Assessments and student expectations can drive learning: students selectively study and learn the content and skills they believe critical to passing an exam in a given subject. Evaluating the nature of assessments in undergraduate science education can, therefore, provide substantial insight into student learning. We characterized and compared the cognitive skills routinely assessed by introductory biology and calculus-based physics sequences, using the cognitive domain of Bloom's taxonomy of educational objectives. Our results indicate that both introductory sequences overwhelmingly assess lower-order cognitive skills (e.g., knowledge recall, algorithmic problem solving), but the distribution of items across cognitive skill levels differs between introductory biology and physics, which reflects and may even reinforce student perceptions typical of those courses: biology is memorization, and physics is solving problems. We also probed the relationship between level of difficulty of exam questions, as measured by student performance and cognitive skill level as measured by Bloom's taxonomy. Our analyses of both disciplines do not indicate the presence of a strong relationship. Thus, regardless of discipline, more cognitively demanding tasks do not necessarily equate to increased difficulty. We recognize the limitations associated with this approach; however, we believe this research underscores the utility of evaluating the nature of our assessments.