1 Introduction
Humanoid Virtual Influencers (HVIs) are AI-generated characters that closely mimic human appearance and behavior, possessing not just human-like physical features but also personalities, hobbies, and personal interests (Mutani 2022). Their engaging presence on social media (SM) platforms has captivated audiences, prompting retailers such as Amazon and Prada to integrate HVIs like Lil-Miquela into their marketing strategies (Rasmussen 2022; Douglass 2022). As HVIs endorse a vast array of products across industries—from fashion (Shudu for Vogue) to electronics (Lil-Miquela for Samsung)—they significantly drive consumer engagement, generating three times more likes and shares compared to human influencers (Baklanov 2021). With over 200 active VIs globally and a market estimated at $4.6 billion and growing, HVIs represent an increasingly influential facet of social media marketing (Hiort 2022; Mutani 2022).
Despite their potential, not all HVIs achieve lasting popularity; while many HVIs have achieved significant fame and influence (e.g., The 2018 edition of TIME Magazine recognized Lil Miquela among the top 25 influential personalities on SM), others have faded away (Bowin 2021; Patrick 2023). This is often the case when VIs lack appeal and fail to resonate with their intended audience (McCoy 2023). As a case in point, in 2021 almost 57% of VIs had a negative follower growth (Baklanov 2021). Therefore, studying HVIs is crucial, as understanding their appeal and the factors contributing to their longevity can assist creators in finding a balance between high aesthetic standards and relatable storytelling, ultimately attracting and retaining followers in a competitive landscape. Creating a sense of being alive and present, and sharing bonds with others, would attract more followers and contribute to HVI’s longevity. Yet this seems at odds with the uncanny valley theory (Mori 1970) contending that eerie feelings may arise when interacting with highly anthropomorphized VIs (Lou et al. 2023). To mitigate the eeriness risk, most renowned VIs have been meticulously crafted to uphold high fidelity based on commonly accepted standards of aesthetics (Yap and Ismail 2022). For instance, some research points out that a significant number of VI followers tend to find those with humanlike characteristics highly visually appealing as compared to other influencers; individuals are drawn to VIs because of the captivating blend of aesthetic appeal and genuine storytelling (Choudhry et al. 2022). In particular, VIs that possess lifelike features and an appearance resembling humans are more likely to be considered attractive and amass a substantial number of followers (Black 2019). Likewise, qualities such as talent, creativity, knowledge (e.g., competence) as well as friendliness, beauty, sensuality, and style (e.g., warmth), all are found to contribute to the appeal and attractiveness of VIs (Black 2019; Da Silva Oliveira and Chimenti 2021). While some studies (e.g., Zhou et al. 2024; Li et al. 2023) suggest that HVIs are not as effective as their human counterparts, intriguingly enough, other studies (e.g., Stein, Linda Breves, and Anders 2022) argue that individuals respond cognitively, emotionally, and behaviorally to VIs in a manner similar to human influencers. Nevertheless, the intent of the current study is not to compare HVI to human influencers but rather to examine the relationships between factors that contribute to the effectiveness of HVIs that is purchase intention and intention to follow recommendations.
In fact, there is a lack of research examining the interplay between perceptual factors that shape VI’s appearance and the underlying mechanism that elicits a feeling of being with others (Ahn, Cho, and Sunny Tsai 2022). Additionally, it remains unclear how these dynamics might influence perceptions of VI’s attractiveness and elicit consumer responses (Deng and Jiang 2023). A significant research gap exists regarding audience perceptions of VIs and their effects on consumer attitudes and behaviors. Understanding these dynamics is essential for elucidating the mechanisms that contribute to the effectiveness of VIs (Laszkiewicz and Kalinska-Kula 2023). Additionally, there is a lack of research validating the effectiveness of VIs in terms of their influence on consumer perception of VIs, particularly in terms of establishing verified cause-and-effect relationships (Laszkiewicz and Kalinska-Kula 2023). To address these gaps in prior studies, the current research posits that human likeness, lifelikeness, warmth, and competence build up a social presence. This, in turn, affects HVI’s perceived attractiveness and subsequently, consumers’ willingness to follow HVI’s recommendations and purchase intentions. Specifically, our research aims to achieve two objectives. Firstly, to understand how HVI’s appearance in terms of human-likeness (i.e., anthropomorphism), lifelikeness (i.e., animism) as well as communal (warmth) and agentic (competence) traits influence social presence and affect consumers’ perception of HVI’s attractiveness. Secondly, to examine how attractiveness influences consumers’ willingness to follow recommendations (WFR) of HVI and their intention to make a purchase.
Building on the stimulus-organism-response (S-O-R) framework, we theorize that the interplay between anthropomorphism, animism, and the stereotypes of warmth and competence acts as a stimulus. HVI displays anthropomorphism and animism and conveys warmth and competence, when perceived by consumers they influence their sense of social presence and perception of HVI’s attractiveness (i.e., organism), which in turn drive his willingness to follow HVI’s recommendations and his purchase intention (i.e., responses). The S-O-R framework is used here as a theoretical underpinning for mapping the relationships among the key concepts and delineating the direction of those relationships. Moreover, this research brings forth several contributions to the burgeoning work on HVI. Firstly, while a bulk of existing research has predominantly examined anthropomorphism as the primary characteristic of HVI’s appearance, our research introduces animism as an additional trait. By doing so, we address the call by Karpinska-Krakowiak and Eisend (2021) to differentiate between animism and anthropomorphism as separate yet interconnected concepts. Secondly, this work adds to the growing literature on VIs by examining how animism influences the stereotypical evaluations of the VI’s warmth and competence. Prior research has solely investigated the association between anthropomorphism and these two dimensions (El Hedhli et al. 2023). Our research takes a further step forward, by shifting the focus towards animistic elements. Understanding how these two traits interact with consumer perception of HVIs can provide valuable insights for designing HVIs that resonate positively with consumers, ultimately influencing their engagement and their intention to make a purchase. Lastly, this research sheds light on the underlying psychological processes that influence individuals’ WFR of HVI’s and their purchase intention. By uncovering the perceptual and cognitive processes that guide these behaviors, we provide a deeper understanding of why consumers may feel inclined to endorse the recommendations of HVIs.
