How much of the brain have we mapped? 21 connectomes, 1986 to 2026

01 Key findings

1. Whole nervous systems are mapped only for small animals. The roundworm (302 neurons), sea squirt larva (177), fruit fly larva (3,013) and adult fruit fly (166,700 in the male CNS) are complete.White 1986; Ryan 2016; Winding 2023; Berg 2026
2. Mammals are mapped in blocks of one cubic millimetre. The mouse block holds 82,247 detected neurons (0.12% of a mouse brain), the human block about 16,000 (0.00002% of a human brain).MICrONS 2025; Shapson-Coe 2024; IQ Metrics calculation
3. The record has grown about 4.8-fold per decade since 1986, and about 63-fold since 2013. Continuing that line would reach a mouse brain around 2040 to 2070; it is arithmetic, not a forecast.IQ Metrics calculation
4. The Wellcome Trust puts a whole mouse connectome at $7.5-21 billion and a decade off, with about 100,000 person-years of proofreading at today’s tools, against 33 for the fly brain.Wellcome Trust 2023; Dorkenwald 2024
5. Data, not microscopes, is the wall. The 1 mm3 human block is 1.4 petabytes; a whole human brain imaged the same way would be about 2.1 zettabytes.Shapson-Coe 2024; IQ Metrics calculation
6. Teams grew with the maps. The 1986 roundworm paper had 4 authors; FlyWire has 293 and the mouse block 96. Neurons, synapses and authors correlate at r = 0.84 to 0.96.OpenAlex; IQ Metrics calculation

02 Statistics to quote

03 How much of the brain have we mapped?

Scientists have mapped the whole nervous system of animals up to the fruit fly, 166,700 neurons, but only about 0.12% of a mouse brain and 0.00002% of a human brain ^{[12] [21] [20]}. A connectome is a wiring diagram: every neuron and every synapse between neurons, read from slices of tissue imaged with an electron microscope. We collected every published reconstruction of that kind we could read, 21 of them from 20 papers between 1986 and 2026, and wrote down for each one the number of neurons, the number of synapses, the tissue it covers, the microscope, the team and the data size.

The picture has two halves. Small animals are finished: the roundworm Caenorhabditis elegans has 302 neurons in the hermaphrodite and 385 in the male ^[2], a sea squirt larva 177 neurons in its central nervous system ^[4], and the fruit fly Drosophila melanogaster, with 139,255 neurons in the female brain ^[9] and 166,700 in the male brain and nerve cord together ^[12]. Mammals are not: the largest mouse and human maps each cover one cubic millimetre of cortex, about 1 in 1,500,000 of a human brain by volume ^[24].

Figure 1

From 302 neurons in 1986 to 166,700 in 2026: every connectome with a neuron count

Each dot is one reconstruction in our ledger that states how many neurons (or cells) it holds, by publication year; neuron counts on a log scale, each step ten times more. The highest dot is the male fruit fly central nervous system (166,700 neurons, 2026). The zebrafish whole-brain dot (176,810 cells) is a preprint and counts cells, not proofread neurons. As of 5 november 2026.

Roundworm (C. elegans)Sea squirt and marine worm larvaeFruit flyZebrafish larvaMouseHuman

Neurons (or cells) in the reconstruction (log scale)

1001k10k100k

C. elegans 1986Mouse retinaSea squirt larvaC. elegans hermaphrodite 2019C. elegans male 2019Fly hemibrainZebrafish 2022Fly larva brainHuman cortex (H01)Mouse cortex (MICrONS)Fly optic lobeZebrafish whole brain

19851995200520152025

Publication year

The shape of the chart is a staircase with long flat treads. The roundworm map of 1986 ^[1] held the record for 27 years; the next record-setter in our ledger is a piece of mouse retina with 950 neurons in 2013 ^[14], then the fly hemibrain with about 25,000 in 2020 ^[7], FlyWire with 139,255 in 2024 and the male fly with 166,700 in 2026. Each step needed a new way of working, not just more of the old one: automated segmentation of the images, then crowds and professional teams to proofread it ^[9].

Figure 2

Forty years of wiring diagrams: the fruit fly took the lead in 2020

Every reconstruction in the ledger by publication year and animal. The roundworm map of 1986 stood alone for 27 years until the mouse retina (2013); two thirds of the entries (14 of 21) have appeared since 2020. As of 5 november 2026.

