Designing a course on AI/ML in cognitive #neuroscience. First lecture's goal: keep them hooked with fun examples! What are your favourite ML/DL applications in neuroscience and psychology that made you stay until the end? 🧠🤖
Isomap
Deep-Q learning using Atari games as the environment :). I don’t do anything at all with that now as a computational linguist but I loved it as a way to talk about what it means to learn a policy gradient (or way more simply, how to react to things) in a continuous changing world :)
It’s slightly dated now (being updated this year!) but you should check out Chris Summerfield’s 3rd year course at Oxford on this topic if you don’t know it already. humaninformationprocessing.com/teaching/
I teach the AI side of this, so may or may not transfer. One of my hooks is juxtaposing game theory and cog sci research that both look at what it means for something to be “difficult” or an “optimal decision” or similar. This recent paper is my current go-to, but I haven’t explored the space much:
To study cognition, researchers have traditionally used laboratory-based experiments, but games offer a valuable alternative: they are intuitive and enjoyable. In this Perspective, Schulz et al. discu...
DeepLabCut has legitimately transformed how we do behavioural neuroscience, and the tracking videos look great!
I’d mention Jane Wang’s prefrontal cortex as Meta RL paper, the rise of deep RL in 2010s, and David Silver’s team’s achievements alphastar, alphago, alphafold, alpha proof.