Role of Animism Tendencies and Empthy in Adult Education

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

The Role of Animism Tendencies and Empathy

in Adult Evaluations of Robots


Mako Okanda Kosuke Taniguchi Shoji Itakura
Department of Psychology Center for Baby Science Center for Baby Science
Otemon Gakuin University Doshisha University Doshisha University
Ibaraki, Osaka, Japan Kizugawa, Kyoto, Japan Kizugawa, Kyoto, Japan
mako.okanda@gmail.com kosuket314@gmail.com sitakura@mail.doshisha.ac.jp

ABSTRACT 1. Introduction
We investigated whether Japanese adults’ beliefs about friendship In the 21st century, robots have begun to speak and opportunities to
and morality toward robots differing in appearance (i.e., humanoid, interact with these new machines are increasing. Robots are now
dog-like, and egg-shaped) related to their animism tendencies and available to help people, including children and elderly individuals,
empathy. University students responded to questionnaires around our daily life [see also 11, 12]. Some people believe that
regarding three animism tendencies (i.e., general animism or a robots can be their friends or companion animals and should be
tendency to believe souls or gods in nonliving things, aliveness treated as moral beings, but others never feel this way. Why this
animism or a tendency to consider nonliving things as live entities, individual difference occurs? The present study investigated
and agentic animisms or a tendency to attribute biological, characteristics of people who are likely to believe that robots could
artifactual, psychological, perceptual, and naming properties) and be social agents (or living-like beings): the results should support
empathy. We found that friendship and morality were related to building suitable robots for different individuals.
slightly different animism tendencies and empathy even though Experience may be one of the possible factors to explain why one
they shared some major factors. Aliveness animism, as well as a treats a robot as a social agent while another does not. Konok,
tendency to attribute perceptual and name properties toward Korcsok, Miklósi, and Gácsi [19] reported that majority of naïve
robots, might be necessary for an individual to believe that robots people preferred real dogs over robot dogs, and they showed
could be social agents. Participants who responded that robots could negative attitudes toward robot dogs more than those toward real
be their friends showed a tendency to feel a soul in manmade objects dogs. On the contrary, investigators have reported that Aibo was
and a strong self-oriented emotional reactivity, whereas treated very similarly to a real dog by members of online Aibo
participants who answered that robots were moral beings showed communities who had owned Aibo for certain period [15]. Kahn et
a tendency to exhibit strong emotional susceptibility. We discuss al. [17] also reported that 9- to 15-year-old children who had
implications of these results and reasons why people feel that robots interacted with a humanoid robot previously stated in a subsequent
have a mind or consciousness. interview that the robot was a social agent that could be their friend
and it should not be psychologically harmed.
CCS CONCEPTS
• Applied computing → Law, social and behavioral sciences → However, experience cannot fully explain why there are individual
Psychology differences in evaluations of robots. Although majority people
preferred real dogs, some naïve people (12%) answered that robots
could be loved as much as real dogs [19]. Individual characteristics
KEYWORDS
might also be able to explain the individual differences. Some
Animism, empathy, human-robot interaction, robots’ perception
investigators have uncovered reasons for people’s negative
evaluations of robots; for example, personality traits (e.g.,
ACM Reference format:
Mako Okanda, Kosuke Taniguchi and Shoji Itakura. 2019. The role of neuroticism and anxiety) and other psychological variables,
animism tendencies and empathy in adult evaluations of robots. In including levels of distress, religion, and sensitivities to animal
Proceedings the 7th International Conference on Human-Agent Interaction reminders, predicted one’s tendency to feel androids, defined as
(HAI’19), October 6–10, 2019, Kyoto, Japan. ACM, NY, NY, USA, 8 pages. robots with extremely humanlike appearance, uncanny [22].
https://doi.org/10.1145/3349537.3351891
To date, limited studies have identified which people evaluate
Permission to make digital or hard copies of all or part of this work for personal or robots positively. Animistic beliefs could be one of the other
classroom use is granted without fee provided that copies are not made or distributed
for profit or commercial advantage and that copies bear this notice and the full citation
possible factors to explain why one treats robots as social agents
on the first page. Copyrights for components of this work owned by others than ACM while others do not. In the other words, it is possible that people
must be honored. Abstracting with credit is permitted. To copy otherwise, or republish, who are likely to believe that robots are social agents could have
to post on servers or to redistribute to lists, requires prior specific permission and/or a
higher animism tendencies. In fact, Chikaraishi, Yoshikawa, Ogawa,
fee. Request permissions from Permissions@acm.org.
