, there “communication partners or playmates,” and between “ani-
was a moderate level of positive correlation between “act- mals” and “housework.”
ing huge objects” and “guard or battle,” between “ani-
mals” and “communication partners or playmates,” between Relations of Gender and Assumptions about Robots with
“arm manipulators” and “construction or assembling tasks,” Attitudes toward Robots: Second, to investigate the re-
and between “others” and “others.” Moreover, there was lations between attitudes toward and assumptions about
a low level of positive correlation between “human–size robots, two–way ANOVAs were executed with the indepen-
humanoids” and “housework,” between “human–size hu- dent variables of gender and robot type, and variables of
gender and robot task, respectively 2 . In this analysis, the
subgroups of respondents who selected “stationary objects,” Table 3: F –Values in the Two ANOVAs for the Scores of
“arm manipulators,” and “others” were integrated into one NARS Subscales
Gender Robot Type Interaction
subgroup due to their small numbers of respondents and
S1 13.910∗∗∗ 3.386∗ 0.498
correlations with robot tasks. Furthermore, the subgroups S2 0.420 1.063 0.361
of respondents who selected “office work,” “public service S3 10.237∗∗ 5.406∗∗∗ 1.285
such as education,” “medical or welfare service,” “the ser- Gender Robot Task Interaction
vice trade,” and “others” were integrated into one subgroup, S1 8.460∗∗ 1.899† 0.875
due to their small numbers of respondents, correlations with S2 3.576† 2.774∗ 0.543
robot type, and the similarity of their contents. S3 0.914 1.421 0.360
Figures 1 and 2 illustrate the means and standard devia- († p < .1, ∗ p < .05, ∗∗ p < .01, ∗∗∗ p < .001)
tions of the scores of the NARS subscales based on gender
and robot type subgroups, and based on gender and robot
task subgroups, respectively. In addition, Table 3 shows Table 4: Peason’s Correlation Coefficients r between NARS
the F –values in the two ANOVAs for the NARS subscale Subscales, STAI, and PRCA–24
scores. PRCA–24 STAI–S STAI–T
For the ANOVA of gender and robot type, there were sta- S1 Male r 0.008 0.181∗ 0.185∗
tistically significant effects regarding both gender and robot N 160 171 171
type on the scores of S1 and S3, although there was no in- Female r 0.160∗ 0.333∗∗∗ 0.282∗∗∗
teraction effects. It was revealed that the female respondents N 176 182 183
had more pronounced negative attitudes toward situations of S2 Male r 0.041 0.043 0.092
interaction with robots and lower negative attitudes toward N 160 171 171
emotions in interaction with robots, than the male respon- Female r 0.166∗ 0.106 0.136
dents. Moreover, the post–hoc tests with Tukey’s method N 176 182 183
revealed with a 5% significance level that the respondents in S3 Male r -0.020 -0.076 -0.135
N 160 171 171
the subgroups of “small–size humanoids” and “huge acting
Female r 0.138 0.102 -0.092
objects” had stronger negative attitudes toward situations of N 176 182 183
interaction with robots than those in the integrated subgroup
(∗ p < .05, ∗∗∗ p < .001)
consisting of “stationary objects,” “arm manipulators,” and
“others.” In addition, they also revealed that the respondents
in the subgroup of “small–size humanoids” had more pro-
S, and STAI–T for the female respondents, although that
nounced negative attitudes toward emotions in interaction
for the male respondents was low. Moreover, it also re-
with robots than those in the subgroups of “human–size hu-
veals that there was a low level of correlation among S1,
manoids” and “animals.” There were no statistically sig-
S2, and PRCA–24, although there was no such correlation
nificant effects of gender, robot type, or interaction in S2,
for the male respondents. There was no correlation among
negative attitude toward the social influence of robots.
S3, STAI, and PRCA–24.
For the ANOVA of gender and task type, there was a sta-
tistically significant effect of task type on the scores of S2.
