This study presents knowledge diffusion analyses for researchers who have been active in the social sciences and humanities. We compare a network based on switches between the main disciplinary classifications of the documents authored throughout their careers to a discipline similarity network. We find that researchers are not exclusively switching between disciplines that are most similar cognitively. Only less than a third of the authors do not switch between disciplines. On the level of the individual researchers, we also study how the cognitive distance travelled relates to boundary crossing. The cognitive distance authors travel is approximated by calculating cosine distances between the vectors of title records for different periods of activity. Moving to new disciplines is found to be positively related to the cognitive distance an author travels throughout her career. Increased specialisation leading to different types of disciplinary boundary work is suggested as a potential explanation for this finding.