It seems the full paper makes it clearer that 1. cookies were only provided in one session 2. there was no effect on grades / learning outcomes despite a change in evaluations

https://twitter.com/emollick/status/1318737313497812996