Since I studied psych in college I often think of games in terms of operant conditioning - we set up systems of reinforcement and punishment in order to guide the player into having the experience we want them to have. Mark Nau, who studied econ in college, often comes to the same conclusions about design via a different engine of thought.
*Spider-Man 2* demonstrated an interesting phenomenon - for the most part, people's favorite thing to do with that game was to A) free play around New York, swinging and exploring and B) find tall buildings and throw themselves off of them.
Neither of these things have built in game rewards in the game economy. Throwing yourself off a building is even a bit punishing, in theory. (Of course, behaviorists know that you can't tell if something's punishment until you watch subjects' response to the stimuli. For some people, seeing Spider-Man swan dive into concrete is apparently quite rewarding.)
Now that's great. We spent a lot of effort on the swinging and sense of speed and it's good that people are playing with it, even though we didn't really set up the game to encourage that sort of behavior.
Which got me thinking: maybe we should have explicitly encouraged that sort of behavior. Watching focus tests, it seemed like people were towing the line, playing the story missions, making progress - and in that sort of rigid setting, maybe they were forgetting to have fun. Maybe a lot of our players forgot to have fun. Maybe we should have found ways to reward pure swinging.
But that got me thinking about a line of research that we only barely touched on in my psych classes: http://www.ntlf.com/html/lib/suppmat/1303a.htm and http://naggum.no/motivation.html are good summaries.
In short, play is intrinsically motivated. Putting extrinsic rewards and conditions on it turns play into a chore. (Which may be why I find MMO grinding such drudgery. Of course, that's just me. The rest of the world loves the grind.) And maybe that's one of the reasons why free swinging is fun; we never made it into a chore.
But how does this theory fit in with games like Zelda (where various kinds of power ups reward your side quests?) and Tony Hawk (where being able to do wicked aerial stunts and grind long rails should be a reward in itself, but the designers saw fit to build a whole scoring system around it?)
Exploring a dungeon in Zelda is great fun; it's my favorite part of the game. The reward is you progress the story and get to go to the next dungeon. There are smaller, side dungeons that you can explore for various sorts of power ups. What would happen if they took the progression and rewards away? "Here are a bunch of dungeons. Explore them in any order you please."
And what if Tony Hawk took away the high score system?
My untested theory is this: a given kind of play is going to be an intrinsic motivator for some period of time. The more interesting the play, the longer the time. Drawing could be fun forever, for example, but exploring Zelda dungeons might not last that long. At some point, people do get bored. Probably at a different point for each person. You can extend the life of your play by building a reward system around it. This encourages people to explore the play more deeply, and possibly discover rewarding things about it that they wouldn't have discovered without the carrot leading them on.
The most important kind of reward you can give is recognition. "You made it to the end of the dungeon"; "You found a new area"; "You did something you never did before." Points and power ups and unlockables are nice but at some point a player will realize that they're ultimately meaningless, and it puts you on a slope to Hell - each reward has to be bigger than the last; the rewards have to be commesurate with the challenge; or else you hear the fearsome "That's all I get?" When you hear "That's all I get?" it means the player isn't playing your game for intrinsic motivations anymore.