Saucy Necromancers And the Psychology of Self-Perception

I recently started playing the massively multiplayer role-playing game Guild Wars 2. The character creation process is extensive, letting you mix and match 5 races, 8 professions, and 2 genders. According to my math, that’s like more than five different possible combinations. Maybe a lot more.

After trying many combos, I settled on the Necromancer class since I liked its use of pets, debuffs, and damage-over-time effects. Since I had been semi-randomly combining races and classes, this character happened to be a human female. Not too unusual for me, but the thing is that the armor skins are unique to each race/gender combination and after 40 levels the female Necromancer garb has ended up looking a bit …saucy. Like to a ridiculous degree, as if my character was part of a Mardi Gras parade that took a detour through the clearance rack at Victoria’s Secret:

"I am the blessed disciple of  Grenth, dread lord of darkness, death, and --HEY, MY EYES ARE UP HERE!"

“I am the blessed disciple of Grenth, dread lord of darkness, death, and –HEY, MY EYES ARE UP HERE!”

A long line of research around self-perception that has shown that people’s attitudes will change to match their behavior, not just the other way around. One famous study1 showed that National Football League and National Hockey League athletes were more likely to earn penalties when they wore jersies that were black –a color associated with villany and unscrupulous cat burgalers everywhere. And it wasn’t just that referees reserved harsher judgments for them. A follow-up study showed that participants wearing black gravitated towards more violent games when given the choice. The researchers argued that these behaviors could be traced back to self-perception theory: I’m wearing black → I’m expected to be aggressive → I’m aggressive → I’m picking the game where I get to shoot the other guy in the face with a dart gun. Even if that whole chain of thoughts wasn’t conscious, the results were there.

In a way, video game avatars are like uniforms that we slip on. Researchers looking at self-perception theory’s implications for video games and virtual reality avatars have coined the term “the Proteus effect” to describe how players will make inferences about their expected attitudes and beliefs based on their in-game avatar’s appearance, and then act in accordance with those expectations. For example, one study2 found that subjects controlling more attractive, taller avatars were more outgoing and more assertive in a bargaining exercise, and another3 found that people were more likely to help clean up an accident after playing a game as a super hero.

So what about my sexy necromancer and titillating avatars in general? The degree to which this kind of character design is acceptable and/or harmful is the focus of much debate in video games and other discussions about popular culture, but it’s also been the subject of research by psychologists studying the Proteus effect and self-perception theory in the context of virtual reality. My Necro’s attire reminded me of a paper I had read about on VGResearcher by Jesse Fox, Jeremy Bailenson, and Liz Tricase, entitled “The Embondiment of Sexualized Virtual Selves: The Proteus Effect and Experience of Self-Objectification Via Avatars.”4 The researchers were interested in the effects of sexualized avatars on women’s self-objectification and what they called rape myth acceptance (e.g., believing that a rape victim shares responsibility for an attack because she was drinking or flirting would show acceptance of a rape myth). They were also interested in what effect it would have if the avatar’s face resembled the player’s own.

The setup was pretty simple. Subjects (all college age women) wore virtual reality gear and used a female avatar wearing either sexy attire that showed skin, or non-sexy attire that did not. In addition, for some subjects the avatar’s face was made to resemble their own using a digital photograph. Here are some examples from the study:

Sexualized avatars (top row) and nonsexualized (bottom row). Taken from Fox et al. (2013).

Sexualized avatars (top row) and nonsexualized (bottom row) avatars. Taken from Fox et al. (2013).

Participants then completed a simple task with the aid of another person (a male) within the VR environment. After the task, the subjects were asked to write down their current thoughts and complete questionnaires measuring rape myth acceptance. Results showed that those using a sexualized avatar were more likely to include thoughts about their real body (as opposed the avatar’s body or anything else) in the “write down your thoughts” task. Furthermore, those who were using the sexualized avatars with their own faces superimposed on them were more likely than other subjects to accept rape myths. That is, they were more likely to agree with statements like “In the majority of rapes, the victim is promiscuous or has a bad reputation.”

