Bigger picture

[JanBiermann]

👯‍♀️ 👯‍♂️

I was trying to figure out what we need to be ready to push for a meeting with John. And to figure out what our results are so far.

To do this, I started reading the text you have written - hot stuff ! I have (only!) read through 3.1. I have read through everything.

I took notes (first for myself), but why not share with you.See comments in this pdf: [adm_JB_comments.pdf] (https://github.com/JohannesWalter/BostonAdventures/files/9780799/adm_JB_comments.pdf). Updated file: adm_JB_comments.pdf I didn't want to change anything in the written text. If you want to, you can look at my notes and act upon them if you find anything interesting. Or include them as comments in LaTeX, so that we can discuss with John.

Re what next: (I lost track of this a little :D, because I became to busy reading). But my feeling is we should rather "submit" to John whatever we have (and try to describe to him what is not done). Feels like we could continue working productively without him, but it also seems that time has come to coordinate with him. Maybe we will manage to "submit" to him tmrw? Not sure if that's realistic....

Re what are main findings: Based on this subsections, here's my first try to summarize the main findings:

• With feedback: Start ignoring the algo + become good at the task
• Without feedback: Continue trusting the algo
• Algo_Rec from the previous round is an important anchor for this round. But only if the correct answer is not known (nofeedback). Otherwise correct answer becomes the anchor.
• Transparency: Some people disregarding the algo because of the explanation (peak is smaller compared to baseline) (contrary to what we have believed in the earlier version).
• [More generally: A lot happens between rounds (maybe even the most important mechanisms?). Implies: Obeseravtions are not at all independent. Implies: WoA maybe not the best measure.]

😺

[JohannesWalter]

I read until "Based on this subsections, here's my first try to summarize the main findings:"

I'll try to find the main findings myself and then we can compare. Could be fun to do it this way

[JanBiermann]

Sure, do this, I have updated the issue/my first comment

[JanBiermann]

I have revised the list of main findings I see. I have put it directly in the LaTeX file as bullet points.