Clone images elicited higher eeriness than individuals with different faces; related to distinguishableness of each face, the duplication of identity, avoidance reactions based on disgust sensitivity, inter alia

Yonemitsu F, Sasaki K, Gobara A, Yamada Y (2021) The clone devaluation effect: A new uncanny phenomenon concerning facial identity. PLoS ONE 16(7): e0254396. Jul 13 2021. https://doi.org/10.1371/journal.pone.0254396

Abstract: Technological advances in robotics have already produced robots that are indistinguishable from human beings. This technology is overcoming the uncanny valley, which refers to the unpleasant feelings that arise from humanoid robots that are similar in appearance to real humans to some extent. If humanoid robots with the same appearance are mass-produced and become commonplace, we may encounter circumstances in which people or human-like products have faces with the exact same appearance in the future. This leads to the following question: what impressions do clones elicit? To respond to this question, we examined what impressions images of people with the same face (clone images) induce. In the six studies we conducted, we consistently reported that clone images elicited higher eeriness than individuals with different faces; we named this new phenomenon the clone devaluation effect. We found that the clone devaluation effect reflected the perceived improbability of facial duplication. Moreover, this phenomenon was related to distinguishableness of each face, the duplication of identity, the background scene in observing clone faces, and avoidance reactions based on disgust sensitivity. These findings suggest that the clone devaluation effect is a product of multiple processes related to memory, emotion, and face recognition systems.

General discussion

Our six studies investigated the impressions formed by clone faces and factors to influence the impression of clone faces. The participants evaluated six individuals with clone faces as eerier and more improbable than those with different faces and a person with a single face (Study 1). We called this negatively emotional response to clone faces the clone devaluation effect. This effect was stronger as the number of clone faces increased from two to four persons (Study 2). Moreover, this effect did not occur when each clone face was indistinguishable like animal faces (Study 3). It was also shown that the duplication of identity rather than facial features has an important role in this effect and clone faces with the duplication of identity were eerier (Studies 4a & 4b). The clone devaluation effect became weaker when clone faces existed in the lower reality of the context (Study 5). Furthermore, the eeriness of clone faces stemming from improbability would be positively predicted by disgust, in particular animal-reminder disgust (Study 6). Taken together, these results suggest that clone faces induce eeriness and that the clone devaluation effect is related to the realism and disgust reaction. We discuss the details of an assumed mechanism below.

Our findings provide evidence for the internal mechanisms of the clone devaluation effect. The results of Study 6 imply a relationship among realism, the eeriness of clone faces, and animal-reminder disgust sensitivity. The domain of animal-reminder disgust sensitivity includes the lack of the ideal appearance in a human body such as death or a damaged exterior form [30]. In addition to this idea, a previous study indicated that animal-reminder disgust is also involved in the eeriness of humanlike objects such as androids [31], the appearance of which is often improbable and strange. Objects with clone faces, even though they have an absolute appearance in human bodies, might be judged as strange because the clone faces are improbable. As a result, the eeriness is elicited in order to avoid any harm from such improbable and strange objects (i.e., clone faces). Taken together, the improbability of the clone faces is assumed to trigger a defensive reaction stemming from animal-reminder disgust, which in turn plays key roles in emotional reactions (i.e., eeriness).

Considering that improbability and strangeness are the keys to the phenomenon of the clone devaluation effect, a lack of humanity is another of the important parts of the effect. What is involved in the lack of humanity? The results of Studies 4a and 4b indicate that it is likely that this lack of humanity stems from duplication of identity, not facial features. Faces are important information for identifying individuals because human beings have a one-to-one correspondence between face and identity in principle. However, clone faces violate this principle, which may make us misjudge that the identity of people with clone faces should be the same. Thus, the duplication of identity is contrary to the principle of human beings and thus could be considered as a lack of humanity. In most of the present studies except for 4a and 4b, the clone devaluation occurred without the manipulation of identity; observers deduced duplication of identity from clone faces and, as a result, eeriness was elicited. However, we showed that the eeriness of the clone faces was low when clone faces were judged as multiplets; they have the same facial features. In this case, the participants were able to assign clone faces to different identities and thus eeriness was not evoked. Considering this, faces are the most critical cue of personal identity when there is no other special cue (e.g., details of siblings). This idea is consistent with the findings of previous studies [32, 33]. Taken together, the lack of humanity in the clone devaluation effect may stem from the duplication of identities inferred by clone faces.

