What we know about the tardigrade nervous system
We read the nervous-system literature into a graph, one statement at a time, each tied to its page or figure and to the species it was seen in. Pivoting that by body part gives this: every published claim about each piece of the nervous system, side by side, including where the papers disagree.
Three quarters of it is a different animal
Almost everything done in a tardigrade laboratory uses one species, Hypsibius exemplaris: the genome, the transcriptomes, the cultures, the embryos, and the electron-microscopy connectome now being built. The anatomy those results will be read against is mostly not that animal.
This is not a complaint about the literature. It described whichever animal could be collected and sectioned, often decades before anyone chose a model organism. It is a label worth attaching to the anatomy anyway, because some of the disagreements below are probably two people looking at two different animals and both being right.
Where the field disagrees
Every part, and everything said about it
Search by body part, species or the text of a claim. Ours marks a measurement we made from someone else's released scan; everything else is their observation, with the page or figure it came from.
How this is built
- Data
- The project's own knowledge graph: 28 nervous-system sources read in full or in part, every claim carrying a page, figure, table or quote, and an
in_speciesedge naming the animal it was observed in. Our own rows are measurements from the CC BY label fields released by Gross et al. 2019. - Built with
tools/brain/concordance.py, which pivots the graph by structure. The rule that picks out "nervous-system" structures is a keyword match plus thepart_ofchain from a handful of roots, and it prints what it selected so the selection can be checked rather than trusted.- Checked
- Nothing here is a new observation, so there is nothing to test: it is other people's statements, rearranged. What can go wrong is the rearranging — a structure wrongly included, a species tag missed, a claim attached to the wrong part. The selection is auditable and the underlying claims are all in the graph.
- Limits
- Corrected 20 September 2026: this page previously reported 24%, and
docs/FINDINGS.mdreported 28%, by summing each structure's species counter across all structures — which counts a claim once per structure it is attached to. Comparative work on other species touches more structures per claim (2.58 against 2.20), so that understated the H. exemplaris share. Counting each claim once gives 42%. The 38-of-104 figure below is a property of each structure separately and is unaffected. Species shares still depend on what we have read, and we read H. exemplaris papers first; we have now measured that bias rather than just flagging it (tools/brain/reading_order_bias.py): sources read later do carry less H. exemplaris (Spearman −0.36 against reading order, p = 0.03 by permutation), so 42% is more likely an over-estimate than an under-estimate — though the trend is lumpy rather than steady, because reading went in topic batches. A review that reports someone else's disagreement appears as the source on both sides of it. Several structures are named differently by different papers and we have not yet built the table that maps the naming schemes onto each other. - More
- the brain page · the glass animal · the claims · how the project is built