We address the extent to which students developed expert attitudes in the individual disciplines of biology, chemistry and physics during their degree programme. By attitudes we mean the way students think about the sciences compared to discipline experts. We used the standard CLASS survey instrument to compare student attitudes to biology, chemistry and physics with those of experts. The programme featured interdisciplinary modules involving biology, chemistry and physics delivered by a form of problem-based learning across a three-year (BSc) or four-year (MSci) degree. The survey was administered at the start of the programme and at the end of each year of the programme allowing us to execute a longitudinal study of changes over the three- or four-year degree. The survey was run over a total of six years to students on the Interdisciplinary Science degree at the University of Leicester. We find positive results in the formation of expert-like attitudes in biology and chemistry and for higher performing students in physics. We note that the nature of science, that is the beliefs of experts about science with which our students' responses are compared, are not taught explicitly in this programme, but acquired through experience of the problem-based pedagogy. Our conclusion is that students generally develop more expert-like attitudes in the separate disciplines, and hence a greater understanding of the sciences, over the first years of the degree. The numbers of students on the degree programme averaged around 20 and the questionnaires were completed by about 60 students in total. This imposes a limitation on the power of the survey. Nevertheless, we find some statistically significant results. This is the first use of the CLASS tool to investigate the development of attitudes across the disciplines with the same set of students and over an extended period. We believe this is an important contribution to the argument for greater interdisciplinarity at all levels in STEM education. [Note: The volume number (19) and publication year (2024) shown on the PDF are incorrect. The correct volume number is 20 and the correct publication year is 2025.]