Peer-reviewed paperPreprint

Roundworm4 reconstructionsC. elegans 1986: 302 neuronsC. elegans, 8 animals: 204 cells

Larvae of other animals2 reconstructionsSea squirt larva: 177 neurons

Fruit fly7 reconstructionsFly hemibrain: 25,000 neurons

Zebrafish larva3 reconstructionsZebrafish 2022: 208 neurons

Mouse4 reconstructionsMouse retina: 950 neurons

Human1 reconstructionHuman cortex (H01): 16,000 neurons

19851995200520152025

Connectome

The complete map of connections in a nervous system or a piece of one: which neuron talks to which, and through how many synapses. The term was proposed in analogy with the genome; the first, for a roundworm, appeared in 1986 ^[1].

Neuron

A nerve cell. Counts in this study are neurons that the authors reconstructed, or “cells” where the paper counts non-neuronal cells and graph nodes as well (marked in the table).

Synapse

A junction where one neuron signals to another. Chemical synapses release a transmitter; gap junctions pass current directly. Papers count them differently (see Limits), so only the order of magnitude compares.

Electron microscopy (EM)

Imaging with electrons, fine enough to see single synapses (voxels of 4 to 8 nm in the large maps). The tissue is cut into slices or milled away layer by layer and every layer is imaged.

FIB-SEM

Focused-ion-beam scanning electron microscopy: an ion beam shaves off a thin layer, an electron beam images the new surface, and the cycle repeats. Used for the hemibrain, the nerve cord, the optic lobe and the male fly ^{[7] [12]}.

Proofreading

Human correction of the computer’s tracing of neurons, where branches were wrongly split or joined. It is the costliest step: 33 person-years for the FlyWire brain ^[9].

Voxel and teravoxel

A voxel is a three-dimensional pixel; a teravoxel is a trillion of them. The male fly dataset has 160 teravoxels at 8 nm ^[12].

Petabyte and exabyte

One petabyte is 1,000 terabytes; one exabyte is 1,000 petabytes. A 1 mm3 block of human cortex is 1.4 petabytes ^[20].

Person-year

One person working full time for a year. Used by the authors to state proofreading effort ^{[9] [12]}.

04 Which animals have a complete connectome?

Five kinds of animal have a connectome of the whole nervous system or whole brain: the roundworm (302 and 385 neurons), a sea squirt larva (177), the fruit fly larva (3,013), the adult fruit fly (166,700 in the male) and, as a body-wide graph, a marine worm larva (2,675 cells in the graph) ^{[2] [4] [6] [12] [5]}. The roundworm map took, in the Wellcome Trust’s words, “a decade” of work by John White and colleagues at the Medical Research Council in Cambridge ^{[24] [1]}. Automation changed the effort: the hemibrain of 2020 needed 50 person-years of proofreading for about 25,000 neurons ^[9].

Seven of the 21 entries are insects. The fly larva of Winding and colleagues has 3,013 neurons and 544,000 synapses in a brain small enough to be read as a whole ^[6]; the adult male central nervous system has 55 times as many neurons. Between them sit the partial adult maps: the hemibrain (about 25,000 neurons, about 20 million chemical synapses ^[7]), the male nerve cord (about 23,000 neurons ^[8]) and the right optic lobe (about 53,000 ^[11]). The FlyWire brain (139,255 neurons, 54.5 million synapses) was proofread by a community of more than 50 laboratories ^[9].

The newest entry is the largest. The male fruit fly connectome, published in Cell in September 2026 by a team of 111 authors led from the Janelia Research Campus, holds 166,700 neurons of the brain and the nerve cord together, grouped into 11,710 types; compared with the female brain it finds 8,069 types shared by both sexes, 138 that differ, 289 found only in males and 71 only in females ^[27]. It took 13 months of imaging on seven FIB-SEM machines to produce 160 teravoxels and 44 person-years of proofreading ^[12]. The female FlyWire brain was proofread on an older volume, the 2018 FAFB images of 7,062 slices of about 40 nanometres ^[10].

