The Psychology of Video Games

A few months ago I wrote an article for GamePro magazine about what effects deindividuation and anonymity had on gamers. For those of you who aren’t subscribers or who didn’t pick up the magazine, GamePro recently published the article on their website for your clicking pleasure.

Nifty artwork by Andrew Yang

I turned Bobo the Quote Monkey loose on the article and he came back with this:

A recent comprehensive review of the whole body of deindividuation research appeared in the journal Psychological Bulletin. The review confirms that studies where there’s a strong, external message about how to behave were the most likely to elicit the deindividuation effect-but it didn’t always result in antisocial behavior. For example, in one study researchers repeated the electric shock experiment described previously, but had some anonymous subjects dress up like Ku Klux Klan members and others dress up as nurses. The people in the white Klan robes shocked more, while those dressed as nurses-a profession associated with helping and healing-shocked less. Why? While the people under those uniforms knew they were anonymous, part of a group and likely experienced an “I am not who I normally am” feeling, they still took some of their cues on how to behave from the environment. By understanding the results of this study, it’s not hard to see how expectations were placed on the subjects to behave the way they did when under the influence of deindividuation.

The same logic applies to the gaming world.

At any rate, enjoy.

Back in January 2010 when I launched this site, I laid out the things it could lead to on a continuim from low to high. On the low end was “Nobody likes it, everybody dies.” On the very top of the high end was “Book deal, everyone lives” and close behind that was “Someone hires me to write magazine articles about this stuff.” Well, there’s still no book deal but a few months ago GamePro’s John Davison contacted me saying that he liked the site and wanted to know if I was interested in writing for the GamePro print magazine.

After I finished fist pumping, I said that I most definitely was.

Fast forward to today and if you pick up this issue of GamePro you’ll see my article on the psychology of anonymity starting on page 49 and accompanied by some awesome artwork by Andrew Yang. Here’s the cover of the issue:

Just look for the murderous Alice and you'll find it.

Here’s a snip:

Psychologists actually have models of what anonymity tends to do to people because they’ve been studying its effects long before the first person ever rage quit a game of Pong. While little of that early research involved video games, it did employ painful electric shocks, children in Halloween costumes, and college co-eds dressed up as nurses –sometimes two of those things at the same time.

…But is “antisocial” our default mode when we bring up a web browser or multiplayer menu? Is donning a virtual version of Jack’s face paint by adjusting the “brow height” slider on a character creation tool sufficient in and of itself to make us punt all morals out the window? Psychologists say no, it’s not. According to recent research on the topic, there are additional factors at play, which redefine the whole issue.

I had written a bit on deindividuation and anti-social behavior here, but while conducting actual research for the article I found out that I didn’t know the entire story with the current state of research on deindividuation and anonymity. You can read the GamePro article for the whole thing, and if they ever put ito n GamePro.com I’ll certainly link to it there as well.

All in all it was a fun experience writing the article, with the not unsubstantial bonus that they paid me to do it. When the anonymity piece was done my editor Patrick went on to assign me a second article, which has at this point also been written, turned in, and scheduled to appear in next month’s issue.¹ And as of this moment I’m working on a third piece, so apparently they like what I’m doing well enough so far. If you like the anonymity piece or have something else interesting to say about it, please let GamePro know at feedback@gamepro.com. If you don’t like it, then write your Senator or something. I dunno.

While we’re on the topic, though, I really like changes that GamePro has made to its print magazine in the last several issues. They seem to realize that they can’t compete with websites for timely content or reviews, so they seem to be going for more in-depth stories that require some actual research and reporting. Stuff like the history of GameStop, the impact of Metacritic on the gaming scene, the nature of bug testing in games, the impact of piracy on the games industry, and more.

I’ve been published in “legitimate” outlets both online and in print before, but I’m still happy to be in the pages of GamePro. Blogging dreams do come true! Plus it’s nice that nobody has to die.

