1 Introduction
The influence of humans’ physical appearances on other humans has been displayed in previous psychology studies (Mims et al.
1975; Walster et al.
1966). Similarly, the physical appearances of robots have a significant effect as well. In fact, it has been shown that android-type robots (robots that look exactly like humans) are preferred as communication partners to mechanical-type robots (robots that are similar to humans in figure but have a metallic or mechanical appearance) and humans in sports/exercise situations(Suzuki et al.
2021a,
b). Other studies have also mentioned the effect of the physical appearances of robots in many situations (Goetz et al.
2003; Hinds et al.
2004; Nishio et al.
2012; Abubshait and Wiese
2017; Suzuki et al.
2021a,
b). Thus, it is important to consider the physical appearance of robots when introducing them into our daily lives because their appearance can be an important factor when facilitating interaction with humans.
There are numerous aspects to physical appearance; the one that influences gender assignment cannot be ignored. Robots are characterized by their physicality, which makes it easy for humans to assign gender arbitrarily. Therefore, the problems associated with gender-specific properties (so-called ‘gendered robots’) should be carefully examined and discussed. The interaction between robots and humans in daily life applications should be considered to determine whether humans prefer robots with the same or a different gender. In addition to existing studies suggesting that males generally express positive responses towards robots more than females do (Nomura et al.
2006,
2008; Kuo et al.
2009; Lin et al.
2012), the effect of the gender of robots on human psychological and behavioral reactions toward these robots has been examined. For example, a comparison experiment between a mechanical humanoid robot and an android with a female appearance indicated that university students preferred female robots for domestic use (Carpenter et al.
2009).
Furthermore, previous studies have indicated that men and women tend to prefer robots with female features and male features, respectively, suggesting cross-gender effects. For example, in an experiment in which a humanoid robot had a non-gendered appearance and its gender was manipulated by voice quality, visitors were asked to make a monetary donation and answer a questionnaire. The results showed a cross-gender effect in that male participants were more likely to donate money to the female robot than women were. Moreover, male participants rated female robots as more credible and trustworthy than male robots, whereas the converse was observed with female participants (Siegel et al.
2009). In another experiment in which participants solved puzzles while cooperating with a doll robot that had a non-gendered appearance and gendered voice and name, a cross-gender effect was similarly observed; participants interacting with the robot of a different gender felt more comfortable than those who interacted with the robot of the same gender (Alexander et al.
2014). This effect was also observed in an experiment in which a navigation task was performed in a town using computer-mediated communication between human instructors and robot followers (Koulouri et al.
2012).
Although robots assigned to people of a different gender were evaluated more positively, some studies have shown that robots assigned to people of the same gender were evaluated more preferably instead. For example, an experiment in which items were sorted on a touch-screen table using instructions given by a small-sized humanoid robot, whose gender was changed using computer-generated voices and names, indicated that female participants completed the task equally fast regardless of the gender of the robot, whereas male participants were faster in completing the task when the robot was male (Kuchenbrandt et al.
2014). This means that the interaction between humans robots may differ depending on various types of interaction factors, such as the nature of the situation or task.
Furthermore, the reason behind people’s preference for a certain gender of robot may lie behind the gender stereotyping of people toward one another. For example, a common gender stereotype is based on hair length, and female robots with long hair are considered more communal than male robots with short hair (Eyssel and Hegel
2012). They also reported that tasks that were considered stereotypically male, such as heavy lifting and machine operation, were perceived to be more suitable for male robots. Furthermore, a gender stereotype existed in occupations; male robots were preferred in security-related occupations and female robots were preferred in healthcare-related occupations (Tay et al.
2014).
The effect of a robot’s gender on a user’s decision to use it is assumed to depend on these factors: the characteristics of the robot and the user's cultural expectations of the robot. The preferred gender of a robot may vary depending on differing factors such as the nature of situations and roles. The situational factors related to a robot’s gender need to be taken into account when considering introducing robots into daily life.