Now, I think this is preliminary research. The increase in body-related thoughts can probably be accounted for by a simple priming effect –if the subjects had used chickens as avatars they probably would have written more about barnyards and sandwiches for the same reason. But the fact that use of these avatars causes increased thoughts about body image seems clear regardless of the cause. The rape myth acceptance seems more troubling, and the authors note that this is “a dangerous attitude for a woman to have as a potential juror, confidante, voter, family member, or even victim.” 5 I’d love to see more research that teases out the reasons for this, as the authors don’t really offer a solid one.

So what do we make of this? I’m not going to wade in and say that these kind of avatar options shouldn’t be available. After all, I continue to play my Necromancer. But it’s worth noting that people are doing real research on what effects these self representations have on the people who use them, and you should be aware of it as you play, design, and interact. Video games are unique relative to other media in the amount of interactivity and customization they offer and the ways that people use them.

UPDATE: Jessie Fox, one of the study’s authors, posted a blog entry about some of the misrepresentation the research has gotten in the press (thankfully, not here). Worth a look.


1. Frank, M. G., & Gilovich, T. (1988). The Dark Side Of Self- And Social Perception: Black Uniforms And Aggression In Professional Sports.. Journal of Personality and Social Psychology, 54(1), 74-85.
2. Yee, N., & Bailenson, J. (2007). The Proteus Effect: The Effect Of Transformed Self-Representation On Behavior. Human Communication Research, 33(3), 271-290.
3. Rosenberg, R., Baughman, S., & Bailenson, J. (2013). Virtual Superheroes: Using Superpowers in Virtual Reality to Encourage Prosocial Behavior. PLoS ONE, 8(1). Retrieved October 19, 2013, from
4. Fox, J., Bailenson, J., & Tricase, L. (2013). The Embondiment of Sexualized Virtual Selves: The Proteus Effect and Experience of Self-Objectification Via Avatars. Computers in Human Behavior, 29, 930-938.
5. Ibid., pg. 936

9 thoughts on “Saucy Necromancers And the Psychology of Self-Perception

  1. Wait. This is a 2×2 study:

    (M=1.95, SD=.57) than those whose avatar is sexualized with another person’s face (M=1.54, SD=.36) or those with the non-sexualized avatars (M=1.82, SD=.42 with own face & M=1.72, SD=.45 with other face).

    So Means:
    Self Sexualize = 1.95
    Other Sexualized = 1.54
    Self Non-Sexualized = 1.82
    Other Non-Sexualized = 1.72

    This indicates that seeing ones own face on an Avatar increases Rape Myth Acceptance, more than Avatar sexualization.
    It also indicated that seeing someone else sexualized decreases Rape Myth acceptance.

    Also, all results are within 1 standard deviation. And the Standard Deviation increases along with the mean. Indicating that the result is selective.

    • Good comment, Shea . Yep, the effects are admittedly not large, and as I said I’m not satisfied with the authors’ explanations of the RMA findings.It could be something as simple as another priming effect where people who are made to think about sexually provocative dress become primed to accept the myths. I’d like to see more research on the topic to tease out how the effects happen (or disprove them altogether).

      • Since they were interacting with a male participant during the experiment, it is possible the person they were interacting with could also have caused the effect.

        What I’m really confuse by is why seeing their own face on an avatar /regardless of avatar sexualization/ has such a noticeable effect on RMA…

        • Possibly it made it more ‘real’?

          I mean, in my experience, when you hear about a rape victim, even if they are described to be wearing a revealing dress, or a bikini for that matter, it’s clear and obvious they are still a victim (unless you are a true subscriber to the rape myth), and many women would strongly claim that the women is not responsible for the incident.