Figure 3

What each map holds: neurons, synapses, authors, tissue volume and proofreading effort

Powers of ten (log10) of five facts per reconstruction, oldest first; a blank means the paper does not state it. Reading a cell: 5.2 under Neurons is 166,700 neurons; 0.0 under Volume is 1 mm3. Only 3 of 21 entries state a proofreading effort. Synapse counts follow each paper's own definition, so they are not strictly comparable.

FewerMore

Neur.NSyn.SAuth.AVol.VPYPYC. elegans 19862.53.70.6Mouse retina3.00.8Mouse cortex 20151.3-2.8Sea squirt larva2.23.80.5C. elegans hermaphrodite 20192.51.1C. elegans male 20192.61.1Mouse layer 41.0-3.3Fly hemibrain4.47.32.01.7C. elegans, 8 animals2.33.91.2Zebrafish 20222.33.01.0Fly larva brain3.55.71.3Fly brain (FlyWire)5.17.72.51.5Human cortex (H01)4.28.11.50.0Fly nerve cord4.41.9Zebrafish brainstem3.51.5-2.6Fly brain + cord (BANC)2.1Mouse cortex (MICrONS)4.98.72.00.0Fly optic lobe4.71.9Marine worm larva3.44.40.8Zebrafish whole brain5.27.41.6Fly male CNS5.22.01.6

The heatmap shows where the information is thin. Of the 21 entries, 15 state a count of neurons, 3 more a count of cells, 11 a synapse count and 3 the proofreading effort. The human and mouse blocks are the only entries that cover a full cubic millimetre ^{[20] [21]}; the cortex pieces of 2015 and 2019 cover 0.0015 and 0.0005 mm3 and state no neuron count ^{[15] [16]}.

Figure 4

More neurons, more synapses, more authors: three things that grow together

11 reconstructions that state both counts, by neurons (across) and synapses (up), both on log scales; bubble area = number of authors on the paper. The two cortex blocks (human H01 and mouse MICrONS) sit above the line of whole animals because their synapses belong to thousands of neurons cut by the edge of the block.

Roundworm (C. elegans)Sea squirt and marine worm larvaeFruit flyZebrafish larvaMouseHuman

Authors on the paper530100290

Synapses as the paper counts them (log scale)

1k100k10M1B

Fly brain (FlyWire)Fly hemibrainMouse cortex (MICrONS)Human cortex (H01)Fly larva brainC. elegans, 8 animalsZebrafish 2022Marine worm larvaC. elegans 1986

1001k10k100k

Neurons or cells (log scale)

Synapse counts rise with neuron counts, but not at a constant rate. A roundworm neuron has about 17 chemical synapses on our arithmetic (5,000 over 302), a sea squirt larva neuron 37, a fly larva neuron 181, an adult fly neuron about 390 (FlyWire) to 800 (hemibrain). More neurons also means more connections per neuron. The two cortex blocks have far more synapses than neurons (6,400 per detected neuron in the mouse block, 8,400 in the human block) because the block cuts thousands of neurons at its edge; their ratios are not comparable to those of whole animals.

Figure 5

Most connectomes hold hundreds or thousands of neurons; only 8 of 18 reach 10,000

18 reconstructions with a neuron or cell count, in half-order-of-magnitude bins of log10(neurons): 2.5 is 316 neurons, 4.0 is 10,000, 5.0 is 100,000. The roundworm, sea squirt, larval fly and zebrafish maps fill the left; the adult fly and the two cortex blocks fill the right.

52332310,000 neurons

2.02.53.03.54.04.55.05.5

log10 of neurons (or cells) in the reconstruction

05 Has the human brain been mapped?

No: the largest human map covers about 16,000 of the 86.1 billion neurons of a human brain, one in 5,380,000, in a single cubic millimetre of cortex ^{[20] [22]}. The block, called H01, came from a piece of temporal lobe surgically removed and was cut into more than 5,000 slices of about 34 nanometres; it holds 57,216 cells, of which glia outnumber neurons two to one, 133.7 million synaptic connections, and 1.4 petabytes of images. Only 104 cells were proofread by hand ^[20].

