Sage Journals: Discover world-class research

Abstract

Immersive multi-user virtual reality (VR) enables users to embody a first-person avatar and through them enact agency over their virtual self-representations and identities. Moreover, these visual representations can profoundly impact users’ thinking and behaviour. Despite this, there is a dearth of understanding of how opportunities to create an avatar versus using a preassigned one might affect users. One of the first pre-registered multi-user collaborative VR experiments, this mixed-method study investigates how customisation agency affects self-perceptions, and what role individual and avatar style differences play in this relationship. Results are limited but suggestive of VR avatars potentially serving as an empowerment tool through taking control over shaping one’s self-image, with interacting influences from self-discrepancy perceptions and avatar styles. Moreover, this article highly emphasises the complex relationships between intra- and interpersonal aspects of experiences. Implications are considered in terms of various contextual aspects and individual experiences and differences.

Keywords

Avatar customisation embodiment metaverse self-discrepancy self-representation user-modelling virtual identity

Introduction

Virtual reality (VR) interactions involving embodiment are increasingly widespread, ranging from mental health treatments to recreational activities, offering various levels of potential and advantages. (e.g. Maloney and Freeman, 2020; Mayer et al., 2022; Slater and Sanchez-Vives, 2016). Despite being hindered by the novelty, user inexperience and the early stage of the technology and its applications, one of its most intriguing features is arguably the sensation of ‘stepping into another’s shoes’. This involves inhabiting a virtual self-representation by tracking the user’s body movements and translating them into the avatar’s (Maselli and Slater, 2013). This phenomenon, while strange and curiosity-provoking, has also been shown as a potentially powerful tool for gaining perspective and self-compassion (Osimo et al., 2015). Embodiment has the potential to greatly enhance social encounters compared to non-immersive on-screen interactions, given the physical nature of interactions afforded by immersive systems, such as social touch, social presence, non-verbal communication and other (Maloney and Freeman, 2020). Furthermore, these self-representations can impact how we think, feel and act, much like physical ones (Adam and Galinsky, 2012; Banakou et al., 2018; Huang et al., 2011). Consequently, this straightforward yet remarkable mechanic has the potential to serve as a powerful empowerment tool, as suggested by controlled studies and social VR users alike (Freeman et al., 2022; Osimo et al., 2015).

Despite a substantial body of research investigating outcomes related to user representation in both on-screen and virtual reality contexts, an overview of immersive gameful social interaction literature indicates limited overlap between these two technological or context domains (Bujić et al., 2022). Research on social interactions within such environments has predominantly focused on interactions with computer agents, highlighting particular aspects of representation, such as anthropomorphism, while at the same time often neglecting multi-user scenarios. A recent systematic review of avatar visualisation supports these findings (Weidner et al., 2023), noting a strong preference for realistic, anthropomorphic and personalised avatars that resemble the user’s physical appearance. Conversely, examining the diverse landscape of freely available social VR applications like VRChat (VRChat Inc.) shifts the focus towards the voluntary choice of non-realistic, performative avatars (Freeman and Maloney, 2021) that exploit VR embodiment affordances. This concept of avatars as playful yet powerful self-expression tools is well documented in gaming contexts (Sibilla and Mancini, 2018), though with notable limitations such as reduced immersion and the absence of body transfer illusion (Slater et al., 2010), mediated by VR body tracking technologies that connect the self with the avatar. Further insights from a recent meta-review suggest that these new immersive technologies enhance the impact avatars might have on their users (Beyea et al., 2023). While recent social VR literature has underscored the significance, potential and variety of motivations and purposes behind avatar choices (Freeman et al., 2022; Maloney et al., 2020), there is a noticeable lack of controlled VR studies that systematically examine the mechanisms and implications of customization agency over visual representation through avatar appearance, particularly in terms of VR avatar customization as a tool for empowering users and enhancing self-perception.

This preregistered laboratory experiment study aims to fill this gap by investigating whether and how using a customised self-representation through an avatar in VR affects users’ self-perceptions. Using a mixed-methods between subjects repeated-measures design, we examined potential differences in self-perception within a controlled setting. Two groups of participants (N = 66) used an avatar they created themselves or were assigned a simple one, and worked to solve 3D puzzles that required communication and collaboration. Besides using psychometric tools and subsequent analyses of participants’ traits and self-perceptions, we conducted interviews to explore their motivations for avatar creation, analysing these motivations along with the psychometric data using a mixed-method approach. This study thus adds to the interdisciplinary research on avatars and embodied VR by providing a novel, controlled multi-method investigation into the motivations behind avatar creation and the effects of avatars on both intra- and interpersonal self-perceptions.

Background

Self-representation in virtual environments

Individuals often subconsciously or intentionally alter their self-representation to fit a specific occasion or goal (Baumeister, 1982; Human et al., 2012). Digital environments remove physical and biological constraints opening up new possibilities, although many users still prefer to personalise their self-representation, or make them in their image (Freeman et al., 2020; Messinger et al., 2008). The creation of avatars enables people to address or rectify perceived flaws in the real world (Bessière et al., 2007; Morie, 2014) and experiment with playful forms that may be socially inappropriate or physically impossible (Mancini and Sibilla, 2017).

Interactive media in particular, such as video games, often provide the opportunity to customise these digital self-representations, or avatars, which are potentially linked to users’ identification with them (Sibilla and Mancini, 2018) by aligning the virtual self with one’s own image or matching the needs and context of the experience (Hudson and Hurter, 2016). The reasons for designing an avatar in a specific manner at a given time depend on numerous factors, similar to any other self-representation choice such as appropriate attire. However, both physical and digital representations affect how we think and act (Adam and Galinsky, 2012; Banks, 2015; Frank and Gilovich, 1988; Huang et al., 2011), similarly, people primarily conform to the biases, stereotypes and other meanings associated with their avatars, a phenomenon known as the ‘Proteus effect’ (Praetorius and Görlich, 2021; Yee and Bailenson, 2007). The attractiveness of the avatar can influence intimate behaviours, interpersonal distance and self-disclosure during an interaction, while the height of the avatar can affect aggressive behaviour and confidence during a negotiation task (Yee and Bailenson, 2007).

Comparable and even stronger effects occur with the body ownership illusion, the phenomenon of confusing a VR avatar’s body with their own (Maselli and Slater, 2013), where avatar traits have been found to influence individuals’ behaviour, perception and cognition, such as antisocial tendencies Peña et al., 2009), racial bias (Peck et al., 2013), size estimation (Banakou et al., 2013) and cognitive task performance (Banakou et al., 2018), among others. Collectively, this research forms a foundation for understanding the extensive possibilities and implications, highlighting the importance of ‘virtual skins’ and self-perceptions. However, most of these insights within the VR context are based on controlled and imposed representations that are frequently stereotypical and serve as proofs of concept, rather than those freely selected by users.

