Anthropomorphism refers to the inclination of individuals to ascribe human-like characteristics, motivations, intentions, and emotions to both imagined and actual behaviors of non-human entities (Epley, 2018). For robots to facilitate meaningful social interactions with humans, they must exhibit a certain level of personification in either their form or behavior (Duffy, 2003). Anthropomorphism refers to the extent to which consumers ascribe human-like traits to service robots (Blut et al., 2021). Empirical studies suggest that this phenomenon is shaped by consumers’ evaluations of the technology’s human-like attributes, including its physical appearance, nomenclature, facial characteristics, and perceived level of autonomy (Belk et al., 2023). The anthropomorphic design of robots is primarily manifested in two dimensions: appearance and behavior. The anthropomorphism of appearance pertains to the extent to which a robot visually resembles a human. This may include features such as a face or a human-like body, which shape human perceptions of robotic appearance (Aggarwal and McGill, 2007). Walters et al. (2008) categorized service robots into three classifications--mechanical robots, humanoid robots, and android robots--based on the degree of anthropomorphism in their appearance. Among these categories, mechanical robots exhibit the lowest level of anthropomorphic appearance, while android robots display the highest level. Conversely, the anthropomorphism of behavior relates to the extent to which a robot emulates human behavior and is capable of expressing emotions through body movements, vocalizations, and linguistic content during interactions, thereby demonstrating human-like characteristics (Bates, 1994). Kim et al. (2019) differentiated robot behavior into high and low anthropomorphic states based on the degree of anthropomorphism. In a highly anthropomorphic behavioral state, service robots are capable of executing limb movements, nodding, altering body posture, modulating vocal intensity, and conveying emotions through language (Kim et al., 2019). In contrast, a low anthropomorphic behavioral state is characterized by minimal physical movement and emotional expression. In the research undertaken by Lu et al. (2021), the notion of humanlike appearance is confined exclusively to the physical attributes of the service robot, which consequently limits the scope for diverse interpretations of its visual representation. Furthermore, the concept of human-likeness in linguistic style refers to the extent to which the robotic server replicates the behavior of human service staff by employing a literal rather than a figurative mode of communication.

Anthropomorphism plays a significant role in facilitating human acceptance of robots and serves as a fundamental framework for understanding consumer responses to robotic entities (Kim et al., 2019). From a design perspective, robots can elicit immediate human reactions to life-like or social cues without necessitating extensive cognitive processing. Consequently, individuals tend to apply established social interaction patterns and norms, typically reserved for human interactions, to their engagements with robots (Lee et al., 2005). From a human-centered cognitive standpoint, individuals articulate their comprehension of robotic functionality through specific psychological models. The anthropomorphism of both the appearance and behavior of service robots reflects consumers’ social cognition regarding these entities, viewed through the emotional lens of static impressions and the interactive lens of dynamic behavior. Specifically, appearance personification conveys static visual cues, while behavior personification communicates dynamic visual cues to consumers. Research indicates that dynamic visual cues are more effective than static cues in enabling consumers to extract nuanced information (Tidbury et al., 2016). The humanoid design of robots invokes associations with human physical characteristics; thus, the presence of more human-like features enhances the robot’s attractiveness (MacInnis and Folkes, 2017). Furthermore, when robots exhibit outgoing and cheerful behaviors, they are more likely to be accepted and followed by individuals (Goetz et al., 2003). Human-like behavior fosters the attribution of human cognitive processes to robots (MacInnis and Folkes, 2017), suggesting that if robots mimic human actions, individuals may apply their psychological models of human behavior to interpret robotic behavior (Lee et al., 2005). Consequently, consumers may respond to robotic services in a manner analogous to their responses to human services (Kim et al., 2016). Empirical studies suggest that robots possessing human-like traits elicit more pronounced social responses from individuals than those that are less anthropomorphic, potentially affecting perceptions of robotic agency (Kiesler et al., 2008). In particular, there is a positive correlation between the extent of a robot’s human resemblance and the propensity of individuals to ascribe agency to it, as well as to acknowledge its ability to engage in social interactions (Garvey et al., 2023). The anthropomorphism of service robots can engender a heightened sense of immersion for consumers, thereby establishing a social presence within the framework of human-computer interaction. However, an overreliance on anthropomorphic designs, particularly in the case of humanoid robots that closely resemble humans yet remain artificial, may lead to the uncanny valley phenomenon (Mori et al., 2012). This effect can evoke feelings of discomfort or alienation among individuals and may even incite technological biases, such as “speciesism” (Kim et al., 2019).