The post–hoc tests with Tukey’s method revealed with a 5% Discussion
significance level that the respondents in the subgroup of The results of the social research mentioned in the previous
“tasks in places hard for humans to go or hazardous loca- section are based on Japanese respondents, that is, on one
tions” had stronger negative attitudes toward the social in- specific culture. Here, we carefully discuss the results’ im-
fluence of robots than those in the subgroup of “communi- plications, separating those limited to Japanese culture from
cation partner or playmates.” Moreover, there were statisti- results that may be generalized to other cultures, particularly
cally significant trends regarding task type in S1 and gender those related to alternation of attitudes toward robots.
in S2.
Implications from the Social Research Results
Correlations between NARS, STAI, PRCA–24: Third,
to investigate the relations between attitudes toward robots, The results for assumptions about robot type suggest a
general anxiety, and communication apprehension, we cal- bias of respondents toward humanoid–type robots. On the
culated Peason’s correlation coefficients r between the other hand, φ–coefficients between this type and assump-
NARS subscales, STAI, and PRCA–24. Since there is a tions about robot tasks suggest that “humanoid robots” were
possibility of gender difference with respect to anxiety and not strongly related to specific tasks. Moreover, the corre-
communication apprehension, this calculation was done for lations between “acting huge objects” and “guard or battle,”
each gender subgroup. between “animals” and “communication partners or play-
Table 4 displays these coefficients. The table reveals that mates,” and between “arm manipulators” and “construction
there was a moderate level of correlations among S1, STAI– or assembling tasks” suggest conservative images of robots
that have been reconstructed via the media. This trend is
2
No ANOVAs with robot type and task were conducted due to similar to the results by Nomura et al. (2005a). These sug-
the existence of cells in which the number of respondents was zero. gestions imply that more Japanese people are more biased
toward humanoid–type robots than other types, but are un- can extend our discussion to other cultures to some ex-
clear about what tasks this type of robot does. tent. While bias toward and relations between certain as-
The results of the ANOVAs for the NARS subscale sumptions about robots may be specific to each culture,
scores suggest that there the respondents assuming “small– what is important is that attitudes toward robots may de-
size humanoids” and “acting huge objects” had more pro- pend on specific assumptions dominant in a given culture.
nounced negative attitudes toward interaction with robots In other words, if the dominant assumptions about robots are
than those assuming “stationary objects,” “arm manipula- changed, the whole trend of attitudes toward robots can be
tors,” and “others.” We assume that the robot types “sta- altered in that culture. This may be caused by commercial-
tionary objects,” “arm manipulators,” and “others” lead peo- ization of really acting robots and media information about
ple to have conservative images of robots, such as being big them.
computer, industrial robots, and animated robots. Thus, this Joinson (2002) pointed out that people tend to have either
suggestion implies that novel types of robots or robots re- extremely positive or extremely negative attitudes toward
lated to battle evoke negative attitudes toward human inter- novel technologies. As mentioned in the previous section,
action with robots. people may have negative attitudes toward robots unfamil-
Moreover, the results of the ANOVAs suggest that the re- iar to their culture, but as information about robots spreads,
spondents assuming “small–size humanoids” had stronger their assumptions may change and attitudes toward them
negative attitudes toward emotions in interaction with robots may also alter.
than those assuming “human–size humanoids” and “ani- An important problem is that gender differences may af-
mals.” This implies that emotional reactions toward robots fect the alteration of attitudes toward robots. If there are cur-
are different between robot types, depending on interaction rently gender differences in attitudes toward robots and their
effects between design and size. relations to some personal traits in a culture, as implied in
In addition, the results of the ANOVA suggest that the the previous section, these differences may affect the nature
respondents assuming “tasks in places hard for humans to of attitude change toward robots; for example, males may
go or hazardous locations” were more negative toward the develop more positive attitudes toward humanoids whereas
social influence of robots than those assuming “communi- females in the same culture may come to have more nega-
cation partner or playmates.” We estimate that the image of tive attitudes toward them. Of course, the trend of attitude
danger in the former task assumption evoked negative atti- change may depend on the cultures.
