I was unprepared when opening the latest edition of favorite industry magazine Game Developer, when a blink on the cover should have already told me what was revealed on page 21 in this October 2008 issue. As my eyes went into saccadic overload, I could feel my lacrimal glands kicking in leading to an overlubrication of my cornea, which almost prevented me from reading the biofeedback article conveniently titled “Shoot to thrill”.
It was hard to believe at first: A major article on how the American EmSense Corporation was using biofeedback to evaluate engagement in games, obviously showcasing some results from their recent studies of an accumulated mass of 300 hours of data (physiological and gameplay). Ding! Ding! Did I not want to write a PhD about this topic? No reference to academic work (well, my papers might be a bit too new, but hey there is a whole crowd of game researchers, who have been into metrics and physiological measurement for a few years now)… Fighting an initial state of cognitive arrest resulting from the usual anxiety PhD students get when they have the feeling that their research has already been done by somebody else, I was later able to read the article in detail.
They have used a number of modern, mostly first-person, shooter action games, like Battlefield 2141, Call of Duty 3, F.E.A.R., Gears of War, Ghost Recon AW 2, Resistance: Fall of Man, Halo 2, and Half-Life 2. Then, they claim to use their own proprietary headset to measure the following: EEG, heart activity, breathing, blinking, temperature, and motion. Ok, so I am guessing here, that their bandana includes maybe four dry electrodes (opposed to the 32 channels we usually use in our research) and maybe also incorporates galvanic skin response (even though the head is not the perfect location to measure this).
The scientific question that remains: How accurate can this be to assess the factor they claim to be able to measure with this? Engagement, emotion, adrenaline and cognition. I already sometimes feel like I try reading someone’s palm when I sit in front of elaborate high-resolution EEG data, how would it be the data were in smaller resolution? Although, I have to admit that I do not possess equal powers as a team of 7 committed MIT graduates and they seem to use elaborate algorithms to calculate their engagement profiles. In a sigh of relief, I reminded myself that it is industry stuff that I am reading here, not a scientific journal. Therefore, giving them the benefit of the doubt, I continued to look at their findings:
- Cut scenes engage players only by sticking to a thematic emotion or referring to characters to which the player has already established emotional attachment. This sounds perfectly reasonable; however, I strongly wonder how they would evaluate emotional attachment without taking a facial EMG.
- Tutorials are more engaging if players experience real combat scenarios. When players are threatened in the tutorial phase already, they will likely learn the interaction mechanics faster than when they have to listen to an instructor. They also refer to an “emotional rollercoaster” here, a moment of calm instruction is followed by intense combat, which is explained in the next point.
- The alternation between brief calm moments and intense events or combat is keeping players more engaged. This is a hypothesis I can completely back up from the science side as in a study that I will be presenting at this year’s Future Play conference, I found that a level I designed around the concept of sequences of combat in alternation with regeneration/resting elicited more signs of players being the flow state than a regular level. As scientists explore the concepts behind game experience and what levels of engagement are elicited by what game elements, I think there could be mutual benefit.
- Players find close combat more engaging than shooting. This I find a promising premise for a future study as their argument seems reasonable, even though I still wonder how they asses positive emotion for a energy sword kill in Halo 2, using the equipment they described.
- Little goodies in games can sum up to a generally greater experience. This was in my opinion their weakest point. As of course they were analyzing game events in context, much like we do in our current studies, but they only giving a few examples here of what those little goodies can be (e.g. grenades in Halo 2). But what I think they are really trying to say here is that by designing engaging single player combat, you are more likely to have an engaging multiplayer experience, if that builds on your single player combat mechanics.
- Cut scenes need to be more Michael Bay and less Steven Soderbergh. Meaning that dynamic action scenes, especially when they involve game characters are likely to be more engaging for players than cut scenes that try to inform the player about something.
- Tutorials should grab you from the start. Meaning that if you are only trying to simulate the game during the first minutes of gameplay, players will likely be less engaged. And losing your players in the tutorial phase might result in your game not succeeding at all. This sounds obvious, but it is good to see it backed up by their data.
- Too much constant intensity will result in detachment of player engagement. This is something I would like to put to the test myself, as it seems an interesting hypothesis that players will lose interest if the game is too challenging. Getting challenge optimally right might only be possible in the aforementioned alternation between intense combat and cooldown areas.
- Repetitive actions result in players losing interest, so does assured outcome. This is basic wisdom of a good game designer, adding a random element to your games ultimately results in more fun. Something we see in board games using dice mechanics every day.
- Use of interesting weapons can have a large impact on gameplay. This is a more difficult conclusion, because it is hard to generalize. I would say that any kind of usable item in a game that has a high novelty value for the player is likely to keep them engaged as they might find the payoff of using it a driving factor for gameplay itself. The perfect example of this is Portal, where the whole gameplay mechanic was designed around a novelty weapon. I would like to study this in more detail as well.
They conclude their article with a statement that I find perfectly sound: The major benefit of using bio-sensory evaluation techniques for games is that you can find behavioral trends of players that allow you to estimate what parts of your game design are working and what parts are not. This makes this kind of analysis an invaluable tool for improving game design. Case in point, these methods will change the game industry for the better. With a mixed feeling of excitement and pressure, I will return now to analyzing the studies that consolidate my PhD. If you are into these kinds of studies, next week is a great chance to catch one of my talks at Future Play in Toronto, Canada and iTec, Darmstadt, Germany. In addition, all encouragements for finishing it soon are greatly appreciated.