Self-discrepancy and self-representations

Humans consistently define themselves by observing and analysing their behaviours and the contexts in which they occur (Bem, 1972; Robak, 2001). This process of self-perception involves not only how individuals view themselves (intrapersonal), but also how they are influenced by the reactions, feedback and expectations of others (interpersonal), whether these are real or imagined. Through both intra- and interpersonal pathways, individuals construct their personal and socially mediated self-concepts, encompassing ideas and beliefs about who they are in different contexts (Sanitioso and Wlodarski, 2004; Wallace and Tice, 2012). Moreover, the self-discrepancy theory (Higgins, 1987) identifies three domains of self: actual (how one perceives their true self), ideal (how one perceives they should be) and ought (how one perceives others think they should be). These domains are in constant negotiation, and the discrepancies between actual-ideal and actual-ought have significant effects on both intrapersonal aspects such as mental health (Mason et al., 2019) and interpersonal aspects such as interaction, shyness and external shame (Amico et al., 2004).

The importance of self-concepts and self-discrepancies extends to virtual settings as well. Individuals with less stable self-concepts may present differently online compared to offline (Fullwood et al., 2016; Strimbu and O’Connell, 2019). Digital self-presentation choices can be context-sensitive, with users customising avatars while taking into account the platform and social experiences (Freeman et al., 2020). In this context, self-discrepancies influence customisation choices and the resulting avatar style (Loewen et al., 2021; Mancini and Sibilla, 2017). Therefore, customising one’s virtual self-representation could serve as a tool to explore various self-perspectives, reinvent yourself or cope.

Hypotheses and exploratory research questions

Due to the limited grounds and complexity of the research model (Figure 1), this study employs a combination of confirmatory and exploratory research methods. Confirmatory hypotheses consider the role of Agency vs No-Agency in self-representation and self-perceptions as outcomes (H1 and H2), as well as the interaction effects of individual differences in Self-discrepancies (H3). Here, Self-evaluation and Actual-Ideal discrepancy present intrapersonal and Other as shamer and Actual-Ought interpersonal dimensions of the model.

Figure 1.

The model of the study representing trait, manipulation and outcome levels with the proposed hypotheses and exploratory research questions.

H1a: Self-evaluation is higher after the manipulation in the Agency condition, compared to a baseline taken before the experiment.

H1b: Self-evaluation is unchanged after the manipulation in the No-Agency condition, compared to a baseline taken before the experiment.

H2a: Other as shamer is lower after the manipulation in the Agency condition, compared to a baseline taken before the experiment.

H2b: Other as shamer is unchanged after the manipulation in the No-Agency condition, compared to a baseline taken before the experiment.

H3a: Interaction effect of the actual-ideal discrepancy and core self-evaluation is smaller for the Agency than the No-Agency condition.

H3b: Interaction effect of the actual-ought discrepancy and other as shamer is smaller for the Agency than the No-Agency condition.

As a further exploration of these mechanisms, exploratory research questions investigate the relationship between self-discrepancies and customisation in terms of avatar style preference (eRQa), and the relationship between the customisation style and self-perception outcomes (eRQb).

Method

A repeated-measures between-subjects experiment (N = 55) was conducted in order to compare two groups 1) Agency: utilising a self-customised avatar; and 2) No-Agency: utilising an imposed premade ‘plain’ avatar in VR. To facilitate a social VR context, each experiment session was conducted with a participant pair or dyad engaged in a collaborative task. Short interviews with the Agency group examined their motivations and preferences for avatar creation and consequently whether and how the avatar styles affected the study’s outcomes. The data collected includes a) self-report measures (quantitative; N = 55) and b) interviews with voluntary participants from the Agency group (qualitative; n = 26). As such, the mixed-method approach employs both confirmatory research of existing literature in a novel context and further exploration providing alternative explanatory solutions and veins for future studies.

The design included considerations of sample size against the study’s aims as well as multiple other relevant factors such as resource constraints and effect sizes of interest. As such, a sample of N = 64 was set as the goal, with the assumption that n = 60 will be collected without technical or validity issues and used in the analyses. With an alpha level of.05 and using a sample of 32, a one-tailed paired samples t-test has a statistical power of 80% to detect the effect size of Cohen’s d = .449. Considering that the primary foci of the study are the applied implications of the main treatment factor, smaller effect sizes are not necessarily of interest, prompting the acceptance of the planned sample size.

For this study, we followed national and University-specific guidelines for research integrity and ethics (Finnish National Board on Research Integrity TENK), obtained the University’s permission to conduct this research including a data management plan compliant with EU regulations, and followed best practices for the ethical involvement of human participants. Following open science practices and contributing towards reliable and reproducible science, the design, hypotheses and confirmatory analyses of this study were registered prior to data collection^*1 and data are made openly available alongside this article. All tests conducted on this dataset have been reported in this article.

Participants

A non-probability convenience sample (N = 66) was recruited through university channels in the fall of 2022. During recruitment, participants received information on the research, followed by demographic data and personality traits questionnaires. Participants chose their preferred time slot for the experiment and were instructed not to choose a slot together with a friend to control for prior familiarity with the collaborator outside of VR.

The validity of the data collected was controlled in multiple ways. For example, attention cheques were implemented in the questionnaires using items that required a specific answer, and responses were discarded if the collaborators met in person. The dataset used in the analysis contains N = 55 participants, with n = 28 (self-reported gender: 14 females, 12 males, 2 other) in the Agency and n = 27 (13 females, 13 males and 1 other) in the No-Agency condition (age: M = 27, SD = 5.5 and M = 28.6, SD = 6.22, respectively).

Procedure

The procedure consisted of several identical steps in the Agency and No-Agency conditions (Figure 2), with the sole difference of whether the pairs of participants were assigned their customised avatar or a premade ‘neutral’ one. Recognising that no single representation can truly be neutral and devoid of perceived characteristics, we have designed it through an iterative process with the goal of maximising ambiguity. The process included 10 researchers creating avatars and rating the selection, while aiming for the lowest consensus on the avatar’s perceived gender (Zhang and Juvrud, 2024) and ethnicity (the panel consisted of 3 females and 7 males of 3 ethnicities). Although an alternative approach would have been to assign a different randomly generated avatar to each participant, this would have introduced significant noise and variation to the data and, more importantly, would render the experiment not replicable. Thus, the same avatar was assigned to all participants in the No-Agency condition, reflecting the intended lack of customisation and personalisation alike.

Figure 2.

Procedure steps.

First, participants completed the pre-survey at their convenience and booked the on-site experiment. On-site, they were first familiarised with the customisation and collaboration tasks. The avatars were created using the online avatar creation tool Readyplayerme for 7 minutes. Afterward, the participants were equipped with VR. After meeting their partner and engaging with the puzzles for 15 minutes, they were redirected to the on-screen post-questionnaire and asked to participate in an optional short interview. In addition to the adopted task, tutorial rooms and a meeting room were implemented to familiarise the participants with body tracking, their avatar, controls and each other (see Figure 3).

Figure 3.

The virtual environments of the in-VR procedure and view of a No-Agency avatar and the collaborative task.

VRChat was selected as the experimental platform because it facilitates the creation of a private multi-user environment for the implementation of the CoBlok task, and allows for the quick and reliable import of custom avatars. In addition, full-body tracking and embodiment were realised through a 6-point tracking system as supported by the inverse kinematics solution in VRChat (IK 2.0). Head movements of the avatar were tracked by the HTC Vive Pro Eye head-mounted display, hands by two Valve Index controllers, while the tracking of the pelvis and feet were provided by three HTC Trackers 3.0. For enhancing immersion and intimate connection with the self-representation, participants explored their avatar and its movements in front of a virtual mirror and they were placed along most of the walls of the environment so that participants could often anchor themselves as their avatar (Mottelson et al., 2023; Spanlang et al., 2014).

Materials

Task

The collaborative scenario was based on an adapted version of CoBlok (Wikström et al., 2020). It is a puzzle task requiring collaboration to construct a composite shape corresponding to two individual 2D perspectives of the final composition. The participants could manipulate 5 different geometric primitives (cube, sphere, cylinder, pyramid and cone) and assemble an arrangement of 2–4 objects congruent to both 2D images (see Figure 3). For example, a sphere would be seen as a circle by both participants, while a pyramid would be seen as a square by one participant and a triangle by the other. Without both perspectives, it was impossible to be certain of the correct arrangement. Instead of an automated check, participants were required to negotiate whether the solution was correct. A screen capture video from a participant’s perspective in the Agency condition is available alongside this article as supplementary material.

Measures

Data consist of a combination of psychometric and interview data. Psychometric instruments were selected based on their conceptual relevance and demonstrated rigour in terms of validity and reliability throughout the literature due to the lack of opportunity to conduct a full validation study with the current experiment setup and sample size. However, internal reliability was assessed using McDonald’s Omega coefficient (w) as needed.

Self-perceptions (Figure 1) were measured on two levels that could conditionally be presented as referring to intrapersonal and interpersonal aspects. The levels represent an individual’s perception of self but in relation to them as an individual (Self-evaluation; internal shame, intrapersonal) or as a social actor (Other as shamer; external shame, interpersonal). Although they can be highly related in the conceptual and empirical literature, they are considered to be different aspects of self-perception. Both were collected before and after the experimental manipulation and task (pre-post repeated-measures design), with their difference variables computed as post minus pre scores. This resulted in a positive score difference (SEVdiff and OASdiff) demonstrating higher post-test scores than pre-test scores. This is important to note due to the different conceptual bases of the two measures. Although, arguably, a higher Self-evaluation difference is seen as desirable, the opposite would be true for Other as shamer as it would indicate higher reported external shame after the experiment task.

Self-evaluation (SEV) was measured using the original 12-item 5-point Likert-type scale with no changes (Judge et al., 2003) (e.g. ‘I am capable of coping with most of my problems’). Pre-test w = .87, post-test w = .89.

For Other as shamer (OAS) sum scores of the adopted 8-item 5-point Likert-type scale were used (Matos et al., 2015) (e.g. ‘People see me as unimportant compared to others’). Pre-test w = .91, post-test w = .90.

Collection of Self-discrepancy profiles (Figure 1) was carried out using a version of the Schwartz Value Survey (SSVS) (Lindeman and Verkasalo, 2005; Loewen et al., 2021) only during the pre-test. The 10-item 16-point sets were repeated three times with differing instructions–specifically, ‘Please rate the importance of the following values as guiding principles from the perspective of how.’.. a) ‘you truly see yourself’ (Actual), b) ‘you would like to be’ (Ideal) and c) ‘others think you should be’ (Ought). The magnitudes of Actual-Ideal; a-i and Actual-Ought; a-o discrepancies were calculated using sum scores of absolute individual differences of the 10 items (e.g. Actual-Ideal discrepancy score = ABS(actual1–ideal1) + ABS(actual2–ideal 2) . . . + ABS(actual10–ideal10)), consistent with previous computations (Loewen et al., 2021).

Finally, agentic preferences and choices regarding Self-representation (Figure 1) in the Agency group were assessed through a structured interview. Specifically, through two questions reflecting their avatar creation motivation and the perceived self-similarity with the final avatar, respectively: 1) ‘What were your guiding ideas, thoughts and impressions when creating the avatar? What did you think or feel the avatar should be like?’, and 2) ‘How similar or different to you do you think your avatar was?’. The interviews were recorded and transcribed to be used in the analyses. For a more nuanced view that combined intentions and outcomes in the avatar creation process, the two answers were jointly coded against the Actual, Ideal, Ought and Other self-categories (see section ‘Analyses’ for details).

Analyses

Analyses were conducted using a multi-method approach combining confirmatory null hypothesis testing with data exploration techniques and qualitative interview data transformed into a categorical variable. Ad hoc hypotheses were probed using (a) one-tailed or two-tailed paired samples tests FVGas appropriate, and (b) thorough critical examinations of graphical data representations when statistical inference tests were not suitable or were underpowered, such as for more nuanced moderation tests. All analyses, statistics and graphical data presentations were made using the open statistical platform Jamovi v2.3,² including the package Flexplot.

The qualitative data from the voluntary interviews (n = 26) were coded with pen and paper to create a categorical variable that reflected the participants’ motivated avatar styles. As the style is presumed as a mediator of the effect of self-discrepancies on self-perceptions in the study model (see Figure 1), only those in the Agency group were interviewed for this study. The analysis was carried out using a deductive approach and through coding participants’ avatar customisation motivations and intentions against the self-discrepancy theory, using categories of Actual, Ideal and Ought (Loewen et al., 2021). An additional category, Other, was added to reflect avatar styles that did not necessarily or explicitly rely on any of the three mentioned SDT categories, similar to the so-called differents from a study by Loewen and colleagues (Loewen et al., 2021).

First, three authors coded all of the interviews independently with over 90% consistency, while five interviews were discarded due to lack of information or clarity regarding their motivated avatar styles. Second, initial disagreements in coding were discussed and resolved with the additional exclusion of one respondent as all three coders agreed that the interview response has not provided enough information for reliable categorisation. The final results (n = 20) were added to the quantitative dataset as a categorical variable and included in the analyses.

Results

Before examining the proposed hypotheses, we inspected the basic descriptive statistics and properties of the data composition and relationships between variables. The overview of all psychometric variables (Table 1) suggested prevalent non-normal distributions based on the Shapiro-Wilk test and prompted the use of non-parametric tests in all analyses due to the violated assumption of normality. Thus, the primary focus is on medians (Mdn) and interquartile ranges (IQR), while means (M) and standard deviations (SD) are presented as complementary information. The Wilcoxon signed-rank test was used in place of the student’s paired t-test, whereas extensive descriptive and graphical results were presented when no adequate replacement inferential test was available.

