Abstract

Nearly alll of us will know the traditional model for teaching imaging analysis. It consists of a series of lectures giving the big picture, combined with workshops using specialized software such as SPM or FSL . I’ve taught that way myself, for many years, but it gradually became clear to me that I was leaving my students with very superficial understanding. Meanwhile, my work in open source software was teaching me more efficient and reproducible practice for analysis, but I was not passing this on in my teaching.

One might defend the traditional model by asserting that we don’t have time to teach students to program, to understand the basic ideas, and to use efficient practice for computation. But – is that true?

We tested that defense by running courses on basic brain imaging that did teach programming and fundamental ideas from the linear model, as well as efficient tools and practice for reproducibility [1, 2]. In general the courses worked very well [1]. I believe they should inform a new standard model of teaching in brain imaging, and other fields with high demands for cross-disciplinary technical work, and the pervasive use of computing.

[1] https://www.frontiersin.org/articles/10.3389/fnins.2018.00727/full [2] https://bic-berkeley.github.io/psych-214-fall-2016/classes_and_labs.html