HAI ‘19, October 6–10, 2019, Kyoto, Japan Hirata, and Ishiguro [4] examined characteristics of people who
© 2019 Association for Computing Machinery. sensed a mind or emotions in an android that appeared in a real
ACM ISBN 978-1-4503-6922-0/19/10…$15.00
https://doi.org/10.1145/3349537.3351891
stage play. They reported that this perception was related to a We tested level of empathy and three animism tendencies in
tendency toward animism. relation to participants’ evaluations of robots. As noted above, one
was the construct of general animism as tested by the Animism Scale
The term “animism” has been defined in different ways in the
for Adults [9] and employed by Chikaraishi et al. [4], which includes
literature. Animism was most commonly been defined in the fields
three factors: 1) soul in nature (i.e., whether one believes in souls or
of cultural anthropology and theology as the tendency to believe in
gods in nature); 2) other self (i.e., whether one feels souls or spirits
the existence of gods, spirits, or souls in nature, personal
in others’ personal belongings); and 3) anthropomorphism (i.e.,
belongings, or manmade objects [see more in 9, 31]. This type of
whether one treats their own personal belongings as living
animism could be defined as general animism. In addition to
creatures). Second, we investigated aliveness animism, which we
animism in adults, Piaget [29] indicated that children are likely to
defined as one’s tendency to categorize ambiguous things (e.g.,
believe that nonliving moving things, such as clouds, the sun, or
moving objects or plants) as alive. This is more like animism
clocks, are actually alive. This second animism may reflect a
tendency Piaget [29] originally pointed out; however, it is not like a
cognitive error exhibited by young children due to unconsolidated
simple error when it was produced by adults who should have
biological knowledge about living and nonliving things. This type
enough knowledge of living and nonliving things. A previous study
of animism could be defined as animism error. However, animism
has indicated that plants were less likely to categorize as living
may not fade away completely in adulthood [e.g., 9, 21], especially
things even by adults [7]; therefore, individuals who are more likely
when an individual is asked to make a judgment about robots.
to believe that plants are alive should have stronger aliveness
Robots are perceived as intermediate between living and non-living
animism. Last, agentic animism, defined as whether one would
things [14], perhaps because they have faces and living-like
sense a mind in robots, was evaluated. In the present study, “mind”
movements [see also 13] and/or cognitive process similar to humans
means one’s tendency to attribute living-like properties, namely
[26]. Therefore, it is not surprising that some adults with developed
perceptual or psychological properties (or intensions) to robots, or
biological knowledge treat robots as if they are alive [15] or
even give the robot a name. Agentic animism should differ from the
anthropomorphized robots in various ways [5]. This tendency, or
other two types of animism since it requires one to sense a target as
belief, that robots and other objects are intentional agents that have
an intentional agent that has psychological features, whereas
psychological features could be considered a third type of animism
general animism and aliveness animism only focus on whether an
or agentic animism.
individual attributes living-like forms or behavioral characteristics
Another possible factor that may determine whether an individual to a target, and do not always require to attribute psychological
believes that robots are social agents is their level of empathy. That features.
is, one who is likely to treat robots as if they are alive should also
Agentic Animism
understand others’ emotions or point of view well. Chikaraishi et (Perceptual, Psychological,
al. [4] examined this issue with an android; however, they did not Empathy Naming)
find any relationship between people’s tendency to sense a mind or (Fantasy)

feel emotions in the android and their level of empathy. However,


this research used an android as stimuli, and there are other types Friendship Moral
of robots that have different appearances. People could have
different evaluations of these robots. For example, Gray and Empathy
Wegner [8] reported that American adults evaluated a robot with a (Other-/Self-
Oriented Emotional
human-like face as more uncanny than a mechanical appearing Aliveness Animism
Reactivity
robot without a human-like face. Li, Patrick, and Li [20] reported
that Chinese, Korean, and German adults rated a dog-like robot Figure 1. Hypotheses in the present study: possible factors that
more positively than a machine-like robot. Thus, it remains unclear explain ones’ tendency to believe that robots are social agents.
whether or not individuals who are likely to believe or treat other
types of robots as social agents have strong levels of empathy. We have three hypotheses in the present study (Figure 1). First,
individuals who believe that robots could be their friends and
In the present study, we investigated whether people who are likely should be treated as moral beings would be likely to attribute
to believe that robots are social agents (i.e., a belief that robots can psychological, perceptual, and naming properties to robots because
be their friends or that robots should be treated as moral beings) we usually consider these properties for friends or living beings
have higher animism tendencies and stronger empathy than people including animals. Second, higher level of fantasy in empathy would
who do not. Because a robots’ appearance could affect this belief, explain friendshipness toward robots, and higher level of other-
we used three different kinds of robots in the present study: a oriented and lower level of self-oriented emotional reactivities
humanoid robot, a dog-like robot, and a white egg-shaped robot that would explain morality toward robots. Third, people would be more
has animated eyes. Since we are interested in people’s natural likely to believe that humanoid and dog-like robots that have living-
impressions of real robots, we chose robots that were commercially like appearances are social agents than egg-shaped robot.
available on the consumer market. Humanoid and dog-like robots
were used as representative of living beings (i.e., humans and dogs)
and an egg-shaped robot was used because its appearance was very
different from real living animals.