Counting by volume gives the same answer. A human brain is, by the Wellcome Trust’s arithmetic, 3,000 times a mouse brain, which is 500 times the 1 mm3 volumes imaged today: 1.5 million blocks like H01 ^[24]. The adult human brain contains 86.1 +/- 8.1 billion neurons in the count of Azevedo and colleagues, and the mouse brain 67.9 million ^{[22] [23]}. Our own count of how many fruit fly connectomes make a human brain, by neurons alone, is 618,000 FlyWire brains.

Figure 6

A human brain is 5.4 million times the neurons of the biggest human map; a mouse brain 825 times

For each brain, the amount mapped (left) and the whole brain (right) as powers of ten: neurons (16,000 in the human block H01 against 86.1 billion; 82,247 detected in the mouse MICrONS block against 67.9 million) and tissue volume (1 mm3 against 1,500,000 mm3 for a human and 500 for a mouse, the ratios Wellcome Trust prints). Each unit on the axis is a factor of ten; the gap is 6.7 orders of magnitude for human neurons.

Mapped in the largest blockWhole brain

Human: neurons4.210.9

Mouse: neurons4.97.8

Human: tissue volume, mm30.06.2

Mouse: tissue volume, mm30.02.7

0246810

log10 of the count (neurons) or of the volume (mm3)

The gap between 1 mm3 and a whole brain is wider than the gap in neurons because a neuron’s wiring is spread through the brain. The authors of H01 and MICrONS both reconstruct neurons that leave the block, so a block-level connectome is a sample of local circuits, not a whole. For other counts of the brain that people quote, see our brain statistics fact check, how much data the human brain can store and the weights of famous brains.

06 Has a mouse brain been mapped?

No: the largest mouse connectome, MICrONS, covers a block of about one cubic millimetre of visual cortex, with 82,247 neurons detected in its larger sub-volume (0.12% of the mouse brain’s 67.9 million) and 524 million synaptic clefts ^{[21] [23]}. The MICrONS Consortium imaged 2 petabytes of raw data over about 6 months and co-registered it with calcium imaging of around 75,000 neurons in the living animal ^[21]. Only a subset of the cells is proofread.

Earlier mouse maps were far smaller. Helmstaedter and colleagues reconstructed 950 neurons of the retina in 2013 ^[14]; Kasthuri and colleagues a cortical fragment of about 1,500 cubic micrometres in 2015 ^[15]; Motta and colleagues a block of about 500,000 cubic micrometres in 2019 ^[16]. MICrONS is about 2,000 times the volume of Motta’s block (our arithmetic). For a different view of the mouse as an animal, see our comparison of animal brains.

07 Which other animals have been mapped, and how much?

The larval zebrafish is the vertebrate closest to a whole-brain map, with about 176,810 cells and 25.25 million synapses found algorithmically in a 6-day-old larva in a 2025 preprint that has not been peer reviewed ^[19]. The peer-reviewed zebrafish maps are partial: 208 reconstructed neurons with 1,079 synaptic contacts from a whole imaged brain ^[17], and about 3,000 neurons of a brainstem block of 250 x 120 x 80 micrometres ^[18]. We found no synapse-level connectome of a bird, a reptile, a monkey or any other mammal beyond mouse and human; absence in our search is a lower bound, not proof.

Figure 7

From 29 papers screened to 17 with a neuron count

How the ledger was built: leads from our plan and from searches, the papers whose primary text or abstract we could read, those that report a synapse-level wiring reconstruction with its own dataset, and those that state how many neurons or cells it holds. The 21 rows come from 20 papers (the 2019 roundworm paper gives two).

Papers screened (leads from our plan and searches)29all

One paper's full text was not available to us

Text or abstract read in the primary source2896.6% of all

8 papers were reviews of existing data, imaging volumes without a wiring diagram, or analyses of a dataset already counted

A synapse-level wiring reconstruction with its own dataset2069.0% of all

3 papers state a count of synapses or edges but not of neurons or cells

… that states a neuron or cell count1758.6% of all

08 How fast is connectome mapping growing?

The largest connectome has grown about 4.8-fold per decade over 1986-2026 and about 63-fold per decade since 2013 (a log-linear fit through the record-setting maps, IQ Metrics calculation). The first figure uses all 5 record-setters (302 in 1986 to 166,700 in 2026; R-squared 0.77); the second only the 4 since 2013 (950 to 166,700), a sharper but shorter line. Neither is a law: a record set by one laboratory depends on a decision to spend years on a single animal.

