Prep: weekly mtg 20160504 (Byron, Steve, Matt, Aaron, me) - mobeets/nullSpaceControl GitHub Wiki

discussion points

function of null space activity:

• Aaron: method for keeping carbon-copy, an information channel for redundancy

ideas:

• decode cursor velocity from null space activity? [kind of an orthogonal question, but might be interesting]
  • plot cursor movement as guessed by null space, vs. true cursor movement

to do:

• no linking group for sfn

first class: uncontrolled/unconstrained second class: minimal energy third class is habitual AND minimal intervention habitual/pruning: "intervene minimally on the habitual behavior" what "intervening minimally" means is left open now

using learning focus is not just that there IS task-related activity it's that we're predicting it using a previous mapping

learning allows us to dissociate hypotheses main result is about distribution of null space activity

manner in which redundancy is handled at the neural level may be different than the way it's handled at the level of the muscles