1. Print scheduling lag is weird. [↩]

Earlier this week Blizzard dropped a big AoE by announcing that it was greatly reducing user anonymity on its Starcraft II and World of Warcraft forums. Everybody who posts on those boards will soon have their real first and last names displayed. So Trolly McTrollpants will no longer be able to post under that name …unless that’s what it says on his credit card. Which seems doubtful.

(Update: A few days after making the announcement, Blizzard recanted due to rather loud and displeased reactions from players. They have noted, though, that they still have other plans for changing the behavior of their forum posters.)

Aside from wanting to make Battlenet (the system that handles matchmaking and other socially oriented tasks for players of Blizzard games) more of a social networking tool, the intent of this change seems to be to bail out some pretty nasty bathwater from the forums, even if that means a few smiling babies get tossed, too. Trolls, flames, and vitriol in 48 flavors seem to be a big problem in the official forums. Given WoW’s stupefyingly large player base, the company has too big a community, which is a weird but apparently real problem. I imagine its forum moderators and community managers are overwhelmed and this nuclear option of sharing real names is meant to do two things: 1) reduce the number of people using the forums, and 2) make people be nicer to each other by robbing them of their anonymity. I think it will succeed at both, though it’ll be a bumpy ride.

As I’ve written about before, anonymity has some well known psychological effects. People who feel like less of an individual because they’re an anonymous part of a group may be more likely to look to social cues and the behavior of other group members to determine their own behavior, and this often results in their being antisocial jerks. Reducing anonymity can in turn reduce this kind of behavior and make people a little more likely to be courteous and hold their tongue if that’s how they’d behave in face to face interactions. So I think Blizzard is likely to see results from this. It’ll also be magnified by the number of thoroughly anti-social jerks who withdraw from the forums entirely because they don’t want to risk reprisal for their actions.

(Personally I think they’re making the right move, though it’ll cost them. Also among those leaving will be those who value their anonymity for other reasons –because they’re women who don’t want to deal with other players’ knowing it, because they don’t want others to know about their hobby from a Google search, or because of any other plausible reasons. But I guess that Blizzard has done the social arithmetic and decided that those are acceptable losses in the face of making their forums usable.)

What’s fascinating to me, though, it to wonder how Blizzard might be trying to measure the impact of this decision internally. I mean, technically Blizzard isn’t really doing anything to increase accountability for your posts; they’re just changing the social and psychological context. If I were working there and put in charge of this task, I’d first identify some metrics that would be of interest. Some of them are no brainers, like the number of new posts/replies and new account registrations. But you could get creative, too. You could look at random posts and get measures of how frequently certain words are used, ranging from rude phrases (“stupid” “shut up” “noob” and the like) to curse words and euphemisms for curse words. Maybe you’d want to look at average word count per post, too, since derogatory and pithy one-line replies might be less frequent than more thoughtful replies. There are probably already lots of existing models of verbal aggression out there in the psychology literature that you could draw from to make your predictions.

Once I had those metrics figured out, I’d ideally do a phased rollout of different forums or different regions that would allow me to have a control group under the status quo and an experimental group that gets the Real ID treatment. You could then collect data for a week or a month, then compare the two groups on those metrics. Or, if you just had to do a simultaneous rollout of the changes across the board (pardon the pun) then you could use archival data to do before/after comparisons.

So there you go, Blizzard, if you’re not already doing something along those lines, you should and you should even look to publish it in one of the new academic journals addressing video games and electronic media. People have been hypothesizing about this stuff before, but nobody has ever had a chance to actually test it on this scale. My consulting fee can be paid in purples and epic mounts.

While doing research for an article on the effects of anonymity on player behavior, I came across a fascinating study that I couldn’t find a place for in that piece, but which I wanted to share somewhere.

In an article appearing in the Journal of Personality and Social Psychology¹ cultural anthropologist R. J. Watson considered the question of what makes it easier for young men to kill enemies² while at war. Deindividuation theory holds that people who lose their sense of self-identity are more likely to lay aside internally held morals and look to the situation or the dominant morality of a crowd for guidance. You see this kind of thing most starkly in people who become anonymous and blend into groups.