Figure 8

20 of 21 reconstructions are from 2013 or later, and the fruit fly leads since 2020

Reconstructions in the ledger per publication year, by animal. Years between 1986 and 2013 are not drawn: the ledger holds none. 2025 is the busiest year (5), with the first whole-brain zebrafish preprint, the fly brain-and-cord preprint and the mouse cortex block. As of 5 november 2026.

Roundworm (C. elegans)Sea squirt and marine worm larvaeFruit flyZebrafish larvaMouseHuman

11986

12013

2014

12015

12016

2017

2018

32019

12020

12021

12022

12023

42024

52025

12026

Counting every entry rather than records tells the same story. Only 1 of the 21 entries appeared before 2013, 6 between 2013 and 2019 and 14 from 2020 on. Teams grew too: the median entry had 4 authors before 2013, 14 in 2013-2019 and 71 from 2020 ^[25]. Neurons, synapses and authors are strongly correlated across the maps that report all three (r = 0.84 to 0.96), because the big maps are team projects.

Figure 9

Neurons and synapses move together (r = 0.96); time alone explains less (r = 0.52)

Pearson correlations across the 8 reconstructions that state neurons, synapses and authors, using log10 of each count. Linked, not cause: bigger maps have more of everything, and newer maps are bigger. Eight points are a small sample.

-1 (opposite)+1 (move together)

YearYNeuronsNSynapsesSNeurons (log10)+0.52Synapses (log10)+0.53+0.96Authors (log10)+0.59+0.95+0.84

Speed of proofreading matters as much as speed of imaging. The male fly CNS was imaged in 13 months on seven FIB-SEM machines (160 teravoxels) and then proofread over 3 years by 29 expert proofreaders, 44 person-years ^[12]; the hemibrain needed 50 person-years for about 25,000 neurons, FlyWire 33 for 139,255 ^[9]. In neurons per person-year, that is 500 for the hemibrain, 4,220 for FlyWire and 3,789 for the male CNS (IQ Metrics calculation).

09 When will a whole mouse brain or the human brain be mapped?

The Wellcome Trust’s 2023 report calls a whole mouse connectome “probably still a decade off”, plausible within 10 to 15 years, and expects human brains to “remain off limits to connectomics for several decades” ^[24]. If the record-setting line since 2013 simply continued, the neuron count would reach a mouse brain around 2040; the line since 1986 puts it around 2070, and a human brain around 2057 to 2116 (IQ Metrics arithmetic, not a forecast).

Figure 10

A mouse connectome: 2033 by the Wellcome Trust, 2040 or 2070 if past growth simply continued

Years in which the neuron count of the largest connectome would reach a mouse brain (67.9 million) or a human brain (86.1 billion). Wellcome Trust (2023): "probably still a decade off", in the next 10 to 15 years. Ours is arithmetic, not a forecast: a straight line through the 5 record-setting maps since 1986 (x4.8 a decade) and through the 4 since 2013 (x63 a decade). Wellcome expects human brains to "remain off limits to connectomics for several decades".

Mouse brain: Wellcome Trust, 10-15 years from 20232033 (2033–2038)

Mouse brain: our trend, record-setters since 20132040 (2040–2040)

Mouse brain: our trend, record-setters since 19862070 (2070–2070)

Human brain: our trend, record-setters since 20132057 (2057–2057)

Human brain: our trend, record-setters since 19862116 (2116–2116)

202020702120

The cost is the reason for the spread. The Wellcome report estimates $7.5 to $21 billion for one mouse brain, calls manual proofreading the most resource-intensive step and puts it at about 100,000 person-years with today’s tools; it also notes that automation cut the whole fly brain from about 2,000 person-years to about 40 ^[24]. At FlyWire’s rate of 4,220 neurons per person-year a mouse brain would take 16,084 person-years, one sixth of the Wellcome figure (IQ Metrics arithmetic).

Figure 11

Proofreading a mouse brain: 100,000 person-years by Wellcome, 16,000 at the rate FlyWire reached

Proofreading effort (across, log scale) against neurons (up, log scale). Blue: three fly maps, whose papers state the effort (33, 44 and 50 person-years). Grey: the Wellcome Trust estimate for a whole mouse brain. Orange: our arithmetic, a mouse brain at FlyWire's 4,220 neurons per person-year. The fly rate has risen 8-fold from the hemibrain (500) to FlyWire.

