05. Dopamine: The Wanting Machine
In the spring of 1969, on the wards of Beth Abraham Hospital in the Bronx, roughly eighty people who had been unable to begin a movement for decades were started on L-DOPA. They were survivors of encephalitis lethargica, the sleeping-sickness epidemic of the 1920s, left in post-encephalitic parkinsonism — awake and aware inside bodies that would not start. Their physician was Oliver Sacks. Over the following weeks the wards came back: patients moved, spoke, and were recognisably themselves again. Then the effect frayed into tolerance and dyskinesia — movement arriving unbidden and refusing to stop — and most of them settled back toward stillness. Sacks published the account in 1973 as Awakenings.
That ward is the most complete demonstration this book has of what a signalling molecule does for a human being, and it ran backwards. Nobody there was shown dopamine working; they were shown its absence, then its brief return, then its absence again. This is the molecule's peculiar biography: we know it best by subtraction.
The subtraction was a laboratory method before it was a clinical one. In 1957 Arvid Carlsson, with Lindqvist and Magnusson, gave rabbits reserpine — a drug that strips the brain of its dopamine — and watched the animals go rigid and immobile. Injecting L-DOPA brought them back. Injecting 5-hydroxytryptophan, the precursor of serotonin, did not. That second injection is the whole result: it separates dopamine from the molecules beside it on the same chemical shelf and shows it driving movement in its own right, not merely as a way-station toward something else. Published in Nature, the finding carried a Nobel Prize in 2000 and is the direct ancestor of the bottle that reached Sacks's ward twelve years later.
What Sacks handed his patients was not dopamine but the step before it. The line is two enzymes long: the amino acid tyrosine is hydroxylated by tyrosine hydroxylase into L-DOPA, and L-DOPA has a carboxyl group cut away by DOPA decarboxylase to become dopamine. Vol. I traced the first of those enzymes into a pterin-dependent hydroxylase family older than animals, and watched a water flea run both in the wall of its gut. Plants build dopamine by their own convergent routes; a ciliate with no nervous system at all makes it and secretes it. None of this is vertebrate equipment, and it is why the treatment could work at all: Sacks supplied the intermediate and let each patient's own decarboxylase perform the final step. He was not delivering a signal. He was restocking a shelf.
Making a molecule and hearing one are different achievements, and they carry different dates. Vol. I's chapter on the vertebrate bargain found the full mammalian-style D1/D2 architecture, direct and indirect pathways included, already complete in a lamprey, whose lineage parted from ours around 560 million years ago. But the receptor inventory is a hundred million years younger. A whole-genome duplication near the root of Vertebrata, placed in the Ordovician around 450 million years ago by comparative genomics, took a single ancestral D1/D2 pair and made four D1-like receptors and several D2-like ones. Architecture at 560. Count at 450. One word — receptor — two separate commits underneath it.
One lead belongs in the margin rather than the ledger. A claim surfaced during this book's sourcing, in search summaries only, that D1-like and D2-like receptors are not sister lineages at all but independent arrivals at the same ligand — convergence rather than duplication. Two attempts to reach the primary source were blocked, and this book's rule is that an unverified claim is recorded as a question or not at all. This one stays a question. Note only what it would cost if it held: not the 450-million-year row, which counts duplications, but the assumption beneath it that "the ancestral D1/D2 pair" is one object worth counting duplications of.
Below the vertebrates the receptor's floor gets harder to find, and the sharpest evidence is negative. Ctenophores — the comb jellies of Vol. I's chapter on the first animals — run a working nervous system with no dopamine signalling in it whatsoever, which puts a ceiling on how deep the receptor can reach. A 2023 phylogenetic reconciliation by Goulty, Botton-Amiot, Rosato, Sprecher and Feuda presses the same direction in its title, calling the monoaminergic system as a whole a bilaterian innovation — a claim about the family, not a direct test of the dopamine receptor. Nor is the receptor's shape fixed above that line: Vol. I describes an octopus dopamine receptor, DopC1, that is ionotropic — a channel opening directly and fast — where every vertebrate dopamine receptor is a slow G-protein-coupled one. It stands as a 2025 preprint, not yet peer-reviewed, a caveat worth carrying alongside it. Meanwhile the synthesis line was already running, in gut walls and in single cells with no neurons to run it for.