Table 1.

Descriptive statistics of self-evaluation, other as shamer and self-discrepancy measures per condition.

		SEVpre	SEVpost
M (SD)	Agency	3.35 (.55)	3.46 (.66)
	No-Agency	3.46 (.63)	3.53 (.71)
Mdn (IQR)	Agency	3.58 (0.63)	3.64 (0.79)
	No-Agency	3.42 (0.63)	3.58 (1.00)
W (p)	Agency	.83 (< .001)	.89 (.008)
	No-Agency	.98 (.732)	.98 (.822)
		OASpre	OASpost
M (SD)	Agency	6.68 (5.29)	7.25 (5.8)
	No-Agency	7.52 (4.67)	7.37 (5.27)
Mdn (IQR)	Agency	5.5 (8.5)	5.5 (7.25)
	No-Agency	6.0 (4.5)	8.0 (6.0)
W (p)	Agency	.92 (.031)	.89 (.006)
	No-Agency	.94 (.119)	.91 (.025)
		Actual-Ideal	Actual-Ought
M (SD)	Agency	24.6 (13.5)	36.3 (15.9)
	No-Agency	19.9 (10.3)	35.7 (19.1)
Mdn (IQR)	Agency	22.0 (13.5)	36.0 (19.8)
	No-Agency	18.0 (11.0)	33.0 (15.0)
W (p)	Agency	.91 (.021)	.93 (.065)
	No-Agency	.95 (.239)	.90 (.016)

Moreover, a non-parametric correlation test, Kendall’s Tau B, was used to examine relationships between variables for the entire sample (Figure 4). Primarily, there was a strong correlation seen between the intra- and interpersonal aspects of Self-perception–Self-evaluation and Other as shamer. On the other hand, it was surprising that no base relationship was found between self-discrepancies (Actual-Ideal and Actual-Ought) and measures of Self-perception.

Figure 4.

Correlations (Kendall’s Tau B) of all primary variables in the entire sample.

Avatar customisation and self-perception

H1: customisation and self-evaluation

Paired tests of pre- and post-scores for Self-evaluation were conducted for each of the conditions separately (Agency and No-Agency) corresponding to H1a and H1b respectively. As hypothesised (H1a), those in the Agency condition reported higher Self-evaluation after the experiment than before it, with only one pair of values tied and a moderate effect size in a one-tailed test (W = 110, p = 0.28, r = −.421). When looking at the No-Agency condition, a two-tailed test was used since the hypothesis considers changes in both directions for participants with no freedom in selecting, or customising, their self-representation (H1b). This hypothesis was supported with five pairs of values tied (W = 87.5, p = .211, r = −.308).

H2: Avatar customisation and other as shamer

Wilcoxon signed-rank tests were again used to test for changes in Other as shamer depending on whether participants used the avatar of their choice or a default one (H2). Again, a one-tailed test was used for the Agency condition as it was hypothesised that the prominence of external shame as measured via Other as shamer would be lower when having freedom to use the preferred self-representation (i.e. avatar) (H2a). However, this hypothesis was not supported (W = 84; p = .921, r = −.336). However, the hypothesis suggesting that no difference in either direction for Other as shamer would be detected in the No-Agency condition (H2b) was supported by a two-tailed test (W = 178, p = .686, r = .095).

H3: interaction effects of self-discrepancies

Together with the main effects of Agency or No-Agency in avatar customisation, we hypothesised about the mechanisms for the change or lack thereof. Mainly, we related what we conditionally call intra- (Actual-Ideal) and interpersonal (Actual-Ought) Self-discrepancies to corresponding Self-perception outcomes of Self-evaluation (H3a) and Other as shamer (H3b). As a starting point, Self-discrepancies did not appear to influence baseline Self-perceptions (Figure 4). Due to the properties of the data and the inappropriateness of applying parametric inference tests to them, H3a and H3b are examined in depth using exploratory data techniques. To this end, Self-discrepancy values were transformed into dichotomous variables (high and low values) divided around the corresponding median values, and presented using panel-based visualisations where the two panels represent the two conditions (Figure 5). Using this approach, it is relatively straightforward to inspect whether outcomes were different for different moderating values in each of the conditions.

Figure 5.

Graphical examinations of the interaction between the conditions, self-discrepancies and self-perceptions.

For H3a and H3b, it was hypothesised that any effects of the Actual-Ideal and Actual-Ought discrepancies on the Self-evaluation and Other as shamer change, respectively, would be smaller in the Agency than in the No-Agency group. This was based on the premise that the opportunity to use the customised avatar allows users to self-regulate Self-discrepancies. Although conclusions cannot be drawn without appropriate tests since the current data do not allow truly distinct effects, there appear to be trends in the relationship between Self-discrepancies and outcomes depending on the condition (Figure 5). For self-perception, Self-evaluation and Other as shamer, it seems likely that the changes are more stable (as reflected through median values and variance) in the Agency than No-Agency condition regardless of the magnitude of the corresponding Self-discrepancy (Actual-Ideal and Actual-Ought, respectively).

In addition to investigating the hypothesised interaction effects, there was a follow-up exploratory cross-examination of self-discrepancies and self-perception due to intrinsic relatedness between the intra and interpersonal levels through binned visualisations (Figure 1). This vein suggested a relevant interaction relationship in this sample between the condition, Actual-Ought discrepancy and Self-evaluation pre-post difference (Figure 6). Whereas the pre-post difference in Self-evaluation (SEVdiff) appears to be relatively stable in the Agency condition for any magnitude of the Actual-Ought discrepancy, a higher discrepancy in the No-Agency condition tended to be related to a lower or even negative change in the self-reported Self-evaluation.

Figure 6.

Actual-ought discrepancy and self-evaluation (left) and actual-ideal discrepancy and other as shamer (right) pre-post differences per condition.

On the other hand, no trends could be concluded when looking at the conditions, Actual-Ideal discrepancy and Other as shamer difference scores (Figure 6). Although the middle panel, showing mid-magnitude of the Actual-Ideal discrepancy, suggests a sudden striking difference between the conditions, adjacent panels do not provide support for interaction effects. Taken together and for the purpose of not overinterpreting the data, the middle panel is thus considered a statistical artefact rather than indicative of possible true effects.

Exploring self-discrepancy, avatar style preferences and self-perceptions

Finally, a mixed-method approach was used to explore the following relationships in a collaborative VR task: (a) between the preexisting Actual-Ideal and Actual-Ought discrepancies on one side and avatar style preference on the other (eRQa), and (b) between the avatar style preference and Self-perception outcomes (eRQb). Differing preferences in customising the avatar could shed light on the mechanism of the Agency condition’s effects on self-perceptions through the use of their customised avatar in representation. Surprisingly, there was a balanced distribution of Actual, Ideal and Other styles, but disproportionally few reported that their motivations were guided by appeasing another (Ought style). The following are example quotations in each of the 4 categories:

– Actual (n = 6): ‘I was just indecisive if I should do something very different or something similar to me, and I chose to do the most discrete because I felt it would be better . . . I didn’t want to look too different. I don’t know why, just preferred it’.

