Reflections
Because of various constraints, my analysis of student performance is limited to informal (virtual) classroom observation. There are promising signs that students not only gained knowledge of the intersection of AI and neuroscience but also developed motivation to conduct research in that intersection.
This figure appears in an assigned video, Principles of fMRI Part 1, Module 5: Basic MR Physics.
In terms of knowledge, one indicator of positive gain is the difference in students’ interaction with guest speakers. In the beginning of the semester, I arranged for a guest, who is a professor in neuroscience, to visit our class. Students seemed less comfortable to engage in a scholarly conversation with the guest speaker, even though we already read and discussed papers published by the guest prior to the visit. In the second half of the semester, I also arranged for a number of guests to visit. Students were substantially more engaged in discussing deep concepts involving neuroscience and AI; they were able to use a much richer technical vocabulary, such as names of brain regions and analysis techniques.
In terms of motivation, a number of guests expressed interests in offering research opportunities for students in this course to take a deeper dive into cutting-edge research in the intersection of AI and neuroscience. Regarding those opportunities, students expressed high-levels of enthusiasm. To the best of my knowledge, some students have already taken the initiative to pursue those opportunities, beginning to participate in interdisciplinary research projects in neuroscience labs where they are able to contribute their technical skills.
The biggest challenge would be the need to rapidly change to an online learning environment as necessitated by the COVID-19 pandemic. This unplanned change impacted this course in two ways. First, as discussed in the Implementation section, the course uses active learning and gives students many peer-teaching and learning opportunities. A specific method is to share and present one’s Jupyter notebook. While this method works well during an in-person session, it was not easy to replicate that experience remotely. It took a few weeks for us to adapt and become accustomed to the online environment. Though I made efforts to approximate the active learning activities as originally planned, making use of features such as breakout rooms and annotations, the experiences are not the same.
Second, the course project in the original plan would let students collect new data using a portable neuroimaging device my lab has access to. However, after the campus was closed, this plan had to be scraped. When it became clear the course project needed an alternative, I involved students in figuring out what we should do instead. At the end, we decided to switch to a consumer-grade EEG headband that can be purchased cheaply. I was able to tap into a fund to purchase a unit for each student. Because each student has access to their own device, we were able to operate under the social distancing measure. The tradeoff is that students could no longer capture research-grade data for their project. Nevertheless, it was the alternative we agreed to as a class that can approximate the original plan for students to go through the entire process of AI and neuroscience research.