触覚技術：ジェンダーの分析

課題

エンジニアは、ハプティックデバイスを介して触覚をバーチャルで伝える新しい方法を設計している。これは、ロボット工学において触覚をどのように採用すべきか、という疑問を提起する。ロボットの触覚は人間の慣習に従うべきだろうか？人間とロボットの間の触感やハプティックデバイスを介した触感は、人間同士の場合の触感と同じ意味を持ち得るだろうか？

方法：ジェンダー分析

人間とロボットの触覚に関する研究はまだ初期段階にあり、ロボットの「ジェンダー」と人間の「ジェンダー」を社会的に検討した研究はほとんどない（「ジェンダー化されたロボット」を参照）。ロボット研究者は、ジェンダーが人間とロボットの触覚にどのように影響するかを検討する必要がある。このような研究には次の要素を含める必要がある。1）男性、女性、多様な性の人々を代表する被験者と、異なるジェンダー構成を持つロボット、2）ジェンダー別（男性、女性、多様な性）に細分化されたデータ、3）ジェンダー別の分析結果。効果的な相互関係を構築するためには、人間とロボットのジェンダーの相互作用を分析することが不可欠である。

ジェンダード・イノベーション：

1. 人間同士の触感のジェンダーダイナミクスを理解する

2. ロボットやハプティックデバイスにジェンダーに適した触覚を組み込む

次のステップ

1. 人間とロボットの触覚における人間のジェンダーの影響を分析する

2. 人間とロボットの触覚における認識されたロボットのジェンダーの影響を分析する

 

The Challenge

Social robots are entering our lives in healthcare, elderly care, teaching, and entertainment (Li et al. 2017). In these settings, robots interact and communicate with humans by following rules of appropriate social behaviors. Social touch is an important part of human non-verbal communication. Emotions such as anger, fear, and happiness can be accurately communicated by touch, and more complex emotions like envy can be communicated by people who are close, such as romantic couples (Hertenstein et al., 2006; 2009; Thompson & Hampton, 2011).

Engineers are designing new ways to communicate touch virtually through haptic devices. Van Erp and Toet argue that haptic interactions help humans bond with robots (2013). This raises questions about how touch should be employed in robotics. Should robot touch follow human conventions? Can touch between a human and a robot or mediating haptic device have the same meaning as it does between two humans?

Gendered Innovation 1: Understanding the Gender Dynamics of Touch Among Humans

How does gender factor into the equation of social touch? As engineers seek to reproduce human social touch as closely as possible, it is important to understand the strong (largely unwritten) rules of etiquette governing human social touch (van Erp & Toet, 2013). Affective touch is a subset of social touch that communicates either positive or negative emotions. When appropriate, affective touch can modulate physiological responses, increase trust and affection, establish bonds, and moderate behavior. When inappropriate, affective touch can induce fear and repulsion (van Erp & Toet, 2015; Burleson et al., 2018). Here we focus specifically on how affective touch among humans varies with gender, keeping in mind that gender norms differ with intersecting factors, such as ethnicity and age.

The social aspects of touch lie in the interaction between two humans. Each human is made up of different social characteristics; these include gender, age, educational background, etc. Click on the highlighted elements in Figure 1 to see how the meaning of touch between humans depends on: 1) the overall social context in which humans touch; 2) the relationship between the people touching; 3) the purpose of the touch; and 4) the broader gender norms & relations governing social interaction in particular cultures.

Figure 1. Touch between humans depends on who the person is in terms of gender, age, occupation, ethnic background, education, language, culture, ability, etc. It also depends on the social context within which a touch takes place—for example, at work or at home—and the gender norms governing social interactions in that space. Touch is governed by the purpose of the touch, gender relations between individuals, and their social relationship.

Next Steps in Studying Human/Human Touch
Studies of human social touch show that participants’ age, gender identity, and cultural backgrounds—as well as social context and purpose of the touch—all change touching behavior and response. To date, most studies analyze binary gender, i.e., men and women, with the assumption that people are heterosexual. We recommend that studies be expanded to consider gender-diverse individuals, including diverse sexual orientation, such as lesbian, gay, bisexual, queer, asexual (LGBQA) or diverse gender, such as transgender, genderqueer, or gender fluid.

Gendered Innovation 2: Building Gender Appropriate Touch into Robots and Haptic Devices

Some dimensions of human social touch may translate to human/robot interactive touch. Three aspects of human/robot touch will be considered here: 1) human touch mediated by haptic devices, 2) humans touching robots, 3) robots touching humans.