– Ideal (n = 6): ‘I wanted to create this avatar similar to me but not the same because in the real world you cannot have everything you want on yourself. In the virtual world I wanted the avatar to look like me but not exactly the same as me’.

– Ought (n = 2): ‘First of all I liked the fact that it first let me choose feminine or masculine, so not male or female. I wanted myself to have the hair I have and then just look chill and easily approachable, cause I wanted the person that plays with me to look at me and be like hey that’s nice-looking . . . I’d like to get to know them, or that it’s not scary at least!’

– Other (n = 6): ‘I was kind of falling under a stereotype so creating an avatar such as like someone who is kind of doctor, like an experiment researcher, so I just chose a white coat because of that and . . . yeah, in general, there was a good range of selections how avatar could look like and to be honest, I didn’t really go kind of deep cause I was just looking several seconds and just choosing right away (. . .)’.

Due to the scarcity of data points for each avatar style category, it was not feasible to conduct any inference tests, so exploratory techniques were used to analyse the data instead. A significant limitation is the insufficient number of ought-related motivations, as only two interviewed participants cited their motivations as being influenced by others’ perceptions of them.

First, the relationship between Self-discrepancies and avatars was examined using the high-low dichotomous discrepancy variables and avatar styles (Table 2). Notably, those who were aiming to create idealised or wishful avatars were likely to report relatively higher Self-discrepancies, both in terms of Actual-Ideal and Actual-Ought. Otherwise, low discrepancies appear to somewhat support Actual and Other styles, although neither of those motivations is characterised by prominently low discrepancies, and the differentiation between the two does not seem definitive in this regard.

Table 2.

Frequency table of high and low discrepancies and avatar styles.

	Actual-ought		Actual-ideal
Avatar style	high	low	high	low
Actual	3	3	2	4
	(25%)	(37.5%)	(18.2%)	(44.4%)
Ideal	5	1	5	1
	(41.7%)	(12.5%)	(45.5%)	(11.1%)
Ought	1	1	2	0
	(8.3%)	(12.5%)	(18.2%)	(0%)
Other	3	3	2	4
	(25%)	(37.5%)	(18.2%)	(44.4%)
Total	12	8	11	9
	(100%)	(100%)	(100%)	(100%)

Second, potential relationships between the chosen avatar style or self-representation and self-perception changes were examined using boxplots of pre-post Self-evaluation differences per avatar style (Figure 7). However, since no significant change was previously suggested in the Other as shamer measure (see 0.0.0), only the pre-post difference in Self-evaluation was considered. It appears that the Ideal avatar style predicts a higher post Self-evaluation when compared to the pre-test; interestingly, the Other style, representing playful and exploratory motivations, appears to also demonstrate a positive trend; finally, the Actual style shows a high variance in Self-evaluations but overall an arguably neutral one on average (Figure 7).

Figure 7.

Boxplots of self-evaluation pre-post difference per avatar style.

Discussion

Due to the increasing prevalence of immersive virtual reality applications in almost all forms of social interaction, it is increasingly relevant to investigate whether and how the opportunity to use a customised self-representation through an avatar in VR influences self-perceptions. This is one of the first multi-method collaborative embodied VR task experiments on the role of self-representation through avatar customisation, contributing to the broad literature of computer-mediated communication, psychology in interactive media, games studies and human–computer interaction. In this study, the agency over self-representation was reflected by using either the preferred self-customised avatar or a pre-made simple one. Users’ individual discrepancies between their perceptions of actual, ideal and ought selves (Bessière et al., 2007; Loewen et al., 2021) were analysed along with differences in self-perception as assessed with repeated measures of Self-evaluation (Judge et al., 2003) and Other as shamer (Matos et al., 2015) (corresponding to internal and external shame). Finally, interview data were used to understand avatar style preference categories (Actual, Ideal, Ought, or Other) and their outcomes. To our knowledge, this is the first comprehensive study that employs a controlled laboratory experiment using a mixed-method approach and collaboration with embodied avatars in this context. It uniquely sheds light on the motivations in customisation and the consequent implications on self-perceptions, and additionally scrutinises the relationships between the internal and external levels of these mechanisms.

Taken together (see Figure 1), the confirmatory and exploratory results suggest a limited but positive effect of self-representation agency on users’ self-perceptions before and after the VR task where they embodied their preferred avatar (H1a). However, the effects of self-discrepancies on different aspects of self-perceptions were not clear, but implied a strong interaction between the interpersonal and intrapersonal layers of discrepancies and outcomes, suggesting further investigation of this interaction while accounting for other traits depending on the context in question (H2). Furthermore, there was evidence that using a customised avatar played a role in diminishing the negative effects of the actual-ought discrepancy and was related to a higher positive Self-evaluation compared to the group that could not customise their avatar (H3). Moreover, our results indicated that participants’ avatar style preferences could to some extent be explained by the magnitudes of Actual-Ought and Actual-Ideal discrepancies (eRQa), and in turn moderately predict self-evaluation outcomes, with Actual style (i.e. created in their own image) appearing to be the most neutral in this regard (eRQb). However, it should be emphasised that these results support implications of users’ self-selected rather than imposed avatar styles and as such are not generalisable to the latter scenario. Ultimately, these results highlight the importance of the interaction of individual differences and styles in eliciting certain effects, rather than suggesting that imposing a certain style on all users necessarily yields favourable results in regards to self-perception outcomes. Overall, this study provides support for self-representation as a potential tool for empowerment in virtual worlds while emphasising a complex interaction of the intra- and interpersonal aspects of this relationship.

Differing contexts, differing minds

It should be considered that the evidence and its generalisation are bound by the study setup and context. Although such procedurally randomised trials are necessary for rigorously controlled comparison of a manipulation’s effects, they unavoidably impose an arguably serious framing on the participants. Participants are aware that they are ‘observed’ and might feel as if their personal actions will be evaluated, influencing their experience (see Hawthorne effect (McCambridge et al., 2014)). Thus, the experiment frame poses two possible broader issues in the context of the study objectives, namely freedom and privacy, reducing the potential for playfulness, exploration and exaggeration in avatar creation and the resulting experiences (Freeman and Maloney, 2021). This and possible similar studies should then be analysed integratively with other approaches, such as surveys and in-game interviews (Freeman and Maloney, 2021), and perhaps even series of smaller impromptu experiments within existing immersive virtual worlds’ users (Saffo et al., 2020). When contrasting such approaches, special care should be given to the contexts and their explicit or implicit constraints in players’ mindsets while deliberating their self-representation style and motivations for it.