2. Method We included objects with faces or motion because these features
would make both children and adults to judge these objects as living
2.1 Participants things [e.g., 13]. We also included plants because these were less
Fifty Japanese university students from 18 to 21 years of age (M = likely to be categorized as living things by both adults and children
18.90, SD = .735, 24 males and 26 females) participated in the present [e.g., 1, 6, 7, 28]. Participants marked the name of the object if they
study. Participants were given questionnaires at the end of their believed it was alive.
classes (experimental psychology), and all of them responded
voluntarily. Two participants (both males) did not complete Table 1. The Animism Scale for Adults used in this study, adapted
questionnaires and were excluded from further analysis. The design and translated from Ikeuchi [9].
and purpose of the study were reviewed by the ethical approval Factors Questions
committee of Otemon Gakuin university and were explained to the Soul in Nature I believe that a god is living in large old rocks and trees.
participant. The participants completed all questionnaires when I believe that a god is living in the ocean and the mountains.
they agreed to participate the study. I believe that we should not cut large old trees or we will be cursed.
Other Self I believe that a handmade item may have the creator’s soul.
2.2 Materials I believe that relics may have the former owner’s soul.
I believe that used clothes or tools may have the former owner’s soul.
2.2.1 Target Robots and individuals’ evaluations of these robots. Personification I sometimes feel attached to my personal belongings.
Three robots that differed in appearance were used in the present I sometimes name my personal belongings.
study. Participants were shown black and white photographs of a I sometimes feel sorry when I throw my personal belongings away.
humanoid robot (Toyota’s KIROBO mini), a dog-like robot (Sony’s I sometimes feel my personal belongings are actually myself.
I believe that personal belongings have a mind like we have.
aibo), and an egg-shaped robot (MJI’s Tapia). Participants were
asked a question about friendship (“Do you want to be a friend of
this robot?”), and a question about morality (“Is it okay to hit this Table 2. The Multidimensional Empathy Scale (MES) used in this
robot?”). These questions were previously used by Melson et al. [25] study, adapted and translated from Suzuki and Kino [30].
and Kahn et al. [16]. Factors Questions
Other-Oriented I like to comfort someone who is feeling sad.
2.2.2 Agentic questions (agentic animism). For each robot,
Emotional I cannot share my friends' distress.
participants answered 11questions for the three types of robots: A Reactivity (OO)
I do not feel pity when others fail at something.
naming question (“Is it okay for this to have a name?”) and four
When I see or hear that someone is trying very hard, I want to support
property projection questions, as used in Jipson and Gelman [13], him/her even if it doesn’t concern me.
and questions about additional property. The property projection When someone near me is getting into trouble, I hope he/she will solve the
problem as soon as possible.
questions included biological (“Does this one eat?” and “Does this
Self-Oriented I do not want to repeat others' failures.
one grow?”), perceptual (“Can this one see things?” and “If I tickled Emotional When I see someone who has a serious problem, I am relieved that it is not
this one, would this one feel it?”), psychological (“Can this one Reactivity (SO) my problem.
think?” and “Can this one feel happy?”), and artefactual (“Did a I am impatient when I hear about others' success.

person make this one?” and “Can this one break?”) properties, as I sometimes feel unhappy when I hear about someone's success.
Emotional I am likely to agree with the people around me.
well as questions about aliveness (“Is this one alive?” and “Can this
Susceptibility (ES) I do not change my beliefs or opinions based on my friends.
one die?”). Participants marked “yes” or “no” to each question.
I cannot decide things myself.
2.2.3 The Animism Scale for Adults (general animism). We used My feelings can be changed by the people around me.

the Animism Scale for Adults [9] to measure participants’ general I do not change my feelings when they are different from others' feelings.
Fantasy (F) When I read nice stories or novels, I imagine that I am also in the stories.
animism beliefs (Table 1). It comprises 11 questions in three factors:
I never imagine that things that happen in novels are my business.
soul in nature, other self, and, anthropomorphism. The exact wording
I like to fantasize.
of all items in the Animism Scale for Adults is provided in Table 1. I repeatedly daydream about or imagine my future.
Participants responded to each item on a five-point scale from 1 (not I have a hard time coming back to reality after watching a nice film.
at all) to 5 (very likely). Perspective I try to understand someone who has different opinions from mine.
Taking (PT) I try to understand my opponent’s point of view.
