Research dossier · 20 September 2026

Parasite origin of one of the two brains

Testing an idea: that one of the two developmental "brains" reported in 2026 began with an ancient parasite-host relationship.

An open hypothesis, not a finding. Maggie Hall with Claude. Most sources were read as excerpts; the tests listed have not been run.

Working hypothesis

The idea as stated: one of the "two brains" evolved due to a parasite infection. This dossier tests the working version below, which needs confirming or correcting.

Working version (Claude's framing): part of the anterior brain, the Hox-free sensory territory patterned by genes such as Otx and Six3, traces to an ancient parasite-host relationship at the head end. The Hox-patterned trunk and tail sit at the other end of the same axis.

Pieces raised so far:

The two-parallel-progenitors result: forebrain/midbrain and hindbrain arise from separate progenitors.
The tail and the anterior brain as a pair.
The snail eyestalk parasite Leucochloridium.
Snails in early human art.

Open questions:

Which half is the candidate? Anterior is assumed here.
Is the agent an animal parasite, a virus or retrotransposon, or a microbe?
Is the claim about the origin of the territory, or about parasites shaping it?

Status: no published paper proposing this was found (see Gaps and novelty).

What makes it more plausible

Infection-derived genes have built organs and neural machinery before, and the composite-brain idea is already mainstream; none of it shows a parasite building a brain territory.

Infection-built organs are real. Retroviral envelope genes called syncytins were captured independently in primates, rodents, rabbits, carnivores and ruminants. Knockout mice fail to form a placenta without them.
It has happened in nervous systems. Tetrapod Arc and fly dArc1 come from separate retrotransposon lineages, and both form capsids that move RNA between neurons. Arc is absent from fish and the other deuterostomes examined. A retrovirus-derived RNA, RetroMyelin, is reported as essential for vertebrate myelination (preprint). These are single genes, not whole brain territories.
A composite brain is mainstream. Tosches and Arendt argue the forebrain fuses an apical, sensory-neurosecretory system with a blastoporal one. The Stanford result adds a second layer. Nothing says either part is foreign. The apical, sensory side is where a parasite story would sit (Claude's inference).
Trematodes keep targeting eyes. Diplostomid eye flukes each occupy one site in the fish eye: lens, vitreous humour or retina. The lens shows low immune responses. Leucochloridium sits in the same order, Diplostomida, so this is one lineage's habit, not independent evidence.
Parasites as a selective pressure on brains. A 2019 review argues that parasitic manipulation has shaped nervous and endocrine systems. That concerns countermeasures such as access to the brain, not the origin of a brain lineage.

What constrains it

The core patterning genes look native and ancient, the split predates the fossil record of parasites, and known domestications are late and partial.

Timing. The Stanford authors say the anterior/posterior split arose about 550 to 600 million years ago. The oldest convincing parasite-host fossil is early Cambrian, with none convincing from the Ediacaran. Soft-bodied parasites rarely fossilise and a virus never would, so fossils cannot settle it.
Otx and Gbx look native. Anterior Otx and posterior Gbx domains were probably overlapping in the ancestral deuterostome. Otx2 forebrain enhancers are conserved from skate to mouse. Acorn worms already show vertebrate-like signalling centres.
Gbx sits with Hox. In amphioxus, Gbx is on the same chromosome as the Hox cluster. Vertebrate Hox clusters are compact, with little or no interspersed repetitive DNA, which argues against invader insertions inside the posterior patterning cluster.
The body is two-part in acorn worms too. The adult acorn worm forms by adding a Hox-patterned trunk to an anterior larval territory. On developmental order, the Hox-patterned posterior is the added piece, not the anterior.
Chaetognath tail: native duplications. The chaetognath genome shows tandem gene duplication and phylum-specific genes, with no foreign input reported. Its Hox set is the largest reported for protostomes without a whole-genome duplication.
Parasitism simplifies the parasite. Parasitic flatworms have lost many Hox genes. A tapeworm survey found no Post1/2 orthologs. No case was found where a parasite adds posterior patterning to its host.
Late, partial domestications. Arc is tetrapod-specific, far younger than the split.

