1 Introduction

Even before the Internet, cyberspace existed. It was formed of text-based networks that laid the groundwork for modern Internet communication, which is becoming more embodied through graphical representations of the self online and, potentially, in virtual reality (VR). This article discusses the evolution of digital embodiment—from the early days of cyberspace and VR, as they developed contemporaneously but separately, to today—and the narratives being promoted about it, which often differed. In the physical world, our bodies impact our experiences as we are judged along lines of gender, race, disability, and other facets of our identities. While the form and extent of embodiment in cyberspaceFootnote 1 and VR have changed, what has not changed is that virtual embodiment also creates different experiences for different people, often along the same lines as in the physical world. As we live more and more of our lives online, our on- and offline lives merge into an “onlife” (Floridi 2015), and our experiences in virtual spaces significantly influence our well-being.

This article analyzes historical perceptions of virtual embodiment,Footnote 2 examining its implications for our experiences online and in extended reality (XR), which encompasses VR, augmented reality (AR), and mixed reality (MR) (Mann et al. 2018).Footnote 3 Ideas of virtual embodiment have evolved from a Cartesian dualist perspectiveFootnote 4 where a user’s cyberspace embodiment was seen as independent from physical characteristics, to acknowledging that physical characteristics could be embodied in cyberspace, to insisting that they should be celebrated. Throughout this development, this article will argue that narratives promoted by the proponents of emerging technology—mainly able-bodied white men—have overlooked the experiences of marginalized communities, who are often embodied in cyberspace in ways that create disparate experiences. By analyzing the predominantly AmericanFootnote 5 origin of cyberspace, consumer VR, and the embodiment debate, this article underscores the importance of recognizing how our bodies impact our experiences in virtual environments and the need for vigilance so that harms to marginalized communities are not overlooked by those controlling the narratives of emerging technologies.

This article’s embodied concept of experience is drawn from phenomenology, following a long line of phenomenologists including Husserl, Heidegger, and Merleau-Ponty who argued that experiences are formed by interactions of the body, mind, and surrounding environment. Both Husserl and Heidegger emphasized the relational nature of the body through their concepts of the “lived body” (Landgrebe 1973; Carman 1999) and Dasein (Heidegger 1996, p. 153), holding that the body’s interfacing with the environment is what forms our experiences. Merleau-Ponty elaborated on the idea of the “body schema,” a proprioceptive and pre-reflective model of our body’s abilities, limitations, and position. It is often confused with the “body image,” which includes the conscious perception of the body, the conceptual construct of the body, and emotional attitudes and feelings towards the body (Gallagher 1986). Self-presentation is a third important Merleau-Pontian concept, covering the way someone shapes perceptible aspects of themselves to convey a desired self-presentation (Schlenker 2003; Goffman 1956), which becomes more malleable in cyberspace as individuals can construct multiple identities within the confines of digital media (Jensen Schau and Gilly 2003). Any theory of embodied experience online must account for these three aspects, which combine to impact our experiences. These theories deal with the biological reality of physical embodiment, but this article will apply them to virtual environments and also discuss the social impacts of different forms of embodiment.

This article’s analysis will proceed chronologically, covering cyberspace and embodiment through three distinct eras. Sections 2 through 4 each address one of these eras; Sect. 5 concludes.

2 Early years: text, disembodiment, and textual disembodiment

Cyberspace and virtual reality (VR) were developed in parallel and before widespread accessibility to the Internet, and certainly before the creation of social media. Early discourse around cyberspace portrayed it as a place where people were fully disembodied and liberated from the bonds of their physical bodies, but this vision was not supported in reality. This Cartesian discourse was promoted by early adopters who, in many cases, were also building this new technology, and it accorded with technological constraints that limited interactions to text. In contrast to early proponents, many users showed that they wanted aspects of their physical identities to be relevant online, foreshadowing future evolution in this direction. VR discourse also included a thread of disembodied escapism, but its earliest advocates saw it as a way to connect people where the transcendence of the body was a necessary step rather than the primary purpose, and then appreciate the physical world more on return. I will first discuss cyberspace, followed by VR.

2.1 Cyberspace

The first iteration of cyberspace, emerging in the 1980s,Footnote 6 was text-based and characterized by bulletin board systems (BBSs), multi-user domains (MUDs), and, later, MOOs (“MUD, object-oriented”) where people exchanged information, chatted, and role-played. At the time, the term “cyberspace” was still developing. William Gibson is often credited with coining the word, using it to refer to a 3D immersive dataspace in his 1982 short story Burning Chrome and its 1984 sequel Neuromancer (Gibson 1986, 2000), but it was in fact first used by Danish artist Susanne Ussing and architect Carsten Hoff in the 1960s as they produced cybernetic-inspired works under the name “Atelier Cyberspace” (Lillemose and Kryger 2015). Under our definition, though, cyberspace was a text-based medium at the time. It was heralded by early adopters as a place where the mind and body could be totally separate in a version of Cartesian dualism and was portrayed as a world of “total representation” (Heim 1993, p. 100) where “minds are connected to minds, existing in perfect concord without the limitations or necessities of the physical body” (Gunkel and Gunkel 1997).Footnote 7 This Cartesian split was promoted by culture-shapers like Stewart Brand and Larry Brilliant, founders of the Whole Earth ‘Lectronic Link (WELL) BBS that was, while small in terms of userbase, enormously influential on early cyberculture (Turner 2010, p. 142) and, reflecting their links to counterculture and communal movements, promoted cyberspace as a “disembodied, peer-to-peer utopia” that was “the route to new forms of equality and communion” (Turner 2010, p. 16; 2010, 261).

