Summary: weekly mtg 20170426 (Byron, Steve, Aaron, Emily, me) - mobeets/nullSpaceControl GitHub Wiki

look at this angle during perturbation:

• the angle between velocities being decoded through both mappings being similar
• this should predict how much of a washout effect there is; i.e., how poor performance is when washout starts

what's interesting:

• don't want "washout is just not complete"
• we want "he knows it's the intuitive mapping again but the activity is still consistent with the perturbation mapping"

maybe start focusing on really the strongest sessions that show this washout effect and dig in there

• need to look at behavior on first three or so trials--he should be immediately good here

also look at the cursor trajectories

• plot dots per timestep so we see at which timestep the position was there

how do we rule out him just not being motivated? long-term WMPs.

options for showing a memory trace of the WMP:

• prune your cloud so that the activity still works for WMP
• or, your IME during washout is biased to work with the WMP

maybe there won't be a memory trace on OMPs?

look at sessions where washout performance just immediately drops to intuitive levels

• say on some sessions for all targets it looks like he's pruning, and other sessions he shows no pruning
• do these sessions correlate?

maybe he can't help but learn a WMP that still works with the intuitive

• see 20160722

overall, let's focus on some cherry-picked sessions for a while