Considering the findings of our six studies, it was possible to some extent to speculate on a mechanism of the clone devaluation effect (Fig 9). Initially, when one observes clone faces, they are analyzed based on facial features. In this case, because they have identical facial features, they are judged as the same faces. This point is supported by Study 3, which showed that the clone devaluation effect did not occur when clone faces could not be analyzed and distinguished. Several models of aesthetic and facial processing argued that visual features were analyzed and abstracted from objects in the first stage [34–36]. Therefore, it is appropriate to assume that facial features are analyzed as the first stage of the facial processing system. After the judgment of clone faces, their information is processed in the memory system. In this stage, the realism of encountered clone faces is evaluated based on prior knowledge and experience. The results of studies 4a and 4b reflected this processing because the eeriness of famous or little-known multiplets’ faces was diminished even if they have clone faces. This stage is related to the reality of the background scene in which the clone faces exist, which supported the results of study 5’s that the eeriness and improbability of the photographic clone faces was higher than the cartoon clone faces. Although the present study only used the cartoon scene as less realistic scene than the photographic scene, we can investigate the influence of the background scene in further detail if the virtual reality space in which 3D avatars with clone faces existed was used as a more realistic scene. Finally, the improbability of clone faces induces avoidance reaction to strangers, which stems from disgust. Through this processing, unpleasant emotions (i.e., eeriness) toward clone faces are finally evoked.

Fig 9. The mechanism of the clone devaluation effect proposed in the present study.

The emergence of objects with clone faces is not a mere fantasy; they can be expected to appear in the near future. For example, when humanoid technology has further developed, the appearance of humanoids will be the same as that of humans, and the mass production of such humanoids will be possible. Further, the uncanny valley may disappear in humanoids. The present study indicates that clone faces elicit eeriness. This suggests an irony: even if this new technology can bridge the uncanny valley, a new eerie phenomenon (i.e., the clone devaluation effect) would appear alongside new technologies like human cloning techniques or robotic development. The results of a previous study that investigated public attitudes towards human cloning in the United Kingdom found that people have relatively negative responses to reproductive cloning technology [37]. However, unlike this survey, the present study created an experimental situation in which new technologies such as cloning human were realized and found that people tended to evaluate clone faces negatively. Therefore, our findings shed light on the negative aspects of the development of new technology and we urge the reconsideration of the rapid introduction of such technologies into society from the perspective of psychological responses instead of bioethics.

The present study has some limitations, which suggest the need for further research on the clone devaluation effect. First, we used dog faces in Study 3 to investigate the influence of distinguishableness of faces on the clone devaluation effect. However, there is another effect that enables human beings to easily distinguish faces that belong to their own racial group, which is known as the other-race effect [e.g., 38]. To explore the influence of distinguishableness in detail, this effect may be helpful. If other-race faces are used as clone faces, the clone devaluation effect would be weaker than own-race faces or not occur. Second, we should also set up an experimental scene to simulate real-life situations in the near future. In most of our studies, we presented clone faces of humans mainly because we assumed a future where technology overcomes the uncanny valley. However, it is possible that robots with features less similar to human faces, which have not fallen into the uncanny valley, will become popular before more human-like robots will be available on the market. As the results of Study 3 imply, the clone devaluation effect is unlikely to occur in non-human clone faces. Considering this, we speculated that if robots with features less similar to human faces are clearly perceived as robots, not humans, the clone devaluation effect would not occur in clone faces of such robots. Third, it is possible that the observation time of the clone faces varied depending on the participants, which may have affected the results of the present study since we did not control the exposure time of clone faces. Observing clone faces for a long time may lead to habituation, resulting in reduced clone devaluation effects. Therefore, to reveal a temporal aspect of the clone devaluation effect (e.g., minimum time for the clone devaluation effect to occur) future studies should manipulate the exposure time of clone faces.