          But, if you walk in the street at 2 AM in a tight dress, and get guys following you and making suggestive remarks, you are likely to sigh and tell yourself “I should have wore a coat/taken a cab”. In other words, I know many women who wouldn’t dream blaming a rape victim for the crime, but would blame themselves if they have gotten into that situation. For many women, empathy and compassion is mostly reserved for other people, all the abuse goes inside (how many women tell other women they are fat, compared to the number that looks at the mirror and calls themselves that?)

          This is only from my personal experience, though.

          My problem with sexualized avatar outfits is the connection between ‘evil’ (or at least naughtiness) and sexual power. It’s part of the slut shaming concept, of considering women who wear sexual outfits as ‘bad girls’, of associating sexual confident with being ‘bad’. Evil man do sometime show their abs, but it’s rarely to the same extent.

          It’s also not fair that male avatar get’s cool armor, and female ones gets ridiculous chainmail bikini. Even worse, sometime the armor’s design is pretty cool, but have those odd boob windows or lacking pants. So frustrating!

  2. I’m very interested in follow-up research on this. Thanks as ever for these interesting studies you do on the psychology of games. I know many players like myself always feel affected by games but it’s impossible for us to tease out the real reasons for that.

    These results are definitely strange.

  3. Pingback: Friday Thinking by John Verdon – Organizational culture/control, Big Data, Robotics, Sensors, Design, Small worlds — Institute for Knowledge Mobilization

  4. Pingback: Links for November 5, 2013 | Andrzej's Links

  5. Tracked this “study” down to the Stanford press release, that gives a huge blaring endorsement to outrage exploiter and self-proclaimed non-gamer game critic Anita Sarkeesian. I’m a Marxist feminist, but I think there’s a lot of largely un-addressed, uncritically accepted “social theory” passing for feminism and social justice activism, that really does not deserve to be considered so. It’s alarming to see the name of someone like Sarkeesian, and a for-profit website not affiliated with the “study” being cross promoted in such a press release:

    “Many female gamers assert that gaming culture is not welcoming to women. The website collects user-submitted accounts of sexual harassment women experience in online platforms such as Xbox Live. When women critique sexism in games and gamer culture, they are often dismissed or even bullied. Pop-culture critic Anita Sarkeesian faced a barrage of cyber-bullying – including threats of rape and death – for announcing a project examining common tropes of female characters in video games.”

    I’m a scientist by training, and find a study that inserts such a blatantly value-laden, normative moral-philosophical concept like “rape myths” into “research”, to have no place in a serious scientific study. Its subtle, but perhaps even worse than seeing Shell or British Petroleum sponsor a study on “the global warming is man-made myth acceptance among players of post-apocalyptic games”.

    It seems like a fairly blatant attempt to insert a foregone moral conclusion into a study of psychology and self perception that would have been better served by examining self-perception in value neutral ways.

    Whether I have an opinion or moral stance on the question of an individual’s personal responsibility for their own safety, in what circumstances and to what degree, I wouldn’t set out to study how much they regarded themselves according to “rape myths” — basically a notion echoing the political bias that “the rapist must be emphasized, not the victim’s responsibility, otherwise you’re blaming the victim”. This is essentially the critical subtext of such a non-scientific concept as “rape myth”. This is assuming the validity of a political opinion, then trying to pass it off as a universal psychological category. Frankly, a scientist should be checking and disclosing their bias, not trying to pass opinions into “research” this way. Regardless of what I believe, I am a scientist first, and this matters.

    Science is not a vehicle for pet political ideology; making it so hurts science, and it hurts feminism. This kind of “research” is neither, and rather cynical in its attempt to exploit both.

    • Thank you for your comment.
      I have to agree with this, I didn’t consider it before – I think that all topics should be debatable, even when Sarkeesian is doing almost a joke of her “crusade”. But yeah, science should be taken way more seriously. Maybe take this example as a start on how not to do things, and try and do it the right way, with a proper study.

Leave a comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.