Impressive keynote lecture by Peter Dayan Today at #pug2025 illustrating that we need to combine recurrent neural networks with characteristics of simpler cognitive models to describe a looooot of choices in 🐁- important implications for modeling human data as well