Although the issue of gender stereotyping in robots has been highlighted and its effects are becoming increasingly recognized, the roles and situations in which it occurs remain unclear, raising the question of what the gendered expectations for robots are. In addition, there is limited knowledge on whether these gender stereotypes are related to the existing gender stereotypes of people. The impact of a robot’s gender may depend on many factors, such as situational factors (e.g., nature of tasks), human factors (e.g., gender), and culture. Therefore, gender preferences in robotics designs also may depend on the nature of the interactions between humans and robots and the situations in which they interact.
The purpose of this study is to clarify whether people prefer robots with male or female physical appearances in regular communication situations and whether egalitarian gender role attitudes are related to this preference. Furthermore, in this study, we compared these tendencies for robots and humans.
4 Discussion
In this study, we aimed to clarify whether people prefer robots with male or female appearances in communication situations and whether equality orientation in gender role attitudes is related to this preference. We also compared the results obtained with the gender preferences of humans.
4.1 Tendency of gender preference
First, we checked the ratios of human, android-type robots, and mechanical-type robots that were selected, and the results showed that, in any situation and for any subject, “males” was not selected, and “females” or “neither” was selected. Furthermore, the number of respondents who chose “either” was higher than that who chose “female”; This was more noticeable in situations with simple tasks included in situation set B and more for robots than for people. In other words, in terms of users' needs, genders do not need to be assigned to robots when introducing them into daily life. It may be a good idea to introduce a robot with a gender-neutral physical appearance to prevent the reproduction of gender stereotypes in robots.
When the relationship between participants’ genders and gender preferences was examined, we confirmed that the effect of gender on the gender preference of a robot weakened when the human factor was eliminated. In situation set A, even if the object was a human, the participant’s gender had a slight effect on the preference, but if the object was an android-type robot, the effect was weaker. In addition, in situation set B, there was almost no preference based on participants’ genders, even if the object was a human. This means that regardless of one's own gender, the expected gender differed depending on the content of the task. Even if gender preference may differ in different situations, if the physical anthropomorphic properties of the robot are reduced, the influence of the gender preference of the user will become weaker. Therefore, to prevent the reproduction of gender stereotypes in robots, making this elimination in the design may be a simple solution.
In both situation sets, the association between the gender preference for humans and both types of robots was found to be moderate, and the strength of the association was comparable. In other words, those who tended to prefer a particular gender of robots also tended to prefer a particular gender of humans. Similarly, we suggest that people who do not have a gender preference for humans are less likely to have one for robots. Thus, it is possible that assigning genders to robots may not necessarily increase gender bias.
4.2 Egalitarian gender role attitudes according to participants’ genders and gender preferences
In addition, in some situations for android-type robots and in all situations for machine-type robots, the equality orientation in gender role attitudes was shown to be higher for those who were not specific about their preference for a robot’s gender. Therefore, people with a low gender bias were less likely to have a gender bias toward robots, and the introduction of a robot with a specific gender did not increase any gender bias. In this regard, the problem is not associated with the design of the robot but the gender bias that people already have toward one another. Therefore, to intervene in and solve the problem of gender bias against robots, educating people about gender bias is an effective solution. Moreover, for people with a pre-existing gender bias, the introduction of gender in the design of a robot may increase this bias. It is important to examine this point in the future.
Although it has been reported that gendered robots may lead to the reproduction of gender stereotypes (Marchetti-Bowick
2009; Robertson
2010; Weber and Bath
2007), the present results do not suggest this, at least not in the context of this empirical study. An investigation into whether this is related to the culture of Japan or whether it is limited to the situation examined in this study will need to be conducted in the future.
The results also showed that numerous respondents did not prefer the gender of their communication robot. Based on this viewpoint, there is no need to introduce a robot that specifies its gender, thereby preventing the reproduction of gender stereotypes.
4.3 Implications
In Japan, when robots are introduced into daily life, there is no need to give them a specific gender appearance. Although providing a robot with a feminine appearance may be preferable to some due to past traditions, presently, it may be more appropriate to introduce robots with a gender-neutral appearance to avoid reproducing any gender bias. There is also the possibility that introducing these kinds of robots will be more acceptable abroad. Certainly, there may be cases where it is better to assign a gender depending on the characteristics of the situation or purpose, but it should be done with careful consideration.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.