2.2.4 Living/nonliving distinction (aliveness animism). Participants
I listen to others while thinking about their implications.
were presented with a list of words that included eight objects (lit I always put myself in others' shoes.
candle, phone, clock, doll, teddy bear, banana peel, cloud, and mud) I do not care about others' points of view when I criticize them.
and four plants (tree, flower, grass, and vegetable) and were asked to
choose the living things from among these ambiguous targets. Some
of the objects were used in the studies of Ikeuchi [9] and MacDonald 2.2.5 Multidimensional Empathy Scale (MES). We used the
and Stuart-Hamilton [24]. We used categorical names of plants to Multidimensional Empathy Scale (MES) to evaluate participants’
prevent participants from responding on the basis of personal ability to feel empathy [30]. The MES, shown in Table 2, includes 24
preference for particular plants or flowers (unlike Inagaki and questions with the following five factors: other-oriented emotional
Hatano, [10], who tested children’s recognition of plants and reactivity (OO), self-oriented emotional reactivity (SO), emotional
animals by using specific names such as tulip). susceptibility (ES), fantasy (F), and perspective taking (PT).
Participants responded to each item on a five-point scale from 1 (not from 3 to 15; other self scores ranged from 3 to 15; anthropomorphism
at all) to 5 (very likely). scores ranged from 5 to 25).

2.3 Procedures Table 4. Average of scores in general animism and empathy


All participants answered questionnaires in the following order: 1)
agentic questions for a humanoid robot; 2) agentic questions for a
dog-like robot; 3) agentic questions for an egg-shaped robot; 4)
living/nonliving distinctions; 5) the Animism Scale for Adults, and
6) the Multidimensional Empathy Scale (MES) because all
participants answered paper-based questionnaires in a group. The
agentic questions were asked along with friendship and morality
questions. Photographs of the robots appeared one at a time on a
sheet of a paper. A within-subjects design was used: all participants
answered questions about three robots.
3.3 Living/nonliving distinction (aliveness
animism)
3. Results
The number of marks given by a participant was counted and
3.1 Agentic animism questions summed to produce an aliveness animism score (ranging from 0-
Preliminary, we examined whether participants’ responses to 12). The participants’ mean aliveness animism score was 3.54 (SD =
agentic question for three robots were different from each other. 1.584). Most of the participants marked plants as alive, and few
Each “yes” response received a score of 1 and each “no” response participants chose manmade objects or natural things (two for a lit
received a score of 0 for the aliveness, biological, psychological, candle, three for a doll, one for a teddy bear, five for a banana peel,
perceptual, and naming questions. Each “no” response received a three for a cloud, and two for mud) as alive. As noted in
score of 1, and each “yes” response received a score of 0 for the introduction, the results might indicate that an aliveness animism
artifact questions. For each participant, scores were summed to in the present study was almost equal to a tendency to believe that
produce an agentic score for each robot (ranging from 0-11). We plants are alive.
conducted one-way ANOVA for mean agentic scores of the three 3.4 MES (Multidimensional Empathy Scale)
robots; however, there was no main effect, F (2, 94) = 1.90, p =.16, η2
= .04. We concluded that the participants’ agentic animism Table 4 summarizes participants’ mean scores for sub-component
tendency toward the three robots were similar and further analyses factors of MES. The scores were summed to produce empathy
combined all three robots’ scores together. scores for each factor. Scores for each factor ranged from 5 to 25,
with the exception of self-oriented emotional reactivity that
We calculated mean scores of each property for each robot, as included four questions ranging from 4 to 20.
shown in Table 3. For these questions, all “yes” responses received
a score of 1 and all “no” responses received a score of 0 according to
the procedures of Jipson and Gelman [13]. Because naming Table 5. Correlations of three animism tendencies
questions included one question, scores for this question ranged Agentic Aliveness General Empathy
from 0 to 1, whereas scores for other properties that included two Animism Animism Animism
questions ranged from 0 to 2. Aliveness Animism .07 −
General Animism .07 .27 −
Empathy -.16 .03 .48 −
Table 3. Mean agentic scores on each property for each robot.
Parentheses indicate standard deviation.
Robot types Live/Alive Artifact Biological Psychological Perceptual Name 3.5 Correlation between three animism
Humanoid 0.35 (0.53) 2.00 (0.00) 0.25 (0.53) 0.56 (0.65) 0.90 (0.42) 0.81 (0.39)
Dog-like 0.31 (0.51) 1.96 (0.20) 0.21 (0.50) 0.42 (0.61) 0.94 (0.48) 0.81 (0.39) tendencies and empathy
Egg-shaped 0.29 (0.50) 1.90 (0.31) 0.21 (0.46) 0.46 (0.54) 0.84 (0.53) 0.77 (0.42)
We analyzed whether there were any correlations between the
three animism tendencies and empathy to confirm agentic animism
was independent from other measures as we noted in introduction.