Predictions and tests

Five genomic checks could support or sink the idea using data that already exists; these are Claude's suggestions and none has been run.

Test	Data	Would support	Would count against
Are conserved non-coding elements at the anterior programme (Otx, Six3, FoxQ2, nk2.1) derived from transposons or viruses, compared with the posterior programme (Gbx, Hox)?	Hemichordate genomes list deeply conserved non-coding sequences as candidate regulatory elements; chromosome-level assemblies exist	Enrichment on one side only	No enrichment on either side
Do any anterior-programme genes have closest relatives outside animals, indicating horizontal transfer?	Acorn worm, amphioxus and chaetognath genomes	A foreign-branching gene with a conserved neural role	All genes root within animals
Is gbx present in the chaetognath genome?	The chaetognath chromosome-level genome; the earlier search used only a transcriptome	Absent in a complete genome, suggesting the posterior programme was added later	Present, so the earlier gap was a sequencing miss
Do hemichordate Hox clusters carry more repeat DNA than vertebrate ones?	Chromosome-level hemichordate assemblies	Repeat-rich clusters, allowing insertions	Compact, repeat-poor clusters as in vertebrates
Did other lineages independently capture retroviral or capsid genes for anterior neural tissue, as Arc and syncytin were captured repeatedly?	Neural transcriptomes across phyla	Convergent captures in anterior tissue	Captures only in unrelated tissues

A null to beat: duplication and co-option of native genes, the pattern the chaetognath genome shows.

Snails and early art

Snail shells are among the earliest human ornaments, but the record shows shell beads, not depictions of snails, and it speaks to human history, not to the origin of the brain.

The idea: snails appear in the earliest art, so a snail parasite is the likely candidate. What the record shows:

Earliest ornaments are snail shells. Beads from Bizmoune Cave, Morocco, are dated to at least 142,000 years ago. Two beads from Skhul are dated to 100,000 to 135,000 years ago. Blombos Cave beads are around 75,000 years old.
No snail depictions confirmed. The oldest engraved marks found are on shells, by Homo erectus around 500,000 years ago, and experts dispute whether those count as art. A depiction of a snail at the start of mark making was not confirmed.
The beads are marine snails. They were carried inland, for example more than 40 km from the sea. They are not the land or freshwater snails that host the parasites discussed here.
Exposure is plausible. Land snails were cooked about 30,000 years ago in Iberia, and in the Maghreb from about 20,000 years ago. Rat lungworm, caught from raw snails, sends larvae to the brain, spinal cord and nerve roots. Humans are dead-end hosts, and no Palaeolithic case was found.
The scale is the problem. Snail use spans at least 142,000 years; the brain split spans about 550 million. This could support a separate human-scale hypothesis, such as snail-borne parasites shaping human perception or symbolism, which would need its own tests.

Gaps and novelty

No paper was found that proposes a parasite or infection origin for either half of the brain, so the idea looks untested rather than refuted.

Searched for and not found:

Any parasite, symbiont or horizontal-transfer origin proposed for the anterior/posterior brain split.
Reports on whether the Otx2, Six3 or Gbx regulatory regions are transposon-derived.
Histology showing whether Leucochloridium contacts the snail's optic nerve or brain, and a Leucochloridium genome.
Snail depictions from the start of mark making.

Conflicts and caveats:

Sources disagree on the snail's eye in Leucochloridium infection. One says the sac replaces and blinds the eyestalk. Older work summarised here says infected snails can keep using their tentacle eyes for at least a year.
The evidence for behaviour manipulation is thin: one field study found infected snails sat higher and more exposed, and its authors could not explain a divergence from an earlier study.
Most sources were read as search excerpts, not full text. Only the chaetognath paper was read at length, and its final discussion section was cut off. The Stanford paper was read through its abstract and press coverage.
Everything under Predictions and tests is a suggestion, not a result.

Reading list

Sources found while researching, grouped by topic; most were read as excerpts.

The two-brains result

Composite brains and axis patterning

Chaetognaths, Hox and the tail

Infection-derived genes

Parasites, eyes and snails

Deep time

Snails and early humans