Patrick MineaultReally interesting talk at #main2022 from @SussilloDavid about reusing motifs in RNNs https://www.biorxiv.org/content/10.1101/2022.08.15.503870v1 https://www.nature.com/articles/s41593-018-0310-2
Ben Cuthbert@patrickmineault @SussilloDavid
I asked a question about the impact of encoding choice (response: shouldn't change re-use of dynamical motifs). I should have clarified.. what I'm really curious about is whether @SussilloDavid thinks that dynamics could be shared *across* stimulus encodings. In other words, do task variables need to be projected into a universal encoding space in order to re-use dynamical motifs?
David Sussillo