Social presence pertains to the subjective experience of social agents, including humanoid intelligence, and encompasses the sensation of co-presence with others (Biocca et al., 2003). Within the context of robotic service environments, it is defined as the degree to which robotic technology engenders a perception of another social entity’s presence among customers (Van Doorn et al., 2017). The process of anthropomorphism transforms robots from mere emotionless machines into entities capable of forming social and emotional connections with human counterparts. In the realm of human-computer interaction, the absence of a robust sense of social presence reduces the experience of social robots to a mere physical interaction with artificial entities (Fong et al., 2003). The primary objective in the design of social robots within this field is to facilitate a sense of social presence for users during their interactions (Lee et al., 2011). To achieve a genuine social experience in human-computer interaction, social robots must demonstrate a range of behaviors akin to those of real social actors. The level of immersion serves as a critical indicator of social presence, with the capacity to facilitate face-to-face interactions being a significant criterion for the successful establishment of social presence (Catherine et al., 2018). The visual representation of interactive objects can signify their degree of “authenticity” and “humanization,” whereby increased interaction with these objects correlates with heightened social presence. Furthermore, tactile feedback plays a vital role in enhancing the sense of social presence (Catherine et al., 2018).

The concept of social presence exerts a multifaceted influence on consumer attitudes and behaviors related to consumption. According to social impact theory, both the quantity of social participants (referred to as social scale) and the degree of social directness (or proximity) inherent in social presence can significantly affect consumer emotions and self-representation behaviors, ultimately shaping purchasing decisions (Argo et al., 2005). An increase in the scale of social presence is associated with heightened negative emotions and a greater emphasis on impression management among consumers. In response to these negative emotions and the need to maintain a favorable self-image, consumers tend to engage in a broader range of choices during the purchasing process. Furthermore, social presence has been shown to enhance consumers’ perceptions of safety and positively influence their purchasing attitudes (Wang et al., 2015), thereby affecting their behavioral intentions and trust within service contexts. Research conducted by Van Doorn et al. (2017) indicates that social presence impacts consumers’ social cognition and psychological ownership of interactive objects, which subsequently influences consumer satisfaction, loyalty, and repurchase intentions. Additionally, Xie et al. (2019) highlighted the role of social presence in shaping consumer conformity behaviors. Clearly, social presence has emerged as a significant variable affecting consumer decision-making processes. In the context of human-robot interaction, perceived social presence refers to the degree to which individuals perceive a machine as genuinely present (Heerink et al., 2010). van Doorn et al. (2017) proposed the concept of perceived social presence, defining it as “the extent to which machines (such as robots) evoke in consumers the sensation of being in the presence of another social entity.” Social presence holds significant implications in human-robot interaction (HRI), as fostering a robust sense of social presence during these interactions is often considered the primary objective in the design of socially interactive robots (Fong et al., 2003). Social interactions are not solely confined to human beings; rather, technologies have the capacity to replicate human characteristics in both appearance and behavior (Caić et al., 2020). Such interactions between technology and humans are frequently characterized as quasi-social interactions (Carlos et al., 2024).

Variety-seeking is a significant characteristic and selection strategy employed by consumers during the acquisition of products (McAlister, 1982; Liu et al., 2015). This behavior reflects consumers’ inclination to seek diversity in their choices of goods and services (Olsen et al., 2016). Academic research offers two primary interpretations of this phenomenon (Sevilla et al., 2019). The first interpretation emphasizes decision-making over time, characterizing diversity seeking as the extent to which consumers switch between various options within the same product category (Mcalister and Pessemier, 1982). For instance, Kahn et al. (1986) posited that the pursuit of diversity represents consumers’ tendency to diverge from their previous brand purchases. Similarly, Sevilla et al. (2019) contended that diversity seeking manifests as a preference for selecting alternatives that differ from prior choices, suggesting that even alternating between two options can elicit change-seeking behavior. The second interpretation focuses on the quantity of distinct items selected within a consumer’s purchasing history or product portfolio (Ratner and Kahn, 2002). Gullo et al. (2019) defined diversity seeking as the ratio of unique product categories acquired within a specific category relative to the total number of items purchased, which can also be conceptualized as a function of the combinations of consumer choices (Sevilla et al., 2019). Consequently, consumer variety-seeking can be articulated as the degree of transition between new and existing products or different brands within the same product category, as well as the quantity and proportion of varied items in consumers’ decision-making processes.