Similarly, this study design involved a pronounced social aspect, basing the VR task on the collaboration of user pairs. Privacy and intimacy of the player-avatar relationship are thus limited not only by the experimenters as observers but also by their collaborators as their social judges. For this reason, we expect that the customisation preferences would to an extent be shaped by the social factor. When it comes to outcomes, although the system provided no indication of whether the provided puzzle solutions were correct, participants’ perceived failure or lack of contribution could have influenced feelings of external shame. While this potential effect is presumably equally distributed across the two conditions, it might be interesting for future research to manipulate the system’s feedback on the provided solution so as to further investigate the relationship between avatar customisation and self-perceptions in adverse scenarios.

In addition, the pairs of participants were physically close by, not completely anonymous and distant (Thibault and Bujić, 2022), with the possibility of meeting each other or being recognised after the experiment. All of these circumstances probably contributed to the detected but not hypothesised interaction effects involving Actual-Ought discrepancy and Self-evaluation (i.e. intra- and interpersonal layers). Thus, these results should be taken together and expanded with single-player scenarios or, for example, even through the design of between-subject studies with identical procedures while omitting the social aspect.

However, it is clear that users’ experience in collaborative VR environments can be improved and even made more pleasant by enabling them to customise their avatars (e.g. playful avatars in the Other category), as it might not only be fun in itself, but also improve how they perceive themselves (Table 2). Although the research focus is often on personalised avatars as the technologies enabling hyperrealism are developing (e.g. Koek and Chen, 2024), these results suggest that idealised or playful self-representations might be more beneficial, emphasising the benefits of diversity in customisation options (Table 2; Figure 7).

Avatar as a tool

As embodied gameful virtual reality experiences are not yet as common as on-screen, involving different approaches as described above is relevant for potentially differing effects of repeated exposures on sustained changes and particularly the novelty of embodiment. Players may not be aware of the unique sense of the illusion of body ownership of a VR avatar (Maselli and Slater, 2013) and how it might affect them. Consequently, they would not be as proficient in, consciously or subconsciously, customising the avatar so as to reap the highest benefits of the embodiment phenomenon by using the customisation agency as a tool. Moreover, as seen here to some extent, even experienced users often seem to exhibit avatar style preferences for the Actual self (Mancini and Sibilla, 2017) and it would be interesting to investigate whether it would be due to certain individual differences that leave them not particularly affected by, for example, Proteus effects (Praetorius and Görlich, 2021; Yee and Bailenson, 2007), low discrepancies, or something not yet considered. Ultimately, as our understanding of the affordances of a system grows, so do our behaviours and habits. First-time or short-term use might differ from experienced or long-term use as users become aware of the possibilities and related experiences. On the one hand, as immersive applications are becoming more widespread, it is increasingly important to develop representational literacy in users and provide them with tools for exploration and building strategies when it comes to their digital selves; on the other, particularly serious contexts such as education and mental well-being could benefit from understanding these relationships between individual differences, agency in avatar customisation and users’ self-perceptions.

However, both conditions in this study arguably contain a kernel of self-representation agency, or at least its imagination. Implementing the procedure step where both groups had the task of considering their choice of self-representation and creating the desired avatar was necessary to control the effects of using the preferred self-representation, but at the same time could produce its own effects. It is unclear whether and how that task in itself could have affected participants’ mindsets, whereas this effect could be positive in individuals, such as those with high immersive tendencies or role-playing experience, or negative in those whose focus is more on the visual representation as an anchoring point and who might have been disappointed or even frustrated that they could not use their preferred self-representation.

Moreover, avatar customisation was limited to human and human-like avatars, with some possibilities to select a costume or otherwise clothing that conceals physical attributes and becomes the centre of the self-representation. The participants’ agency was then limited to an extent, as there were relatively few opportunities for metaphorical expressions such as through fantastical and supernatural representations, or animals (Krekhov et al., 2019). Although some restrictions in this regard will probably always exist based on the system or the range of premade avatars available, metaphorical representations might be particularly suitable for the empowerment of some players and especially so for playful exploration of self-expression (Yee, 2006).

Future issues for studies on VR avatars

The field of user– or player–avatar studies and especially in the context of embodied VR would greatly benefit from novel conceptualisations and frameworks that review, contrast and integrate existing perspectives. Some existing relevant veins, as mentioned previously, include avatar preferences, player-avatar parasocial relationships, player types and the Proteus effect. Current studies on embodiment effects are predominantly based on replicating or exploiting individual psychological theories and effects that are sometimes not even accessible in physical reality (Beyea et al., 2023; Pan and Hamilton, 2018). However, the more complex and dynamic mechanics and effects related to the user, embodiment & avatar and the contextual in-between of the self are largely missing meaningful considerations (Bujić et al., 2022). Some attention to this was given, for example, as a response to reductionist player typologies and suggested differentiation between the person (user), personage (character) and persona as the self that is in-between and shaped by both the person and personage (Vahlo, 2018), although with limited development. However, the field lacks suitable frameworks and the following tools to fully understand the dynamics between the user and an embodied avatar. This investigation is necessary for enabling user literacy on what they truly engage with within these experiences, as well as for predictable, meaningful and sustainable use of virtual reality technologies.

Similarly, even if a boundless avatar customisation system were available, there is a lack of comprehensive conceptual frameworks and operationalisations to appropriately examine the possible effect of nuanced self-representations, and particularly so with the added complexity of individual cultural, social, embodied and psychological differences. This study is thus limited by the existing concepts and instruments that are used as a composite collection that combines different approaches and theories rather than examining them within an existing framework that integrates the physical and virtual self. As such, other individual differences could be examined in future studies, such as cultural differences in the perception of visual self-representations and their appropriateness in different contexts, subculture identities that could influence their virtual behaviour such as related to gaming or transhumanism, or the role of body image or social anxiety. Similarly, other outcomes that would be relevant in this context could relate to changing one’s schema of self-concepts as ideas of what constitutes or shapes the individual and how the processes of self-appraisal and perceptions of others’ appraisal of oneself interact with embodied virtual self-representations.

Conclusion

This study contributes to our understanding of user–avatar relationships through a mixed-method experiment in a collaborative embodied VR setting. It builds on existing veins of research on embodiment, avatar customisation and the implications of avatars on self-perceptions by studying these aspects in a controlled social VR environment and further differentiating between intrapersonal and interpersonal layers of the dynamics between them. Key findings suggest that allowing users to customise their avatars has a positive effect on their self-perceptions during social interactions but that these effects are highly individual and likely context dependent and require extensive further research. This study also shifts the focus from personalised to idealised and playful versions of self-representations as most beneficial, although it is unclear whether imposing such styles on all users would have the same effect. In sum, this work suggests that providing opportunities for customisation enables better social VR experiences, reinforcing the importance of the choice of avatars in full-body VR and the complexity of the user–avatar relationship.

Footnotes

Data availability statement

The anonymised dataset used in this study is openly available alongside this article.