The paper’s structure is arranged as follows. Firstly, building on the S-O-R framework, a conceptual model is developed, along with the establishment of theoretical foundations and hypotheses. The subsequent section involves the empirical testing of these hypotheses through four cross-sectional studies. The findings of the studies are then reported. Finally, the paper concludes by delving into the theoretical and practical implications, outlining the study’s limitations, and providing recommendations for further research endeavors.
2 Conceptual Background and Hypotheses Development
Drawing on the S-O-R framework (see Figure 1), we suggest that the interplay between HVI traits, namely anthropomorphism, animism, and the stereotypes of warmth and competence act as a stimulus that influences consumers’ sense of social presence and perception of HVI’s attractiveness. This, in turn, impacts their WFR of HVI and their intention to make a purchase. The next sections will further elucidate the interrelationships between these concepts.
FIGURE 1
2.1 The Interplay Between HVI’s Anthropomorphism, Animism, and Social Presence
2.1.1 The Effects of HVI’s Anthropomorphism and Animism on Social Presence
Anthropomorphism, or ascribing human traits to nonhuman beings (Epley, Waytz, and Cacioppo 2007), has been shown to have a positive relationship with social presence which refers to the sense of bonding with another being (Biocca, Harms, and Burgoon 2003). For instance, recent research by Blut et al. (2021) found that the anthropomorphic appearance of nonhuman agents such as robots plays a crucial role in establishing a sense of connection with users. Similarly, Tsai, Liu, and Chuan (2021) point out that anthropomorphic profiles of chatbots convey positive social presence effects. In the context of HVIs, anthropomorphism is highly effective in shaping the sense of social presence. In this vein, Ahn, Cho, and Sunny Tsai (2022) found a direct correlation between the anthropomorphic HVIs and consumers’ perception of social presence. Likewise, in their recent study, Kim and Park (2024) report a positive association between HVI’s anthropomorphism and social presence. Therefore, it can be argued that incorporating humanoid cues, such as facial expressions and human appearance, into HVIs would elevate their perceived humanness and ultimately increase the extent to which HVIs convey a feeling of being with another entity.
To sum up, the existing literature on anthropomorphism and social presence suggests a clear link between assigning human traits to nonhuman entities and fostering feelings of connection with them. As such, it is safe to argue that integrating features like facial expressions and a human-like appearance into HVIs may bolster their perceived humanness and strengthen the bond with users. The present study seeks to explore the effects of anthropomorphism on social presence, proposing the following hypothesis:
H1.HVI’s anthropomorphism has a positive effect on social presence.
While current studies on HVIs have focused on a single aspect of appearance, that is anthropomorphism, no study has attempted to reach beyond humanlike physiognomy to investigate how HVI’s lifelikeness referred to as animism would affect the perception of social presence, although, in many instances, the perception of being alive has been recognized to play a key role in triggering a sense of presence and interacting with nonhuman entities such as social robots (Bartneck et al. 2009).
Intriguingly, Karpinska-Krakowiak and Eisend (2021) emphasize the need to differentiate between anthropomorphism and animism as these concepts have been mistakenly used interchangeably in marketing literature; animism pertains to the cognitive inference that non-living objects possess life, while anthropomorphism specifically refers to the belief that these objects possess human-like qualities. In this vein, visual cues like a human-like face, hands, and physical bodies would evoke anthropomorphism. Whereas, gestures and movements would elicit animism (Karpinska-Krakowiak and Eisend 2021; Bartneck et al. 2009). Recent studies contend that entities with strong animistic characteristics, like robots and virtual streamers display lifelike qualities that facilitate customer interaction, thereby enhancing social presence (Blut et al. 2021; Gao, Jiang, and Guo 2023). In other words, anthropomorphism is rooted in visual cues—such as facial expressions that evoke human traits—which influence consumers’ perception of connections and engagement levels with HVIs. In contrast, animism involves deeper cognitive processes, prompting consumers to ascribe mental states to HVIs based on perceived intentionality and agency. While anthropomorphism stems from visual elements designed to forge a sense of connection, animism invokes cognitive inferences about HVIs, shaping consumers’ perceptions of intentionality in their interactions with HVIs.
In fact, gestures and movements portrayed in videos, animated images, or stories on SM posts can evoke animism (Karpinska-Krakowiak and Eisend 2021). Many HVIs on platforms like Instagram are depicted crying, brushing their teeth, eating ice cream, and socializing with friends (Ahn, Cho, and Sunny Tsai 2022). These portrayals of “slice of life” convey a certain level of lifelikeness and infer a sense of connection with others.
Taken together, existing research on HVIs has primarily centered on anthropomorphism. However, there is a rising interest in investigating the influence of animism on social presence. Recent studies indicate that entities with pronounced animistic traits can improve customer interaction and social presence. Therefore, the present study aims to examine the impact of animism on social presence, resulting in the following hypothesis:
H2.HVI’s animism has a positive impact on social presence.
Overall, examining the influence of animism alongside anthropomorphism presents an opportunity for a more nuanced understanding of social presence in the context of HVIs. Given that much of the current research has centered on anthropomorphism, there is a compelling need for studies to investigate the broader implications of animism. Understanding the role of animism in fostering deeper connections with audiences can help brands attract followers and refine their marketing strategies, ultimately enhancing their effectiveness.
In summary, while the established literature supports the significance of anthropomorphism in enhancing social presence, the emergence of animism as a distinct and influential construct warrants dedicated research. This study aims to contribute to the understanding of how both anthropomorphism and animism shape social presence in HVIs, ultimately fostering richer interactions between consumers and nonhuman entities in the SM environment.
2.1.2 The Mediating Effect of Animism
The role of animism as a mediator has not been thoroughly investigated in previous research (Karpinska-Krakowiak and Eisend 2021). Among a few studies, Castro-González et al. (2018) found a positive correlation between the perceived human-likeness of a robot’s mouth and its perceived liveliness; as a robot becomes more humanized, the perception of its liveliness increases. This suggests that the perception of animism is influenced by anthropomorphism. Likewise, Blut et al. (2021) found that anthropomorphism has a positive impact on animism, furthermore, animism strongly mediates the association between anthropomorphism and the intention to use robots. In a more recent study, Malhotra and Ramalingam (2023) found that the impact of AI anthropomorphism on the intention to make a purchase is mediated by perceived animism.
