Aeron Laffere @aeronlaffere@qoto.org

Let me introduce myself! My name is Ilenna Jones, I am a computational neuroscientist with keen interest on how dendrites contribute to a singular neuron's ability to compute functions and learn tasks. I build biophysical models of neurons in #pytorch and use #deeplearning principles to investigate how #neuron models can learn and compute given their biologically realistic constraints.

I'm looking for postdocs right now! Feel free to connect if you're looking for someone like me!

I'm very interested in helping other students find resources and guidance as they consider science and #neuroscience in general. Feel free to connect if you're looking for advice/perspectives!

#computationalneuroscience #theoreticalneuroscience #neuroscience #neuralcomputation #learning #modeling #BlackInNeuro #FirstGen #Questbridge

ilenna.com

I realized I never did an #introduction!

I'm Grace Lindsay. I'm a new Asst. Prof in #Psychology and #DataScience at New York University. I wrote a popular science book on how & why we use #math to study the #brain.

In my research in my lab I hope to model how #attention works in the brain. I'm also separately working on applications of #MachineLearning to #ClimateChange!

Find out more here! 👇 https://gracewlindsay.com/

Our work showing that the locus coeruleus broadcasts unsigned visuomotor prediction errors to facilitate cortical sensorimotor learning is now on bioRxiv!
Please check it out, would love feedback! https://www.biorxiv.org/content/10.1101/2022.11.08.515698v1

#introduction I'm a neuroscientist at Oxford and UCL. I spend most of my time thinking about the front of the brain and how it controls behaviour. Sometimes this leads me backwards, but never really further back than hippocampus. I am also deputy editor at eLife so you can shout at me for that if you like.

Some Mastodon thoughts, for bird-site expats (which include myself). I'm aggregating these from posts I've boosted before, so little of this is my own brain.

- There's no algorithm here. That means favoriting/liking doesn't do anything except communicate approval to the OP and others (which is still nice!).

- No algorithm means boosting ("retweeting") is the true method to increase a post's visibility. Do that more than you did on birdsite.

- There's no post-quoting here, and that's by design. Look at quote-tweets on the birdsite; it's a feature primarily used for toxicity.

- There's no direct word-search here either; that means you want to use hashtags to make posts more searchable. This is also intended, since word-searching posts was often used to harass/stalk on the birdsite and elsewhere, so that was left by the wayside here. This also means hashtags are much more a thing here than any of the algorithm-powered sites.

- It's encouraged to put in text descriptions when you post images; a lot of Mastodon users use screen-readers due to various disabilities, and getting an image description read out loud helps them immensely.

- Speaking of screen-readers: using capitalization in your hashtags allows the screen-readers to read them more easily, especially if you're smashing multiple words together. #rockmusic = unreadable. #RockMusic = readable.

- The best way to make threads is to make set your first post as public, but "unlist" all of your replies. This prevents your whole thread from clogging up feeds.

- Content Warnings should be used more liberally here. If you haven't gotten the impression yet, much of Mastodon was built and populated by marginalized groups who were harassed/bullied off of other platforms. This is the culture they built, to respect each other's mental health. It's not a rule, but it's well-appreciated.

- Consider chipping a few bucks towards whomever runs the server you're on; the strain is real, and most server admins were likely paying out of pocket before so don't have an existing donation base. The growth here has been extremely fast, and that means money's needed.

- DMs are just posts with privacy settings. So if you @ someone in a DM, you pull them into the thread. That could be embarrassing.

- Also, no, DMs aren't end-to-end encrypted, but they aren't on Twitter either. Don't use either if you want true privacy.

- Including your Mastodon handle in your birdsite profile will help people find you here; there's a tool (pruvisto.org/debirdify/ is one of them that's used) people can use to pull Mastodon handles from Twitter profile.

- Use the blocking and reporting features liberally, if needed. This should go without saying, but they work, and work well!

- If there's an entire Mastodon server you don't want to hear from, you can block the whole thing too.

- Preferences -> Appearance -> "Slow Mode": this can make larger "Local" feeds and any "Federated" feed much more readable.

I'll reply with some more as I see them, or reply here too. I've only been here 4 days but I'm loving it so far.

Hakwan Lau and I have a new paper out in Structure and Function on how we should try to understand the brain, and prefrontal cortex in particular.

The online link is:
https://rdcu.be/cZbZg

If you click on this link you can read the paper online.

Last night Colleen Gillon, a PhD student in my lab, passed her PhD defence with flying colours! No corrections!!!

On that note, as part of her thesis, Dr. Gillon made a beautiful, easy-to-use code base for the data from our
Allen Institute OpenScope project (see a pre-print here: https://www.biorxiv.org/content/10.1101/2021.01.15.426915v2).

Check it out, analyse the data, let us know what you find! Here's the code:

https://github.com/colleenjg/OpenScope_CA_Analysis

@ejoftheweb @jburnmurdoch @t0nyyates
The German Data Commissioner is advocating that all public bodies should leave Facebook (and presumably Twitter) and use open platforms instead.

Various German bodies and authorities have already created their own Mastodon servers, as have many EU agencies and the Commission.

https://mstdn.social/@TheEuropeanNetwork/109307607062425853

#introduction

I am a PhD student @SaxeLab at the University of Oxford studying how brains can preserve previously acquired knowledge during learning.

My research aims at developing mathematical tools and using simulation studies to understand algorithms that describe and analyse learning in the brain. In particular, I am interested in the learning dynamics of gradient-based algorithms and how they apply to learning in biological organisms.

Be cautious in interpreting the results from fitting cognitive/computational models!

This is, in some ways, obviously true. But it is quite common in the field (and even in my own work) to "over-interpret" results from model fitting. I think the real long term solution to this is pre-registration, so that we are not using models to "discover" effects but to make specific predictions.

https://elifesciences.org/articles/75474

The move Twitter -> Mastodon reminds me of the move Matlab -> Python. I like Matlab, but Python makes more sense. So, I just signed up to pay a small monthly amount to Mastodon (info at https://mastodon.social/about). Not super clear if the funds go to my "instance" (mastodon.social) or to Mastodon in general (if such a thing exists?), because this fediverse thing is, hmm, complicated. But it seems worth it. As someone else nicely put it, I'd rather be the customer than the product.

#introduction I am an associate professor at the Gatsby Computation Neuroscience Unit and Sainsbury Wellcome Centre at UCL and visiting scientist at #MetaAI.

I think we're pretty amazing learners and my goal is to unravel the computational principles governing learning in artificial and biological systems. I'm broadly interested in the theory of deep learning and applications to neuroscience and psychology.

#machinelearning #ai #neuroscience #psychology #deeplearning #ucl #cifar