Profile cover photo
Profile photo
Jason Fleischer
8,045 followers -
Computational neuroscience, robotics, soccer, photography, and grilled cheese sandwiches
Computational neuroscience, robotics, soccer, photography, and grilled cheese sandwiches

8,045 followers
About
Jason's posts

Post has attachment
With great sadness I share the news of Gerald Edelman's death.

Back in 1972 he had already accomplished more in science than most would in their whole lives. Not content to rest on his Laurels (pun whole-heartedly intended), Dr. Edelman decided to switch from Immunology to Neuroscience, where he made breakthroughs in theory that are every bit as fundamentally important as those that netted him the Nobel prize. He was tireless, even relentless in his pursuit of science. But he was also a lover of the arts and a student of the human condition. Every time he told a joke, it was really a life lesson he was trying to get you to learn. Because we never seemed to learn those lessons he kept telling the same jokes, over and over.

I feel deeply privileged to have had him as a mentor. My deepest sympathies to his family, our thoughts are with Maxine, David, Judith, and Eric.
Photo
Photo
2014-05-19
2 Photos - View album

Post has attachment
Don't judge a study by its cover, and especially not by its abstract. Also, a little digression into the history of scholarly publishing which provides an nice bit of perspective on our system's troubles   

This afternoons headline (for me) at COSYNE: Ulanovsky showed us that fruit bats (being rodents) have the same kinds of things going on in their brains when they navigate that rats (so well studied) do.  It's fascinating is what happens to hippocampal place and grid cells and head direction cells in 3D, and what it means that the theta oscillation in bats is intermittent at best.  

Some things that I liked this afternoon at the COSYNE meeting: Ostojic talked about a new means of looking at stability simulated networks with asynchronous irregular firing.  Most analyses look at homogenous perturbations to firing, which means that they are looking at changes for all neurons in the network.  This new method looks at heterogeneous changes in both neurons and time, opening up some much more interesting possibilities in dynamics.  Gjiorgjieva looked at a model of how changes in neural gain nonlinearities might explain how young (just-born) animals have travelling waves of activity across the brain that disappear as the animal matures. 

This morning at the COSYNE meeting:  Mrsic-Flogel showed us that the inputs to mouse L2/3 V1 neurons pretty much determine their receptive fields, and that the brain is like the economy: the top 10% of synapses explain 70%+ of the activity.  Britto and Gerstner showed us that every kind of learning rule that will generate simple cell RFs can really be reduced to non-linear Hebb; even more fun is that it's almost hard to pick a non-linearity that won't give you simple cell RFs.  So now your visual models better try to fit something else other than Gabor's, since just about anything can do it. Schneidman showed us that weak correlations between pairs of neurons still add up, and that once again we can apply the brain is like the economy metaphor: that only the most common firing patterns need be used to fit a good model of neural correlations (ignore the 75% in his case). Gao and Ganguli showed that adding more neurons to your data set doesn't necessarily increase the dimensonality of your data;  that's dictated by the animals task complexity.  Freeman's lab has a neat parallel processing toolkit to do big data analysis on TB neural datasets.  Many many more, that were interesting, but those are the ones that I think are of interest to many.

If you're going to be at the COSYNE meeting (Computational and Systems Neuroscience) in Salt Lake City let's meet up at some point.  Also I will be posting a few thoughts about the presentations as they happen. Stay tuned for the next few days...

Post has attachment
This is exactly the kind of science writing we need more of: 1) it's in the right place to make non-scientists notice; 2) it's thorough and doesn't try to oversimplify, but it does provide metaphors that help a non-expert grasp the idea; 3) it focusses on providing a broad overview of an important field, not hyping some random "advancement" fresh off the news release wire

Post has attachment
Humor and science education, two great tastes that taste great together.  Hopefully this will lead a bunch of people to google the term synaesthesia and read some of the classics of pop-neurosci by +Oliver Sacks  or  V.S. Ramachandran :)

Post has attachment
Female?  You have two X chromosomes, one from your mom and one from your dad.  It turns out that in most cells only one of those X chromosomes is active, the other has been turned into a useless lump. And it might be the case that some of your organs are operating on mom's X while others are operating on dad"s.  It's fascinating stuff, and it might have big implications for understanding how genes make proteins to make you.  Oh yeah, and curing cancer, etc..  Although to be honest, if you do any research into basic science it's a good idea to figure out how eventually your work will help cure cancer ;)
Wait while more posts are being loaded