Loading today’s stories
Science2 distinct publishers3 min readPublished
The wiring of a male fruit fly's brain and nerve cord is now public, and the payload for anyone planning a bigger map is a claimed thousandfold gain in connectome efficiency whose denominator nobody has stated.
The Scientist · Science desk
Compiled by The ScientistSomething wrong?How this is made
Start with the arithmetic on Janelia's own efficiency claim, because that is the part of this work that travels. The opening estimate was 500 people for ten years [4], call it 5,000 person-years. Gerry Rubin says the campus's key contribution was raising the efficiency of generating connectomes by more than a thousandfold [5]. Divide one by the other and you get five person-years [21], which describes no project anyone ran; eighteen years separate the 2008 start from this month's publication [20]. So the multiple is not a statement about total labor. It most plausibly refers to throughput in imaging and image interpretation, which is exactly where Harald Hess's microscopy scaling and the computational work with Google Research collaborators were aimed [6]. The source does not define the denominator, and the denominator is the whole question for anyone costing a bigger brain.
The other number to slow down on is the cross-sex comparison. The male central nervous system totals more than 166,000 neurons [1]; the female brain map released in 2024 covers about 140,000 [9]. The 26,000-neuron gap [22] is mostly scope, since the new map includes the ventral nerve cord, the fly's counterpart to a spinal cord, and the 2024 map covers the brain [1][9]. Read that gap as a difference between the sexes and you are reading an anatomical boundary instead.
What the sex-comparison paper found is more useful than a tally. Alongside a network of cells specific to the male brain, sensory and motor circuitry turns out to be largely shared, with particular switches inside those shared circuits sending signals to different destinations in males and females [17]. That locates part of the difference in routing rather than in separate hardware, and it is testable against behavior: males lunge where females headbutt [25].
The vision paper shows the limits of the format. Visual processing reaches more than half of the roughly 11,000 neuron types identified in the map [15], which tells you how widely visual signals are distributed and nothing about what any of those cells compute. The taste study likewise traced receptors on the legs, wings, mouthparts and throat back into the brain and examined where those paths meet the circuits for swallowing and walking [19]. The map narrows which neurons to perturb. The perturbation is still the experiment.
The 2008 skeptics were not being unreasonable: the previous complete connectome had 302 neurons and had taken more than a decade to finish [3][2]. The scaling since is real, from the hemibrain's 25,000 neurons in 2020 to 166,000 now, a factor of about 6.6 [24], with the fly running roughly 550 times the worm's neuron count [23]. Next targets are larval zebrafish and adult Danionella [14], toward what Janelia describes as the first mechanistic account of how a vertebrate brain generates behavior [8]. Whether the cost curve survives vertebrate tissue is untested, and Carlos Ribeiro's further suggestion that fly architecture could inform more efficient artificial systems [12] is a hope, not a result in these papers [11].
Ranked by verification strength, evidence, and original report placement.
Janelia researchers published in Cell the connectome of the full central nervous system of a fruit fly: a wiring diagram of more than 166,000 neurons making up the fly's brain and ventral nerve cord, and the millions of connections between them.
Janelia is applying lessons from the fly connectome to new animal models with the ultimate goal of providing the first mechanistic account of how a vertebrate brain generates behavior.
The new map charts the male fruit fly's brain and the insect's equivalent of a spinal cord, and joins a map of a female fruit fly brain unveiled in 2024 that covers about 140,000 neurons.
Rubin said it is the first time both sexes of an animal with complex social behavior can be compared, and that this allows researchers to home in easily on the neurons causing behavioral differences.
The new fly brain map, initially released as a preprint, was published in the journals Cell and Current Biology on Thursday, Sept. 3, alongside three other papers each using the new data to explore a specific aspect of fruit fly neurobiology.
In the near term the scientists aim to map the brains of larval zebrafish (Danio rerio) and adult danionin fish (Danionella).
Distinct publishers with included, body-backed reporting in this cluster.
livescience.com
1 article · September 3, 2026
phys.org
1 article · September 3, 2026
Follow any of these and your For You feed starts watching them — no settings page required.
science
Salk's fly driver lines narrow octopamine access to single pairs of neurons1 distinct publisher
science
Cats leave a 13-compound calling card in urine, and the kidney may be the warehouse3 distinct publishers
invest
Google says frontier models already know the facts they get wrong. That is a budget decision.1 distinct publisher
science
Tagged blue whales pinpoint their feeding along the ocean's narrow thermal fronts1 distinct publisher
Evidence-backed comparisons of source perspectives and observed adoption signals. Read the methodology
Which Builder, Operator, and Investor concerns the observed source mix emphasized—not a truth score.
Evidence, demonstrated adoption, hype gap, incentives, and confidence are assessed independently, each on its own current evidence. How these are measured.
Solid on the map, thin on the multiplier
The physical result is about as well anchored as a same-day story gets: peer-reviewed papers in Cell and Current Biology, the same 166,000-neuron total in both accounts, and companion findings specific enough to be checked against the data. The framing device of the story is weaker. Janelia's headline achievement — better than a thousandfold gain in connectome efficiency — appears once, as a quote, with no baseline, no unit and no independent measurement, and Live Science does not carry it at all.
Public resource, producer-reported uptake
Real usage exists and is dated: three papers published the same day already run on this data, and the map itself is released to the community rather than announced. The longer adoption record — hundreds of findings from the 2020 hemibrain, teams worldwide now attempting bigger brains — comes from Janelia describing its own resource, and nobody in this reporting counts it independently. Downstream vertebrate work is stated as intent, not deployment.
The multiplier outruns the map
The core result is if anything under-sold: 166,000 neurons and their connections, done and published. The overstatement sits in the two things layered on top. A thousandfold efficiency gain with no denominator cannot be assessed by a reader, and the leap from a poppy-seed brain to a technical roadmap for mouse and human connectomes — offered by a collaborator in a press statement — skips several orders of magnitude that the fly project itself took eighteen years to cross at its own scale. Note also that the comparison being sold as the payoff, male versus female, is presented with mismatched neuron counts.
Two stories, one press moment
Phys.org is running HHMI's institutional account of its own eighteen-year wager, complete with the founding director who placed it explaining what the key contribution was; Live Science's quotes come from Janelia's statement and a separate Champalimaud statement. Rubin appears in both. Nobody who doubted the project in 2008 is asked whether they were wrong, and no third party is asked whether the multiplier is real — which matters most for exactly the claim that has no numbers behind it.
Facts firm, framing unaudited
Two independent outlets, same-day peer-reviewed publication and consistent headline figures make the factual spine reliable. What holds confidence down is narrow provenance on the interpretive layer — a single institutional source for the efficiency and legacy claims, no external assessment of either, and no visibility into cost per neuron then versus now.