This doctoral dissertation describes an experiment of computer-supported collaborative learning that investigates the effects of variables of discussion task specification on higher-order learning observed in asynchronous text discussion of an online higher education course. In addition to presenting a new model of collaborative information processing, this dissertation describes an experiment in which discussion groups completed versions of discussion tasks that varied in degree of task context and required outcome. Content analysis was used to assess conceptual conflict and knowledge generation, two factors associated with higher-quality discussion for higher-order learning. Conceptual conflict was measured through message disagreement and knowledge generation was assessed via the level of information processing displayed in discussion messages. The experimental results do not indicate that conditions of high or low task context and task written product have large effects on the quality of online discussion for higher-order learning.