Reported by the fly papersWellcome Trust estimateIQ Metrics arithmetic

Neurons (log scale)

10k1M100M

Fly hemibrainFly brain (FlyWire)Fly male CNSMouse brain: WellcomeMouse brain at FlyWire rate

303003k30k300k

Proofreading effort, person-years (log scale)

Storage is the second wall. The Wellcome report puts raw images of a whole mouse brain at 0.5 exabytes at one byte per voxel ^[24]. Scaling the two block datasets by volume gives 700 to 1,000 petabytes for a mouse (H01 at 1,400 TB and MICrONS at 2,000 TB per mm3) and 2.1 to 3.0 zettabytes for a human brain, (our arithmetic, labelled; compression would reduce it). Compare estimates of what a brain stores.

10 Who builds connectomes, and with which microscopes?

Most connectomes come from teams in the United States, Germany, the United Kingdom and Canada, and four of the seven fly papers come from one campus, the Howard Hughes Medical Institute’s Janelia Research Campus ^{[7] [8] [11] [12]}. The ledger classifies each paper by the country of its last author’s first-listed institution in OpenAlex ^[25]; 11 of 20 papers have a United States senior author. China appears once, with a zebrafish preprint ^[19].

Figure 12

11 of 20 connectome papers have a United States senior author; two imaging families carry the fly maps

Papers in the ledger by country of the last author's first listed institution (OpenAlex), by animal and by imaging method. Width = number of papers (20). The fruit fly papers split between serial-section transmission EM (FlyWire) and focused-ion-beam scanning EM (hemibrain, nerve cord, optic lobe, male CNS); one fly preprint does not state its method in the text we read.

Country of senior authorCountryAnimalAnimalImaging methodMethod

United StatesUnited States 11GermanyGermany 5CanadaCanada 2United KingdomUnited Kingdom 1ChinaChina 1

RoundwormRoundworm 3Sea squirt and marine worm larvaeSea squirt and marine worm larvae 2Fruit flyFruit fly 7Zebrafish larvaZebrafish larva 3MouseMouse 4HumanHuman 1

FIB-SEMFIB-SEM 4Multibeam SEMMultibeam SEM 1Not statedNot stated 1Serial block-face SEMSerial block-face SEM 3Serial-section EMSerial-section EM 8Serial-section TEMSerial-section TEM 3

Two imaging families dominate. Serial-section electron microscopy cuts tissue into thousands of slices first: the roundworm, sea squirt, marine worm and FlyWire maps ^{[4] [5] [9]}. FIB-SEM mills the sample away layer by layer, giving cubes of 8 nanometres on every side: the hemibrain, nerve cord, optic lobe and male fly ^{[7] [12]}. Serial block-face SEM, used for the mouse retina and cortex, and multibeam SEM for H01, are the other two ^{[14] [20]}.

Figure 13

Mouse maps grew from 6 authors in 2013 to 96 in 2025; the roundworm map from 4 to 16

Authors on the first and the latest paper in the ledger for each animal (OpenAlex author counts). Roundworm: White 1986 and Witvliet 2021; mouse: Helmstaedter 2013 and MICrONS 2025; zebrafish: Svara 2022 and the 2025 Fish-X preprint; sea squirt / marine worm: Ryan 2016 and Verasztó 2025; fruit fly: hemibrain 2020 (108) and the male CNS 2026 (111). Human has one paper (31).

RoundwormMouseOther animals

First mapFirstLatest mapLatest

Roundworm 4Sea squirt / marine worm 3Fruit fly 108Zebrafish 11Mouse 6

16 Roundworm7 Sea squirt / marine worm111 Fruit fly39 Zebrafish96 Mouse

Figure 14

FlyWire leads on neurons and authors, the mouse block on synapses

Rank (1 = most) of the 8 reconstructions that state neurons, synapses and authors, on each of the three counts. FlyWire ranks 1, 3 and 1; the mouse cortex block (MICrONS) 2, 1 and 3. Ranks of counts, not of quality.

Mouse cortex (MICrONS)Fly brain (FlyWire)Other maps

NeuronsNeur.SynapsesSyn.AuthorsAuth.

