How will task outcome (i.e., success or failure) and goal-setting figure in the constructivist learning environments propounded by Dave Jonassen? If learning is optimal when it is a kind of play, then we should be interested in the evidence from research on academic, and other, risk-taking that people are most strongly attracted to games and learning activities presenting a moderate (35% to 50%) chance of success.
Activities can have either learner-chosen or environment-chosen goals and success/failure criteria (taxonomists awake!). From Cynthia Leshin's [1 May 94] description, MUSE would appear to have self-chosen goals. If it's anything like the role-playing games I have encountered (RuneQuest), having the players/learners choose goals in no way diminishes the excitement of the game. People tend to choose goals that are of moderate difficulty or challenge relative to their self-perceived ability. OK, so far this all fits neatly into the constructivists' paradigm.
But how do environment-chosen goals fit into the picture? Should we regard them as forbidden and alien to the contructivist way of life? The strongly motivating effects of externally defined goals are obvious in any video game center or track meet. In role-playing games, a good gamemaster or dungeon master will embed attractive goals within the game, apparently to stimulate and guide the goal-setting of the players. It seems to me that effective "phenomenaria" (Perkins term) and learning tool designers will not shrink from similarly designing goals of moderate difficulty into the learning environment. External and internal goal-setting can be made compatible.