Funding

The author(s) disclosed receipt of the following financial support for the research,authorship and/or publication of this article: This research was funded by the Academy of Finland (342144;‘POSTEMOTION’),Foundation for Economic Education (16-9394) and Finnish Cultural Foundation (00230774).

ORCID iDs

Mila Bujić

Anna-Leena Macey

Bojan Kerous

Oğuz Buruk

Author biographies

Mila Bujić is a Postdoctoral Research Fellow at the Faculty of Communication Sciences,Tampere University. With a background in anthropology (BA),game studies (MSc) and humans and technologies (games and gamification;PhD),her interests focus on the enmeshment of humans and digital technologies and its impact on individuals and broader societies. How do our perceptions and understanding of self and others change in these new technological environments and how could these changes be reflected in our collective daily practices? Her current major projects entail embodied interactions in social VR and futures research in transhuman digital technologies. Bujić relies on highly interdisciplinary phenomenon-based research and a wide range of methods. Her work has previously been published in venues such as Internet Research,International Journal of Human-Computer Studies and Interacting with Computers .

Anna-Leena Macey is a Doctoral Researcher at the Gamification Group,Tampere University,in Finland. Her current research focuses on emotional management and social interaction in immersive virtual reality environments,in particular the use of embodied and environmental cues and affective design to evoke emotional and attitudinal responses in users. Her aim is to examine the ways of utilising virtual reality and virtual environments to empower and increase the well-being of their users. She holds a MA in the field of Communication Sciences from the University of Helsinki.

Bojan Kerous is a postdoctoral researcher at the Gamification Group at Tampere University,Finland. He obtained his engineering degree in Computer Engineering and Informatics from Electrotechnical Faculty at the University of Sarajevo. His doctoral research,focusing on the examination of stress exposure and self-regulation in VR in healthy and patient populations,was conducted at the Visual Computing Department of the Faculty of Informatics,Masaryk University,Czechia. His current research interests lie in affective and multimodal human–computer interaction,with a particular focus on XR technologies and esports research.

Oğuz Buruk is an Assistant Professor at Tampere University. He is an industrial product designer and has completed his PhD in Interaction Design at Koç University – Arçelik Research Center for Creative Industries (KUAR). His work focused on creating novel interactive environments for gaming purposes. Specifically,he worked on topics such as gaming wearables or social touch in games. His research on these topics was published in high-quality venues of HCI and Design such as CHI,DIS,Visual Communications and the International Journal of Art and Design Education.

Juho Hamari is a Professor of Gamification at Tampere University and head of the Gamification Group at the Faculty of Information Technology and Communication Sciences. He is leading research on phenomena related to gamification and extended realities under the Academy of Finland Flagship Center of Excellence and University Profiling programs. Dr Hamari has published c. 200 articles on these topics,on varying perspectives such as information systems,consumer psychology,human–computer interaction and game studies.

References

Adam

Galinsky

(2012) Enclothed cognition. Journal of Experimental Social Psychology 48(4): 918–925.

Amico

Bruch

Haase

, et al. (2004) Trait shyness, actual-ought self-discrepancy and discomfort in social interaction. Personality and Individual Differences 36(7): 1597–1610.

Banakou

Groten

Slater

(2013) Illusory ownership of a virtual child body causes overestimation of object sizes and implicit attitude changes. Proceedings of the National Academy of Sciences 110(31): 12846–12851.

Banakou

Kishore

Slater

(2018) Virtually being Einstein results in an improvement in cognitive task performance and a decrease in age bias. Frontiers in Psychology 9: 917.

Banks

(2015) Object, me, symbiote, other: a social typology of player-avatar relationships. First Monday.

Baumeister

(1982) A self-presentational view of social phenomena. Psychological Bulletin 91(1): 3.

Bem

(1972) Self-perception theory. In: Advances in experimental social psychology, volume 6. Elsevier, pp. 1–62.

Bessière

Seay

Kiesler

(2007) The ideal elf: identity exploration in world of Warcraft. Cyberpsychology & Behavior 10(4): 530–535.

Beyea

Ratan

Lei

, et al. (2023) A new meta-analysis of the proteus effect: studies in VR find stronger effect sizes. PRESENCE: Virtual and Augmented Reality 1–14.

10.

Bujić

Macey

Järvelä

, et al. (2022) Playing with embodied social interaction: a thematic review of experiments on social aspects in gameful virtual reality. Interacting with Computers 33(6): 583–595.

11.

Frank

Gilovich

(1988) The dark side of self-and social perception: black uniforms and aggression in professional sports. Journal of Personality and Social Psychology 54(1): 74.

12.

Freeman

Maloney

(2021) Body, avatar, and me: the presentation and perception of self in social virtual reality. Proceedings of the ACM on Human-Computer Interaction 4(CSCW3): 1–27.

13.

Freeman

Maloney

Acena

, et al. (2022) (re) discovering the physical body online: strategies and challenges to approach non-cisgender identity in social virtual reality. In: Proceedings of the 2022 CHI conference on human factors in computing systems. pp. 1–15.

14.

Freeman

Zamanifard

Maloney

, et al. (2020) My body, my avatar: how people perceive their avatars in social virtual reality. In: Extended Abstracts of the 2020 CHI Conference on Human Factors in Computing Systems. pp. 1–8.

15.

Fullwood

James

Chen-Wilson

(2016) Self-concept clarity and online self-presentation in adolescents. Cyberpsychology, Behavior, and Social Networking 19(12): 716–720.

16.

Higgins

(1987) Self-discrepancy: a theory relating self and affect. Psychological Review 94(3): 319.

17.

Huang

Galinsky

Gruenfeld

, et al. (2011) Powerful postures versus powerful roles: which is the proximate correlate of thought and behavior? Psychological Science 22(1): 95–102.

18.

Hudson

Hurter

(2016) Avatar types matter: review of avatar literature for performance purposes. In: Virtual, augmented and mixed reality: 8th international conference, VAMR 2016, held as part of HCI international 2016, Toronto, ON, Canada, 17–22 July. Cham: Springer, pp. 14–21.

19.

Human

Biesanz

Parisotto

, et al. (2012) Your best self helps reveal your true self: positive self-presentation leads to more accurate personality impressions. Social Psychological and Personality Science 3(1): 23–30.

20.

Judge

Erez

Bono

, et al. (2003) The core self-evaluations scale: development of a measure. Personnel Psychology 56(2): 303–331.

21.

Koek

WJD

Chen

VHH

(2024) My avatar makes me feel good? The effect of avatar personalisation and virtual agent interactions on self-esteem. Behaviour & Information Technology 1–16.

22.

Krekhov

Cmentowski

Emmerich

, et al. (2019) Beyond human: animals as an escape from stereotype avatars in virtual reality games. In: Proceedings of the annual symposium on computer-human interaction in play, pp. 439–451.

23.

Lindeman

Verkasalo

(2005) Measuring values with the short Schwartz’s value survey. Journal of Personality Assessment 85(2): 170–178.

