Editing PositiveReinforcement

Even scratching the surface of ["Psychology"] (say, by attending ["Maggie Tai Tucker"]'s talk at ["GDC 2001"]) reveals that different types of behavior have different levels of effectiveness. Here they are, with the most effective first: 1. Reward added (increase behavior) 1. Reward removed (decrease) 1. Punishment removed (increase) 1. Punishment added (decrease) So, the best way to encourage behavior is to add rewards, and never add punishments if you can help it. And do it with a RandomReinforcementSchedule. And use a TokenBasedEconomy. ---- Examples: In Tony Hawk, the game encourages you to vary your routine by giving you more points for different stunts. I think the game industry has only made small inroads into the territory of games that reinforce rather than punishing. If we want to reach out more to the casual gamer, we'll have to keep exploring. (And if Ed's right, it will help us appeal to women, as well.) After you've explored behaviorism enough, you discover that the distinctions between the various kinds of reinforcers and punishments are gray and fuzzy. (You give a kid candy when he's good; is giving the kid candy reinforcement or taking it away when he's bad punishment?) The same goes for videogames. (In Tony Hawk, you get fewer points for doing the same trick again, but you still get points. Is that less reinforcement, or is it punishment?) We may find that once players get used to games that don't kill them when they fail, they may start complaining about games that don't give them enough points when they succeed. But we can burn that bridge when we come to it. ---- CategoryGoodIdea

Summary:

This change is a minor edit.

This wiki is licensed under a Creative Commons Attribution-Share Alike 3.0 License.

Username:

Replace this text with a file