These views were reflected not only by culture leaders but also by early adopters of cyberspace. Cyberspace was initially occupied primarily by tech-savvy, young, single white men (Driscoll 2022, 21–23). This gave them significant influence on the culture that lasted for years. Many of these early adopters were self-styled “hackers”Footnote 8 who developed their own culture, laid out in the “Jargon File.” An appendix entry on “Gender and Ethnicity” acknowledged the gender and racial composition of “hackerdom” and its general belief in textual disembodiment:

Hackerdom is still predominantly male… In the U.S., hackerdom is predominantly Caucasian… Racial and ethnic prejudice is notably uncommon and tends to be met with freezing contempt. When asked, hackers often ascribe their culture’s gender- and color-blindness to a positive effect of text-only network channels, and this is doubtless a powerful influence (Gender and Ethnicity n.d.).

The “hackers” were the pioneers of cyberspace and had a significant role in shaping its culture and surrounding narrative. Their views were shared by some contemporaneous observers, including critical race, disability, and feminist scholars who saw cyberspace as a revolutionary way to break down societal inequalities (Rudnicki 2017), for if people were judged only on their minds, discrimination based on bodies would have to cease.

To some extent, some early cyberspace participants did feel disembodied, or at least anonymous. For instance, a MUD participant stated to a researcher that “In RL [real life], if I’m black, or handicapped, or don’t have a college degree (or even a highschool [sic] degree), that will affect how I’m treated. That matters for nothing here, really” (Kendall 1998). However, other accounts and studies showed that racist, classist, and ablest harassment (and thus embodiment) did exist in cyberspace. One could try to maintain anonymity and have these factors not matter, but anonymity also emboldened those who were inclined to harass other people online (Lea and Spears 1991). Harassment, especially sexual harassment, became so prevalent that women joining BBSs and MUDs were warned about “netrape” where female personae were textually violated (Cherny 1994), showing that the narrative of academics and early adopters did not apply to those who did not share the dominant and “default” identities of whiteness and maleness of the majority of early cyberspace adopters (Driscoll 2022, 22–23) and did not, or could not, maintain anonymity. Those who were not members of the “default” group had to be vigilant to avoid harassment and thus were not leaving their physical bodies behind.

Identity could be communicated directly when users chose to disclose it, or indirectly. Physical identity could be inferred through typed speech, whether through words and vernacular or the way one interprets experiences, and “passing” as another gender or race required a deep understanding of the ways that one’s identity affects speech (Turkle 1997, p. 212), with men’s and women’s online communication styles differing enough to be considered “different cultures” (Herring 1996). This is more reminiscent of phenomenology than Cartesian dualism; it shows that one’s body impacts experiences in the physical world, which impacts how the mind thinks and thus communicates in cyberspace, and so even if someone knew nothing about the body of the person they were communicating with, their mind was still shaped by their bodily experiences. So, even if bodily anonymity was maintained by not disclosing their physical identity, a mind-body split was not achieved.

Furthermore, people in cyberspace were not afforded no identity—that is, they were not a brain floating in the ether—but were instead granted an assumed identity of whiteness and maleness (Nakamura 2002, 47–48; Moser and MacLeod 1996) because this was the identity of the majority of cyberspace users at the time, and people wanted to tie the text on their screen to a physical body. Boler (2007) noted that “A/S/L?” (“age/sex/location?”) became one of the most common greetings in cyberspace and was a way to seek an essentialist grounding for the other person. MUD users gave “greater weight to offline identity than online identity,” prioritizing information about the physical body over a virtual embodiment, and reported having more trust in those they had met face-to-face (Kendall 1998). Indeed, early cyberspace was tightly coupled to the physical world. Many BBSs catered to mostly local communities, partly because of the high cost of long-distance phone calls (Driscoll 2022, p. 16), and often facilitated face-to-face meetings, community gatherings, and even dates (Driscoll 2022, 153–56; Kendall 1998). Attending a physical meetup would then tie the physical body even more closely to one’s online presence. Even absent physical meetings, anonymity was not disembodiment—users were still bound to a body, but not necessarily their own. If they disclosed aspects of their physical identity, or if it was inferred based on their typed speech, they could embody a digital persona more like their physical identity, but no longer enjoyed the “neutrality” they previously had.