From a social psychology perspective, individuals’ behaviors in seeking diversity can be categorized into two types: true variety-seeking and derived variety-seeking (Mcalister and Pessemier, 1982; Van Trijp et al., 1996). The distinction between these two types lies in the underlying motivation for the observed transformational behavior, which can be either intrinsic or extrinsic in nature (Mcalister and Pessemier, 1982). True variety-seeking is posited to arise from intrinsic motivations, where individuals pursue diversity driven by a desire for change, novelty, or satisfaction with product attributes. In contrast, derived variety-seeking is influenced by external motivations that are not directly linked to an individual’s intrinsic desire for diversity (Mcalister and Pessemier, 1982). Furthermore, research indicates that various factors, including personal characteristics of consumers, product attributes, and environmental or stimulating factors, significantly influence variety-seeking behavior (Gullo et al., 2019; Song and Liu, 2020). Notably, elements such as the composition and size of the product selection set, the complexity of the consumer decision-making context, and the dimensions of the spatial environment are recognized as critical environmental factors that impact consumers’ variety-seeking behavior (Gullo et al., 2019; Levav and Zhu, 2009), garnering increasing attention from scholars in the field.

A service robot functions via an integrated control system that enables it to replicate human behavior, convey emotions, and deliver services to clients (Seo, 2022). Despite the growing utilization of service robots in recent years, the concept of service robots functioning as service representatives remains a novel experience for the majority of customers. Novelty can be defined as the perception of originality that arises from the juxtaposition of current awareness with past experiences (Kim et al., 2021). Customers encounter a unique experience when a service engages their curiosity and satisfies their desire for knowledge within an enhanced context (Sheth et al., 1991). Consequently, novelty can be understood as a divergence in the perception of prior experiences compared to current realities, signifying an experience that is new and unfamiliar, distinct from previous encounters (Albaity and Melhem, 2017). Consumers engage in comparative evaluations of current stimuli against those encountered in the past, recognizing novelty when their present perceptions exceed prior experiences (Lee and Crompton, 1992). Consequently, the pursuit of novelty is intrinsically linked to the quest for diversity and serves as a significant determinant in consumers’ choice motivations and decision-making processes (Petrick, 2002). This phenomenon reflects consumers’ propensity to alter their selections based on recent experiences (Ratner et al., 1999).

Empirical studies have substantiated the influence of anthropomorphism in service robots on customer behavioral intentions. For instance, Mende et al. (2019) demonstrated through experimental research involving humanoid robots that a heightened level of anthropomorphism can elicit discomfort among customers, subsequently prompting compensatory consumer behaviors. Similarly, Liu et al. (2021) demonstrated that, in the realm of tourism service interactions, a higher level of anthropomorphism in service robots is positively associated with customers’ readiness to participate in the value co-creation process. Furthermore, Li and Zheng (2021) highlighted that, from the perspective of appearance anthropomorphism, the extent of anthropomorphism in service robots positively influences customers’ diverse seeking behaviors. Additionally, Zhang and Wang (2022) identified an inverted U-shaped relationship between the level of anthropomorphism in the appearance of service robots and consumers’ willingness to utilize them, indicating that both low and high levels of anthropomorphism may deter usage. Collectively, these studies suggest that the anthropomorphism of service robots can elicit various customer intentions and behavioral responses. However, the current body of research predominantly emphasizes the differential impacts of varying degrees of appearance anthropomorphism, with limited exploration into the effects of differing levels of behavioral anthropomorphism.