Fly brain (FlyWire) 1Mouse cortex (MICrONS) 2Fly hemibrain 3Human cortex (H01) 4Fly larva brain 5C. elegans 1986 6Zebrafish 2022 7Sea squirt larva 8

1 Fly brain (FlyWire)3 Mouse cortex (MICrONS)2 Fly hemibrain4 Human cortex (H01)5 Fly larva brain7 C. elegans 19866 Zebrafish 20228 Sea squirt larva

Figure 15

Fly teams are the largest: a median of 108 authors against 14 for roundworm papers

Authors per paper by animal, one dot per paper (the 2019 roundworm paper counts once). Box = middle half, line = median, whiskers = fewest and most authors. Worm: 4 to 16; fly: 20 to 293 (FlyWire, 293 authors including its consortium). Groups with fewer than 3 papers show the range only.

Middle halfMedianWhiskersauthors

Roundworm3 papers14

Larvae2 papers5

Fruit fly7 papers108

Zebrafish3 papers31

Mouse4 papers16

Human1 papers31

050100150200250300

Authors per paper (OpenAlex).

Authorship counts also reveal how maps are shared. FlyWire lists 293 authors because it was built by a consortium in which researchers from more than 50 laboratories took part ^[9]; the 1986 roundworm paper had 4. The maps are open: the fly, mouse and human datasets are public for anyone to download, and we link each paper. For the memory side of the brain see our memory and IQ guide and memory world records study.

11 Does a bigger connectome mean a smarter animal?

No: a connectome counts neurons and synapses, which are not intelligence, and nothing on this page measures the intelligence of any species. The fruit fly has the largest complete connectome of any animal and no animal is smarter for it. Brain size has a weak link with IQ within people (see brain size and IQ) and no use as a ranking between species. What a connectome gives is a parts list and a wiring diagram, a base for models of how circuits compute, which scientists then test with recordings ^{[6] [21]}. To see which part of the brain intelligence tests tap, read which part of the brain IQ tests measure; for a famous brain that was weighed and cut, Einstein’s brain.

Connectome science also feeds other questions we cover. Neurons grown in a dish, organoids, are now wired to computers; our organoid intelligence study counts the experiments. For brain implants, see how many people have a brain implant. And for how fast a brain handles information, how fast do humans think.

12 Every connectome in our dataset

Table 1

Every reconstruction in our ledger (21 rows)

Oldest first. Neurons are reconstructed neurons unless the cell says "cells". Synapses follow each paper's own definition (see the CSV column synapse_basis); blank = not stated in the text we read.

