Neuroscience in Conversation: Tracking Scientific Change Through an Open Seminar Archive

Abstract

Scientific talks capture how researchers frame questions, debate ideas, and communicate emerging methods, providing a complementary record of scientific change to the published literature. We investigated whether an open neuroscience seminar archive could reveal shifting scientific attention and contain predictive information about subsequent discourse. We analysed 470 English-language transcripts from the Clinical Neuroanatomy Seminars YouTube channel, spanning 2010-2026 and approximately 2.6 million words, using longitudinal vocabulary analysis, unsupervised topic clustering, conceptual comparisons, and rolling temporal forecasting. The archive captured sustained discussion of language, white matter anatomy, connectivity, clinical neuroscience, and neurobiological mechanisms, alongside changes in their relative prominence. Network-related terminology became more prominent than localisation-related terminology, references to individual variability and computational approaches increased, and open science practices became more visible in scientific discussion. Beyond retrospective description, a regression model predicted keyword frequencies one and two years ahead with approximately 12% and 17% lower mean absolute error, respectively, than a persistence baseline. This modest gain is better explained by smoothing of year-to-year fluctuations than by genuine prediction: terms showing the strongest recent growth tended to fall back rather than continue rising. These findings suggest that spoken scientific discourse shows clear longitudinal change but offers limited evidence of information about subsequent shifts in attention. Interpretation remains grounded in the archive curated scope and changing content composition. By making nearly two decades of freely accessible neuroimaging education amenable to systematic analysis, this study connects open educational practice with the study of science itself. Open seminar archives can serve as reusable research resources, preserving the conversations through which scientific communities develop and revise their ideas.