tudes toward the social influence of robots performing such Furthermore, it is not clear which personal trait affects at-
tasks. titude change toward robots. Although currently there may
The ANOVA results also suggest that the female respon- be no strong relation between attitudes toward robots and
dents had more pronounced negative attitudes toward inter- communication apprehension in a given culture, as implied
action with robots and less negative attitudes toward emo- in the previous section, the increasing ability of robots, in
tions in interaction with robots than did the male respon- particular those related to communication, can change as-
dents. Furthermore, the correlation coefficients among the sumptions about robots, and as a result can change the con-
NARS, STAI, and PRCA–24 suggest that there is also a gen- nection between attitudes toward robots and personal traits
der difference regarding relations between negative attitudes related to communication.
toward robots and personal traits related to anxiety. This The above discussion merely focuses on a possibility. To
suggestion implies that there is a gender difference in atti- investigate it further, we should explore the psychological
tudes toward robots, depending on which factor we focuse relationships between assumptions about robots, attitudes
on in interaction with robots, and gender–based difference toward robots, and concrete emotions evoked in interaction
in their relations to some personal traits. with robots in more detail (Nomura et al. 2004). Moreover,
On the other hand, the correlation coefficients between we need to investigate cultural differences in assumptions
the NARS subscales and PRCA–24 suggest that there is a about robots.
low level of correlation between attitudes toward robots and
communication apprehension as a personal trait related to Summary
interaction. This suggestion is consistent with the exper-
imental results by Nomura, Kanda, & Suzuki (2006), in This paper reported the results of a social research project on
which there was only a low level of correlation between people’s attitudes toward robots by using the Negative Atti-
negative attitudes and communication avoidance behaviors tudes toward Robots Scale. The results revealed that atti-
toward robots. This implies that attitudes toward robots tudes toward robots are different depending on assumptions
are not directly connected to personal communication traits about robots, and there may be gender differences associ-
at the present, since robots such as humanoids are not yet ated with them. Based on these results, we discussed the
widespread in daily life and images of their tasks are not yet possibility of attitude changes toward robots, including the
fixed. influence of the changing ability and commercialization of
robots, media information about them, and personal traits.
Implications about Alternation of Attitudes toward
Robots References
Although the implications outlined in the previous section Bartneck, C.; Nomura, T.; Kanda, T.; Suzuki, T.; and Kato,
should be limited to Japanese culture in a strict sense, we K. 2005a. A cross–cultural study on attitudes towards
 robots. In Proc. 11th international conference on Human– Osgood, C. E., and Tannenbaum, P. H. 1955. The principle
Computer Interaction. (CD–ROM Proceedings). of congruity in the prediction of attitude change. Psycho-
Bartneck, C.; Nomura, T.; Kanda, T.; Suzuki, T.; and Kato, logical Review 62:42–55.
K. 2005b. Cultural differences in attitudes towards robots. Pribyl, C. B.; Keaten, J. A.; Sakamoto, M.; and Koshikawa,
In Proc. Symposium on Robot Companions (SSAISB 2005 F. 1998. Assessing the cross–cultural content validity of
Convention), 1–4. the Personal Report of Communication Apprehension scale
Chaplin, J. P., ed. 1991. Dictionary of Psychology. Dell (PRCA–24). Japanese Psychological Research 40:47–53.
Pub. Co., 2nd edition. Shibata, T.; Wada, K.; and Tanie, K. 2002. Tabulation
Friedman, B.; Kahn, P. H.; and Hagman, J. 2003. Hard- and analysis of questionnaire results of subjective evalua-
ware companions? – what online AIBO discussion forums tion of seal robot at Science Museum in London. In Proc.
reveal about the human–robotic relationship. In Proc. CHI 11th IEEE International Workshop on Robot and Human
2003, 273–280. Interactive Communication (ROMAN 2002), 23–28.
