Student and teacher perceptions of course content are critical components of the teaching and learning process. Studies that have investigated such perceptions offer conflicting evidence of both convergence and divergence between the perceptions of students and teacher. These studies used data modalities such as interviews and surveys, which have different granularities of information in them. We analysed student and teacher perceptions of difficulty in a mechanical engineering lecture with multiple data modalities capable of capturing perceptions at different granularities. Our analysis revealed a multiscale nature of perceptions, wherein the students' and their teacher's perceptions can appear to both converge and diverge depending on the scale of interpretation, which is a function of the modality and granularity of data used in the study. At macroscale, the students and the teacher agreed about difficult sections of the lecture and the underlying reasons. However, their reasoning diverged when probed at finer scales. Furthermore, we note that convergence observed between the perceptions of the teacher and students at a coarse scale can sometimes hide differences in finer scale perceptions. We discuss the implications of these results for the interpretation of results from past studies and for practitioners.