Although even at the time there was evidence that the narrative of disembodied equality was not true for everyone (or anyone), those who promoted it were reluctant to engage with contradictory viewpoints. For the majority, “default” identity, it was convenient to believe that the same applied to everyone. This was evident at a 1991 BBS conference, where many speakers “hailed the coming of an egalitarian virtual community, free of bias” (Driscoll 2022, p. 174). However, the only Black woman at the conference, Lisa Downing, afterward published an open letter in Boardwatch magazine where she described how she was “snubbed and ignored” all weekend, cautioning that “[technology] will not erase the prejudices of the user” (Downing 1991, as cited in Driscoll 2022, 173–74). The editor of Boardwatch responded, insisting that she was the cause of any conflict and saying, “I doubt you were as unique and out of place as you apparently delight in thinking you were” (Rickard 1991, as cited in Driscoll 2022, p. 174). No one else in the community engaged after this. Regardless of exactly what happened at the conference, the way that the mostly white, mostly male community refused to engage with a critique of the culture they promoted was telling, reflecting a reluctance to engage on issues of race and gender when they contradicted their narrative of egalitarianism. White men were the majority of those online and thus the assumed identity who did not have to consider how their identity impacted their experiences (Driscoll 2022, 22–23), but believing that it was the same for everyone was erroneous.

Early cyberspace reflected the reality of the American society where it was born, which was awash with narratives of colorblindness and loath to discuss race (Kendall 1998). While the founders of early cyberspace likely genuinely believed in its potential as an equalizing force, the affordances of the technology at the time meant that only text-based communication was possible. It is logical that they would be trying to identify the supposed benefits of the technologies that they were pioneering. In general, they could be confident that, as a majority, they would not be harassed based on their gender or race, but they were often willfully blind to the fact that it did not carry over to all identity groups.

What was perceived as disembodiment by the primarily white, primarily male pioneers of cyberspace was instead a form of semi-anonymous embodiment, just in a default, “invisible” identity that one could exist in unchallenged because it was the dominant identity. Not everyone could access this form of uncomplicated embodiment, but many users desired it. One way this manifested was in identity-oriented spaces. Identity-oriented BBSs, including for women, people of color, and disabled users, and specific sections of general-purpose BBSs sprung up to provide users with safe cyberspaces (Driscoll 2022, 151–54). By making specific non-dominant identities the majority, these shared spaces allowed them to think about physical identity less and experience the same bodily anonymity that white men did in most of cyberspace, escaping what could otherwise feel like a “nightmare” (Driscoll 2022, p. 151). Since inclusion in these spaces was predicated on one’s physical identity, virtual identities were still grounded in physical ones.

Even outside of affinity spaces, people showed a desire to bring a physical body into cyberspace. MOOs and MUDs often involved role-playing that required users to create a character description. Notably, following the color-blind discourse of the day, many of these eschewed formal categories for race, even though some included them for gender (Nakamura 2002, 37–38). These self-descriptions allowed people to try on new identities. Rather than exist as floating minds in the ether of cyberspace, they actively portrayed themselves not only as their physical identities truly were, but as other identities. Some claimed they did so for practical reasons; one man who played in MUDs as a woman said, “I do it to improve the ratio of women to men. It’s just a game” (Turkle 1997, p. 212). Sometimes this was to explore aspects of their own identity; individuals questioning their gender identity could virtually embody different genders online in a way they could not necessarily do in the physical world. Regardless of motive, identity switching was a common practice; a study of MOOs found that 40% of participants in social MOOs had engaged in gender-switching, and more than half of participants in role-playing MOOs had (Roberts and Parks 1999).

Sometimes this led to problematic forms of embodiment. Many female personae, especially those played by men, were extremely sexualized (Nakamura 2002, 43–47). Furthermore, users often adopted highly stereotyped racial identities in cyberspace, a kind of “identity tourism” that entrenched those racial stereotypes (Nakamura 2002, 43–44).Footnote 9 Ironically, these stereotyped embodiments were encouraged by the text-based medium; Nakamura noted that cyberspace being “disembodied” meant that “the need to create very clear, recognizable personae is thus a practical one” (Nakamura 2002, p. 39). In creating these “clear, recognizable” characters, people often fell back on stereotypes. This reality differs significantly from the Cartesian belief of early adopters that minds could exist merely as minds, with no ties to the physical body. Users wanted to bring some kind of body with them—whether their own, someone else’s, or an invented one—and so “in constructing this necessary difference, the subject has recourse only to those markers of difference that already exist within the symbolic order… racial stereotypes provide familiar, solid, and reassuring versions of race” because of their presence in media (Nakamura 2002, p. 40). By bringing a body that was not theirs in the physical world into cyberspace, people also brought racial stereotypes of the physical world into cyberspace.

Those who tried on different identities found that inhabiting them convincingly was difficult for the abovementioned reasons of how physical identity influenced textual communication; to “pass,” one could not merely change their self-description. However, it could also teach people about identity in the physical world. For example, “[w]hen men playing females are plied with unrequested offers of help on MUDs, they often remark that such chivalries communicate a belief in female incompetence” (Turkle 1997, p. 214). Bruckman (1996) noted the same, and also that sexual favors were often expected from female personae in exchange for assistance (320–21). Even in textual environments, the divorce of mind and body was not always total. In a phenomenological exchange, the body informed interactions and experiences in cyberspace, which in turn shaped how people saw physical identities.

Although users could be anonymous in early cyberspace under specific circumstances, they could never be truly disembodied, and this was not universally accessible even to those who could access cyberspace. Nor was it universally desired. This Cartesian “forgetting” of the body (Stone 1991, p. 113) was a narrative propagated by the cultural discourse and technical limitations of the time. Based on a naïve understanding of biological embodiment, it promised liberation but ignored the social realities of textual interactions and the fact that people wanted to anchor the people on the other side of the screen in a physical body. Instead, people’s physical identities impacted their experiences online, and people used their experiences in cyberspace to shape their perceptions of others in physical space.

2.2 Virtual reality

Virtual reality research was being conducted at the same time that cyberspace was developing, but the two were largely separate due to technological limitations.Footnote 10 Two independent narratives about its development arose: one from one of the founders of the field, and one from academia and science fiction that was picked up in popular discourse. Although now VR headsets are the most common VR device, the most prominent VR development company, VPL Research (founded in 1984), initially focused only on VR gloves and suits because computers were not fast enough to support headset graphics (Lanier 2017, 226–27). They eventually released the “EyePhone,” a VR visor, but the initial focus on gloves and suits shaped VPL’s VR approach to focus on experiences through creative methods of input rather than emulating physical bodily interactions with the physical world and high-fidelity depictions of it. VPL’s founder, Jaron Lanier,Footnote 11 saw VR as a way for people to connect on a small scale, overcoming discord through experiences so engaging and unlike physical reality that they transcended differences (Egan and Lanier 2022). Some of these involved manipulating the body schema, such as by inhabiting a lobster body (Won et al. 2015) or by being without an avatar and experiencing the “whole world [as] your body” (Barlow 1990b).Footnote 12 These examples of “homuncular flexibility” (Won et al. 2015) featured diverse input methods, but all used the physical body as input in the style of reality-based interaction (RBI)Footnote 13 (Jacob et al. 2008), and were intended to enhance appreciation of the physical world, not replace it. Foreshadowing the narrative schism, some who tried VR interpreted it as a dissolution of the physical body (Barlow 1990a). However, for its founders, VR was a very embodied experience, and any leaving behind of the body was a side effect rather than the primary goal, in contrast to the goals of the pioneers of cyberspace.

The founders of VR saw it as serving a very different purpose than early promoters of cyberspace, although its purpose was similarly shaped by technological limitations. It was not intended to separate the mind and body in the Cartesian style. It was instead supposed to promote an appreciation of the physical world and connect minds, but on a small scale (which was all the technology could support at the time), and by engaging using the body, not abandoning it. As Lanier wrote, people working on VR at the time were thinking that “maybe we’d collectively create a global virtual space, but even then, the connection with another individual should remain more cherished” (Lanier 2017, p. 191). This required instilling a sense of presence, or “being there,” created through the “place illusion” that made the user feel like they were in a real place, combined with the “plausibility illusion” that what was being depicted in the virtual environment was actually occurring (Slater 2009). Thus, in focusing on physical interaction, it was reminiscent of the phenomenological view of embodiment and experience. The focus on small-scale connection differs technologically and philosophically significantly from the “metaverse” of today’s discourse, which is intended to be this “global virtual space.” This evolution will be discussed further in Sect. 4.

Early VR did not have avatars due to technical constraints and so were first-person experiences in a “floating point of view” (Balsamo 1996, p. 126). Users of VPL’s system employed the DataGlove to manipulate the virtual environment, tying the physical body to the virtual experience. VR researchers quickly implemented avatars, although early ones were limited to a floating head or single hand due to computer speed limitations (Lanier 2017, p. 247). Eventually, blocky full-body avatars that could be driven with VPL’s DataSuit were developed (Lanier 2017, p. 247). These served practical purposes in games, but also in VPL’s mission to connect people. The “Reality Built for Two” (RB2) included avatars so that multiple people could cohabitate in a “virtual world” simultaneously (Lanier 2017, p. 14). These avatars were not meant to be a replacement for the physical body, but a different incarnation of them—a “thought body” or “virtual body” (Barlow 1990b) that changed the body schema and created a “transfixing” experience (Lanier 2017, 15–16). Furthermore, the RB2 system focused not on emulating real-world activities, but on building virtual worlds together (Blanchard et al. 1990), showing its goal to transcend the limitations of the physical world, rather than the physical body.

Even though it was being developed in small Silicon Valley labs, VR quickly captured the cultural imagination, and it became associated with the same rhetoric of disembodiment as early cyberspace, facilitated by science-fiction and cyberpunk novels that appealed to both hackers and the masses. Early works in the 1980s like Tron and Neuromancer featured characters entering immersive dataspaces of Gibsonian “cyberspace,” but it was not until 1992’s The Lawnmower Man that VR was shown as an immersive virtual world experienced through headsets and VR clothing (provided by VPL) with the user inhabiting an avatar (Drummond et al. 2014). The proliferation of VR-related media in the 1990s culminated in 1999’s The Matrix (Drummond et al. 2014) and cemented the image of VR as a virtual world intended to emulate the physical world, rather than as a tool for small-scale connection. These media promoted the Cartesian mind-body split by merging Gibsonian cyberspace and VR and took a contemptuous view of the physical body. Neuromancer describes the protagonist’s loss of ability to enter immersive “cyberspace” as “the Fall,” invoking the Biblical ban from Eden (Gibson 1986, p. 8). The catastrophe is justified as so: “The body was meat. Case fell into the prison of his own flesh” (Gibson 1986, p. 8). This disdain for the physical body was common in cyberpunk literature and painted one’s virtual embodiment as more important than the physical body with the goal of “reaching the eventual obsolescence of the body” (Ajana 2005)—indeed, it was often intertwined with the egalitarian discourse of hacker culture—and portended the later focus on full-body VR avatars.

