Linux Gaming News

GDC 2011: Valve’s Experiments With Biofeedback in Game Development

At GDC 2011, Valve researched Mike Ambinder delivered a fascinating lecture on how Valve measure player physiology during gameplay and their experiments with the data.
Valve always deliver a fascinating glimpse of the behind-the-scenes world of their games development processes at GDC, and this year was no exception. Mike Ambinder, an experimental psychologist that works at Valve, delivered a fascinating lecture titled “Biofeedback in Gameplay: How Valve Measures Physiology to Enhance Gaming Experience”.
The title is pretty self-explanatory – basically, Mike’s research involves measuring physiological readings and trying to figure out if those quantifiable metrics can be used constructively to alter your gaming experience.

So he’s measuring things like heart rate, skin conductance level (related to how much you sweat), facial expressions, eye movements, and even brain activity. The measurements are then correlated with recordings of the gameplay, allowing him to investigate factors like how much a subject’s heart rate changes when faced with major challenges in a game.

There are several different applications for this. The first one Mike mentioned was dynamically changing aspects of a game based on your physiological response – for example, anyone that has played Left 4 Dead will be familiar with the attack of a horde and the resulting extra stress that causes, especially if it is timed with the attack of a Tank. A possible use case for this sort of technology would be to notice when the player has been exposed to a long period of stress (‘arousal’, is the technical term) and then hold off on some zombie horde until the player has time to relax a little bit. Conversely, if the player is at a low state of arousal, it might indicate now is the time to throw them a new challenge like, like a Tank.

The video below shows an example – in the top right you can see the measured galvanic skin response of a player as they play through a level in Left 4 Dead 2, along with some commentary (apologies for the quality!)


Obviously, a big part of this sort of process is trying to understand if there is what Mike calls “an optimal engagement pattern” – that is, is there a magical combination or sequence of these high/low arousal states that results in a gaming experience that more gamers report as being enjoyable?

There was one example given in which the answer was a clear “yes”, and it’s one that anyone that has played a multiplayer game will almost certainly be able to instantly relate to. Turns out they tested a little old game called DOTA 2 – Defence of the Ancients 2 – and discovered that people really, really, really like knowing the physiological reactions of people on the other team.

Players loved to find out that what they did caused an intense frustrated reaction in the other player; normally you only get a tiny window into their rage through simple text chat, but actually seeing their vitals spike in reaction to something you do is – for obvious reasons – quite the enjoyable event.

This creates a lot of opportunity for developers as well in terms of game mechanics – achievements or other in-game rewards for causing a certain level of emotional response in a player, for example. You could potentially also use it in co-operative games to help detect when a teammates is in trouble – if their pulse spikes you might get a warning on the screen that could obviate the need for a painfully-typed text message or scrabbling for the voice key.

Another possible advantage that isn’t directly related to gameplay but might be related to a gamer’s overall experience is using player biofeedback for matchmaking. If you have player history that can indicate whether a player is high-strung and volatile, you might want to avoid matching them up with a placid, calm player and instead put them against someone else to ensure everyone is getting an experience more matched to their personality.

So far, the research indicates that there is a quantitative difference – player surveys reveal that gamers playing a game that is altered using this mechanism end up having a better time than a game with no biofeedback contribution, as you can see in the below slide. But there’s still work that needs to be done to better define “enjoyment” in a way that will lead to better measurements so researchers can gain more understanding in this field.

Another demonstration video was shown of a player running through a level in Portal 2 – using their eyes as an input device. Using some fancy eye tracking hardware, the crosshair of the game was aimed using the eyes – allowing the player to move independently of aim, something that any FPS gamer would certainly be interested in. It seemed surprisingly responsive and accurate, but the question still remains – is it more enjoyable. Their preliminary results indicate that decoupling movement and aim is a good thing.

There’s clearly a lot of potential for this research in terms of modifying the gameplay experience. There’s a lot of obstacles too – getting people to even put on 3D glasses is proving to be a bit of an effort, so asking people to wire themselves into their computer before playing a game could be perceived as a bit of a stretch – but even as a research tool it will no doubt provide extremely valuable information into how people play games. It’s an excellent example of the science of game development, and it’s great to see Valve thinking about these sorts of things in order to make the best games possible.

You Might Also Like

%d bloggers like this: