evaluatingdesign.com

An interesting article about using physiological measures with game preference:
http://www.usatoday.com/tech/products/cnet/2007-08-21-video-game-brain-scans_N.htm

From the article:

“So game maker THQ tried an unusual research method last year to evaluate people’s early emotional response to its in-development shooter game Frontlines. Instead of asking a test group how it liked the game, as with most market research, the company hired technology specialist EmSense to measure people’s brain waves, heart rate and sweat responses while they played the military-theme game. Armed with that data, THQ took Frontlines in a whole new direction developmentally, said Bob Aniello, chief marketing officer at THQ.”

For many years, psychologists have investigated the relationship between subjective preference and physiological data, with some conflicting results. There is a relationship between arousal and subjective preference (in many cases, arousal can increase subjective preference for unrelated decisions - see the following article on arousal and partner preference.)

Heart rate and galvanic skin response can be used to measure arousal, making physiological measures appropriate.

It’s important to recognize, however, that several limitations of physiological data make it a questionable sell for discerning subjective preference. Consider the following issues with physiological data:

1. Sometimes, there’s no correlation between physiological data and subjective preference.
   In Heart and Emotion(Myrtek, 2003), several studies show differences in heart rate - but do not show differences in subjective preference. Further, differences may be moderated by the user’s experience level - for example, experts may actually show no difference in heart rate but still show increases in game preference. Similarly, novices may show a difference in heart rate that may not be correlated with actual game preference.
2. Physiological measures can be slow to respond to external stimuli.
   Heart rate is slow; Skin response is even slower. Earlier this week, I was talking with a fellow researcher about the potential of using HR with changes in camera angles. The issue? Changes in angles occur quickly - far too quickly for HR to “catch up”.
3. Teasing apart differences in physiological data can be difficult (and sometimes misleading).
   Heart rate can be affected by a number of issues - the individual’s age, amount of physical activity, etc. If a subject had caffeine earlier, their heart rate can be affected. What participants do outside of the experiment can affect the results that occur in the experiment. More to the point of this study, subtle changes may not be immediately obvious from physio data (see point #2 on the time lag of stimuli and physio response).

This is just to illustrate the very real challenges that occur in interpreting physiological data, and trying to incorporate it into preference data. Consider the procedure used with THQ:

“THQ is looking to EmSense for more information on how people respond to its content before it hits stores. The game maker will test an early-in-development game, or “pre-alpha,” by creating a minute-long CGI video that captures the essence of the game. Showing the clip to a small audience, THQ gets a read on the general response of would-be players, without them actually playing the game. In later stages of development, THQ might evaluate player responses to the game by exposing them to 20 minutes of play.

THQ struck a deal with EmSense last year; and with the technology, the company gets insight on play up to 50 minutes of the game, with comparisons to games in its genre.

“The tests helped us position the game — it was originally a cooperative-play, squad-style military game and now it’s more of an open world, multiplayer game.”"

I’m especially interested to see how physiological measures pointed towards a drastic design change. For now, a promising start might be in more controlled, incremental steps. One study we’re currently collecting data on at GMU is evaluating the effect of cinematics on both arousal and subjective preference. If we can show a significant difference using a very small measure, we may be able to begin to quantify the effect of different aspects of games and their impact on physiological measures.

More on this as our research develops.

If you enjoyed this post, make sure you subscribe to my RSS feed!