As VR entered the public consciousness, it also entered the academic discourse, and the pop culture dialogue on the body was replicated. Lanier’s borderline-spiritual views fell by the wayside after VPL filed for bankruptcy in 1990 (Kim 2015). The result was a promotion of disembodiment similar to that surrounding cyberspace, facilitated by the “representational” disappearance of the physical body (Balsamo 1996, p. 126). Penny (1993) wrote that “Virtual reality replaces the body with two partial bodies: the corporeal body and an (incomplete) electronic ‘body image’… Virtual reality leaves the meat body on the chair. It is a confirmation of, rather than a liberation from, Cartesian dualism.” This “electronic body image” is distinct from Lanier’s “virtual body,” which was linked to the physical body rather than independent. Though a dualist rather than phenomenologist approach, it also echoes Merleau-Ponty’s body image, foreshadowing the attachment people would form to their avatars.

Just as quickly, criticism of this dualist perspective arose. Balsamo (1996) noted that “Even though some games may soon allow players to design personal avatars or puppets—simulations of oneself—more frequently VR is promoted as a body-free environment, a place of escape from the corporeal embodiment of gender and race” (Balsamo 1996, p. 123), similar to the discourse of early cyberspace. She went on to argue that even without avatars, VR is an inherently gendered concept because the promises of disembodiment, like in cyberspace, were more easily experienced by those of dominant identities. Indeed, she posited that white men may have desired disembodiment as an “enticing retreat from the burdens of their cultural identities” that they had to grapple with as American society increasingly confronted issues of gender and race (Balsamo 1996, p. 131).

While the founders of VR—who saw VR as quite embodied—did not explicitly express a desire to escape the inconvenient realities of their white male bodies, it was true that not everyone had the same experience in cyberspace, and whether or not we accept Lanier’s claim that VR is an embodied medium by default, it is clear that it was embodied to some extent. First, one’s physical body impacted how one responded to entering VR. Women more often experienced “cybersickness” (also called “VR sickness”) (Hayles 1996), a problem that persists today (Kelly et al. 2023). There were also representational issues that meant that some people could not embody an avatar that looked like their physical body; many “generic” avatars still read masculine and, often, white (Green 1997, 65–67). There was limited disability representation and, since VR devices were still being developed, little accessible hardware, although some early VR applications were developed to help veterans with memory impairment, facilitate sign language, and provide aphasia therapy (Lanier 2017, p. 381).

While perceptions of cyberspace were shaped by pop culture, individuals often had little practical experience with VR. VR technology was so expensive that there was little consumer market. Some arcades installed VR games (Fowle 2015), but home applications remained limited. VPL’s DataGlove cost $10,000 and required a $250,000 computing system, putting it out of reach of even early adopters. Consumer-targeted devices, like Mattel’s $90 Power Glove and Nintendo’s “Virtual Boy” VR console, did not live up to technological expectations and were rapidly discontinued (Boyer 2009; Rossen 2017). Even VR arcade games were declining by the mid-1990s (Fowle 2015). Thus, people’s perceptions of VR were shaped more by science fiction rather than their own experiences.

In the narrative clash of early VR, the portrayal of VR as a disembodying, Cartesian experience won out over the idea that VR was a pseudo-spiritual experience inherently tied to the physical body and intended to augment the physical world, perhaps because the former was easier to market in pop culture. Regardless of narrative, people’s physical identities did impact their experiences in cyberspace for both biological and social reasons. However, VR proved not advanced enough for widespread adoption, and while it remained present in pop culture, there was little opportunity for most people to use VR in the ensuing decades.

3 Internet summer, VR winter

3.1 Internet

As technologies advanced beyond purely text-based interfaces and the invention of the World Wide Web and web browsers made accessing cyberspace less intimidating, platforms responded to users’ desire to be embodied in cyberspace with different categorization and depiction methods. At the same time, barriers to access increased the digital divide.

The presence of graphical user representations made it clear that people were to some degree embodied online. The narrative reminiscent of Cartesian dualism faded, but the prevailing attitude became that although physical traits could be apparent online, they did not matter—especially race, where an explicit attitude of colorblindness dominated (Kang 2000), imported from American culture where a discourse of color- and class-blindness reigned and transferred into online interactions (Kendall 1998). However, as in the text-based cyberspace days, the reality was more complicated, and the social impacts of embodiment could not be escaped. People desired embodiment online for themselves and others, but this once again meant that people had unequal experiences in cyberspace.

As computers and Internet connections grew faster, transmitting images became feasible. Platforms increased embodiment online through graphical avatars, but mediated forms of embodiment through platform design. The term “avatar,” derived from the Sanskrit “avatara” (referring to the “descent” of a deity adopting a physical, terrestrial form (Junghare 2012), was first used to refer to an on-screen representation of a user in the 1985 video game Ultima IV: Quest of the Avatar (de Wildt et al. 2020). Neal Stephenson also used it in his 1992 novel Snow Crash, after which it began to be used to refer generically to online representations of people, whether pictorial or 3D, anthropomorphic or abstract, and static or dynamic (Nguyen 2023). Platforms like The Palace allowed users to upload their own images, while others like Second Life had them customize a humanoid body. Eventually, static icons and pictures that were previously called “avatars” became “picons,” “userpics,” and “profile pictures” (Nguyen 2023), and “avatar” was reserved for dynamic full-body representations. An archive of “picons” shows that many were the faces of actual people (Picons Archive n.d.). In The Palace, using a “real face av” was seen as a gesture of intimacy, a way to “step out” of “anonymity” (Suler 2001) that implied that someone was embodying a character more like their physical self. Even though avatars that were not of a “real face” could be perceived as equally embodied—and there was no guarantee that what someone presented as a “real face av” was their real face—the physical body was still prioritized above virtual embodiments, just as it was in the BBS and MUD days when meet-ups in the physical world were ways to build trust in cyberspace. Eventually, social networks like Facebook would take advantage of this by encouraging people to use their real identities online, coupling the on- and offline identities tightly together; this will be discussed more in Sect. 4.

Although avatars may have offered a form of anonymity, even lower-fidelity representations were a form of embodiment—albeit an “alternative mode” (Ajana 2005)—and this promoted differential experiences in cyberspace. A study of pictorial avatars found that people preferred avatars that matched their gender and that more anthropomorphic avatars were seen as more attractive and more credible (Nowak and Rauh 2005). This shows that not only did people pick avatars to represent themselves, but they also made judgments about other users based on their avatars. Thus, there was still meaningful virtual embodiment, as there was in textual cyberspace, which allowed people to form genuine connections, but also increased the potential for negative experiences. Ajana (2005) argues that the Merleau-Pontian phenomenal body was extended into the virtual sphere, which implies that harm to one’s online persona would be experienced as an affront to the physical self. Trauma reactions to harassment, even in text form, show that this was the case (Dibbell 1994); if the virtual persona was an insubstantial thing independent of the self, online experiences would not trigger such responses.

Having visual representation made harassment easier, although specific statistics are difficult to come by. Still, in 1998, 41% of female Internet users reported being sent unwanted pornographic material or being harassed or stalked online (Griffiths 2000), with many women reporting being deluged with sexual messages just because they had a female screenname (Döring 2000). To extrapolate, having a pictorial representation of an individual rather than having to read a textual description or sift through their posts would make targeting them much easier. Racial harassment in these spaces was also common (Kang 2000). Furthermore, sometimes platform affordances enabled new forms of harassment. For instance, in The Palace, one user would adopt an avatar that was “a pair of upside[-down] legs that he insert[ed] down the cleavages of unsuspecting women, giving the illusion of the rest of his body being inside their dresses” (Suler 2001).

More literature has focused on full-body avatars such as those in Second Life, where harassment and assault are acknowledged as two of the biggest issues facing the platform (Bugeja 2008). A database of reported incidents (community standards and terms of service violation) between 8 June 2007 and 24 December 2010 showed 3,778 incidents of nine different kinds of harassment; 392 (10.4%) were classified as “sexual harassment” or “teen sexual harassment” (Second Life Incident Reports Repository n.d.). These fell disproportionately on women, with 60% of women reporting being harassed or stalked compared to 44% of men (Beck 2015). Even in the BBS and MUD days, people faced racist and homophobic comments, with some speculating that the same anonymity that to a certain extent concealed their identities empowered people to say things they would not in the physical world (Moser and MacLeod 1996). In the increasingly embodied Internet, people not only had anonymity, but also more accessible identity markers to use to target their attacks.

Platforms attempted to counter harassment through moderation, but also managed and mediated online embodiment through platform design in a way that implied that not all physical traits should matter or be represented in virtual contexts. While encoding identity options through tools like dropdown menus made categorizing users easier, it also restricted self-expression to the pre-defined categories chosen by the platform. Some platforms had many gender options, while others adhered to a strict binary (Bruckman 1996, p. 318; Fizek and Wasilewska 2011). The move from textual descriptions to picons to avatars also restricted embodiment options; instead of being able to describe themselves or upload any image they wanted, users were restricted to options defined by the platform. This impacted not just representations of gender but also of race and disability.Footnote 14 Linden Labs initially did not provide wheelchairs in Second Life, even as it coded different walking styles for male and female avatars (Fizek and Wasilewska 2011). One virtual space designed for teens, Whyville, was noted for having some non-white avatar heads but few bodies to match (Kafai, Cook, and Fields 2010). This encoded—literally—who could be embodied online and how, and non-white avatars were under-represented in virtual fantasy worlds (Higgin 2009). In some cases, this was due to a lack of representative avatar options. In others, as in the BBS days, it was because of a reluctance to embody an avatar that might subject them to more harassment. Sometimes these were linked; one participant in Whyville was told that her Black avatar “looked wrong” because most avatars were white (Higgin 2009).

