Problem solving prior to instruction (PS-I) involves complex learning mechanisms such as prior knowledge activation across two learning phases. Goal formulations of such learning phases may interact with these mechanisms. We examine how differences in goal type (problem-solving vs. learning goal) and specificity (specified vs. unspecified goal) affect students' prior knowledge activation and learning. In a 2 × 2 experimental design, we varied the goal formulation in a PS-I design. Participants were randomly assigned to four conditions: specified and unspecified learning goals as well as specified and unspecified problem-solving goals. We analyzed the effect of goal type and specificity, and their combination, on students' conceptual learning and prior knowledge activation by analyzing the quantity and relevance of generated solution attempts. We additionally considered how learner characteristics, such as students' mastery-goal orientation, interacted with the experimental manipulation. Specified goals led to higher solution relevance, which mediated effects on learning outcomes. Learning goals resulted in better conceptual knowledge than problem-solving goals but did not affect solution quantity. Students with a pre-existing mastery-goal orientation achieved superior learning outcomes with learning goals, suggesting a motivational effect. Our findings highlight the importance of goal formulation in instructional designs such as PS-I. Learning goals in PS-I can enhance conceptual learning, particularly for students with a mastery-goal orientation. Specified goals that allow for relevant solution generation are crucial for effective prior knowledge activation, especially in traditional PS-I settings with problem-solving goals.