Think about it: soldiers and warriors throughout history have changed their appearance when they go into service, especially when they prepare for active battle. At a minimum, they change their clothes, hair style, and mannerisms. Obviously, there are various benefits to such standardization, including making it easier to outfit an army, build group cohesion, and tell friend from foe. But at the more extreme ends, warriors apply war paint, piercings, masks, or make other drastic changes to their appearances. Watson wondered if soldiers and warriors who most radically change their appearance were more likely to not only get on with the killing, but be more willing to engage in more brutal acts like mutilation and torture. To test this hypothesis, he looked at archival data about world cultures compiled by anthropologists, missionaries, and other first-hand observers.

The results were pretty stark and pretty clear: in 90 percent of the cultures where wartime opponents were mutilated or tortured³ the acts were done by warriors who radically changed their appearance before going into battle. One could see this as support for the supposition that when you make a person feel less like an individual and more like a faceless part of a group, they’re more likely to go whole hog when you sic them on the enemy in a combat situation. Then, when peace time comes, they can step out of that identity by reverting their appearance.⁴

As you can see, the addition of ...well, actually this one does kind of look like me, so maybe it's a bad example.

This got me thinking about character creators in video games. A lot of games allow extensive control over the appearance of your in-game identity, letting you adjust eye size, nose position, cheek height, chin prominence, and many frankly ridiculous other factors. Some of these are so extensive that you might be able to create, if you wanted to, someone that looks quite a bit like YOU instead of applying all kinds of fierce tatoos, face paint, piercings, or bowler hats. (If you actually do have fierce facial tattoos, then, well, that’s cool. That’s cool.) Other games even let you use photographs of yourself to virtually put yourself in the game.

So, given Watson’s findings above, would you expect people who made in-game avatars to look like them act differently, on average, from those who make savage looking avatars that look nothing like them? I think so. This is one of those times when I wish I had an awesome research lab complete with computers and a popcorn machine, because I’d love to do this study. Deindividuation theory suggests that playing as someone who looks like you would lead you to pay more attention to your internal moral compass (whatever that may be) in the same way that losing your identity behind a costume would make you more likely to adopt the morals of those around you or the ones implied by your environment or even the costume itself. Heck, one study⁵ even showed that simply placing a mirror in front of subjects led to this kind of effect, so it make sense. Maybe you’d have a harder time playing that “evil” character in the latest Bioware RPG.

So, anyway, somebody get out there and do this research. Then when you publish it I’ll settle for being listed as the second author.

1. Watson, R.J. (1973) Investigation into Deindividuation Using a Cross-Cultural Survey Technique. Journal of Personality and Social Psychology, 25, 342-345. [↩]
2. Or anyone, really [↩]
3. And killed; don’t forget killed [↩]
4. And, of course, nobody means disrespect to our (or your) men and women in uniform; this study looked at acts far beyond donning simple Army fatigues [↩]
5. Froming, W. J., Walker, G. R. & Lopyan, K. J (1982). Public and private self-awareness: When personal attitudes conflict with societal expectations. Journal of Experimental Social Psychoogy, 18, 476-487. [↩]

What makes people be such jerks in online games and messageboards? And why do these kids keep stealing my Halloween candy? Wait, actually I think I remember reading something about this…

Ah yes, a classic experiment done by Edward Diener at the University of Illinois Urbana-Champaign and his colleagues way back in the 1970s¹ Little did they know it, but Diener and his colleagues laid the foundation for understanding, at least in part, why all those smacktards on Xbox Live keep calling you a “gaytard,” whatever that means. And they did it with candy –delicious, 1970s candy.