Finally, as in the days of BBSs, the digital divide continued. While Internet service providers (ISPs) had sprung up in the US, lowering technological barriers to cyberspace access, they erected further price barriers, perpetuating economic and racial access disparities that widened in the late 1990s (Chon 2000).Footnote 15 Furthermore, cyberspace was most accessible to those who spoke English, with three-quarters of websites hosted in the US and other English-speaking countries in 1996 (Chon 2000). Physical characteristics not only affected experiences in cyberspace and VR but also were correlated with people’s ability to access them. While the existence of virtual embodiment (reminiscent of phenomenological theories) was being acknowledged, it was in a passive way that did not inspire action to ensure equitable representation and experiences.

3.2 Virtual reality

After the hype of the 1980s, VR consumer technology was mostly dormant. Schroeder (1993) identified three strands of VR in the 1980s and early 1990s: art and entertainment, flight simulation and robotics, and military and space-related research. Of these, flight simulation and military applications found the most success, while consumer products and portrayals of VR had diminished, meaning that VR was largely out of the public eye (Drummond et al. 2014). From the perspective of the average Internet user, VR was a technology that existed mostly in science fiction, which promoted the full-body, total-immersion version of VR that was, at the time, still technologically infeasible.

4 VR resurgence and cultural merging

By the 2010s, cyberspace had transformed, dominated by social media sites. These platforms encouraged people to use their real identities online, with Facebook initially implementing a “real name policy” (What Names Are Allowed on Facebook? 2014)—later nuanced to require names users “go by in everyday life”—to promote “authenticity” and community accountability (Account Integrity and Authentic Identity 2023). Although some platforms allowed anonymous accounts, many social media platforms were marketed to facilitate “authentic” connections, with authenticity predicated on real identities, reminiscent of “real face avs” and identity-confirming BBS meetups.

VR experienced a resurgence with the launch of the Oculus Rift on crowdfunding platform Kickstarter in 2012. Advances in 3D graphics and hardware made consumer VR headsets viable. With them came avatars that went far beyond the blocky heads and hands of early avatars, as well as “social VR” platforms called “metaverses.” The term “metaverse” first appeared in Neal Stephenson’s Snow Crash (Zenou 2022) and is often etymologically defined as “beyond universe,” but according to Jaron Lanier in a 2022 interview, the term was an inside joke between Lanier, Stephenson, and Gibson. The use of “meta” referred to the meta key on programmer workstations, which was known as the “god key” for the extra functionality it enabled (Egan and Lanier 2022). Likely unaware of its tongue-in-cheek background, in 2021 Facebook rebranded as “metaverse-first” company Meta and launched its “metaverse” Horizon Worlds (Meta 2021a; Zuckerberg 2021).

Social VR platforms are a merging of cyberspace and VR. Essentially immersive social media platforms, they are now similar to the immersive virtual worlds called “cyberspace” in cyberpunk and other science fiction. As a result, we need to consider modern VR and its narratives in the historical context of both cyberspace and VR. While the technology’s foundations lay in 1980s Silicon Valley labs, its ethos of authenticity and celebrating of identity is an evolution of cyberspace discourse. However, platform actions are often at odds with their rhetoric.

Embodiment in cyberspace has evolved from text to images and two-dimensional avatars to three-dimensional avatars piloted in a virtual world. While MUD research suggested that about half of people saw their online persona as a separate entity from themselves (Turkle 1994), users in VR identify strongly with their avatars and often use them to explore aspects of their identity by modifying their virtual self-presentation (Freeman et al. 2022), showing the connection between the virtual phenomenal body and the physical body. Platforms have taken note and implemented a wide range of avatar options. Social VR platform VRChat is notable for having a wide range of humanoid and more creative avatars, including a hot dog (VRChat 2023). Horizon Worlds, on the other hand, focuses on humanoid avatars, reflecting Meta’s focus on “authenticity,” although Facebook’s real name policy does not extend into Horizon Worlds. Meta has consciously focused on enabling users to create any depiction of themselves that they want.Footnote 16 Chief Diversity Officer Maxine Williams wrote that “you should be able to be authentically present” in VR; there were over 1 quintillion possible avatars in 2022 (Williams 2022). Meta’s Codec Avatars aim to go beyond this by creating a photorealistic model of the face and, eventually, the entire body (Ma et al. 2021), essentially translating bodily self-presentation entirely into virtual reality.

Platforms, acknowledging that displaying physical traits is unavoidable in social VR, now insist that user diversity is something to be celebrated. This is a shift from the discourse of cyberspace that held that physical traits were not conveyed through text and the early Internet that insisted that while they could be displayed, they were irrelevant, and reflects the changing American discourse, which instead of ignoring race and gender now recognizes that it exists and creates different social experiences. The focus on “authenticity” echoes the “bring your whole self to work” movement (Inam 2018), which Sheryl Sandberg, former Chief Operating Officer of Meta, called her “central management message” (Frier 2017). The philosophy endorses showing all aspects of your identity to work and having them be celebrated. For individuals to be able to authentically express themselves in VR, they cannot be subject to differential treatment based on those identity characteristics. However, as in earlier online platforms, they are, and the way platforms have responded to issues of harassment and accessibility cast doubt on their claims of experiential equality and imply that they are adopting an oversimplified position similar to that of early cyberspace pioneers.

