Nicolas Darmanthé

Have you read "crocs in the cabinet"? The NT is an endless stream of shitfuckery

A scatter plot showing environmental impact score and nutrition impact score of different types of sausage. Below the scatter plot are 5 smaller plots, each highlighting one of the categories.

Small multiples can save even the most complicated chart. And the best thing is, you can use this technique for almost any type of chart. A scatter plot, for example!

medium.com/p/3aa0b039410

(Post 4 out of 4 while I'm on holiday, sharing some highlights from previous years)

Agree that this will be globally disruptive in the short term, but I'm optimistic that this will be a catalyst for incredible growth of research/leadership outside of the US in Eurasia/Australasia. It's sad for decent hard working Americans but the rest of the world may actually come out ahead.

What do you use? I've found duckduckgo alright, haven't really tried/encountered other alternatives.

It doesn't look exactly like it looks in this long exposure picture, but it's epic in a way that a camera can't capture. You get a real sense of depth and vastness when you lie down on the ground and look up at the sky in the outback.

The net result is something like starvation in the midst of plenty, with LLMs being the fast food of our online worlds giving us internet diabetes.

The internet has also allowed people who previously didn't have a voice/platform to share their knowledge with the rest of the world. The problem is that in parallel there's been a boom in the amount of "garbage content" that can be generated, shared, paraphrased by LLM and shared again ad infinitum

The internet initially supported the democratisation of information and lowered the barrier to entry for knowledge acquisition. Millions more people now have access to information that previously would have been locked away in a university or state library.

It was already hard to find quality content in the ocean of information before LLMs. I always tell people that the issue isn't the lack of resources/information about a topic, the issue is finding the good quality information (and knowing it is so). AI has made the problem even worse.

Google search peaked 5 years ago. Total garbage now.

Been there good luck 😂

Agree for simple cases, seems to break down when more advanced logic is required

Lecture slide with quotes claiming that migraines are caused by psychiatric issues

Lecture slide showing information on stomach ulcers and how they are not caused by stress

Lecture slide with quotes claiming that AIDS was caused by psychological factors, with original article in which these quotes appeared

Lecture slide with quotes claiming that multiple sclerosis is caused by early childhood trauma and lack of positive bonding experiences, with original source (from 2002)

It's that time of year again! Today I look at the history of so-called "Medically Unexplained Symptoms" with my students (part of our #pseudoscience class).

Migraines? Rooted in "perfectionism"
Ulcers? "stress"
AIDS? "shame/guilt"
MS? "childhood trauma"

Tropes now recycled for #MECFS & #LongCovid

Towards a biologically annotated brain connectome - Nature Reviews Neuroscience High-resolution maps of biological annotations in the brain are increasingly generated and shared. In this Review, Bazinet and colleagues discuss how brain connectomes can be enriched with biological ...

Nature Reviews Neuroscience

Towards a biologically annotated brain connectome
 🧠🟦
www.nature.com/articles/s41...

I typically give extensions no questions asked but keep track in case it becomes a recurrent issue. I also don't like blanket penalty rules like "-2% every day this is late". Instead I might just be slightly less lenient/have higher expectations around the "no excuse for messing this bit up" stuff

I just discovered obsidian yesterday and I love it to bits. I can finally move out of the Evernote ecosystem! It feels very vscodey and I love that the notes are just simple markdown. Will make it a breeze if I ever decide to publish notes to a website

This used to be possible on twitter until their API had its brains blown out by elong

Looking forward to better API and documentation. Would be interested in ability to pull all (or sample of) posts containing a keyword, with the handle it originates from, if there is a target handle (e.g. reply), if it is a repost etc. This would allow us to do basic social network analyses

And yes for sure there's people that are naturally very gifted and just get it quickly (eg Karl Friston?) but I still think "slower learners" can bring super interesting things to the table too and should be encouraged and supported

I wonder how many people there are out there who think they're terrible at maths, but are actually quite gifted if they can find a problem that interests them and can build up some confidence 🤔 I've met a few people (particularly girls) who I thought unfairly saw themselves as bad at maths

Even discrete maths and basic probability maths (which I wasn't so bad at to start with) became so much cooler when I started having interesting & practical problems to solve (rather than abstract textbook exercises)

In fact I got better at maths thanks to coding. I realised "oh, this expression is just a fancy math way of writing a for loop" for example. And linear algebra finally became super interesting (and easier) once I had a reason to compute things on n-dimensional neuroimaging related matrices.

I felt the same about dyscalculia. A few early traumatic experiences in the French education system tanked my confidence. Still not great at reading pure maths, but have good intuition about numbers/can visualise well. Find maths so much easier when I need it for a practical reason and can code it.

Keen to see this site grow but at the same time loving the small tightly knit communities. Content feels more relevant and engaging. Easy to have little discussions with people. Feels cosy/like Twitter used to be back in the day!

Does it get better after residency?

Actually I should say re-discovered because I just realised I retweeted this paper back in October 2021 🙄😂

This was so good! I discovered Justine Hansen's other related paper where she looks at neurotransmitter <-> structural/functional brain organisation and OMG! And she made the neurotransmitter PET receptor maps available and that is just priceless. She is a legend!

I love how easy it is to read/interpret despite no figure legend

We're hosting Eli Muller to talk to us about his paper "Diffuse coupling in the brain - A temperature dial for computation" TOMORROW (Friday 6th October Asia time)
sciencedirect.com/science/arti...

Open invite—please join!

Details: www.world-wide.org/seminar/9586/

#ComplexSystems #Neuroscience