Individuals exhibit primary, secondary, and mechanistic variations in their perceptions of the anthropomorphic characteristics and behaviors of service robots. The anthropomorphism of appearance primarily conveys static visual cues to consumers, whereas the anthropomorphism of behavior provides dynamic interactive cues. Distinctions exist between two types of cognition in specific contexts, with corresponding differences in the cognitive regions and mechanisms within the human brain. When individuals assess the anthropomorphic appearance of service robots from a static impression standpoint, they are generally not influenced by the robots’ behaviors. Conversely, when individuals evaluate the anthropomorphic behaviors of service robots through the lens of dynamic interaction, the robots’ actions and the fluidity of human-machine interaction become the focal point of consumer attention, overshadowing the robots’ appearances. Fink (2012) emphasizes that the efficacy of robots in human-computer interaction is more closely tied to their behaviors, with successful behaviors serving as a fundamental aspect of interaction. A high degree of anthropomorphism in behavior facilitates a more natural, effective, and appealing human-computer interaction, fostering a sense of closeness. When robots engage in social behaviors such as making eye contact, nodding, and utilizing arm movements, users exhibit increased attention, trust, and empathy towards these robots (Bartneck et al., 2007). The anthropomorphism of service robot behavior is more likely to evoke a sense of immersion among consumers, thereby establishing a social presence within the context of human-computer interaction; in contrast, merely observing the static appearance of a robot is less effective in generating such a social presence. Furthermore, the anthropomorphic appearance of robots tends to elicit only a brief period of stimulation, primarily during initial encounters. In contrast, the anthropomorphism of robot behavior can engender a lasting sense of social presence that recurs across multiple service interactions. When customers engage with service robots exhibiting anthropomorphic behaviors, they not only experience curiosity and novelty but also perceive the presence of another social entity through interaction and communication, thereby cultivating a robust sense of social presence (Xie et al., 2019). It is evident that merely observing the anthropomorphic appearance of service robots is insufficient to foster a meaningful sense of social presence; rather, it is through the observation and interaction with anthropomorphic behaviors that a genuine sense of social presence can be established.

The extent of anthropomorphism exhibited by service robots significantly influences customers’ ability to accurately conceptualize the robot’s physical form and its activities, thereby impacting the perceived level of social presence. The embodiment theory posits that individuals engage in multi-channel representation and simulate perceptual, kinesthetic, and introspective information pertaining to non-real physical entities during cognitive processing (Wu et al., 2011). When service robots display a higher degree of anthropomorphism, they are capable of executing smoother limb movements, nodding, altering body posture, modulating vocal volume, and articulating emotions through language. In such scenarios, customers are likely to construct a more authentic experience of the robot’s physical presence and its activities, facilitating cognitive processing that activates multi-channel representations and simulates relevant perceptual, action-oriented, and introspective information. This leads to a heightened perception of the robot as a social entity, resulting in an enhanced sense of social presence. Conversely, when the degree of anthropomorphism in robot behavior is minimal, the service robot struggles to perform fluid limb movements, nod, change posture, modulate volume, or convey emotions effectively through speech. In these instances, customers encounter difficulties in realistically conceptualizing the robot’s physical form and engaging with its activities, which hampers cognitive processing. Consequently, they find it challenging to activate multi-channel representations and simulate the necessary perceptual, kinesthetic, and introspective information, leading to a diminished perception of the robot as a social entity. As a result, customers experience a lower sense of social presence in such contexts.

The concept of social presence exerts a multifaceted influence on consumer attitudes and behaviors. According to social influence theory, the quantity of social entities and the degree of social immediacy associated with social presence can significantly affect consumer emotions and self-representation behaviors, which subsequently influence purchasing decisions (Argo et al., 2005). Empirical research has demonstrated that social presence serves as a fundamental prerequisite for various critical consumer behavioral intentions. It has the potential to enhance trust in service contexts, improve perceptions of safety, and positively shape purchasing attitudes (Lee and Shin, 2012). Furthermore, social presence can augment hedonic value and foster favorable consumer attitudes. It also impacts consumers’ social cognition and psychological ownership of interactive objects, thereby influencing their satisfaction, loyalty, and repurchase intentions. Additionally, social presence may affect consumers’ conformity behaviors and their propensity for diverse seeking behaviors. In light of these findings, the following hypotheses are proposed:

Hypothesis 1

Service robots exhibiting higher levels of anthropomorphism in behavior will elicit stronger variety-seeking behaviors from customers compared to those with lower levels of anthropomorphism.