In conclusion, the present study has identified the clone devaluation effect, which asserts that clone faces elicit eeriness. The improbability and lack of humanity of the clone faces were related to this effect. Moreover, the duplication of identity, not facial features, had key roles in the clone devaluation effect. Furthermore, the clone devaluation effect stemmed from disgust and avoidance reactions (particularly, animal-reminder disgust). The present study suggests it is possible that the introduction of new technology in robotics or the cloning of human beings into society may cause unpleasant psychological reactions in the future.

Life satisfaction does not adapt to high income: After five or more years, people who maintain high incomes still have higher life satisfaction than the average population

Happiness adaptation to high income: Evidence from German panel data. Jianbo Luo. Economics Letters, July 13 2021, 109995. https://doi.org/10.1016/j.econlet.2021.109995

Highlights

• Life satisfaction does not adapt to high income.

• After five or more years, people who maintain high incomes still have higher life satisfaction.

Abstract: This paper is the first to use national representative panel data to demonstrate that individuals do not adapt to high income in the long run: after five or more years, the life satisfaction of high-income people is still higher than that of the average population. Using entropy balancing (EB) matching and Lasso variable selection to reweight the control group yields similar results.

Keywords: High incomeSubjective well-beingAdaptationEntropy balancingMachine learning Lasso

Wearing an opposite-sex virtual body enhanced participants’ perceptions of pleasantness and erogeneity for caresses on intimate areas from a same-sex toucher; this effect was stronger in men than in women

Wearing same- and opposite-sex virtual bodies and seeing them caressed in intimate areas. Manuel Mello et al. Quarterly Journal of Experimental Psychology, July 9, 2021. https://doi.org/10.1177/17470218211031557

Abstract: Immersive virtual reality enables people to undergo the experience of owning an artificial body and vicariously feeling tactile stimuli delivered to it. However, it is currently unknown how such experiences are modified by the sexual congruency between the human and the artificial agent. In two studies, heterosexual men (Experiment 1) and women (Experiment 2) embodied same-sex and opposite-sex avatars and were asked to evaluate the experience (e.g., pleasantness, erogeneity) of being touched on social or intimate areas of their virtual body by a male or female avatar. Electrocardiogram and galvanic skin response were also recorded. Moreover, participants’ implicit and explicit gender biases were examined via a gender-potency implicit association test and the Ambivalent Sexism Inventory. When embodying a same-sex avatar, men and women rated caresses on intimate areas from an avatar of the opposite sex as more pleasant and erogenous. Conversely, body swap—that is, wearing an opposite-sex avatar—enhanced participants’ perceptions of pleasantness and erogeneity for caresses on intimate areas from a same-sex toucher. This effect was stronger in men than in women. Furthermore, physiological correlates of enhanced processing of arousing stimuli predicted behavioural outcomes during the body swap illusion. Wearing an opposite-sex avatar affects one’s own body representations and may have important implications on people’s attitudes and implicit reactivity to touch-mediated interactions. Men seem more susceptible to this type of body swap illusion. Our paradigm may induce profound changes of cross-sex perspective-taking and provide novel tools for promoting empathy and comprehension of sex-related diversity.

Keywords: Body ownership, social touch, virtual reality, embodiment, body swap, sex differences

Yawning Is More Contagious in Pregnant Than Nulliparous Women

Yawning Is More Contagious in Pregnant Than Nulliparous Women. Naturalistic and Experimental Evidence. Ivan Norscia, Lucia Agostini, Alessia Moroni, Marta Caselli, Margherita Micheletti-Cremasco, Concetta Vardé & Elisabetta Palagi. Human Nature, Jul 13 2021. https://rd.springer.com/article/10.1007/s12110-021-09404-w