Diener et al. were interested in what situational factors lead children to engage in antisocial behavior, specifically stealing and breaking rules. So in a clever little quasi-experimental study they had researchers in Seattle, Washington turn 1,352 Halloween trick-or-treaters into unwitting subjects. They set up several houses’ entryways with two bowls –one with fun-sized candies and another with money. When children came up to the door, the experimenter would greet them and tell them to take ONE candy from the bowl. They would then say they had to attend to something out of sight in another room, but would actually go behind a fake wall and spy on the children through a peep hole² to see how much candy and/or money they took.

Under the right conditions, these adorable trick-or-treaters will cuss you out for five minutes on Xbox Live before disconnecting.

That was the general setup, but experimenters at different households also introduced what your Research Methods 101 textbook would call “manipulations” because of how they manipulated or changed one of the factors of interest to the researchers. For the anonymity manipulation, some kids were asked their names and address while others were were not. For the altered responsibility manipulation, the researcher would sometimes single out one kid –the youngest– and state that THAT tyke would be responsible if anyone took too much candy or stole any of the money.

The results were pretty stark. The kids who stole the least were those who came to the door alone (that is, not part of a group) and who were robbed of their anonymity by being asked their name and address. They only engaged in antisocial behavior 7.5% of the time. In contrast, simply being in a group –even when everyone provided identifying information– almost TRIPLED the transgression rate to 20.8%. But behavior was even worse for those costumed kids who believed the researcher didn’t know who they were. Even when alone, 21.4% of those anonymous kids stole, and when in a group that figure more than doubled to 57.7%.

But you guys, wait, that’s not the worst of it. The absolute worst offenders were those groups of kids that were anonymous AND for which the experimenter had singled out one kid and said “I will hold you responsible if any extra candies are missing.” In those groups, the transgression rate was 80% –EIGHTY PERCENT!

Number of little monsters stealing under each condition, adapted from Diener et al. (1976)

The researchers hypothesized that the reason for these increases in antisocial behavior had to do with “deindividuation,” which is when a person is made to lose a sense of individual identity. Remaining anonymous does this, as does being in a group and having a scape goat on whom to pin responsibility for your behavior.

What does this have to do with people being asshats in online games, messageboards, and chat? A lot, I think. When you allow people to be a) anonymous, b) part of a group, and c) not responsible for their actions, then you get more antisocial behavior. As the guys at Penny Arcade accurately (if vulgarly) put it:

How can players and game designers counter this? Well, from the player’s perspective, playing only with people on your friends list obviously eliminates the anonymity factor. If you find someone you like playing with, add him/her to your friend’s list. Or organize games in online communities where anonymity is reduced, like social networking sites or message boards. This is one reason why I think a lot of gamers ground their teeth when Infinity Ward limited the use of party chat in Modern Warfare 2.

Developers and the architects of matchmaking systems can help by making it easier to get extra information about a person and follow up with him/her after the game. Or even better, create tools for players to get together and share personally identifiable information (even something as basic as real name and location, which any message board could do now) so that they can more easily find and form their own groups. Heck, at the extreme end I kind of like the idea of making everyone use their real name in their gamer tags or handles, though that way is fraught with issues related to the privacy of children’s information.

Making it as easy as possible to do things like file complaints and player reviews also helps make people feel more accountable for their behavior.³ Muting and booting systems, if employed by more mature players, also seem like they would lower the likelihood of people modeling the behavior of others and transferring the responsibility of fouling up the group’s behavior.

So there you have it: the missing link between smacktards and Halloween. Just be glad that they can’t reach through Xbox Live and steal your candy.

1. Diener, E., Fraser, S., Beaman, A., & Kelem, R. (1976). Effects of Deindividuation Variables on Stealing Among Halloween Trick-or-Treaters. Journal of Personality and Social Psychology, 33, 178-183. [↩]
2. That feels …strange to write. [↩]
3. Though, ironically, I wonder if low feedback ratings that travel with a player from game to game might have the same effect as did the transferal of responsibility to one child in the Halloween candy experiment above; somebody test that! [↩]