Hypothesis 2

Social presence mediates the relationship between the anthropomorphism of robot behavior and customers’ variety-seeking behaviors.

The context in which consumers make decisions plays a crucial role in determining whether their choices are private or public in nature. In public environments, consumers tend to engage in a wider array of behaviors aimed at projecting a distinctive and open personal image (Ariely and Jonathan, 2000). The presence of others creates a clear social consumption scenario, wherein consumer behavior is readily observable. Under these circumstances, consumers’ emotions, cognitions, and behaviors are likely to shift, leading to an increase in diverse seeking behaviors, impulsive purchases, symbolic consumption, and other forms of social consumption (Inman et al., 2009). The Social Facilitation Theory posits that the presence of others can serve as an internal motivator, influencing individuals’ attitudes and behaviors (Kushnir, 1981), and prompting a greater propensity for herd buying decisions (Xie et al., 2019). Ratner and Kahn (2002) conducted a comparative analysis that highlighted the differential effects of private versus public consumption on diversification-seeking behavior, revealing that individuals tend to incorporate more variability into their consumption choices when their actions are subject to public scrutiny. Jinhee et al. (2006) found that consumers are more inclined to pursue diversified options in the presence of others, as their behavior becomes observable, thereby aligning with social norms and the expectations of others. Furthermore, Wang et al. (2020) discovered that the presence of acquaintances diminishes consumers’ diversification-seeking behavior, whereas the presence of strangers enhances it; this indicates that individuals with varying self-conceptions are significantly affected by the social context. Consequently, the number of individuals involved in decision-making processes influences consumers’ diversification-seeking behavior. Based on these findings, the following hypothesis are proposed:

Hypothesis 3

The context of consumption decision-making moderates the effect of social presence on consumers’ variety-seeking behavior.

Hypothesis 3a

For consumers engaged in public decision-making, their variety-seeking behavior is significantly heightened when social presence is high compared to when it is low.

Hypothesis 3b

There is no significant difference in the influence of social presence on variety-seeking behavior among consumers making private decisions.

In light of the preceding analysis, the theoretical framework is illustrated in Figure 1.

The pre-experimental involves the creation and validation of anthropomorphic video materials intended for the examination of service robot behavior. Following a meticulous selection and editing process, two experimental videos were produced to illustrate varying degrees of anthropomorphism in accommodation service robots. These videos were derived from subtitled footage recorded by Kim et al. (2019), which effectively anthropomorphizes robot behavior while controlling for confounding variables. A highly anthropomorphic state refers to service robots that are equipped with the ability to manipulate their limbs, display nodding behaviors, alter their body posture, and modulate their vocal volume, which enables them to convey emotions during verbal exchanges (Kim et al., 2019). Conversely, a low anthropomorphic state indicates that the physical movements and emotional expressions demonstrated by service robots are relatively limited (Kim et al., 2019). For the purposes of this study, the English dialogue in the videos was accurately translated into Chinese, accompanied by corresponding Chinese subtitles, as detailed in the appendix documentation. The videos used for the study are available as Supplementary Material, Supplementary Video 1 is the high anthropomorphic condition and Supplementary Video 2 is the low anthropomorphic condition. Table 1 illustrates the primary distinctions between high-level and low-level anthropomorphic videos. Preliminary testing with a small sample indicated that these two videos successfully differentiate between high and low levels of anthropomorphism in service robots.