In sum, current research on how animism influences the relationship between HVI’s anthropomorphism and the formation of social bonds remains limited. Yet, recent studies (i.e., Laban 2021; Pandey and Rai 2024) have emphasized the need to consider the mediating role of animism in interactions between humans and anthropomorphized entities. Further investigation in this area has the potential to provide valuable insights for improving the social presence of anthropomorphized entities like HVIs. Therefore, this research aims to examine the mediating role of animism in the link between anthropomorphism and social presence and puts forth the following hypothesis:
H3.HVI’s animism meditates the relationship between HVI’s anthropomorphism and social presence.
2.2 The Association Between Stereotypical Judgments and Social Presence
2.2.1 The Effects of Anthropomorphism and Animism on Stereotypical Judgments of HVIs
Consumers evaluate virtual agents, such as chatbots and avatars, based on two dimensions of the Stereotype Content Model (SCM): warmth and competence (Roy and Naidoo 2021; Yao, Kuai, and Jiang 2023). Warmth is associated with the agent’s ability to demonstrate prosocial intentions and is determined by their perceived honesty, friendliness, and sincerity. Competence, on the other hand, alludes to the agent’s skills and expertise in executing these intentions (Fiske et al. 2007). In the realm of HVIs, El Hedhli et al. (2023) found that anthropomorphism has a positive impact on perceived warmth and competence. However, the relationship between animism and these dimensions has not yet been explored.
In the context of virtual characters, Gao, Jiang, and Guo (2023) argue that virtual streamers high in animacy convey a feeling of warmth and personalness. Interestingly, Nguyen et al. (2015) use animistic cues including mannerisms of gestures (e.g., open gestures, leaning forward, smiling) and gaze behaviors to convey different levels of warmth and competence of avatars. Thus, it is safe to argue that animism, as a second trait of HVI’s appearance, could also influence perceptions of warmth and competence. Collectively, the existing research suggests that both anthropomorphism and animism play key roles in shaping perceptions of warmth and competence. As a result, we propose that both concepts have a beneficial impact on the dimensions of the SCM. Building on this premise, we hypothesize the following:
H4a.HVI’s anthropomorphism has a positive impact on perceived warmth and competence.
H4b.HVI’s animism has a positive impact on perceived warmth and competence.
2.2.2 The Stereotypical Judgments of HVIs: The Mediating Effects
The mediating role of warmth and competence has been studied in various contexts in literature. For instance, Zheng et al. (2023) showed that warmth and competence are mediators of the relationship between anthropomorphism in humanoid robots and user satisfaction. In the sphere of influencer marketing, Kim and Read (2021) argue that SCM dimensions mediate the relationship between the perception of influencers’ appearance (i.e., smiling face) and their followers’ response to persuasive messages on Instagram. Noteworthily, El Hedhli et al. (2023) found that warmth and competence also act as parallel mediators in the impact of HVI’s anthropomorphism on consumers’ disposition to follow HVI’s recommendations.
Considering that HVI’s traits of appearance (i.e., anthropomorphism and animism) have been shown to instill warmth and competence (El Hedhli et al. 2023; Nguyen et al. 2015) and that social presence—defined as “the feeling that one has some level of insight into the others’ intentional cognitive and affective states” (Biocca and Nowak 2001, 409)—inherently stems from competence and warmth, it could be argued that SCM dimensions mediate the relationship between HVI’s traits and social presence. Taken together, the current literature suggests that both warmth and competence are important factors in shaping consumer perceptions and reactions toward influencers and virtual entities. Given the significance of social presence in human-HVI interactions, this study examines the mediating role of warmth and competence in the relationship between HVI traits (i.e., anthropomorphism and animism) and social presence. Based on this premise, the following hypotheses are proposed:
H5.Perceived warmth and competence mediate the impact of HVI’s anthropomorphism on social presence.
H6.Perceived warmth and competence meditate the impact of HVI’s animism on social presence.
2.3 The Interplay Between Social Presence, HVI’s Attractiveness, and Willingness to Follow Recommendations (WFR)
2.3.1 The Effect of Social Presence on HVI’s Attractiveness and WFR of HVI
Social presence has been discussed in several studies on AI agents including chatbots and VIs (Ahn, Cho, and Sunny Tsai 2022). The connection between social presence and attractiveness is well-documented in the marketing literature. For example, Edwards et al. (2015) found that displaying social presence on LinkedIn heightens perceived levels of social attractiveness. In a virtual agents experiment, Skalski and Tamborini (2007) found that increasing social presence led to higher evaluations of agents including their physical attractiveness. Similarly, Verhagen et al. (2014) found that social presence predicts the attraction toward virtual agents. In the context of HVIs, Ahn, Cho, and Sunny Tsai (2022) found that anthropomorphizing HVIs enhances social presence thereby boosting perceived physical and social attractiveness.
There is also evidence suggesting a positive correlation between social presence and consumers’ willingness to follow influencers on SM. For instance, Jin, Ryu, and Muqaddam (2021) argue that when consumers develop a sense of social presence with fashion influencers on Instagram, they become more receptive to their promotional messages. Consumers respond more positively to promotional content that includes individuals with whom they feel connected. In a similar vein, Lăzăroiu et al. (2020) suggest that an online agent’s availability, which is an indicator of social presence, along with perceived personalization, conveying attractiveness, can significantly influence consumers’ trust in an online recommender agent and their willingness to follow its recommendations. In the realm of HVIs, Mirowska and Arsenyan (2023) propose that HVIs use emotional sharing as a strategy to attract and retain followers. Individuals who experience a heightened level of emotional contagion, feel a strong sense of affective social presence and are more likely to become followers of HVIs. Therefore, we hypothesize:
H7.Social presence is positively associated with (a) HVI’s attractiveness and (b) the WFR of HVI.
2.3.2 The Effect of HVI’s Attractiveness on the WFR
The attractiveness of influencers, both in terms of their physical appearance and social appeal, plays a significant role in motivating consumers to follow them. Previous studies have demonstrated that consumers are more likely to be drawn to human influencers who possess physical attractiveness, among other qualities (Farrell, Campbell, and Sands 2022).
In the context of HVIs, recent research has demonstrated that physical attractiveness is equally important for HVIs as it is for human influencers, regardless of the type of product being endorsed (Chiu and Ho 2023). This further emphasizes the significance of physical attractiveness in the endorsement of products by HVIs. In addition to physical attractiveness, the social attractiveness of HVIs also plays a role in shaping consumers’ attitudes towards endorsement posts. Ahn, Cho, and Sunny Tsai (2022) found that consumers are more receptive to product recommendations when they perceive the HVI as socially attractive.