3.2 The animism Scale for Adults (general We summed sub-component scores of each measure to reveal a
animism) general animism score and an empathy score along with the agentic
Table 4 summarized participants’ mean scores for sub-component animism and the aliveness scores. Pearson correlation analyses
factors of general animism. The scores were summed to produce revealed that the general animism score and the aliveness animism
general animism scores for each factor (soul in nature scores ranged score showed a weak positive correlation, r (46) = .27, p = .07, and
the general animism score and the empathy score showed a
significant positive correlation, r (46) = .48, p < .01 (Table 5). score in general animism (β= 0.39, p < .05), the aliveness animism
Therefore, we confirmed that agentic animism was independent of score (β= 0.33, p < .01), and the SO score in empathy (β= -0.45, p <
other factors. .001). Marginally significant effects were found for the artifact score
in agentic animism (β= -0.19, p < .1). For the model of the moral
3.6 Path analysis score, the naming score in agentic animism (β= -0.29, p < .05), the
More than half of the participants responded that they wanted to be soul in nature score in general animism (β= -0.21, p < .05), the
a friend of the robots or they should not hit them (Friendship: 32 vs. aliveness animism score (β= -0.24, p < .05), and the ES score in
16 for the humanoid, 29 vs. 19 for the doglike robot, and 36 vs. 12 empathy (β= -0.27, p < .05) were significantly influenced. The effect
for the egg-shaped robot; Moral: 38 vs. 10 for the humanoid, 38 vs. of the perceptual score in agentic animism (β= 0.22, p < .1) was
10 for the doglike robot, and 39 vs. 9 for the egg-shaped robot). marginally significant, whereas the PT score in empathy was not
significantly influenced. The correlation between the friendship and
We conducted a path analysis to investigate whether the the moral scores was omitted from the best fit model.
participants’ evaluations of robots (friendship and moral) could be
explained by their animism tendencies and empathy by using sem
4. Discussion
package in R. The path model involved the friendship and moral
In the present study, we examined the characteristics that
scores as dependent variables, and the agentic animism (naming,
determine adults who are likely to believe that different types of
biological, perceptual, psychological, artefact, and aliveness),
robots can be their friends or should be treated as moral beings. We
general animism (soul in nature, other self, and
measured participants’ three different animism tendencies, such as
anthropomorphism), aliveness animism, and empathy (OO, SO, ES,
general animism, aliveness animism, and agentic animism, as well
F, PT) scores as predict variables. We conducted a backward
as level of empathy.
stepwise method to get the best fit model. First, we estimated a
saturated model which involved the causal paths of all 15 predicted We proposed that there are different kinds of animism tendencies.
variables to two dependent variables (i.e. friendship and moral) and One type is general animism, or a tendency to believe in gods, souls,
correlation paths among predict variables (total 105 correlations), or spirits in both nature and manmade objects. This term has been
and the correlation path between two dependent variables. The path used in cultural anthropology or theology [9, 31]. A second type is
with the highest p value deleted from the model and re-estimated aliveness animism, or a tendency to believe that ambiguous things
the reduced model. We conducted this reduced procedure until all are alive, which is a tendency often observed in young children [29]
paths were significant. We chose the lowest AIC model from the all as well as some adults [e.g., 9, 21]. Children mistakenly believe that
model estimated by the backward stepwise method as the best nonliving things are alive while adults are less likely to include
fitting model, which is summarized in Figure 2 (χ2 (83) = 44.481, p = plants in living category [7]. We confirmed that people who are
.999 , GFI = .904, NFI = .861, CFI = 1.000, RMSEA= 0.000, AIC = likely to categorize plants as alive in the present study might have
184.482). a higher aliveness animism tendency. The last possible animism is
agentic animism, or a tendency to sense a mind in ambiguous things
such as robots or to treat these things as social agents that have
psychological features [e.g., 13, 18]. We suggested that agentic
animism is the only type of animism that requires one to interpret a
target as an intentional agent, whereas the other two animisms
involve attributing human/animal-like forms or behavioral
characteristics onto a target and do not always require attributing
psychological features. The present results supported this idea:
agentic animism was not correlated with general or aliveness
animisms (Table 5).
As we hypothesized, people’s belief that robots are social agents was
explained by their strong agentic animism, namely a tendency to
attribute perceptual and naming properties to robots; however,
Figure 2: Summary of the best fitting model in path analysis. psychological properties did not relate to this tendency directly. The
The one-side arrows indicate causal effects from starting result might indicate that degree to attribute psychological property
variable to end variable and two-sided arrows show to robots (i.e. thinking and feeling happy) could discriminate agentic
correlations between the variables. The solid arrows indicate animism and an animism error. That is, adults who know that
the positive effects and dashed arrows indicate the negative robots (or nonliving manmade objects) cannot have such
effects. To make high visibility of this figure, the coefficients psychological property still expect that robots behave as if these
of correlation are abbreviated. have this property, whereas children might really believe that
robots can think and feel happy.