Abstract: Contrary to spontaneous yawning, which is widespread in vertebrates and probably evolutionary ancient, contagious yawning—yawning triggered by others’ yawns—is considered an evolutionarily recent phenomenon, found in species characterized by complex sociality. Whether the social asymmetry observed in the occurrence of contagious yawning is related to social and emotional attachment and may therefore reflect emotional contagion is a subject of debate. In this study we assessed whether yawn contagion was enhanced in pregnant women, a cohort of subjects who develop prenatal emotional attachment in preparation for parental care, via hormonal and neurobiological changes. We predicted that if yawn contagion underlies social and emotional attachment, pregnant women would be more likely to contagiously yawn than nonpregnant, nulliparous women of reproductive age. We gathered data in two different settings. In the experimental setting, 49 women were exposed to video stimuli of newborns either yawning or moving their mouth (control) and we video-recorded the women during repeated trials to measure their yawning response. In the naturalistic setting, 131 women were observed in a social environment and their yawning response was recorded. We tested the factors influencing the yawning response, including the reproductive status (pregnant vs. not pregnant). In both settings, yawn contagion occurred significantly more in pregnant than nonpregnant women. By showing that pregnant women were most likely to respond to others’ yawns, our results support the hypothesis that the social variation observed in yawn contagion may be influenced by emotional attachment and that yawning in highly social species might have been coopted for emotional contagion during evolution.

Discussion

The results from both the experimental and the naturalistic data converge in indicating that women’s reproductive status had an effect on contagious yawning, which was more likely to occur in pregnant than in nulliparous women (here defined as women who were not pregnant and had no children). As a matter of fact, pregnant women were more likely to respond than nulliparous women to both video yawns of unknown infants in the experimental trials and live yawns from adults in the naturalistic setting (Tables 2 and 3; Figs. 3 and 5). This finding, presented for the first time with this study, provides support to the Emotional Bias Hypothesis (EBH) because yawn contagion was highest in the category of women characterized by enhanced social attachment predisposition, owing to the biological and psychological changes typical of the gestation period (Barba-Müller et al. 2019; Brandon et al. 2009; Tichelman et al. 2019).

Since yawn contagion has been found to vary across the day (Giganti and Zilli 2011), we checked whether our yawning response sampling could be biased by the time periods during which the data were collected, depending on the availability of the study subjects. In neither setting did we find a significant effect (Tables 1 and 3), probably because the majority of the data was collected in the morning and in the afternoon (with little data collected at the very extremes of the day).

The use of a twofold approach, involving both experimental and naturalistic data collection, allowed the verification of the possible effect of different variables on yawn contagion. The results of the experimental trials show that the yawning response was significantly higher in the yawning than in the control video condition (Table 1; Fig. 2). This finding confirms that yawn contagion was present in the cohort of human subjects considered in this study (nulliparous and pregnant women) since it has been found in other segments of the population (Arnott et al. 2009; Provine 1989, 2005).

Yawn contagion may be affected by selective, top-down attentional biases (Massen and Gallup 2017), in addition to bottom-up, stimulus-driven attention (Attentional Bias Hypothesis, ABH; Palagi et al. 2020). Therefore, in the experimental setting we checked for selective attention to the stimulus and we found no significant influence of the time of attention to the stimulus source (video screen) on yawning (Table 1), which was high overall in both yawning and control video conditions, as well as in pregnant and nulliparous women. This finding reduces the probability that in our sample a selective attention bias may have accounted for the differences between stimulus (yawning/control) and reproductive status (pregnant/nulliparous) conditions. This is line with evidence indicating, directly or indirectly, that contagious yawning in humans may depend on bottom-up more than top-down selective attention (Norscia et al. 2020; for a review see Palagi et al. 2020). Age is another variable known to possibly affect yawn contagion rates (Bartholomew and Cirulli 2014). In our case, in the experimental setting there was a nonsignificant trend of the influence of age in the yawning response, possibly because the women under study fell within the relatively short reproductive age.

In the naturalistic setting we could verify the effect of a social bond between the trigger and the potential responder on the yawning response. Although the bond was restricted to two categories (strangers and acquaintances) owing to data constraints, and despite showing an inverse correlation with reproductive status, the bond had a significant effect on yawn contagion, which was more likely between subjects who knew each other than between strangers. This finding is in agreement with previous literature showing that relationship quality has an influence on yawn contagion, whose likelihood increases as the strength of the social bond increases (from strangers to acquaintances, friends, and lastly to family members; Norscia and Palagi 2011; Norscia et al. 2016). Norscia et al., (2020) found no difference between strangers and acquaintances when the yawns were heard but not seen, although friends and family responded at significantly higher rates than did those in the other categories. In the absence of the visual cue, it is probably more difficult for the potential responders to discern between subjects with whom they have reduced or no familiarity.