Taken together, it is safe to argue that the perceived attractiveness of HVIs can significantly impact consumer engagement and receptiveness toward the influencer’s content and product recommendations. Therefore, we propose the following hypothesis:
H8.HVI’s attractiveness is positively related to the WFR of HVI.
2.3.3 The Mediating Effect of HVI’s Attractiveness
The literature has extensively explored the mediating role of attractiveness in various contexts. Skalski and Tamborini (2007) argue that the attractiveness of computer agents is influenced by the level of social presence and can enhance the processing of the agent’s messages. This, in turn, has the potential to impact individuals’ attitudes and intentions.
In the domain of influencer marketing, the attractiveness of SM influencers has been recognized as one of the key attributes that influence various aspects such as the attitude toward the influencer, online engagement, and word-of-mouth (Taillon et al. 2020; AlFarraj et al. 2021). Ahn, Cho, and Sunny Tsai (2022) further showed that HVI’s attractiveness mediates the linkage between social presence and attitudes toward endorsement posts. Accordingly, social presence can enhance consumers’ positive perceptions of the social and physical attractiveness of HVIs, leading to a more favorable evaluation of endorsement posts. From this H9 is proposed.
H9.HVI’s attractiveness mediates the linkage between social presence and the WFR of HVI.
2.4 The Interplay Between HVI’s Attractiveness, WFR, and Purchase Intention
2.4.1 The Effect of HVI’s Attractiveness on Purchase Intention
The existing literature points to conflicting findings about the influence of attractiveness on the intention to make a purchase. On one hand, several studies suggest that the inclusion of attractive models in advertising campaigns has a significant impact on consumer attitudes toward the advertisement and a strong influence on their intention to make a purchase (Petroshius and Crocker 1989; Patzer 1983; Till and Busler 2000). On the other hand, other studies failed to prove a significant impact of attractiveness on purchase intention (Caballero and Solomon 1984; AlFarraj et al. 2021).
Applied to the context of VIs, Chiu and Ho (2023) suggest that HVI’s physical attractiveness (besides human influencers) is positively related to the purchase intention of Gen Z, regardless of product involvement. Likewise, Yap and Ismail (2022) found a positive correlation between VI’s attractiveness and Generation Y’s purchase intention. However, Wong and Wei (2023) found that while both social and physical attractiveness of influencers positively impact influencer defense, they do not have any effect on purchase intention. Considering these mixed findings, there is a critical need to examine the relationship between the perceived attractiveness of HVI, and the intention to make a purchase. Therefore, in this study, we propose the following hypothesis:
H10.HVI’s Attractiveness has a positive impact on purchase intention.
2.4.2 The Effect of the WFR on Purchase Intention
HVIs’ performance on SM relies on their capability to enhance followers’ likelihood of purchasing (El Hedhli et al. 2023). A survey mandated by Influencer Marketing Factory in the US in 2022 revealed a positive correlation between the tendency of consumers to heed recommendations from VIs and their purchasing intention; about 35% of the respondents reported purchasing a product that was endorsed by VIs.
Furthermore, Thomas and Fowler (2021) argue that VIs possess the same level of persuasive influence as celebrity influencers on consumers’ attitudes and intentions to buy endorsed products. In a similar vein, El Hedhli et al. (2023) found a positive correlation between consumers’ readiness to adopt HVI’s suggestions and their intentions to make purchases. Based on these findings, it can be contended that consumers’ inclination to heed HVI’s recommendations would lead to a greater purchase intention. Hence, H11 is proposed:
H11.WFR of HVI has a positive impact on purchase intention.
2.4.3 The Mediating Effect of the WFR
A scant of studies have examined the potential mediating role of WFR in relation to HVI’s attractiveness and purchase intention. For instance, El Hedhli et al. (2023) found a positive association between consumers’ WFR and their purchase intentions, yet their study did not test the mediating effect of WFR. Along the same line, Kim and Park (2023) found that the desire to imitate or replicate the behaviors of others (i.e., mimetic desire) acts as a mediating factor between VI’s attractiveness and consumers’ purchase intentions. While little research has highlighted the importance of WFR and mimetic desire in influencing purchase intentions, further investigation is warranted to scrutinize the mediating effect of WFR in relation to HVI’s attractiveness. Considering that the inclination to follow HVI’s recommendations is reflective of the desire to imitate or replicate others’ behaviors, the current study proposes that WFR mediates the effect of HVI’s attractiveness on the intention to make a purchase. Therefore, we propose the following hypothesis:
H12.WFR of HVI mediates the linkage between HVI’s attractiveness and purchase intention.
Taken together, all direct hypotheses (H1–H4a,b, H7, H8, H10, and H11) and mediation hypotheses (H3, H5, H6, H9, and H12) previously discussed are depicted in Figure 1, illustrating the conceptual framework of this research.
3 Methodology
3.1 Research Context
While most research on HVIs has predominantly been conducted in Western cultures, such as the UK and United States (e.g., El Hedhli et al. 2023; Alboqami 2023), limited attention has been paid to exploring HVIs in the context of Eastern cultures, specifically in South Korea (Kim and Park 2024; Kim et al. 2024). Recognizing this gap, El Hedhli et al. (2023) have highlighted the need for more research in the Eastern cultural context.
Interestingly, Japan stands out as the second country, following the US, with the highest number of VIs, as indicated in the 2023 AI Influencer Report (Geyser 2023). In fact, Japan presents a relevant context to examine the effects of animism and anthropomorphism on HVI’s effectiveness for two main reasons; first, since ancient times in Japan, there has been a deep-rooted belief in animism or “Yaoyorozu beliefs”, Higaki (1989) uses the term 日本のアニミズム to refer to Japanese animism. Second, people in Japan believe that everything has a soul or a god in it, including objects. Unlike other cultures, where people often conflate animism and anthropomorphism (Karpinska-Krakowiak and Eisend 2021), Japanese culture makes a clear distinction between both concepts. While animism refers to the belief that objects, animals, and even natural phenomena possess a spiritual essence or soul, this belief is an integral part of Shinto, the indigenous religion of Japan. In Shinto, everything in nature, ranging from mountains and rivers to mythical ancestress and holy persons, is regarded as having a Kami (Ono 1962). This concept is often referred to as soul or spirit (Sugimoto 2020) and is grounded in an animistic belief system that views the world as alive with spiritual significance (Ikeuchi 2010). Anthropomorphism, on the other hand, involves attributing human qualities, characteristics, or behavior to non-human entities. This can be seen with “moe anthropomorphism” (Japanese: 萌え擬人化, Hepburn: moe gijinka), a form of anthropomorphism in anime, manga, and games where “moe” qualities are given to non-human beings. Considering these distinctions, Japan provides a relevant context to the study of HVIs.