The best fitting model showed that the friendship score was
significantly influenced by the perceptual (β= 0.23, p < .05) and In addition, people who believed that robots were social agents were
naming (β= 0.22, p < .05) scores in agentic animism, the other self likely to give a name to these. Giving a name could be an index of
one’s positive attitudes toward robots. Moreover, aliveness animism
or a tendency to believe that plants were alive also explained from family dinner?” or “Can you put this one into closet even this
people’s belief that robot are social agents. We did not find common one did not want to?” [see also 17]
sub-component factors of empathy that explain people’s belief that
It should be noted that some questions in different animism
robots are social agents.
measures used similar questions in the present study; for example,
Unlike to our hypothesis, the participants similarly responded to the aliveness animism asked participants to choose one that was alive,
agentic questions toward three different robots. The robots used in and aliveness questions in agentic animism asked participants “Is
this study had a face-like shape (i.e., eye-like or mouth-like shape). this one alive?” and “Can this one die?” In the case of agentic
As Jipson & Gelman [13] suggested, facial features might be animism, most of the participants answered “no” to the former
important for us to attribute agentic properties to robots. Further question; however, more participants answered “yes” to the latter
studies are required to examine whether an individual’s evaluation questions (3 vs. 14 for humanoid, 2 vs. 13 for dog-like, and 2 vs. 12
differs on the basis of a robot’s facial features, such as smart for egg-shaped robots). That is, the adults might be basically aware
speakers or cleaning robots, to answer a question whether people that robots are not alive; however, some adults answered that robots
sense a mind in various robots. could die as a figuratively speaking.
Friendship and morality were related to slightly different animism Lastly, the majority of Japanese participants in the present study
tendencies and empathy. The results supported our hypotheses tended to believe that robots were social agents that could be their
partly. Participants who believed that robots could be their friends friends or should be treated as moral beings. Moreover, their scores
showed a tendency to feel a soul in manmade objects (i.e., other self of psychological and perceptual properties were slightly higher
in general animism) and reported weak self-oriented emotional than those reported in American adults in a study by Jipson and
reactivity. Robots are manmade objects; therefore, people who Gelman [13] that used a dog-like robot in a video. Their robot was
believed that robots coulf be their friends were likely to believe that different from aibo, which we used in the present study, so we
manmade objects could have a soul, and they also had a trait to take cannot treat both results as equivalent but there could also be some
care of others ‘emotion. However, fantasy did not explain this cultural differences in terms of evaluations of robots. Nomura et al.
tendency. Treating robots as social agents might be different from [27] reported that Japanese university students were likely to
fantasy or tendency to project oneself into fictional world. believe that robots can make decisions by themselves and have the
capacity to feel emotions as humans do. Japan is a country in which
Participants who believed that the robots should not experience
interactive robots are everywhere -- at home, in shopping malls for
psychological or physical pain reported strong emotional
entertainment, in films, and in cartoons -- and our results may
susceptibility along with a tendency to believe gods or souls in
reflect this fact. However, some studies have reported that Japanese
natures. People who believed that robots were moral beings were
people do not evaluate robots more positively than people in other
likely to consider others’ feeling or opinions and to respect natures.
countries [2, 23]. Additional studies are in need of replication in
These results suggest that individual differences exist in terms of other countries to investigate whether similar results would be
the degree to which friendship and morality can be attributed to obtained.
robots. Chikaraishi et al. [4] did not find a relationship between
There are some limitations in the present study. First, we only
participants’ level of empathy and their tendency to sense a mind
showed the participants robots using photos. It is possible that their
or emotions in an android. However, in the present study, we asked
evaluations would be different if they were observing real robots.
the participants about their belief regarding friendship and morality
Second, we only tested undergraduate students and, thus, it is still
towards three robots and these more general beliefs, but not toward
unclear whether the present results can be extended to other age
one certain android on a stage, were related to empathy. It is also
groups, such as elderly individuals or children.
possible that agentic animism (or a tendency to sense a mind or
emotions in targets) might not be related to empathy as Chikaraishi 5. Conclusion
et al. [4]; however, a belief that robots could be social agents in term
of friendship and morality issues may be related to it. In this study, we investigated whether animism tendencies and
empathy, instead of experience with robots, can explain individual
It is not surprising that people who were likely to believe robots as differences in positive evaluations of robots (i.e., friendship and
social agents sensed a sort of mind in them. However, the morality morality toward robots). In a sample of Japanese university
question that was asked of the participants (e.g., “It is okay to hit students, we confirmed that animism tendencies and empathy may
this one”) made it difficult to distinguish whether the participants be important factors that influence an individual’s belief that robots
answered “no” because it would harm them psychologically, are social agents. Aliveness animism or a tendency to believe
because they would feel sorry for the robot, or because they were ambiguous things as alive, agentic animism or a tendency to
concerned that the valuable robot could be physically destroyed. attribute perceptual and name properties to robots, and some
The participants who said “no” to this question showed higher factors of empathy explained individuals’ belief of friendship and
emotional susceptibility with regard to empathy, so it is possible morality towards robots. We also suggested that there are three
that they believed that they should avoid harming robots different animism tendencies: agentic animism, or a tendency to
psychologically. Further studies should focus on asking other perceive robots as intentional agents, might be different from
morality-based questions, such as “Is it is okay to exclude this one general or aliveness animism.
