Summary: weekly mtg 20170622 (Byron, Aaron, Emily, Matt, me) - mobeets/nullSpaceControl GitHub Wiki

to do:

• reviewers: get opposed ideas from matt
• keywords sound fine
• cite when saying spiking takes energy
• send Aaron/Emily new version

rewrite of results:

• first para stays as just bland intro?
• two paras on what's the problem we're trying to answer, and what tools we used to answer these
• need to move around or remove parts of intro so it's not redundant with results?

rewrite of intro:

• rephrase so that the three hypotheses are first principles intuition
  • like, here's the question we have, what are the options?
• and then we're going to introduce/apply tools to distinguish between these three
• we find that the fixed repertoire principle is what works

results:

• instead of results first, try writing this like tools first?

• write minimal energy para so that it's clear that the approaches we take here are used for all later hypotheses, that this is the simplest example to start with; make it more explicit when certain steps are going to be used for other hyps

• need to also mention that in the full exp., unlike the example, we consider eight directions and not two

Fig. 1B,1C,1D:

• 1B: green arrow in one direction (v_1), different color in the opposite direction (v_2)
• 1C: color points/lines the two colors
• 1D: same
• play around with it
• or just label 1C's potent axis with "Left" and "Right"
• and add cursor movement arrows on 1D for "Cursor: v_1" to the left
• think about options to have predictions in 1C or 1D...
• 1D: the 0 tick is confusing/unnecessary

For Emily:

• random gaussian sample in 90x10 matrix, then take SVD
  • and maybe swap out the U
• because normally we're just shuffling cols of FA matrix for OMPs