Keeping the Rows Apart
"How old is dopamine" is not one question, and the reason this book keeps a ledger instead of a timeline is that it has at least four defensible answers separated by more than a hundred million years. The first is the synthesis capability — could a cell make the molecule? — and it goes back at least 600 million years, to at or before the last common ancestor of protostomes and deuterostomes. Vol. I is careful with that anchor and this book will be too: conventionally cited above 600 million years, it arrives at 642 to 761 million by molecular clock and around 558 to 575 by fossil calibration, and the distance between rocks and clocks is unresolved rather than split. The second answer is receptor existence — was there anything built to hear it? The third is circuit — was it wired into a loop that picks one action and suppresses the rest? The fourth is the job a human would actually name: wanting.
The fourth is the weakest row here, and the method is to say so rather than let the first three carry it across. Vol. I told how monkey dopamine neurons were caught encoding a reward-prediction error, and how rats made to want sucrose more did not come to like it any better. Both are findings about a computation running on the molecule, not about the molecule. Nobody has recorded that burst-and-dip signature in a lamprey or a hagfish — the two animals whose ancient basal ganglia would settle whether the computation is as old as the circuit or a much later tenant of it. The molecule has a date. The computation has none. An undated row is not an empty row; it is a row with an experiment still owed.
Outside the laboratory this bookkeeping stops being bookkeeping and becomes a defence. The popular account collapses all four rows into a single quantity: a "hit" that arrives as pleasure, drains with use, and can be restocked by abstinence — the dopamine detox. Each clause is wrong at a different row. The pleasure claim is the one Vol. I's rats already put down. The draining claim has no reservoir to drain: dopamine, as The Scientist's 2024 debunking of the detox trend puts it, "is essentially never acting in isolation" — other neuron populations gate the dopaminergic ones, and endogenous opioids, orexin and serotonin carry much of what the slang credits to dopamine alone. There is nothing for a fast to act on. The phone in your hand is misdescribed by the slang too: the firing comes before the notification resolves, keyed to the uncertainty rather than the reward — a prediction system doing its job, not a dose landing. And if you want to see dopamine depletion in a human being, it is not someone scrolling with dulled enthusiasm. It is the ward at Beth Abraham in the weeks before the L-DOPA.
The honest standing is uneven, and the unevenness is the finding. The synthesis rows are solid and very old; the circuit row is solid and younger; the receptor rows are solid and split across two dates the popular question flattens into one; and the row naming what a human actually feels is the weakest line here. That asymmetry is not a gap in the evidence, it is the evidence: this molecule was doing chemistry for hundreds of millions of years before anything used it to want. Dopamine also does not end at dopamine. It sits mid-line — an intermediate that acquired a receptor and got promoted to a signal — and the cheapest way evolution has found to make a new message is to bolt one more enzyme onto a line already running. One enzyme downstream of wanting is noradrenaline, and the body uses it to sound an alarm.
