PhD student at the @Lampl_lab, @WeizmannScience, focusing on sensory systems neuroscience. Ashtanga yogi 🧘♂️
Israel
Joined June 2020
- Tweets99
- Following417
- Followers182
- Likes2.8K
Michael Sokoletsky retweeted
Multiple brain areas synchronize their activity to help a rodent accumulate the evidence it needs to make a choice, two new studies suggest.
By @claulopezneuro
thetransmitter.org/decision-…
Michael Sokoletsky retweeted
Just got back from the @AllenInstitute ‘s SWDB’24. It was intense, challenging & absolutely worth it! Truly amazed by how much can be uncovered when brilliant minds come together. Cracking the brain’s code was never a solo gig. A big thanks to the incredible organizers & TAs! 🧠
I know that from a statistical standpoint we should be extra skeptical of negative results, yet I can't help giving more respect to papers that choose to openly and honestly acknowledge them.
Two observations: (1) much of our time, especially in experimental science, is spent doing various menial tasks, and (2) there is a vast world of incredible free podcasts out there. So get yourself a pair of wireless headphones and make those menial tasks a little more fun!
Michael Sokoletsky retweeted
Very excited to announce the first release of NeMoS (NEural MOdelS) a package for neural data analysis that supports GPU acceleration. Check out our tutorials and try it on your data now!
Docs nemos.readthedocs.io/en/stab…
Github github.com/flatironinstitute…
Profile xitter.cf/nemos_neuro
And yet another mouse training tenet: early in training, inertia is a constant challenge. Mice tend to stick to one behavior, whether it's to always lick, always lick right, etc. Use a protocol that repeats the same trial type when the mouse errs, which will compel it to adapt.
Michael Sokoletsky retweeted
TENSS is almost here, but we’re still short on funds. Every contribution matters, no matter how small!
We need your help for TENSS to carry on and inspire the next generations of neuroscientists. Logistics costs have increased and funding sources are becoming scarcer. We are counting on support from the community. Please donate and spread the word!
gofund.me/a8e6ad34
Another mouse training tenet: Never assume they understand your task. Even with mice that long ago reached expert level, I start off every session advancing through all previous training stages. It's akin to beginning each lesson with a review of the previous material.
Excellent talk and book. Highly recommend sharing with anyone you know who is interested in going into neuroscience!
youtu.be/T4BBsSKkCBA?si=7rAH…
Mouse training is an art form, but there is one tenet I stand by: they naturally go from being hypermotivated to not motivated, and while you may control this to an extent by adjusting reward-to-punishment ratio, it's best to sort trials into states (e.g. Ashwood et al. 2022).
Michael Sokoletsky retweeted
In this new blog, I write about:
1. All that's wrong in academia—including bad papers, broken peer-review, the publishing industry, and precarious contracts.
2. How important it is *not* to become a cynic.
3. And my personal antidotes to cynicism creep.
eiko-fried.com/antidotes-to-…
A rare positive of the Covid pandemic is the explosion in lectures uploaded online - World Wide Neuro, Learning Salon, BIU, etc. Though they're not interactive as in person, they allow you to watch interesting talks without needing to wait for the next conference or symposium.
There's a special kind of frustration in trying to teach animals a task and finding they cannot learn it. Takes several miserable months, it's a negative result you cannot publish, and many people wouldn't believe it anyway.
This paper is quite heavy, but it's one of the best examples I've seen of isolating a purely cognitive variable in the mouse brain. It's easy to underestimate how difficult it is. nature.com/articles/s41467-0…
Want to do multiregion optogenetics? I just put up on GitHub a minimal GUI (written in Python) for optogenetic targeting protocols. All you need is an NI card and two galvanometric mirrors in your beampath. github.com/msokolet/optotarg…
How should one learn machine learning? While formal courses can provide deep theoretical understanding, it isn't necessary when using ready-made libraries. In this video, I talk about the importance of intuition and breadth, rather than rigor and depth: youtube.com/watch?v=Ite8EWH6…
Michael Sokoletsky retweeted
First lab research paper! We explored insula's activity patterns (the infamous "manifold"). We found very consistent manifold structure and dynamics, allowing decoding across different mice and different motivations. Check it out, feedback welcome! biorxiv.org/content/10.1101/…
If even the world's most educated people often fall for digital misinformation, then education isn't doing its job. For years I've thought that schools should teach critical thinking as a core curriculum subject. Nowadays I would argue it should be on par with math and English.