Tim Behrens

This is a cool paper :)

I was explaining to my Ukrainian colleague the phrase ‘There’s no such thing as a free lunch’. She told me the equivalent in Ukrainian is ‘The only free cheese is in the mousetrap’ - which is so much better

Biggest human imaging study scans 100,000th UK volunteer UK Biobank scientists say the human body can be studied in greater detail than ever thanks to people like Steve.

What an absolutely extraordinary milestone. Congrats @fmrib-steve.bsky.social @fmrib-karla.bsky.social and everyone else involved.

www.bbc.co.uk/news/article...

Ooh this is cool. Thank you!!

:) Then ask her whether 5 grains of sand would be a pile if you made a pyramid out of them

Is the five rocks a pile if they are all touching the floor?

Scary. Authors building in hidden instructions for presumed LLM peer reviewers (in white text on white background).

Yeah. I agree it’s a similar idea, but I was thinking we could write a joint assessment of what the two papers and treat them as one story.

There is an opportunity here to be creative and still communicate awesome science

Also given every paper is assessed individually, not every paper has to have the same template.

We definitely want big stories with lots of data. Just want to think how to communicate them in a way that is best for both readers and authors. No longer have to do it in a monolithic pdf.

Yeah. Point taken. Bad choice of phrase.

I totally agree in principle, and when it works well it is wonderful, but it can cause a lot of disagreements. And unfortunately, in cases where it actually was equal, the second or third equal author loses out. Hiring committees treat the two authors differently even if we wish they didn't.

If you have major revisions, and the postdoc has moved on. No problem. Publish the paper with its strengths and weaknesses. Then publish a follow-up (with a new author list) to address the weaknesses. We could even go back and update the first paper's reviews/assessment.

EG: We could get rid of the horrid "These authors contributed equally": Just split into multiple linked papers with different author order. We could link the assessments together.

Let's say you have a journal that isn't worried about protecting an impact factor, so it didn't need to package a million results into a single paper (to maximise citation-to-publication ratio). What would you do? Couple of suggestions below. Others very much appreciated!

:)

I realise I am grump-quoting Dan’s cute tweet. Sorry :)

Or perhaps we could use citations to add evidence to a scientific point?? It is so depressing that even with something as simple as a citation we have confounded the way we communicate science with credit assignment.

Woohoo!!!!!! Congrats Anna!

Here is your last reminder that the application deadline for Imbizo.Africa is nearing quickly, the 1st of July, in fact tomorrow. Still the place where diversity is at its best in the world! Tell all who need to hear. #africa #neuro

Woohoo! Congrats Steve!!

Self-supervised, label-free 3D cell imaging is here EPFL researchers introduce CellSeg3D, a self-supervised tool for 3D cell segmentation in fluorescence microscopy, eliminating the need for manual labeling and enhancing accessibility for various biolo...

🥰 Self-supervised, label-free 3D cell imaging is here

Congrats to Cyril on this very cool first, first author publication 🏆! This started when he was a bachelor's student (and he's now completing his masters!).

And many thanks to my other awesome co-authors🙏

elifesciences.org/articles/99848

I want this superpower though!

Suspect other training times will be v high variance across the population and all oscillations will be suppressed by the 10hz toeplit regressor.

From looking at this plot I slightly suspect the answer will be good. The only diff between forward and backwards is the ~200ms training time. The only oscillation that doesn’t peak at zero lag is the ~200ms training time. It peaks at approx the correct lag (perhaps 10ms early).

It won’t be in the sims code I guess. Simulated data did not include oscillations. We’ll just meet up and then share our standard pipeline. You can test it on data with pre-fixed params. See what it looks like.

Btw- would be interested to see those images plotted as z-scores (across population). Both these ones and f-b (subtract copes b4 z-scoring). Also did you include the toeplitz regressors as covariates or not?

Next week is busy. Prob best in July some time.