Summary: weekly mtg 20160302 (Byron, Steve, Matt, me) - mobeets/nullSpaceControl GitHub Wiki

Discussed

• low angError vs. high angError: good that there's not any relative change in hypothesis performance if you change these things
• behavioral asymptote: they want to see it unbinned by target-cursor angle
• behavior: they want to compare it to other measures, like success rate, and acquisition time
• look through different datasets to find data with obvious behavioral asymptotes
• plot CCA corr of Nul,Row activity over time, unbinned, along with behavioral points (smoothed over time)
• What is corr of Row activity with cursor progress even telling us? (Why shouldn't it be 1, since cursor activity determines cursor progress?) Basically, no need to look at corr of Row activity with behavior. and the plot of this corr for Nul with progress vs corr for Row with progress, with a strong positive correlation, this should also show up in the other plots over time, so there's no need to pursue this route further

Notes from meeting

match up progress with other behavioral metrics

• matched by given target, so x-axes align
• or just for any trial

find some sessions that have clear asymptotes

every timepoint

• cca in every bin of 16 or so trials

• correlation of YN and YR

• cursor progress

• success rate smoothed also

• acquisition time

  • (confirm these match patrick's numbers for the whole block)

(be interesting to compare cursor progress w/ other behaviors over time)