Reconstruction	Year	Scope	Neurons	Synapses	Method	Authors	Source
C. elegans hermaphrodite (White et al.)	1986	whole nervous system	302	5,000	Serial-section EM	4	White 1986
Mouse retina, inner plexiform layer (Helmstaedter et al.)	2013	part of the retina	950	–	Serial block-face SEM	6	Helmstaedter 2013
Mouse neocortex fragment (Kasthuri et al.)	2015	sub-volume of neocortex	–	–	Serial-section EM	21	Kasthuri 2015
Ciona larva CNS (Ryan et al.)	2016	whole central nervous system	177	6,618	Serial-section EM	3	Ryan 2016
C. elegans hermaphrodite (Cook et al.)	2019	whole animal, incl. muscles	302	–	Serial-section EM	14	Cook 2019
C. elegans male (Cook et al.)	2019	whole animal, incl. muscles	385	–	Serial-section EM	14	Cook 2019
Mouse barrel cortex, layer 4 (Motta et al.)	2019	block of layer 4, somatosensory cortex	–	–	Serial block-face SEM	10	Motta 2019
Fruit fly hemibrain (Scheffer et al.)	2020	about half of the central brain	25,000	20,000,000	FIB-SEM	108	Scheffer 2020
C. elegans, 8 animals birth to adult (Witvliet et al.)	2021	brain (nerve ring and ventral ganglion)	204 cells	8,000	Serial-section EM	16	Witvliet 2021
Larval zebrafish brain, 208 neurons (Svara et al.)	2022	whole brain imaged; 208 neurons reconstructed	208	1,079	Serial block-face SEM	11	Svara 2022
Drosophila larva brain (Winding et al.)	2023	whole brain	3,013	544,000	Serial-section EM	20	Winding 2023
Fruit fly whole brain, FlyWire (Dorkenwald et al.)	2024	whole brain	139,255	54,500,000	Serial-section TEM	293	Dorkenwald 2024
Human cortex fragment, H01 (Shapson-Coe et al.)	2024	1 mm3 of temporal-lobe cortex	16,000	133,700,000	Multibeam SEM	31	Shapson-Coe 2024
Fruit fly male nerve cord (Takemura et al.)	2024	ventral nerve cord	23,000	–	FIB-SEM	84	Takemura 2024
Larval zebrafish brainstem (Vishwanathan et al.)	2024	brainstem block (250 x 120 x 80 micrometres)	3,000	–	Serial-section EM	31	Vishwanathan 2024
Fruit fly brain + nerve cord, BANC (Bates et al.)	2025	brain and ventral nerve cord	–	–	–	116	Bates 2025 (preprint)
Mouse visual cortex, MICrONS (MICrONS Consortium)	2025	about 1 mm3 of visual cortex	82,247	524,000,000	Serial-section TEM	96	MICrONS 2025
Fruit fly optic lobe (Nern et al.)	2025	right optic lobe	53,000	–	FIB-SEM	71	Nern 2025
Platynereis larva, whole body (Veraszto et al.)	2025	whole body	2,675 cells	26,881	Serial-section TEM	7	Verasztó 2025
Larval zebrafish whole brain, Fish-X (Xie et al.)	2025	whole brain, retina and anterior spinal cord	176,810 cells	25,250,000	Serial-section EM	39	Li 2025 (preprint)
Fruit fly male CNS (Berg et al.)	2026	whole central nervous system (brain + nerve cord)	166,700	–	FIB-SEM	111	Berg 2026

Table 2

Papers we read and left out (9) and why

A paper is in the ledger only if it reports a synapse-level wiring reconstruction with its own dataset and counts we could read. Leaving a paper out is not a judgement of its value.

Paper	Reason
Zheng et al. 2018, FAFB electron microscopy volume of a whole fly brain	imaging volume only; reconstruction came later (FlyWire)
Hildebrand et al. 2017, whole larval zebrafish brain	imaging volume and myelinated-axon survey, no synapse-level connectome
Eichler et al. 2017, larval mushroom body	one circuit; the text gives no total neuron count
Phelps et al. 2021, FANC nerve cord	full text not available to us
Marin et al. 2024, MANC annotation	analysis of the same dataset as Takemura et al.
Cheong et al. 2024, MANC premotor circuits	analysis of the same dataset as Takemura et al.
Emmons 2024, C. elegans connectome analysis	new analysis of existing connectomes, no new reconstruction
Loomba et al. 2022, mouse, macaque and human cortex	abstract gives no counts; paper is paywalled
Jarrell et al. 2012, male C. elegans tail	partial, folded into Cook et al. 2019

13 How we built the ledger

1. 01

   Start from the leads

   We began with a list of known connectome papers, from the 1986 roundworm paper to the 2026 male fruit fly CNS, and added papers found through Crossref and in the reference lists of the papers we read. Blogs and news were leads only; every figure is from a paper, its preprint or an official release page.

2. 02

   Read the primary text by script

   Each paper was saved (PubMed Central, eLife, bioRxiv, publisher PDFs where open, NLM abstracts where paywalled). Every number printed here was matched by a script against the saved text (86 checks); a number that is not found stops the build.

3. 03

   One row per reconstruction

   A row is one dataset with its own counts: the roundworm hermaphrodite and male are two rows from one paper. Rows keep the paper’s own definition of a synapse, and blank means the text did not state it. We never filled a value.

4. 04

   Add what the papers do not say about themselves

   Author counts and institutions from OpenAlex (CC0), dates and licences from Crossref, retraction flags from both (none found). Brain sizes are published counts: 86.1 billion neurons in a human brain, 67.87 million in a mouse brain.

5. 05

   Compute our own numbers

   Growth fits, shares, ratios and data-scaling arithmetic are done in one script (analyse.py) and labelled IQ Metrics calculation. “If the trend held” dates are arithmetic, shown next to the published statements of the Wellcome Trust, never as a prediction.