Online harassment is not a new issue and was known to be a problem in VR before Horizon Worlds launched (Frenkel and Browning 2021). Despite promises from Mark Zuckerberg that safety would be “built into the metaverse from day one” (Meta 2021b), harassment instantly became a problem on Horizon Worlds. A female beta user was surrounded and groped by male avatars, and Meta responded by implementing a “personal boundary” feature, but an investigation of VRChat found that most harassing behavior did not involve contact (Lawson 2022). Chief Technology Officer of Meta, Andrew Bosworth, questioned whether keeping VR safe is even possible (Frenkel and Browning 2021), which is debatable because VR combines the immersion and presence that makes harassment in the physical world damaging with the anonymity of cyberspace that encourages it. As a result, racist, sexist, and homophobic harassment remain rife on social VR platforms (Outlaw 2018; Frenkel and Browning 2021; S.B., Sabir, and Das 2024).

One thing that users are not harassed about are markers of physical disability—because they often do not exist in VR, despite the emphasis on representation. Meta promoted their release of hearing aids, cochlear implants, and wheelchairs for avatars, but wheelchairs are only available for avatar stickers, not for the actual avatars (Meta Accessibility 2022), and prosthetic limbs and amputees cannot be represented, echoing the initial lack of options in The Sims. Although the fact that there are some representation options is a step up from the early days of VR games and social platforms,Footnote 17 there remains work to be done. Lack of accessibility is often an economic question, as regardless of the ethics of not having representation, if something costs more to implement than the value the platform perceives, the platform will not implement it. However, as Perry et al. (1996) noted, “part of the problem is lack of imagination.” The same applies to hardware. VR hardware and software are not designed for people with physical or cognitive disabilities, meaning that people with disabilities are excluded from bringing their authentic selves to VR, even though they may be more inclined to want to use VR (French 2017). Furthermore, as with computers in the early days of networked computing, price remains a significant factor, with consumers hesitant to spend hundreds or thousands of dollars on still-maturing technology.

Social VR is still emerging, but it seems poised to have an impact on many people even if it does not become the universally accessed virtual world portrayed in science fiction. Because it carries many of the features of social media, including the possibility of anonymity, combined with the immersive nature and embodied interactions that are similar to the physical world, its potential impact on users is significant, making it all the more important that we analyze who is deciding who gets to be embodied and how. Social VR is attempting to overcome the issues of physical society by insisting that everybody and every body is celebrated, but as in the physical world, saying so does not make it so, and active intervention is necessary to ensure that it is an environment where everyone can have an equitable experience.

5 Conclusion

Over forty years, cyberspace and VR have evolved and now, merged. While the modes of interactions and technologies have changed, one thing that has remained constant is that the “male, white, straight, able-bodied and ruling class” cybersubject is also the one with an outsized role in defining the technology and culture (Moser and MacLeod 1996), representing the plurality of technology workers, especially at the executive level (Fury 2023). While the narrative has evolved from one of a Cartesian split between the mind and body to a more accurate phenomenological understanding that online experiences are embodied, the shift from denying how bodies influence cyberspace experiences towards attempting to celebrate them has not changed that the realities of the “shadow halves” of the dominant identities are often overlooked (Moser and MacLeod 1996), including—and especially—when they have disparate experiences that call the dominant narrative into question. This matters because as embodiment in virtual spaces—both immersive and non-immersive—increases, so too does the potential for harm that is not virtual harm, but real harm.

The early days of cyberspace offer opportunities and cautionary tales. On the one hand, social VR might allow for increased trust online if avatars are seen as an acceptable substitute for the physical body, and could facilitate the identity-oriented spaces that provided a haven for marginalized groups. On the other, it may offer the ultimate combination of anonymity and targetability that encourages harassment, a problem that is already developing. Conversations around platform policies and design should acknowledge the embodied nature of XR and how “online” experiences have offline impacts. As XR development continues, more research should explore who is able to be embodied in modern XR, expanding beyond VR to include AR and MR and exploring governance models to address how to balance anonymity, safety, and self-expression. Ultimately, though, the history of embodiment in the virtual sphere and the consistency with which marginalized groups have worse experiences show that these issues are fundamentally human ones. Rather than throw up our hands at the flaws in human nature—or worse, deny their existence—we must take action to counter them in our new virtual worlds just as we try to do in the physical world. Racism, sexism, ableism, and other forms of discrimination will not vanish just because we interact with each other in a virtual space. It must be acknowledged that in 2024 and early 2025, platforms including Facebook and X began moving away from content moderation, leading to increases in hate speech targeting vulnerable groups (Hickey et al. 2025, Kaplan 2025). These trends are concerning and may foreshadow a shift to a new era: one where platforms acknowledge that one's body impacts online experiences, and yet do not consider it their responsibility to address––all the more concerning if it filters into XR as well. Regardless, this article will conclude with a perhaps naive argument: that platforms do have an obligation to moderate content and behavior to protect vulnerable users, and we have an obligation to treat our fellow humans better, whether they are in a physical body or a virtual one.