Human Touch Mediated by Haptic Devices
In many instances, haptic technologies aim to reproduce certain effects of human/human social touch at a distance. Researchers argue that mediated touch should mimic interpersonal human touch. Yohanan has cataloged 30 different haptic prototypes that use force, vibration, and temperature to mimic different human sensations, including a hug and a handshake (Yohanan & MacLean, 2012). Examples of such devices include The Hug, Thermal Hug, HotHands, POKE, etc. (for a list see Huisman, 2017, 396).

Humans are often able to communicate emotions through a haptic device, but not as accurately as in person (Bailenson et al., 2007). In human interactions, for example, social touch can enhance altruistic behavior and willingness to help. This “Midas touch,” such as a touch on the shoulder or elsewhere, often promotes prosocial behavior (Hubbard et al., 2003). For example, restaurant-goers leave bigger tips when touched by waiters. This effect was especially pronounced when servers and patrons were of opposite gender. The results of experiments with mediated Midas touch have been mixed. A meta-analysis concluded that mediated social touch has the same strength as the in-person Midas touch, though the authors’ own experiment found a non-significant increase in helping behavior after a virtual social touch (Haans & Ijsselsteijn, 2009).

Humans Touching Robots
Etiquette for humans touching robots tend to follow human/human interaction, including gender conventions. For example, in one experiment in which participants were asked to “clean” virtual dirt particles from a virtual person, subjects of both genders used less force with female representations than male representations. They also used more force on the person’s torso than on their face (Bailenson & Yee, 2008, 167).

Figure 2. Touch between human and robot depends on who the human is and the domain in which the robot serves—for example, healthcare, service, education, etc. Responses to touch are governed by the purpose of the touch. What has not been studied is the interaction between human gender and perceived robot gender (see Gendering Social Robots).

Importantly, norms governing access to specific body regions is similar for humans and robots. In one study, 31 participants (16 female, 15 male) were asked to touch or point to anatomical regions on a humanoid robot. Participants were told they were engaged in an interactive anatomy lesson. The robot used was the 23-inch NAO, which is made of white plastic with various accent colors. Participants had increased physiological responses (electrodermal arousal) when asked to touch the robot’s buttocks or genitals and less when asked to touch accessible regions, such as its hands and feet. Although the experimenters recruited nearly half men and women, the authors did not analyze sex-disaggregated data (Li et al., 2017). The robot used has no perceivable sex or gender apart from the colored areas (in personal communication we learned that the accents are gray, which we interpret as gender neutral). Participants were prompted by NAO to touch them with a synthesized voice. We were unable to evaluate the voice for gender cues.

Robots Touching Humans
Does etiquette for robots touching humans tend to follow human/human interaction, including gender conventions? Or do other rules apply? Research into robot-initiated touch is in its infancy, and few studies have considered the gender of the participant in the interaction with robot “gender” (as established by social cues). HRI experts caution the effective robot-initiated touch will depend on all the dimensions of the interaction that influence the effectiveness of human/human touch, such as the appearance and behavior of the robot, the user’s personality, including gender, the social context of the interaction, etc. (Willemse et al., 2017).

Robot-initiated touch will be crucial in areas such as nursing care. In an experiment at Georgia Tech, researchers programmed a robotic caregiver to touch and wipe subjects’ forearms. People responded more positively when the touch was instrumental, i.e., to clean the person’s skin, than when it was affective, i.e., to comfort the person (Chen et al., 2011). The experiment used equal numbers of men and women, but an analysis of gender differences in participant response was not reported.

Conclusion

As we have seen above human/human touch is complex, dependent on the overall social context in which touching occurs, the relationship— or degree of familiarity—between the people touching, the purpose of the touch, and gender norms. While well understood in human/human touch, gender has yet to be thoroughly examined in human/robot touch.

Next Steps

• 1. Analyzing the impact of human gender in human/robot touch.: Although most studies we reviewed were designed to include equal numbers of men and women, few: 1) disaggregated data by gender (whether the human is a man, woman, or gender fluid); or 2) analyzed results by gender. Reporting aggregated summaries for men and women may lead to flawed conclusions. For example, in a study of eight emotions, Hertenstein et al. (2009) reported that men and women were able to effectively communicate these emotions by touch. When they reanalyzed their data by sex in 2011, however, they found that only pairs of women touching women were able to effectively communicate happiness. Similarly, sympathy was effectively communicated only when one partner of the pair was a woman, and anger only when one partner of the pair was a man (2011). Analyzing gender in this instance was crucial to understanding the full implications of their results.
• 2. Analyzing the impact of perceived robot gender in human/robot touch. We know of no study that controlled for the perceived gender of the robot used or provided an analysis of perceived robot gender as it interacts with human gender. As robots are increasingly enter our lives, these analyses will be important for effective and safe collaborations between humans and robots.