3.2 HVI Selection
The process of selecting HVI for the empirical study involved four steps. A preliminary list of the top 10 VIs in Japan was examined (Taslaud 2023; AJ Marketing 2023). Since the study focused on HVIs, three cartoonlike VIs (i.e., KAF, Hatsune Miku, and Myra Keiko) were excluded from the list. The selection was then narrowed down to three HVIs with the largest number of followers, namely Imma (403 K followers), Zinn (50.4 K followers), and Ria (38 K followers). Ultimately, our search was narrowed down to Imma and Zinn, both have verified and accessible Instagram accounts, while Ria did not (Travers 2022).
Given that the research focus is on the effects of HVI’s animism and anthropomorphism on behavioral responses, the two HVIs were evaluated for their suitability for the research. To do so, 120 Japanese students were asked to rate each HVI based on familiarity, animistic cues, and human-like appearance.1 Results of independent samples t-tests show that Imma scored significantly higher on familiarity and animism than Zinn (M(Imma)familiarity = 5.41; SD = 0.884; M(Zinn)familiarity = 4.01; SD = 0.783; t = 12.986, M(Imma)animacy = 4.708; SD = 0.737; M(Zinn)animacy = 4.312; SD = 0.476; t = 4.936, all p < 0.05, df = 238). However, no difference was observed between them in terms of human-likeness (M(Imma)human-likeness = 5.85; SD = 0.847; M(Zinn)human-likeness = 5.06; SD = 0.781; t = 0.393, p > 0.05, df = 238). These findings validate Imma’s popularity and suitability for this research. Imma has been frequently mentioned in various research studies (Laszkiewicz and Kalinska-Kula 2023; Miyake 2023) and press articles and news (South China Morning Post 2022; The Sunday Times 2020; Bloomberg 2023), and has been frequently sought after by global brands and advertisers (Domus 2020).
3.3 Participants and Data Collection Procedure
To test the overall framework of the research (see Figure 1), we conducted four studies. Instead of expanding the scope of the S-O-R theoretical framework, this research focuses on testing a subset of parsimonious models. These simplified models have advantages over complex ones, including enhanced generalizability, greater precision, and easier interpretation (Jacoby 2002), which can help address some limitations with testing the entire S-O-R framework (for review, see Vieira 2013). Study 1 investigated the potential mediating effect of animism in the link between HVI’s anthropomorphism and social presence. Study 2 tested the mediating roles of warmth and competence on the effect of animism on social presence, as well as the link between anthropomorphism and social presence. Study 3 investigated the impact of social presence on the WFR of HVI, with attractiveness being a mediator. Lastly, study 4 examined whether the WFR of HVI mediated the effect of attractiveness on purchase intention. As prior studies point to some potential demographic variance, gender, and age were controlled for in all four studies. The designs, analyses, and results of the four studies are detailed in the next sections.
Participants in all four studies were hired via a Japanese marketing firm. To be eligible for the study, they are required to meet the following two criteria: (1) familiarity with the HVI, and (2) possession of an Instagram account with access to the HVI’s posts. In each study, we employed stratified sampling (a method based on probability sampling) to ensure obtaining a sample that is as representative as much as possible of the Instagram users’ profiles in Japan. According to recent statistics, the majority of Instagram users are women—59.9% and 18 to 24 represent the largest age group of users (NapoleonCat 2023)2. Data was collected using an online questionnaire. The questionnaire in all four studies was first translated into Japanese and then translated back into English to ensure the consistency of the meaning of the English and the Japanese versions. Participants were first given a description of the HVI’s profile (see Appendix 1), followed by specific instructions to explore HVI’s Instagram account, and then redirected to complete a survey, including demographic questions such as age and gender.
3.4 Common Method Bias
When creating the survey for each of the four studies, we incorporated various scale formats. For instance, anthropomorphism and animism were assessed on a semantic differential scale, while other measures were captured on a Likert scale anchored by “1/strongly disagree” and “7/strongly agree.” We also arranged some of the questions about the dependent variables before the ones related to predictors. This was done to mitigate the possibility of common method bias, as discussed in prior research (e.g., Podsakoff et al. 2003). To further evaluate the presence of common method bias, we checked the Variance Inflation Factor (VIF) values. All VIF values are reported in the next sections and are below 3.33 indicating that the models are free from common method bias (Kock 2015).
3.5 Study 1
3.5.1 Study 1- Design and Procedures
Study 1 was conducted to examine the potential mediator role of animism in the relationship between HVI’s anthropomorphism and social presence (see Figure 2). One hundred and twenty participants (n = 120) took part in this study (60% female; 49% 18–24, 37% 25–34, 9% 35–44, and 5% 45 and above). Participants were tasked with navigating through the HVI account, reading posts, viewing pictures and videos, and paying close attention to the HVI’s appearance, expressions, gestures, and interactions with followers. When filling up the survey, participants rated anthropomorphism (Cronbach’s alpha = 0.872) on three items (fake/natural, non-humanlike/humanlike, and artificial/lifelike) borrowed from El Hedhli et al. (2023). Likewise, HVI’s animism (Cronbach’s alpha = 0.871) was measured on a scale of four items (“dead/alive, stagnant/lively, inert/interactive, apathetic/responsive”) borrowed from Bartneck et al. (2009). Social presence (Cronbach’s alpha = 0.940) was measured with a 5-item Likert scale assessing the degree to which participants perceived the HVI’s Instagram post as conveying a sense of presence. The scale includes the following items: “there is a sense of human warmth”, “personness” and “human contact” in HVI’s Instagram posts. All were derived from Gefen and Straub (2003) work.
FIGURE 2
The meditating effect of animism.
Note: ***p < 0.001.
