This paper describes the GenScope Assessment Project, a project that is exploring ways of using multimedia computers to teach complex science content, refining sociocultural views of assessment and motivation, and considering different ways of reconciling the differences between these newer views and prior behavioral and cognitive views. The project is refining a computer-supported introductory genetics curriculum and the associated formative and summative assessment practices. Researchers then manipulate motivational conditions in an effort to maximize participation in formative feedback. In the first of three planned annual implementations, four secondary school life-science teachers implemented a month-long curriculum. A within-teacher/between class design contrasted three different assessment conditions: grade-oriented, standards-oriented, and accountability-oriented. Self-report and observational assessments of learning, engagement, and motivation were collected in 16 implementation classrooms and 2 comparison classrooms. The implementation classrooms show significantly larger gains on the new transfer measures and comparable gains on the far transfer measures. In the 26 students whose participation in formative feedback was videotaped and analyzed, engagement was strongly related to learning gains. However, overall engagement in formative feedback was disappointingly modest, and there were no noteworthy effects of the motivational manipulation on learning or motivation. Further modifications to the curriculum and formative assessment practices are being undertaken in response to these findings. Four appendixes contain sample assessments and scoring rubrics. (Contains 5 figures and 50 references.) (Author/SLD)