Importantly, our results from the experimental trials show that reproductive status (pregnant/nulliparous) had a significant effect on the yawning response in the yawning video condition but not in the control video condition (cf. Tables 2 and 3). Therefore, only yawning resulting from contagion—and not spontaneous yawning—was affected by pregnancy in our sample. Historical accounts report an increase of spontaneous yawning in the case of certain diseases (e.g., puerperal fever or hemorrhage; Walusinski 2010), and excessive yawning has indeed been indicated as a possible marker of disease in humans (Thompson and Simonsen 2015). Progesterone increases daytime drowsiness and sleeping time (Won 2015) and so it may increase spontaneous yawning rate during pregnancy. In this respect, we cannot exclude that the yawning stimulus might have preferentially primed the yawning motor response in pregnant women also because they experienced increased fatigue (despite showing similar levels of sleep to those of nulliparous women). An investigation on how spontaneous rates vary within subjects across pregnancy, possibly in relation to fatigue and tiredness, and how contagious yawning varies depending on the stimulus (e.g., babies/adults)—with hormonal and neurobiological correlates—could better clarify the above issues.

Overall, the different yawning response of pregnant women relative to women with no children can fall within the broad range of the behavioral changes that start occurring during pregnancy, such as motor activity and dietary choice variations (Crozier et al. 2009; Gradmark et al. 2011). Compared with childless women, pregnant women show increased sensitivity to emotional signals and facial expressions. For example, pregnant women were found to perceive infant cries in more differentiated ways than women with no offspring (Bleichfeld and Moely 1984; Yoshiaki 1985). As gestation progresses, pregnant women also show enhanced ability to encode and process emotional faces, especially related to distress (an emotional state; Keltner et al. 2019) as an evolutionary adaptation to motherhood, which requires hypersensitivity to emotional threat signals and contagion (Osório et al. 2018; Pearson et al. 2009). Our results fit with this scenario because they indicate enhanced responsiveness of pregnant women to yawning, which has been linked (with various degrees of evidence) to anxiety and distress in human and nonhuman primates (from lemurs to apes: e.g., Baker and Aureli 1997; Coleman and Pierre 2014; Leone et al. 2014; Palagi et al. 2019; Thompson 2014, 2017; Thompson and Bishop 2012; Zannella et al. 2015). Thompson (2014) has posited that cortisol (involved in the stress response) may be involved in yawn contagion, at least under certain situations. Another hypothesis, not mutually exclusive to the cortisol hypothesis, may be that yawn contagion is, to a certain extent, under the influence of oxytocin, considering that enhanced emotional recognition is one of the effects of oxytocin, whose levels largely increase during pregnancy (Domes et al. 2007; Preston 2013). In particular, oxytocin appears to increase the accuracy of the recognition of faces displaying angry and happy emotions, especially in women (Yue et al. 2018). Mariscal et al., (2019) found that yawn contagion in autism spectrum disorder (ASD) children was positively related to the blood concentration of oxytocin. The possible relationship between oxytocin and yawn contagion is supported by evidence that yawn contagion in humans follows the empathic gradient (sensu Preston and de Waal 2002), being highest between closely bonded subjects (Norscia and Palagi 2011; Norscia et al. 2020). Some features typical of mother-infant attachment, such as social recognition, bonding, and affiliation, are maintained in adulthood and promoted by oxytocin, which has been found to increase trust (Kosfeld et al. 2005), generosity (Zak et al. 2007), altruism (de Dreu et al. 2010), and both cognitive and affective empathy (Rodrigues et al. 2009; Shamay-Tsoory et al. 2013; Smith et al. 2014; Uzefovsky et al. 2015). One of the future steps is to evaluate the possible covariation between oxytocin and yawn contagion in both pregnant and nulliparous women. Beyond incorporating hormones, further studies could involve postmenopausal versus pregnant women and check how mothers react when they see their own fetus yawning on the echograph screen.

