About Us | Game Reviews | Feature Articles | Podcast | Best Work | Forums | Shop | Review Game

Researchers create computer game that can sense player frustration

Brandon Erickson's picture

What if a videogame could automatically sense when a player is starting to get frustrated? What if, before the player decides to throw down his or her controller in disgust, a game could detect emerging feelings of impatience and provide just enough support to induce the player to continue? Last month’s issue of the International Journal of Human-Computer Studies details the results of a collaborative study between Microsoft Research and MIT in which researchers created a computer program that could successfully predict player frustration (with nearly 80% accuracy) based on purely non-verbal physical data.

The guinea pigs (24 middle-school students) were hooked up to a variety of sensors designed to pick up everything from number of fidgets to changes in posture to speed of head movement, among other things. Under this mass of monitoring equipment, subjects sat before a plasma TV and played a virtual version of the Towers of Hanoi puzzle. Throughout the activity, players had the option of pressing either of two buttons labeled “I’m frustrated” and “I need some help.” By analyzing the data obtained from “frustrated” subjects, the computer correctly predicted frustration nearly 4 out of 5 times.

Given its complexity, I doubt this kind of device will show up in a commercial videogame anytime in the near future. Nonetheless, it does present some interesting possibilities. Imagine being about to give up on a puzzle in Zelda only to hear “You’re almost there. Just keep looking.” Imagine a sports game that lets players score some points just when they’re ready to quit. Imagine being ready to break the Lair disc when the game suddenly offers an easier control scheme. An appealing fantasy, but don’t good game designers already do enough to prevent frustration in the first place?

Over the years, game designers have learned a great deal about the kinds of things that tend to make players want to give up (e.g., unforgiving save systems, unbalanced difficulty, poor controls, etc.). With the arrival of play testing and focus groups, the most frustration-inducing aspects of a game are usually removed or redesigned long before the game is even released. Some games do such a good job at keeping players motivated that they are far more likely to quit in order to eat or sleep than out of frustration. Indeed, sometimes even eating and sleeping aren’t enough to get players to stop.

Perhaps a more interesting line of inquiry for videogames would be to find ways of detecting various emotional states as a means of driving plot generation and enemy interaction. Imagine a role-playing game that could tell which character the player is most attached to and kill him or her off for dramatic effect, or a survival horror game that automatically shocks players right when they’re the most calm and unsuspecting. The possibilities are endless. Although researchers have not been very successful so far at differentiating specific emotional states using computers, this may one day be possible.

The idea of a game that can dynamically respond or even modify its own structure based on the player’s emotional state opens up some fascinating design possibilities. While research on computerized emotion detection is still in a rudimentary phase, it will be interesting to see where it goes and what affect, if any, its development may have on the evolution of videogames and other forms of interactive entertainment.

Category Tags
Articles: Editorials  

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

Another line of research I'd

Another line of research I'd like to add is that with such study's results, psychologists can reliably measure frustration in relation to aggression in video games. Whether the excitation-transfer theory has some weight in explaining why psychologists found such and such.

I know this is an older

I know this is an older article, but I wonder if you're familiar with the upcoming shooter Left 4 Dead? It's from Valve (half life/the orange box, and there's even a hidden teaser for L4D on the orange box), and it's a cooperative shooter that pits four players as survivors trying to escape a city (the given level, wherever) while fending off hordes of infected enemies, the infected being basically zombies that run fast and hit hard (think 28 days later, or the 2004 dawn of the dead).

What's really cool is what they're calling the AI director. Normally in games (shooters specifically) certain actions trigger the next set of enemies to attack you (or to "spawn," or some other event to happen), be it crossing a certain imaginary line (a door into a room, whatever), picking something up, or interacting with something else in the game. In L4D, there is an AI director that sends waves of enemies after you based upon its interpretation of your perceived stress.

There are no triggers whatsoever: the enemies you face are all dependent on what you and your teammates are going through at the moment. If you're getting your asses kicked, the director scales it back, gives you a chance to cool off. If you've gone a while without seeing action, the director hits you, and hard. Or maybe it doesn't; maybe it dispatches a tiny, steady trickle of enemies the whole time, then shuts off the water (so to speak) to build tension before unloading on you near the end. Or maybe it unloads up front, then shuts it off to leave you wondering when the next attack will come.

I don't know the details on how it does this, but I've heard it's absolutely wonderful to play. It builds tension and enhances the experience by making a bunch of ups and downs while you play, all on the fly.

This sort of technology actually ties in to several of the articles you've written. On shooters feeling better when they died versus when they killed, you mentioned that the temporary release from engagement was one possible reason. The director knows that gamers don't want to be pummeled all the time. You also discussed game length versus quality, and this kind of technology dramatically increases a game's replayability (quality) by changing the experience each time. The playtests suggest that players generally only succeed about 40% of the time in beating a given level. This also ties in to scariness, about what frightens people the most, and how it's sometimes what we don't see or what doesn't happen that scares us more.

Anyway, I just figured you'd be interested in a gaming advancement like this. Definitely one to keep an eye on.

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

Code of Conduct

Comments are subject to approval/deletion based on the following criteria:
1) Treat all users with respect.
2) Post with an open-mind.
3) Do not insult and/or harass users.
4) Do not incite flame wars.
5) Do not troll and/or feed the trolls.
6) No excessive whining and/or complaining.

Please report any offensive posts here.

For more video game discussion with the our online community, become a member of our forum.

Our Game Review Philosophy and Ratings Explanations.

About Us | Privacy Policy | Review Game | Contact Us | Twitter | Facebook |  RSS
Copyright 1999–2010 GameCritics.com. All rights reserved.