Because animism might share variance with anthropomorphism we investigated the possibility of multicollinearity. We found that the Variance Inflation Factor (VIF) for animism and anthropomorphism (VIFs = 1.864) are below the threshold of 3 (Hair et al. 2021). Additionally, the Durbin-Watson value shows no autocorrelation as it stands at 1.914 which is substantially close to the value of 2, and close to neither 0 nor 4, suggesting that multicollinearity is not an issue (Durbin and Watson 1992). We employed the PROCESS macro model 4 (Hayes 2022) to test for the meditation hypothesis; anthropomorphism was entered as the independent variable (X), animism the mediator (M1), social presence as the dependent variable (Y), and age and gender were entered as covariates. We chose to use PROCESS instead of structural equation modeling because of its flexibility in accommodating models with both single (Figures 2, 5, and 6) and parallel mediators (Figures 3 and 4). This allows for more intricate and comprehensive statistical analyses (Hayes, Montoya, and Rockwood 2017; Hayes 2022).
FIGURE 3
The mediating effects of warmth and competence on the anthropomorphism-social presence linkage.
Note: ***p < 0.001
.
FIGURE 4
The mediating effects of warmth and competence on the animism-social presence linkage.
Note: ***p < 0.001, **p < 0.01
.
3.5.2 Study 1- Results
The results displayed in Table 1, indicate a significant direct association between anthropomorphism and social presence (X → Y: β = 0.5617, t = 8.3205, p < 0.001, CI = [0.4280, 0.6954]), lending support to H1. Moreover, we found a significant effect of animism on social presence (M → Y: β = 0.4181, t = 4.4390, p < 0.001, CI = [0.2315, 0.6046]), supporting H2.
3.6 Study 2
3.6.1 Study 2- Design and Procedures
Study 2 aimed to examine how warmth and competence can mediate the path between anthropomorphism and social presence (see Figure 3), and also the linkage between animism and social presence (see Figure 4).
A total of 160 subjects participated in this study (59% female; 46% 18–24, 36% 25–34, 12% 35–44, and 6% 45 and above). Participants were instructed to explore the HVI account by reading posts and viewing pictures and videos, with a primary focus on the HVI’s friendliness and expertise. Additionally, they were asked to pay closer attention to HVI’s appearance, expressions, gestures, and interactions with followers. When filling up the survey, participants rated the HVI on anthropomorphism (Cronbach’s alpha = 0.859), animism (Cronbach’s alpha = 0.871), and social presence (Cronbach’s alpha = 0.938) using the same measures from study 1. In addition, participants rated the HVI on the traits of warmth (e.g., kind, generous) and competence (e.g., efficient, skillful) borrowed from El Hedhli et al. (2023). The five items on each trait formed a reliable scale (Cronbach’s alpha = 0.930 and 0.952 respectively). We checked for multicollinearity and found that VIFs for warmth (VIF = 2.361), competence (VIF = 2.203), as well as anthropomorphism (VIF = 1.786), and animism (VIF = 1.959) fall under the threshold of 3 (Hair et al. 2021). In addition, the Durbin-Watson value stands at 2.040 which is close to the value of 2, signaling no autocorrelation (Durbin and Watson 1992).
We utilized the PROCESS macro model 4 (Hayes 2022) to conduct parallel mediation analyses for each model, as depicted in Figures 3 and 4. In both models, social presence was entered as the outcome variable (Y), and warmth (M1) and competence (M2) were entered as simultaneous mediators, while the independent variable (X) differed between the two models, with anthropomorphism in the first model and animism in the second one. Age and gender were introduced as covariates.
3.6.2 Study 2- Results
The results are presented in Tables 2 and 3. When investigating the effect of anthropomorphism on the SCM dimensions, as depicted in Figure 3, anthropomorphism was found to have positive effects on both warmth and competence, lending support to H4a (X- > M1: β = 0.3538, p < 0.01 and CI = [0.2403, 0.4672]; X- > M2: β = 0.4152, p < 0.01 and CI = [0.2872, 0.5433]). When evaluating the mediating effect of warmth and competence, the findings reveal that the bootstrap intervals for the indirect effects (X ➔ M1 ➔ Y and X ➔ M2 ➔ Y) had no zero in between (respectively CI = [0.1880, 0.5842] and CI = [0.1593, 0.5103]), which means that the impact of anthropomorphism on social presence through warmth and competence is significant. Thus, H5 is supported. Since the path between anthropomorphism and social presence is also significant (X ➔ Y: β = 0.1908, p < 0.01 and CI = [0.0889, 0.2927]), we can conclude that the stereotype dimensions partially mediate the linkage between anthropomorphism and social presence. Besides, the effects of age and gender (covariates) were insignificant as the CIs show a value of zero in between (respectively CI = [−0.0633, 0.3798] and CI = [−0.1747, 0.5924]). Additionally, we checked for the contrast (C1) between the mediating effects of warmth and competence on the anthropomorphism-social presence linkage. The results show no differential effect between the two mediators (CI shows a value of zero in between CI = [−0.1928, 0.2088]).
3.7 Study 3
3.7.1 Study 3- Design and Procedures
In study 3, our objective was to examine the potential indirect influence of social presence on the WFR of HVI, through attractiveness (see Figure 5).
FIGURE 5
The mediating effects of HVI’s attractiveness.
Note: Ns = not significant, ***p < 0.001, *p < 0.05.
A total of 200 subjects (61% female; 44% aged 18–24, 39% aged 25–34, 11% aged 35–44, and 6% aged 45 and above) participated in this study. Participants were given the objective of exploring the HVI account, reading posts, and viewing pictures and videos, with specific attention to the HVI’s appeal, social engagement with followers, as well as any sponsored recommendations. Thereafter, participants answered a survey to rate social presence (Cronbach’s alpha = 0.943) using the same measures detailed in studies 1 and 2. Besides, the WFR of HVI (Cronbach’s alpha = 0.944) was rated on a two-item scale borrowed from El Hedhli et al. (2023), assessing consumer predisposition to follow recommendations made by the HVI. The attractiveness (Cronbach’s alpha = 0.856) was measured using four items (not sexy/sexy, ugly/beautiful, plain/elegant, unattractive/attractive) adapted from Ohanian (1990). In order to evaluate the existence of multicollinearity issues, we calculated the VIFs for the independent variables in the model. The VIF for social presence and HVI attractiveness (VIFs = 1.647) falls below the threshold of 3, as suggested by Hair et al. (2021). Furthermore, the Durbin-Watson value, indicating no autocorrelation, is 2.105, which is close to 2 but not 0 or 4. This suggests that multicollinearity is not a problem (Durbin and Watson 1992).