Commit Ledger
| Function | First commit | Grade | What would move the date |
|---|---|---|---|
| Synthesis capability (tyrosine → L-DOPA → dopamine; TH then DDC) | ≥600 Mya, at or before the protostome–deuterostome LCA; method-split anchor — 642–761 Mya molecular clock vs. ~558–575 Mya fossil calibration | B | Confirmed non-neural dopamine use in a cnidarian running homologous enzymes |
| Receptor existence (something built to bind dopamine) | At or below the bilaterian LCA, inside the same 600–760 Mya window; ceiling set by ctenophore genomes carrying none — comparative genomics | B | A confirmed dopamine-receptor ortholog in a ctenophore or a cnidarian |
| Circuit architecture (D1/D2 direct–indirect basal ganglia loop) | At or before ~560 Mya, late Ediacaran — dated by the cyclostome/gnathostome split; comparative neuroanatomy in lamprey | B | Revision of the cyclostome split date; no earlier direct test of the circuit exists |
| Receptor inventory expansion (one D1/D2 pair → 4 D1-like + several D2-like) | At or near 450 Mya, Ordovician — vertebrate whole-genome-duplication timing; comparative genomics | B | Revision of vertebrate 2R-WGD timing generally, not of anything dopamine-specific |
| Ionotropic (fast, channel-type) dopamine reception | Undated; cephalopod-restricted so far — structural inference from octopus DopC1, 2025 preprint, not peer-reviewed | C | Peer review of the DopC1 preprint, or a second ionotropic dopamine receptor in another lineage |
| Reward-prediction-error computation — the human-salient "wanting" role | Undated. The molecule is ≥600 My old; the computation has no confirmed pre-teleost date located for this book | D | Canonical burst/dip RPE signatures recorded in lamprey or hagfish electrophysiology |
Overturned if: a functional dopamine receptor is confirmed in a ctenophore genome — the one lineage that runs a nervous system with none of this chemistry inside it — which would break the bilaterian ceiling drawn here and reopen every receptor row beneath 600 million years.
Sources
- 1957 · Carlsson, Lindqvist & Magnusson, "3,4-Dihydroxyphenylalanine and 5-Hydroxytryptophan as Reserpine Antagonists" — Nature 180:1200
- 2010 · "Arvid Carlsson, and the story of dopamine" — Indian Journal of Psychiatry
- 1973 · Sacks, O., Awakenings — Duckworth, London (the Beth Abraham L-DOPA trial, begun spring 1969)
- 2023 · Goulty, Botton-Amiot, Rosato, Sprecher & Feuda, "The monoaminergic system is a bilaterian innovation" — Nature Communications 14:3284
- 2016 · "The Basal Ganglia Over 500 Million Years" — Current Biology30680-7.pdf) (lamprey D1/D2 direct/indirect architecture; established in Vol. I, ch. 12)
- 2021 · El-Shehawy, Luecke-Johansson, Ribbenstedt & Gorokhova, "Microbiota-Dependent and -Independent Production of l-Dopa in the Gut of Daphnia magna" — mSystems (Vol. I, ch. 12)
- 2013 · "Evolution of Dopamine Receptor Genes of the D1 Class in Vertebrates" — Molecular Biology and Evolution (dopamine-receptor whole-genome duplication near the root of Vertebrata, ~450 Mya; Vol. I, ch. 12)
- 2014 · Skjærven, Teigen & Martinez, "Structure–Function Relationships in the Aromatic Amino Acid Hydroxylases Enzyme Family: Evolutionary Insights" — Wiley/eLS (the family uniting TH, TPH and PAH; Vol. I, ch. 8)
- 1997 · Schultz, Dayan & Montague, "A Neural Substrate of Prediction and Reward" — Science 275:1593 (dopamine neurons encode reward prediction error; Vol. I, ch. 12)
- 1998 · Berridge & Robinson, "What is the role of dopamine in reward: hedonic impact, reward learning, or incentive salience?" — Brain Research Reviews 28:309; retitled 2007; thirty-year retrospective 2025 (Vol. I, ch. 12)
- 2014 · Moroz et al., "The ctenophore genome and the evolutionary origins of neural systems" — Nature 510:109–114; and 2013 · Ryan et al., "The genome of the ctenophore Mnemiopsis leidyi" — Science 342:1242592 (ctenophores: no dopamine signalling of any kind; Vol. I, ch. 9)
- 2025 · Octopus ionotropic dopamine receptor DopC1 — bioRxiv preprint, not yet peer-reviewed (Vol. I, ch. 13)
- 2016 · dos Reis et al., "Uncertainty in the Timing of Origin of Animals and the Limits of Precision in Molecular Timescales" — Current Biology / PMC4651906 (the protostome–deuterostome split held as a method-dependent range, "rocks versus clocks"; Vol. I anchor, ch. 9)
- 2024 · Thomasy, "Debunking the Dopamine Detox Trend" — The Scientist
- 1976 · Jensen, "Enzyme Recruitment in Evolution of New Function" — Annual Review of Microbiology 30:409–425