The formal experiment is structured to test hypotheses through a series of six sub-experiments, as outlined in Table 2. Experiments 1a and 1b are designed to validate the primary effect by employing different dependent variable frameworks. Initially, participants will be randomly assigned to an intergroup design based on varying levels of behavior anthropomorphism (low vs. high). They will then assess the anthropomorphism level of the manipulated independent variable. Subsequently, participants will be exposed to distinct robot service scenarios and will engage in a survey regarding their consumption decisions related to hotel accommodations, specifically focusing on their inclination to select hotel dinner packages and the quantity of complimentary fruit juice drinks. Finally, participants will provide demographic information. Experiments 2a and 2b are intended to examine the mediating effect of social presence. In these experiments, interactive service scenarios and catering service scenarios featuring service robots with consistent behavioral characteristics will be developed to enhance the visual stimulation experienced by participants. Following their exposure to interactive video materials showcasing the service robot, participants will evaluate their perceived sense of social presence. After interacting with the catering service scenarios, participants will assess their propensity to switch between dinner meal options and the number of juice drinks they opt to select. Experiments 3a and 3b are focused on verifying the moderating effect of the context in which customers make consumption decisions, with participants being assigned to either public or private decision-making contexts.

Regarding sample selection, both the pilot study and the six formal experiments were conducted using the prominent survey platform in China, SoJump, which operates on a payment basis. The experimental subjects comprised ordinary consumers with diverse backgrounds, including variations in occupation, industry, education, and age, thereby enhancing the internal validity of the experiments. Participation was entirely voluntary, with all participants thoroughly reviewing the experimental instructions prior to commencement, thereby gaining an understanding of the study’s background and objectives, and subsequently consenting to partake in the research. On average, each experimental group consisted of approximately 60 effective participants, with specific demographic information about the sample provided in the analysis results section of each experiment.

The manipulation of the independent variable across the six sub-experiments was informed by the methodology established by Kim et al. (2019). This approach enabled participants to engage with textual descriptions of service scenarios as well as service videos depicting various behaviors, thereby enhancing the consistency of the experimental process. The selection of participants and the execution of the experiments were conducted through reputable online research platforms, incorporating individuals from diverse backgrounds, including variations in occupation, industry, education, and age, which bolstered the internal validity of the study. The six experiments employed two distinct methods for measuring the dependent variable: conversion tendency and selection quantity, thereby augmenting the external validity of the findings.

The results of the analysis indicate that hypotheses Hypothesis 1 and Hypothesis 2 are substantiated, whereas hypothesis Hypothesis 3 and its sub-hypotheses receive only partial validation. This discrepancy can be attributed to the varying moderating effects of customer consumption decision-making contexts within high and low anthropomorphic behavior robot service scenarios. In scenarios characterized by high anthropomorphic behavior, customer consumption decision contexts significantly moderate the influence of social presence on variety-seeking behavior. Conversely, in low anthropomorphic behavior scenarios, the moderating effect of customer consumption decision-making context is not statistically significant.

In high anthropomorphic behavior robot service scenarios, customers experience an enhanced sense of social presence. This heightened perception arises from the robots’ ability to exhibit more fluid limb movements, nodding, body posture adjustments, volume modulation, and emotional expression through language. Consequently, customers are able to construct a more authentic experience of the robot’s physical form and its activities, engaging in cognitive processing that activates multi-channel representations and simulates perceptual, action-oriented, and introspective information. This process fosters a greater recognition of the robot as a social entity, culminating in an intensified sense of social presence. In such contexts, customers exhibit a higher degree of immersion, allowing them to more vividly perceive the visual representation of the interactive object, engage in more substantial interactions, and clearly observe the collaborative tasks performed by the robot, thereby reinforcing their sense of social presence.

In contrast, low anthropomorphic behavior scenarios present limitations, as the service robots lack the ability to perform smooth limb movements, nod, adjust body posture, or modulate volume, and they struggle to convey emotions through speech. As a result, customers find it challenging to construct a realistic understanding of the robot’s physical form and activities, which impedes their cognitive processing of the experience. This difficulty in activating multi-channel representations and simulating relevant perceptual, action-oriented, and introspective information leads to a diminished recognition of the robot as a social entity, resulting in a lower sense of social presence among customers. Therefore, in high anthropomorphic behavior robot service scenarios, customers generate a more pronounced sense of social presence, with their consumption decision-making context potentially serving as a moderating factor that enhances this experience. Conversely, in low anthropomorphic behavior scenarios, the inherently lower sense of social presence limits the significant regulatory role of customer consumption decision-making contexts. Future research is warranted to explore the specific moderating effects of customer consumption decision-making contexts in greater detail.