How to read this

Every note opens with a block of fields. They are not decoration: CI rejects a note that gets them wrong, so they mean what they say.

The fields

title

The claim in one line, phrased as the paper would phrase it — not as the note concludes.

date

When the note was written. It is not updated on a typo fix. A note from two years ago that has not been revisited is telling you something.

status

draft — read once. reviewed — read again after a gap, with every source checked line by line. There is no third value, because a scale invites a middle that means nothing.

topics

Lowercase, hyphenated. Used for grouping, not for keywords.

sources

What the note rests on, each with a resolvable link — a DOI where one exists. A note with no source does not build.

The sections

A note runs in the same order every time: what the paper claims, what its evidence shows, what would change the reading. The middle section is the one worth your attention, because that is where a claim and its support are allowed to come apart.

The last section is a commitment. A note that cannot say what would falsify it is describing a belief, and belongs in Essays instead, where that is the point.

What to distrust here

The notes are written by one reader, unreviewed, often outside their author's field. Neuroscience notes in particular are read from the outside: the summary of a method may be right in outline and wrong in the detail that matters to someone who runs it.

So read them the way they are written — as a first pass with its sources in the open, checkable against the originals. If a note misreads a paper, the issue tracker is the right place, and a correction beats a new note.

Reuse

CC BY 4.0. Quote it, teach from it, translate it, disagree with it in print — name the source and you are done. Individual notes are citable by their page URL; the collection has a CITATION.cff and a Cite this repository button on GitHub.