Heider, F. 1958. The Psychology of Interpersonal Rela- Shibata, T.; Wada, K.; and Tanie, K. 2003. Subjective
tions. New York: John Wiley & Sons. (Japanese Edition: evaluation of a seal robot at the national museum of science
Ohashi, M. 1978. Seishin). and technology in Stockholm. In Proc. Int. Workshop on
Ishiguro, H.; Ono, T.; Imai, M.; Maeda, T.; Kanda, T.; and Robot and Human Interactive Communication (RO–MAN),
Nakatsu, R. 2001. Robovie: an interactive humanoid robot. 397–407.
Int. J. Industrial Robot 28(6):498–503. Shibata, T.; Wada, K.; and Tanie, K. 2004. Subjective eval-
Joinson, A. N. 2002. Understanding the Psychology of uation of a seal robot in Brunei. In Proc. Int. Workshop on
Internet Behaviors: Virtual World, Real Lives. Palgrave Robot and Human Interactive Communication (RO–MAN),
Macmillan. (Japanese Edition: Miura, A., et. al., (2004). 135–140.
Kitaoji–shobo). Spielberger, C. D.; Gorsuch, R. L.; and Lushene, R. E.
Kahn, P. H.; Friedman, B.; Perez-Grannados, D. R.; and 1970. Manual for the state–trait anxiety inventory. Palo
Freier, N. G. 2004. Robotic pets in the lives of preschool Alto, California: Counseling Psychologists Press.
children. In Extended Abstracts of CHI 2004, 1449–1452. Suzuki, K.; Kashibuchi, M.; Sakamoto, A.; and Nagata, J.
Kaplan, F. 2004. Who is afraid of the humanoid? : Investi- 2002. Constructing an image scale for robots and investi-
gating cultural differences in the acceptance of robots. In- gating contents of measured images: Part 1. In Proc. the
ternational Journal of Humanoid Robotics 1(3):465–480. 66th Annual Meeting of the Japanese Psychological Asso-
ciation, 114. (in Japanese).
Kashibuchi, M.; Suzuki, K.; Sakamoto, A.; and Nagata, J.
2002. Constructing an image scale for robots and investi- Woods, S., and Dautenhahn, K. 2005. Child and adults’
gating contents of measured images: Part 2. In Proc. the perspectives on robot appearance. In Proc. Symposium on
66th Annual Meeting of the Japanese Psychological Asso- Robot Companions (SSAISB 2005 Convention), 126–132.
ciation, 115. (in Japanese).
Newcomb, T. M. 1953. An approach to the study of com-
municative acts. Psychological Review 60:393–404.
Nomura, T.; Kanda, T.; Suzuki, T.; and Kato, K. 2004.
Psychology in human–robot communication: An attempt
through investigation of negative attitudes and anxiety to-
ward robots. In Proc. the 13th IEEE International Work-
shop on Robot and Human Interactive Communication,
35–40.
Nomura, T.; Kanda, T.; Suzuki, T.; and Kato, K. 2005a.
People’s assumptions about robots: Investigation of their
relationships with attitudes and emotions toward robots. In
Proc. the 14th IEEE International Workshop on Robots and
Human Interactive Communication, 125–130.
Nomura, T.; Tasaki, T.; Kanda, T.; Shiomi, M.; Ishig-
uro, H.; and Hagita, N. 2005b. Questionnaire–based re-
search on opinions of visitors for communication robots
at an exhibition in Japan. In Proc. Tenth IFIP TC13 In-
ternational Conference on Human–Computer Interaction
(INTERACT’05), 685–698.
Nomura, T.; Kanda, T.; and Suzuki, T. 2006. Exper-
imental investigation into influence of negative attitudes
toward robots on human–robot interaction. AI & Society
20(2):138–150.

View publication stats