The possible connection between yawn contagion and increased social and emotional bonding is also suggested by the fact that some of the areas that seem to be involved in yawn contagion (such as the ventromedial-prefrontal cortex, superior temporal sulcus, amygdala, insula, posterior cingulate, and precuneus; Nahab et al. 2009; Platek et al. 2005; Schürmann et al. 2005) are also involved in mother-infant care, in mother’s enhanced sensitivity to the baby, and maternal brain changes occurring during pregnancy (Barba-Müller et al. 2019; Hoekzema et al. 2017; Kikuchi and Noriuchi 2015; Preston 2013; Rifkin-Graboi et al. 2015).

In summary, by showing increased occurrence of yawn contagion in pregnant women—a cohort of subjects that is specifically “programmed” to recognize and respond to others’ emotions—this study provides support for the hypothesis that yawn contagion may, at least under certain circumstances, underlie emotional contagion (EBH; Palagi et al. 2020). This process is considered by some scholars a basic form of empathy and occurs when an emotion is transferred from one individual to another, possibly via a motor perception–action mechanism, involving the matching of facial expressions and the resonance of the emotions that underlie such expressions (de Waal and Preston 2017).

The perception–action and the offspring care model both predict that subjects can preferentially attend the stimuli coming from closely bonded others, particularly caregiving individuals such as pregnant women toward babies (Preston 2013; Preston and de Waal 2002). Visual, top-down attention has limited effect on yawn contagion and does not follow a consistent familiarity trend in hominines because other factors, such as dominance, can come into play (Lewis et al. 2021; Norscia et al. 2020; Palagi et al. 2020). Hence, a possible bonding hypothesis between EBH and ABH is that yawn contagion can be influenced by emotional bonding and attention, mainly directed through bottom-up mechanisms.

Importantly, not all contagious yawning is triggered by emotional resonance, and that is not the point in question here. Contagious yawning also occurs between strangers (Norscia and Palagi 2011), and some people are consistently not susceptible to others’ yawns (Bartholomew and Cirulli 2014; Platek et al. 2003; Provine 1986, 1989). Contagious yawning is a form of yawning and—as such—can be related to nonemotional, individual and/or environmental factors, such as time of the day (Giganti and Zilli 2011), age (Bartholomew and Cirulli 2014), and possibly temperature (Gallup and Eldakar 2011). The perception–action mechanism itself is based on a theory in motor control that assumes that our physical motor acts are primed in the brain by observation of those in others, even if they do not bear emotional cues (Preston and de Waal 2002). Thus, contagious yawning can also be a nonemotional motoric response. The pivot around which this study revolves is the possible mechanism leading to the social variations observed in the occurrence of contagious yawning. Although still under debate (Adriaense et al. 2020; Massen and Gallup 2017), various physiological, neuroethological, and psychological studies sustain the possible connection between the social asymmetry of yawn contagion and emotional bonding. Some of the brain areas that appear to be involved in yawn contagion (Nahab et al. 2009; Platek et al. 2005; Schürmann et al. 2005) seem to overlap with those involved in emotional processing of internal and external stimuli and empathy (Palagi et al. 2020) and—importantly—with the maternal brain (Barba-Müller et al. 2019; Hoekzema et al. 2017; Kikuchi and Noriuchi 2015; Rifkin-Graboi et al. 2015). Highest levels of yawn contagion are associated with increased oxytocin levels (i.e., ASD children; Mariscal et al. 2019), enhanced social bonding (i.e., between friends and family; Norscia and Palagi 2011), and maternal prenatal bonding (i.e., in pregnant women; present study). Lower yawn contagion rates in association with levels of autistic traits were found to be related to attentive rather than background emotional empathy deficits (Helt et al. 2021). Finally, another study found that subjects who yawned in response to observing others’ yawns exhibited significantly higher empathy scores (Franzen et al. 2018).

Hence, although we cannot discard the possibility that other priming and motor mechanisms may also underlie the social asymmetry of yawn contagion, the hypothesis that this behavior has been coopted during evolution for emotional contagion still stands and gains further support.