How the tardigrade eye is wired
A tardigrade has two eyespots, one in each side of its head, and each is about the size of a grain of pollen split a thousand ways — roughly 90 cubic micrometres. For decades nobody could say how they connect to the brain, or even whether they are part of it. In 2026 someone counted the cells and found something odd: there are four, and not one of them has an axon.
This page is about a question being closed. It was posed in print in 2013, sat open for thirteen years, and was answered by a paper that does not cite the people who asked. We noticed because we keep the questions our sources leave open as objects in the graph, and check new reading against them — so when the answer arrived, it matched itself to the question.
What is actually in there
Each eye is an inverse pigment-cup ocellus: a cup of dark pigment with the light-sensitive surface facing into the cup rather than out at the world, so light passes the sensing structure before the pigment stops it. Four cells make it, and they sit inside the outer lobe of the brain rather than on the surface of the head.
- Pigment cell — the cup itself, and the only opaque part of the eye.
- Rhabdomeric cell — the light-sensing surface, facing into the cup.
- Ciliary cell — a second, separate photoreceptor type.
- Support cell — no sensory role described.
- AX-1 to AX-5 — five neurons beside the eye that do have axons, and project forwards. Not part of the eye.
- Lateral PDF neuron — arrives and ends on the rhabdomeric cell. One known input.
One eye, seen from the side, drawn from the published descriptions rather than traced from a figure. Cells 1 to 4 are the whole eye; not one of them carries an axon. Proportions are schematic — the real cup is about 90 µm³ in total. The dashed arrows are drawn dashed on purpose: that is the part nobody has followed.
The finding that makes it strange
A sense organ normally reports to the brain down its own nerve fibre. These four cells have no fibre. Every one of them is a secondary sense cell: it detects, but it cannot itself carry the signal anywhere.
The route out is five separate neurons — AX-1 to AX-5 — which lie immediately beside the eye, on the inner side, within the same brain lobe, and which do have axons. They project forwards. Since no ocular cell bears an axon, these five are the eye's pathway to the rest of the animal.
A thirteen-year-old disagreement, settled sideways
The old question was framed as a choice. Martin et al.'s 2017 review records both sides, each traceable to an earlier author:
We keep both of the older statements in the graph rather than deleting them, with
contradicts edges to the new one. They were reasonable readings of what could be seen at
the time, and the trail of how a question was mis-framed is worth as much as the answer.
One input is known
A separate paper, on a signalling peptide called PDF (pigment-dispersing factor), traced a single neuron that ends on the eye. The lateral PDF-immunoreactive cell of each brain hemisphere splits into two processes; the outer one curves round and terminates on the dorsal surface of the rhabdomeric cell. The rhabdomeric cell carries the PDF receptor, so this is a signal being delivered to the photoreceptor rather than taken from it. In insects PDF is the peptide that carries circadian timing — which makes this a plausible route for a clock to reach the eye, though nobody has shown that in a tardigrade.
Still open, and this is the honest end of the page. Where AX-1 to AX-5 terminate has not been traced. So we know the eye has four cells, that none of them can signal, that five neighbours carry the signal out, and that at least one neuron signals in — and we do not know which part of the brain is listening. Until someone follows those five axons, "how the eye is wired" is answered at one end only.
How this page was built
Every statement above is a node in the project's graph with a source and a locator, and the page
was written from those nodes rather than from memory. The mechanism that caught the answer is worth
describing, because it is reusable: when we read a paper, any sentence in which the authors say
something is unresolved becomes a question node carrying the quote. Later reading is
checked against those questions. This one —
how are H. exemplaris eyes innervated, which brain region receives them, and do they keep an
epidermal connection? — was recorded on 17 September 2026 from Martin et al. 2017, and closed on
20 September when five claims from the two Dutta papers matched it. The question is now marked
answered, with its remaining unknown written into the note.
Sources. Dutta et al. 2026, Extensive opsin gene expansion and non-cerebral origin of the minimalist eye in tardigrades, bioRxiv — four cells and their types (p. 5, Fig. 3c–g, Supplementary Figs 5–8); the five axon-bearing neurons (p. 5, Supplementary Fig. 9); position and volume (p. 5, Fig. 3a,b, Supplementary Fig. 4); the epidermal-vesicle reading (pp. 11–12, Discussion). · Dutta et al. 2025, Pigment-dispersing factor neuropeptides act as multifunctional hormones and modulators in tardigrades, Open Biology 15:240242, CC BY 4.0 — the PDF cell terminating on the rhabdomeric cell (§2.4 p. 7, figs 3a,e and 8b,c, ESM fig. S4). · Martin et al. 2017 — the two older positions and the open question (p. 583). · Mayer et al. 2013, PLoS ONE, CC BY 4.0 — the neuroanatomical context in which the question was posed.
Limits. All of it rests on one reconstruction of one species, and the 2026 paper is a preprint that has not completed peer review. The cell count is from serial STEM-in-SEM of a small number of animals; the volume is from X-ray nanoCT of a whole head. We have not seen the raw data. Where a statement is the authors' interpretation rather than an observation — the epidermal-vesicle reading in particular — it is marked as inferred in our graph and should be read that way here.