24.

Loewen

Burris

Nacke

(2021) Me, myself, and not-I: self-discrepancy type predicts avatar creation style. Frontiers in Psychology 11: 1902.

25.

McCambridge

Witton

Elbourne

(2014) Systematic review of the Hawthorne effect: new concepts are needed to study research participation effects. Journal of Clinical Epidemiology 67(3): 267–277.

26.

Maloney

Freeman

(2020) Falling asleep together: what makes activities in social virtual reality meaningful to users. In: Proceedings of the annual symposium on computer-human interaction in play, pp. 510–521.

27.

Maloney

Freeman

Wohn

(2020) ‘Talking without a voice’: understanding non-verbal communication in social virtual reality. Proceedings of the ACM on Human-Computer Interaction 4(CSCW2): 1–25.

28.

Mancini

Sibilla

(2017) Offline personality and avatar customisation. discrepancy profiles and avatar identification in a sample of MMORPG players. Computers in Human Behavior 69: 275–283.

29.

Maselli

Slater

(2013) The building blocks of the full body ownership illusion. Frontiers in Human Neuroscience 7: 83.

30.

Mason

Smith

Engwall

, et al. (2019) Self-discrepancy theory as a transdiagnostic framework: a meta-analysis of self-discrepancy and psychopathology. Psychological Bulletin 145(4): 372.

31.

Matos

Pinto- Gouveia

Gilbert

, et al. (2015) The other as shamer scale–2: development and validation of a short version of a measure of external shame. Personality and Individual Differences 74: 6–11.

32.

Mayer

Makransky

Parong

(2022) The promise and pitfalls of learning in immersive virtual reality. International Journal of Human–Computer Interaction 1–10.

33.

Messinger

Stroulia

, et al. (2008) On the relationship between my avatar and myself. Journal for Virtual Worlds Research 1(2).

34.

Morie

(2014) Avatar appearance as prima facie non-verbal communication. In: Nonverbal communication in virtual worlds: understanding and designing expressive characters, pp. 77–102.

35.

Mottelson

Muresan

Hornbæk

, et al. (2023) A systematic review and meta-analysis of the effectiveness of body ownership illusions in virtual reality. ACM Transactions on Computer-Human Interaction 30(5): 1–42.

36.

Osimo

Pizarro

Spanlang

, et al. (2015) Conversations between self and self as Sigmund Freud – a virtual body ownership paradigm for self counselling. Scientific Reports 5(1): 13899.

37.

Pan

Hamilton

AFC

(2018) Why and how to use virtual reality to study human social interaction: the challenges of exploring a new research landscape. British Journal of Psychology 109(3): 395–417.

38.

Peck

Seinfeld

Aglioti

, et al. (2013) Putting yourself in the skin of a black avatar reduces implicit racial bias. Consciousness and Cognition 22(3): 779–787.

39.

Peña

Hancock

Merola

(2009) The priming effects of avatars in virtual settings. Communication Research 36(6): 838–856.

40.

Praetorius

Görlich

(2021) The proteus effect: how avatars influence their users’ self-perception and behaviour. In: Augmented Reality and Virtual Reality: New Trends in Immersive Technology. Springer, pp. 109–122.

41.

Robak

(2001) Self-definition in psychotherapy: is it time to revisit self-perception theory? North American Journal of Psychology 3(3).

42.

Saffo

Yildirim

Di Bartolomeo

, et al. (2020) Crowdsourcing virtual reality experiments using VRChat. In: Extended abstracts of the 2020 chi conference on human factors in computing systems, pp. 1–8.

43.

Sanitioso

Wlodarski

(2004) In search of information that confirms a desired self-perception: motivated processing of social feedback and choice of social interactions. Personality and Social Psychology Bulletin 30(4): 412–422.

44.

Sibilla

Mancini

(2018) I am (not) my avatar: a review of the user-avatar relationships in massively multiplayer online worlds. Cyberpsychology: Journal of Psychosocial Research on Cyberspace 12(3).

45.

Slater

Sanchez- Vives

(2016) Enhancing our lives with immersive virtual reality. Frontiers in Robotics and AI 3: 74.

46.

Slater

Spanlang

Sanchez- Vives

, et al. (2010) First person experience of body transfer in virtual reality. PLoS ONE 5(5): e10564.

47.

Spanlang

Normand

Borland

, et al. (2014) How to build an embodiment lab: achieving body representation illusions in virtual reality. Frontiers in Robotics and AI 1: 9.

48.

Strimbu

O’Connell

(2019) The relationship between self-concept and online self-presentation in adults. Cyberpsychology, Behavior, and Social Networking 22(12): 804–807.

49.

Thibault

Bujić

(2022) ‘mirror dwellers’: social VR, identity and internet culture. In: DiGRA ’22 – proceedings of the 2022 DiGRA international conference: bringing worlds together.

50.

Vahlo

(2018) In gameplay: the invariant structures and varieties of the video game gameplay experience. PhD Dissertation, Turun yliopiston julkaisuja, Universitaris Turuensis. Available at: http://urn.fi/URN

51.

Wallace

Tice

(2012) Reflected appraisal through a 21st-century looking glass. Handbook of Self and Identity.

52.

Weidner

Boettcher

Arboleda

, et al. (2023) A systematic review on the visualization of avatars and agents in AR & VR displayed using head-mounted displays. IEEE Transactions on Visualization and Computer Graphics.

53.

Wikström

Martikainen

Falcon

, et al. (2020) Collaborative block design task for assessing pair performance in virtual reality and reality. Heliyon 6(9): e04823.

54.

Yee

(2006) The psychology of massively multi-user online role-playing games: motivations, emotional investment, relationships and problematic usage. In: Avatars at Work and Play: Collaboration and Interaction in Shared Virtual Environments (pp. 187–207).

55.

Yee

Bailenson

(2007) The proteus effect: the effect of transformed self-representation on behavior. Human Communication Research 33(3): 271–290.

56.

Zhang

Juvrud

(2024) Gender expression and gender identity in virtual reality: avatars, role-adoption, and social interaction in VRChat. Frontiers in Virtual Reality 5: 1305758.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

Virtually better: Multi-user experiment on avatar self-representation,self-discrepancies,avatar style and self-perceptions in a VR collaboration

Abstract

Keywords

Introduction

Background

Self-representation in virtual environments

Self-discrepancy and self-representations

Hypotheses and exploratory research questions

Method

Participants

Procedure

Materials

Task

Measures

Analyses

Results

Avatar customisation and self-perception

H1: customisation and self-evaluation

H2: Avatar customisation and other as shamer

H3: interaction effects of self-discrepancies

Exploring self-discrepancy, avatar style preferences and self-perceptions

Discussion

Differing contexts, differing minds

Avatar as a tool

Future issues for studies on VR avatars

Conclusion

Footnotes

Data availability statement

Funding

ORCID iDs

Author biographies

References

Supplementary Material