My statement is strictly local.
-
-
Replying to @KordingLab
Not sure what you mean. You assume that there exists a loss function that decreases over time. No?
1 reply 0 retweets 2 likes -
Replying to @RomainBrette
for, arguably most, problems there exists a loss. I am dealing with the simple scenario: task->learn->retask in such cases animals almost always improve.
2 replies 0 retweets 0 likes -
Replying to @KordingLab @RomainBrette
I always wonder how much animals in the world learn this way, however. My bet is that single-trial learning and generalizing across a variety of mostly disimilar experiences shapes behaviour more than gradual performance increase on repetitive tasks.
2 replies 0 retweets 6 likes -
Replying to @MHendr1cks @RomainBrette
As a movement scientist I fear I will have to strongly disagree. You are literally doing 3 repetitive things (eye movements) per second while you read this tweet.
1 reply 0 retweets 2 likes -
So does this imply that the cerebellum (mostly feed forward network) is doing gradient descent?
1 reply 0 retweets 2 likes -
Replying to @IntuitMachine @KordingLab and
I immediately thought of the classic prism glasses experiment. When vis field is shifted, targeting slowly changes over several attempts, not the instantaneous jump that could be achieved by calculation.
1 reply 0 retweets 1 like -
Replying to @jpmartinsci @IntuitMachine and
we have done lots of such experiments. Its like every time you do it you correct by 30% behaviorally. But attribution is to a broad range of things.
1 reply 0 retweets 1 like -
Replying to @KordingLab @jpmartinsci and
I get where you're coming from but I don't think sensorimotor adaptation and the plasticity around decision making / behavioural choices are ethologically similar.
1 reply 0 retweets 2 likes -
Replying to @MHendr1cks @jpmartinsci and
Hmm. The whole brain is just a movement co-processor. And evolved as such.
2 replies 2 retweets 14 likes
Have you thought about the movement as an *internal* set of oscillators, like the movement in a watch? The oscillators get disturbed by changes in perception and changes in needs, and the brain is the supporting control structure that regulates against these disturbances.
-
-
Replying to @Plinz @MHendr1cks and
sure. Popular idea in movement science. E.g. fig 1 in Churchland jPCA paper. But we have little evidence. Best is maybe the spinal cord work.
1 reply 0 retweets 1 like -
Replying to @KordingLab @MHendr1cks and
For motion loops it seems obvious, but how about the core of mental agency?
1 reply 0 retweets 1 like - 4 more replies
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.