In the near future, socially interactive robots are likely to be more Social Psychology, 25 (3). 167-177. 10.14966/jssp.KJ00006203282
widespread than today. They may be able to help people by working [10] Kayoko Inagaki and Giyoo Hatano. 1996. Young children’s
in schools, hospitals, or nursing homes [see also 11, 12]. The present recognition of commonalities between animals and plants. Child
results suggested that people’s positive evaluations of robots may Development, 67 (6). 2823-2840. 10.1111/j.1467-8624.1996.tb01890.x
vary based on their personal trait or beliefs. This idea could help to [11] Shoji Itakura. 2008. Development of mentalizing and communication:
build a robot that can avoid someone who may harm them or that From viewpoint of developmental cybernetics and developmental
can seek a help to a right person when it is bullied. In fact, Brsci, cognitive neuroscience. IEICE TRANS. COMMUN, E91-B. 2109-2117.
Kidokoro, Suehiro, and Kanda [3] reported that children often abuse [12] Shoji Itakura, Hiraku Ishida, Takayuki Kanda, Yohko Shimada,
robots when they are separate from their parents. Robots can seek Hiroshi Ishiguro and Kang Lee. 2008. How to build an intentional
a help for adults in this situation. android: Infants’ imitation of a robot’s goal-directed actions. Infancy,
13 (5). 519-532. 10.1080/15250000802329503
Moreover, the present results could help to make effective robots
[13] Jennifer L. Jipson and Susan A. Gelman. 2007. Robots and rodents:
that can teach something new to us. In other words, robots could be
Children’s inferences about living and nonliving kinds. Child
“good teachers” if a person believes that robots are social agents, Development, 78 (6). 1675-1688. 10.1111/j.1467-8624.2007.01095.x
whereas robots’ teaching would not be effective for a person who
[14] Jennifer L. Jipson, Selin Gülgöz and Susan A. Gelman. 2016. Parent-
never believe that robots are social agents. It is important to
child conversations regarding the ontological status of a robotic dog.
understand individual differences in the evaluation of robots in Cognitive Development, 39. 21-35. org/10.1016/j.cogdev.2016.03.001
order to choose suitable robots in these circumstances. Our study is
[15] Peter H. Kahn, N. G. Freier, B. Friedman, R. L. Severson and E. N.
a first step in understanding such individual tendencies.
Feldman. 2004. Social and moral relationships with robotic others? in
RO-MAN 2004. 13th IEEE International Workshop on Robot and Human
ACKNOWLEDGMENTS Interactive Communication (IEEE Catalog No.04TH8759), 545-550.
We are grateful to students who participated in this study. This 10.1109/roman.2004.1374819
research was supported by a research fellowship from Center for [16] Peter H. Kahn, Batya Friedman, Deanne R. Perez-Granados and
Baby Science, Doshisha University. Nathan G. Freier. 2006. Robotic pets in the lives of preschool children.
Interaction Studies, 7 (3). 405-436. 10.1075/is.7.3.13kah

REFERENCES [17] Peter H. Kahn, Takayuki Kanda, Hiroshi Ishiguro, Nathan G. Freie,
Rachel L. Severson, Brian T. Gill, Jolina H. Ruckert and Solace Shen.
[1] Andrea G. Backscheider, Marilyn Shatz and Susan A. Gelman. 1993. 2012. “Robovie, You’ll have to go into the closet now”: Children’s
Preschoolers’ ability to distinguish living kinds as a function of social and moral relationships with a humanoid robot. Developmental
regrowth. Child Development, 64 (4). 1242-1257. 10.1111/j.1467- Psychology 48 (2). 303-314. 10.1037/a0027033.
8624.1993.tb04198.x
[18] Nobuko Katayama, Junichi Katayama, Michiteru Kitazaki and Shoji
[2] Christoph Bartneck, Tomohiro Suzuki, Takayuki Kanda and Tatsuya Itakura. 2010. Young children’s folk knowledge of robots. Asian
Nomura. 2007. The influence of people’s culture and prior Culture and History, 2 (2). 111-116. 10.5539/ach.v2n2p111
experiences with Aibo on their attitude towards robots. AI & SOCIETY,
21 (1). 217-230. 10.1007/s00146-006-0052-7 [19] Veronika Konok, Beáta Korcsok, Ádám Miklósi and Márta Gácsi. 2018.
