Thursday, May 24, 2012

The Wii, Game Design and Stroke

I recently took part in a research trial that aimed to determine if use of the Wii correlated with an improvement in dominant hand function, for patients who had dominant side damage from stroke. It was specific to stroke, rather than general traumatic brain damage, and if you're involved in the study and want to stay blind to whether I had a Wii, or not, stop reading now, because I'm about to get all dirty on your research protocol.

Over six weeks we either had some basic stretching to do, or we followed up the basic stretching with some (project-provided) Wii Sports action. I had the Wii. Our playing was supposed to be daily, no longer than 45 minutes a day, and seated. After two weeks, I felt I had to quit because it was so frustrating to play that it raised my blood pressure substantially. Since raised blood pressure correlates with death, the Wii study was a dud for me.

So, what was so annoying? Those of you who know my interest in game design, game systems, and now neurology will probably be unsurprised that I was looking at the experience analytically, and came away with some understanding of why I found it so frustrating, especially given the constraints of having had a stroke.

There are five games in Wii Sports, as well as training modes to help you play each game: Tennis, Baseball, Golf, Bowling and Boxing. Boxing was the first game I abandoned, after the first session I played it, because moving my left or right hand correlated so poorly with corresponding motion in my avatar, Sometimes, a punch with the right would end up coming from my Mii's left hand and vice versa. Since the goal was to improve right hand function, my left hand is just fine, and the correlation was so poor, I had no time to waste on the Boxing game, despite the satisfaction of punching the living daylights out of hapless automata.

Next to go, despite a real effort, was the Baseball: it was too hard. In other words the game itself was too coarse-grained, or required too much coordination for any improvement in my motor control to even register. Even if I was getting better at the hard stuff, and that wouldn't show up, the relatively easy stuff got such inconsistent results that the feedback loop necessary for getting my brain to rewire my right side more effectively was simply broken. A batter who missed two 90mph fastballs would then knock a 150mph fastball out of the park. (Please, I do not want to hear how this is, or is not, a reasonable approximation of how you play your rounders.)

Of them all, Bowling was easiest, and had a pretty consistent and tight relationship between the motion of my hand, the virtual motion of the game, and the score represented. But it was too easy: it was soon apparent that equivalent results could be obtained with far less right hand movement, and the brain wants to be efficient, so beyond a certain point, the game was useless therapeutically, it seemed to me. I still played it, though, because strikes are good.

Driving in the Golf game had a nice correspondence between the action of the arms and hands, and the action of the avatar. With wind and hazards to contend with, it was quite compelling, but the sport's scoring was, again, far too coarse. Since putting was horrible, and had a really bad (it seemed to me) correspondence between the controller and the avatar, it was possible to perform well in terms of motor control in the driving but get a terrible (real world) golf score, so as someone working on recovering motor control, the game was sending the wrong message.

Of them all, Tennis had perhaps the best relationship between hand, controller, virtual representation, and end result, but where I most often came to grief was in one of the study's constraints: I had to remain sitting down. So I fell foul of the chair I was sitting on rather than the game itself. Being of a generally placid mien, this was annoying as all hell. Although the result of the tennis match qua tennis match corresponded to my (perceived) performance less well, the ranking seemed to rise and fall according to how well I had done actually.

The study left me pissed off at Wii sports as a therapeutic tool, more than anything else. It seems like, in service to 'real world' considerations, both the Wii as a device and the Wii as a game engine have lost out. I admit to ignorance about the detailed capabilities of the Wii controller's accelerometer(s?), but the Wii has a USB port, and a Kinect will fit, so on that basis, here's what I can't believe doesn't exist already because it should:

- A game engine that exposes motion from tiny to full range of motion for the body parts of the player as easy-to-use quanta.
- Libraries that recognize good and bad use of hands and limbs that again offers very small quantized results. 
- Games that sit on top of the libraries that reward good motion and improved function only.

Our brains function when challenged appropriately, and we're wired to like it,but instead of using technology that already exists to make all people better, we're putting up with crap that makes Zuckerberg richer. There is no reason that, for example, the spell-casting system of a CRPG could not be built to reward good movement, and the game structured to require improvement for advancement. It would be mathematically trivial to scale advancement to handle the difference between a neuro-typical teen and a 60-year old stroke sufferer. If you get the level of abstraction right, your actual game designers barely need to think about what the player is actually doing: it doesn't matter. If the game is compelling and the mapping between improved function and granted reward is tight enough, the rest takes care of itself.

Of course, making the technology work well enough to genuinely determine whether someone's range of motion has improved in the right way with a Wii controller and a Kinect is not a trivial task. But, damn, it's not that hard and there is a huge market when it works.
Shit, make some of this recovery a bit more fun, please!