Like in study 1, the PROCESS macro model 4 (Hayes 2022) was utilized to examine the direct and mediating effects; social presence was entered as the independent variable (X), HVI’s attractiveness as the mediator (M1), WFR of HVI as the dependent variable (Y). Age and gender were also included as covariates.
3.7.2 Study 3- Results
The results displayed in Table 4, demonstrate a significant effect of social presence on HVI’s attractiveness (X → M: β = 0.5903, t = 11.583, p < 0.001, CI = [0.4898, 0.6908]), supporting H7a. Additionally, there was a significant direct link between social presence and the WFR of HVI (X → Y: β = 0.6572, t = 8.996, p < 0.001, CI = [0.5132, 0.8013]), supporting H7b. Moreover, we found a significant effect of attractiveness on the WFR of HVI (M → Y: β = 0.1918, t = 2.430, p < 0.05, CI = [0.0362, 0.3474]), supporting H8.
4 General Discussion
This research has sought to examine the interplay between HVI’s traits of appearance, specifically anthropomorphism (i.e., human-likeness) and animism (i.e., lifelikeness), and social presence (study 1) and to further shed light on the underlying mediating process, of SCM through which the effects of HVI’s animism and anthropomorphism manifest on social presence (study 2). Consistent with previous studies, the results show that human likeness positively affects social presence (Ahn, Cho, and Sunny Tsai 2022). Similarly, the current research suggests that perceived lifelikeness enhances social presence and partially serves as a mediator in the connection between human likeness and social presence. These results highlight the intertwined nature of animism and anthropomorphism in shaping social presence, adding to the ongoing debate on the role of animism as a different trait from anthropomorphism in shaping consumer perception of virtual entities, particularly HVIs.
The results of the study also demonstrate that both human-likeness and lifelikeness are associated with conveying warmth and competence, which ultimately influences social presence.
These findings align with recent work by El Hedhli et al. (2023), which underscores the role of warmth and competence as mediators in the effect of human-likeness on the responses of HVI’s followers. Additionally, they are consistent with Nguyen et al. (2015)‘s work suggesting that manipulating animistic cues such as gestures and gaze behaviors in virtual characters can lead to varying degrees of warmth and competence, thereby enhancing the awareness and believability of these characters.
Moreover, the research examined the role of attractiveness as a mediator between social presence and the WFR of HVI (study 3). Contrary to Ahn, Cho, and Sunny Tsai (2022) findings, we found that attractiveness does not mediate this relationship. Instead, social presence directly influences the WFR of HVI. The discrepancy in findings could be attributed to differences in the demographic characteristics of the samples. Unlike Ahn, Cho, and Sunny Tsai (2022) who conducted their study exclusively with US undergraduate students, our study considered a diverse range of demographic characteristics including age. Furthermore, previous research has suggested that attractiveness is not a universal concept and can vary based on factors such as age, gender, and specific aspects of attractiveness (He et al. 2021).
Lastly, Study 4 of this research found that the WFR of HVI fully mediates the linkage between HVI’s attractiveness and intention to make a purchase. This finding is consistent with recent research conducted by Kim and Park (2023), suggesting that consumers’ mimetic desire (i.e., the desire to imitate or replicate others’ behaviors, which could be equated to following HVI’s endorsement and recommendations), acts as mediator factor in the influence of HVI’s attractiveness on consumers’ intentions to make a purchase.
4.1 Theoretical Implications
The current research provides a theoretical foundation for understanding how a consumer reacts to HVIs. Specifically, this study validates the nomological network as outlined in the S-O-R framework. This framework illustrates how HVI’s appearance (i.e., the stimulus), influences the consumer as the perceiving entity, through his sense of social presence and his perception of HVI attractiveness (i.e., organism). This, in turn, leads to consumer’s intentional responses, such as the WFR of HVI and the intention to make a purchase (i.e., responses).
In regard to anthropomorphism, the research findings challenge the eerie feeling proposed by the uncanny valley theory (Mori 1970) and align more closely with the “Computers Are Social Actors” paradigm (Nass, Steuer, and Tauber 1994). This suggests that people interact with HVIs in a similar way to how they would with humans, even though they are aware that these interfaces lack genuine emotions and motivations. Despite the valuable contributions of the uncanny valley theory to our understanding of anthropomorphism in robotics, there is still a lack of consistency and conclusive evidence on the uncanny valley effect (Kätsyri et al. 2015). Recent studies have cast doubt on the relevance of this effect in today’s society, especially with the rise of AI and the metaverse, where individuals have become increasingly accustomed to engaging with nonhuman entities such as digital humans in virtual realms (El Hedhli et al. 2023) and anthropomorphic chatbots on e-commerce platforms (Zogaj et al. 2023). It has been recognized that in today’s era, consumers are becoming increasingly acquainted with AI agents; they can request customized advice from virtual agents, interact with chatbots, and ask voice assistants a wide range of questions on their smart devices, to name a few of AI’s affordances (Peltier, Dahl, and Schibrowsky 2023).
Additionally, the results of the current research support existing positions that suggest VI’s human-like attribute (anthropomorphism) positively influences perceptions of warmth and competence and further expand this to include lifelikeness (animism), shedding light on the intertwined nature of animism and anthropomorphism in shaping social perceptions.
Animism and anthropomorphism are related concepts, yet they carry distinct theoretical implications for perception and social presence. The differences between animism and anthropomorphism significantly influence how consumers perceive HVIs, in terms of forming connections, appraising agency, and engaging in social interactions. While anthropomorphism emphasizes visual perception with a focus on human-like appearance to foster a personal connection and convey social presence, animism with animistic elements, such as agency, play a crucial role in enhancing consumers’ cognitive awareness of HVIs. Understanding how animism (i.e., the perception of being “alive” and capable of social interaction) and anthropomorphism (i.e., humanlike appearance through visual cues) each contribute to social presence can help refine theories related to interactions with HVIs.
Animism, as a second trait of appearance, also positively affects SCM dimensions. Our research findings demonstrate that warmth and competence mediate the link between anthropomorphism and social presence as well as the path between animism and social presence.
In fact, this research is one of the first studies examining the importance of animism in HVI’s ability to establish a strong social presence among its followers. It offers important insights into the interplay between animism and anthropomorphism in shaping social presence. Past research in marketing has either focused exclusively on anthropomorphism (e.g., El Hedhli et al. 2023) or mistakenly conflated animism with anthropomorphism (read Karpinska-Krakowiak and Eisend 2021, for review). The current research clarifies how animism contributes to fostering a feeling of connection with others both directly and indirectly by mediating the influence of anthropomorphism on social presence.
Importantly, this research validates these effects in an Eastern cultural context, such as Japan, thereby extending previous work conducted in Western cultures (El Hedhli et al. 2023; Alboqami 2023). Taken together, our study offers insight into the socio-psychological factors that drive the relationship between consumers’ evaluations of HVI’s appearance (stereotypes, anthropomorphism, and animism) and their cognitive and behavioral responses.
4.2 Managerial Implications
The distinction between animism and anthropomorphism in HVIs carries significant implications for various practitioners, including retailers who have begun creating their own HVIs (e.g., Alibaba, Magalu & Ogilvy in Brazil, etc.), as well as SM content developers and agency creators of HVIs.
To enhance a sense of presence and connection among followers, developers should create HVIs that engage in realistic activities. As Millennials and Gen Z continue to use SM to document their daily lives and express themselves to a broader audience, creating and sharing content that features HVIs’ experiences in a way that closely mimics the typical experiences of their human counterparts (e.g., travel, food, and fashion choices) will be effective in crafting a sense of human likeness. By showcasing stories about where they have been, what they have eaten, and what they are wearing, among others, HVIs will convey a convincing facade of a genuine human experience. This human-like representation enables them to forge a deeper connection with their audience, as followers can easily relate to their virtual lives, further heightening their human-like qualities and sense of presence.
Similarly, when selecting HVIs for brand retailers and advertising agencies, it is important to choose HVIs that embody a combination of animism and anthropomorphism. These HVIs should exhibit personality, mannerisms, and emotions similar to their human counterparts, as these characteristics are crucial for conveying a feeling of social presence. For instance, to heighten the perception of social presence, HVI creators can incorporate animism through videos, animated pictures, and instant updates on HVI’s status. Additionally, incorporating motion and conversational functionalities (e.g., asking and answering questions, providing feedback, etc.) into the design of HVIs allows for movement and interactivity, further enhancing their animism.
Furthermore, HVI developers need to consider factors related to social judgment psychology when designing the appearance of HVIs. In this regard, modeling virtual character stereotypes based on warmth and competence traits is crucial, as these are basic dimensions that individuals use to judge HVIs. Accordingly, the appearance of HVIs should strike a balance between human likeness and lifelikeness to convey both warmth (e.g., friendliness, kindness, and trustworthiness) and competence (e.g., skill, efficacy, etc.). For example, by using motion graphics and animation software, HVI designers could incorporate gentle smiles and positive gestures (e.g., nodding and leaning forward) to convey warmth and trustworthiness. Additionally, depicting HVIs maintaining eye contact, displaying good posture, and showcasing professional accomplishments can communicate confidence, assertiveness, and effectiveness, ultimately enhancing the perception of competence.
Moreover, digital marketers should understand that the WFR of HVI is more influenced by social presence rather than attractiveness. When choosing HVIs and creating endorsement messages, it is important to prioritize an influencer’s ability to convey social presence to their audience. This means selecting HVIs based on their capacity to form connections with their followers, engage in meaningful conversations, and foster a sense of community. Additionally, since HVI’s attractiveness alone may not translate into purchase intention unless such an effect is mediated by the WFR of HVI, marketers should focus on fostering the connection between social presence and the WFR of HVI, as this pathway will ultimately drive purchase intention. By prioritizing social presence in their marketing strategies, marketers can enhance the effectiveness of HVIs and drive purchase intention among consumers. By cultivating connections and actively engaging HVIs with their audience, marketers can effectively harness the influence of HVIs to promote their brands and products.
4.3 Limitations and Further Research Avenues
The current research presents some limitations that could potentially open up new avenues for further investigation. Firstly, it is important to acknowledge that the empirical part of the research relied on cross-sectional studies, which aimed to investigate the interrelationships among the constructs depicted in the models. While this methodology has the advantage of enhancing the validity of the findings, it is essential to acknowledge that additional studies that would combine experiments and survey studies are desirable. Indeed, experiments are powerful methods for validating the strength and direction of these relationships, as well as increasing internal validity. Moreover, incorporating a second HVI while controlling for variance in animism, anthropomorphism, warmth, and competence traits would help to replicate the findings and enhance external validity.
Additionally, this study focused solely on a single Japanese HVI (Imma), who is recognized as a macro-influencer due to her large number of followers. Forthcoming research is warranted to investigate whether these findings apply to micro-influencers (those with less than 100 K followers) as well. Furthermore, testing the effect of culture would be an intriguing area of investigation. Some studies suggest that individuals’ perceptions of themselves (independent vs. interdependent self-construal) can influence their perception of social presence and anthropomorphism (Jin and Park 2009; Hsieh et al. 2013).
Another limitation of our study is that it solely focused on Instagram. In fact, video-based SM platforms like TikTok, possess a greater capacity to facilitate animism and should also be considered. As HVIs can engage with their audience on multiple SM platforms, it will be intriguing to compare findings across various types of platforms.
5 Conclusion
The rise of HVIs has attracted global attention, particularly in Japan where virtual characters have a strong presence in popular culture. While virtual characters have certainly left their mark on the entertainment industry in Japan, the advent of HVIs in SM is rapidly reshaping the landscape of both the Japanese and global marketing industries. Despite their surging popularity, the effectiveness of HVIs has come under scrutiny from practitioners and researchers alike, with doubts emerging regarding their ability to sway and impact consumer behavior. In response to these concerns, this research sheds light on how HVIs appearance influences consumer responses. Both human-like and lifelike qualities of HVIs boost social engagement by evoking feelings of warmth and competence. Notably, animism plays a key role in fostering a strong connection between HVIs and their audience, which in turn fosters consumers’ willingness to follow HVI’s recommendations and purchasing intention. To achieve optimal social engagement and positive outcomes, developers, retailers, and marketers of HVIs must strike a careful balance between human-like and lifelike characteristics in the design of these virtual entities.