Should we love robots? --The most liked qualities of companion dogs
[3] Drazen Brsci, Hiroyuki Kidokoro, Yoshitaka Suehiro and Takayuki and how they can be implemented in social robots. Computers in
Kanda. 2015. Escaping from Children’s Abuse of Social Robots Human Behavior, 80. 132-142. org/10.1016/j.chb.2017.11.002
Proceedings of the Tenth Annual ACM/IEEE International Conference on
Human-Robot Interaction, ACM, Portland, Oregon, USA, 59-66. [20] Dingjun Li, P. L. Patrick and Ye Li. 2010. A cross-cultural study: Effect
of robot appearance and task. International Journal of Social Robotics,
[4] T. Chikaraishi, Y. Yoshikawa, K. Ogawa, O. Hirata and H. Ishiguro. 2 (2). 175-186. 10.1007/s12369-010-0056-9
2017. Creation and staging of android theatre “Sayonara” towards
developing highly human-like robots. Future Internet 2017, 9, 75. , 9. 75. [21] William R. Looft. 1974. Animistic thought in children: understanding
10.3390/fi9040075 of “living” across its associated attributes The Journal of Genetic
Psychology, Routledge, 235-240.
[5] Nicholas Epley, Adam Waytz and John T. Cacioppo. 2007. On seeing
human: a three-factor theory of anthropomorphism. Psychological [22] Karl F. MacDorman and Steven O. Entezari. 2005. Individual
review, 114 (4). 864-886. 10.1037/0033-295x.114.4.864 differences predict sensitivity to the uncanny valley. Interaction
Studies, 16 (2). 141-172. org/10.1075/is.16.2.01mac
[6] Nathalie Fouquet, Olga Megalakaki and Florence Labrell. 2017.
Children’s understanding of animal, plant, and artifact properties [23] Karl F. MacDorman, Sandosh K. Vasudevan and Chin-Chang Ho.
between 3 and 6 years. Infant and Child Development, 26 (6). e2032. 2009. Does Japan really have robot mania? Comparing attitudes by
10.1002/icd.2032 implicit and explicit measures. AI & SOCIETY, 23 (4). 485-510.
10.1007/s00146-008-0181-2
[7] Robert F. Goldberg and Sharon L. Thompson-Schill. 2009.
Developmental “roots” in mature biological knowledge. Psychological [24] Lorraine McDonald and Ian Stuart-Hamilton. 2000. The meaning of
Science, 20 (4). 480-487. 10.1111/j.1467-9280.2009.02320.x life: Animism in the classificatory skills of older adults. The
International Journal of Aging and Human Development, 51 (3). 231-
[8] Kurt Gray and Daniel M. Wegner. 2012. Feeling robots and human 242. 10.2190/825y-gwat-9bm8-g5tr
zombies: Mind perception and the uncanny valley. Cognition, 125.
125-130. 10.1016/j.cognition.2012.06.007 [25] Gail F. Melson, Peter H. Kahn, Alan M. Beck, Batya Friedman, Trace
Roberts and Erik Garrett. 2005. Robots as dogs?: children’s
[9] Hiromi Ikeuchi. 2010. Animistic thinking in adults: The memorial
interactions with the robotic dog AIBO and a live australian shepherd
service for dolls as a voluntary loss (in Japanese) Japanese Journal of
CHI ‘05 Extended Abstracts on Human Factors in Computing Systems, living things concept: A microgenetic analysis of conceptual change
ACM, Portland, OR, USA. in basic biology. Cognitive Psychology, 49 (4). 301-332.
org/10.1016/j.cogpsych.2004.01.002
[26] T. A. Mikropoulos, P. Misailidi and F. Bonoti. 2003. Attributing
human properties to computer artifacts: Developmental changes in [29] Jean Piaget. 1929. The child’s conception of the world. Routledge and
children’s understanding of the animate-inanimate distinction, . Kegan Paul, London.
Psychology / Ψυχολογία, 10 (1). 53-64. [30] Yumi Suzuki and Kazuyo Kino. 2008. Development of the
Multidimensional Empathy Scale (MES) : Focusing on the distinction
[27] Tatsuya Nomura, Tomohiro Suzuki, Takayuki Kanda, Jeonghye Han,
between self- and other-orientation. The Japanese Journal of
Namin Shin, Jeniffer Burke and Kensuke Kato. 2008. What people
Educational Psychology, 56 (4). 487-497. org/10.5926/jjep1953.56.4_487
assume about humanoid and animal-type robots: cross-cultural
analysis between Japan, Korena, and the United States International [31] Edward Burnett Tylor. 2010. Primitive culture: Researches into the
Journal of Humanoid Robotics, 05 (01). 25-46. development of mythology, philosophy, religion, art, and custom.
10.1142/s0219843608001297 Cambridge University Press, Cambridge.
[28] John E. Opfer and Robert S. Siegler. 2004. Revisiting preschoolers’

You might also like