12. The Vertebrate Bargain

The nerve cord that ran serotonin and octopamine through a lobster's rise and retreat carries a third ancient amine as well, tuned not to posture but to pursuit: dopamine. Where the nonapeptides bind one animal to another, and the lobster's amines set its rank for only as long as the chemistry lasts, this messenger has a different assignment — not what to feel about another creature, and not where to stand, but whether to go after something that has not happened yet.

A lamprey in a cold river has no jaw, no bone in its skull, and barely a vertebra to its name — just a flexible rod called a notochord and a ring of cartilage. It is, together with the hagfish, one of the two oldest lineages of backboned animal alive today, a living stand-in for what the earliest vertebrates were like before anything grew a hinge in its mouth. Cut open its small brain and something unexpected turns up: a fully wired basal ganglia, the deep-brain structure that decides which movement gets to happen right now, running on dopamine in essentially the same layout a human's does.

The basal ganglia work like a traffic controller for action, holding back competing impulses until one is cleared to go. Dopamine is the chemical that tunes the controller's threshold. Losing the dopamine-producing cells in this exact circuit is what causes Parkinson's disease in humans — the tremor, the stiffness, a body that no longer obeys smoothly. In the lamprey, researchers have mapped the same wiring: dopamine-producing cells in the regions equivalent to the mammalian substantia nigra and ventral tegmental area, connected through the same "direct" and "indirect" pathways that let a mammal's brain both start and stop a movement, gated by the same two receptor types, D1 and D2, acting like two different switches on the receiving side. The comparison was drawn together in a 2021 review, and named outright five years earlier in a paper called, without much modesty, "The Basal Ganglia Over 500 Million Years," published in Current Biology in 2016.

Lampreys and hagfish split from the lineage leading to jawed vertebrates roughly 560 million years ago, in the late Ediacaran, before anything alive had a jaw or a limb. That makes them the most basal living vertebrates available for comparison. Because their dopamine-driven basal ganglia already show the full mammalian-style architecture, the system's origin can be placed at least that far back — to the root of Vertebrata itself. Some uncertainty remains about how much of the lamprey's other brain wiring mirrors the mammalian arrangement exactly, versus representing a simpler ancestral state elaborated on later. The core dopamine architecture is not in question.

The circuitry is old. The chemistry beneath it is older still. Dopamine is built in two steps from tyrosine, an amino acid the cell already had on hand for other jobs: an enzyme called tyrosine hydroxylase converts it to L-DOPA — the same molecule given as medication to Parkinson's patients today — and a second enzyme, DOPA decarboxylase, finishes the conversion into dopamine itself.

That exact two-step recipe turns up in an animal with no backbone at all. In 2021, a team from Stockholm University — Rehab El-Shehawy, Sandra Luecke-Johansson, Anton Ribbenstedt, and Elena Gorokhova — confirmed both enzymes expressed in the gut wall of the water flea Daphnia magna, tracked the second enzyme's activity across the animal's molt cycle, and detected biologically active L-DOPA, though not free dopamine, inside the gut. The paper appeared in mSystems on 9 November 2021.

Daphnia is a crustacean, a protostome — on the opposite branch of the animal family tree from every vertebrate, which are deuterostomes. The simplest reading of two such distant lineages sharing an identical two-enzyme pathway is inheritance from a shared ancestor, not two separate inventions of the same chemistry. That would push the pathway's origin back before the protostome-deuterostome split — and exactly when that split happened is itself an open argument, sitting somewhere in the Neoproterozoic. Molecular-clock studies place it at roughly 642 to 761 million years ago, with one eighteen-gene analysis landing near 670 million; fossil-calibrated statistical models pull it in tighter, to about 558 to 575 million years ago. The disagreement between these two methods — "rocks versus clocks" — is reviewed candidly in a 2016 paper on the limits of precision in molecular timescales.

The chemistry was not idle wherever it came from. In the roundworm Caenorhabditis elegans, a dopamine system built from exactly eight neurons shapes foraging decisions, as reviewed in 2025 in Royal Society Open Science. In the fruit fly Drosophila, a cluster of dopamine neurons called PAM is required for a fly to learn that a smell predicts sugar or water — an appetitive association, meaning one built around approaching something good rather than fleeing something harmful — and fires when the reward itself lands, a result reported in a 2025 preprint. In C. elegans specifically, dopamine seems to act more as a diffuse chemical mood-setter, bathing many neurons loosely, than as the point-to-point wiring vertebrates use, so the word "circuit" fits this invertebrate case more loosely than it does a mammal's brain.

Keeping Score

The clearest demonstration that dopamine is not simply a pleasure chemical came from recording individual dopamine neurons in the brains of behaving monkeys. Wolfram Schultz, working with theorists Peter Dayan and P. Read Montague, found these neurons firing not just when a reward arrives, but in proportion to how wrong the animal's expectation had been. A reward better than expected produces a burst above the neuron's normal rate. A reward that lands exactly as expected produces nothing unusual. A reward that fails to arrive, when one was expected, produces a dip below baseline, timed precisely to the moment it should have shown up. Published in Science in March 1997, the pattern matched, almost term for term, the error signal used in a family of computer learning methods called temporal-difference learning, where a system updates its guess about the future based on exactly how far off its last guess turned out to be.

The same signal has since turned up well beyond one macaque laboratory: prediction-error-shaped dopamine activity has been traced through teleost fish and rodents, and picked up in humans by fMRI scans of the ventral striatum and ventral tegmental area during simple reward-guessing tasks. A computation first isolated in a monkey's midbrain turns out to be a vertebrate-wide habit, not a primate specialty.

Nothing about that computation needed new molecular parts. It ran on wiring already in place for hundreds of millions of years — the same tyrosine-to-dopamine pathway found in a water flea's gut, installed in a circuit built at the root of Vertebrata, and handed a new job: not moving a limb, but keeping score.

The prediction-error story, though, is not the whole story. Kent Berridge and Terry Robinson, at the University of Michigan, found that injecting amphetamine directly into a brain hub called the nucleus accumbens made rats want sucrose more — approaching it, working harder for it — without making them like it more, judged by the same orofacial taste-reactivity expressions, a distinctive tongue-and-mouth pattern used to measure hedonic pleasure in infant mammals. Wanting and liking, they argued, are separable, run through different circuits, and dopamine's main contribution is generating the first: the pull of what they called incentive salience, not the pleasure of the reward itself.

The original theory dates to 1998, in Brain Research Reviews; it was sharpened and re-titled outright as "The debate over dopamine's role in reward" in a 2007 update. Extended into an incentive-sensitization theory of addiction, the same split treats craving that keeps intensifying even as a drug's pleasure fades as exactly what runaway "wanting," unmoored from "liking," would be expected to produce — the account Robinson and Berridge were still defending, thirty years on, in a 2025 retrospective. The three-way split — whether dopamine is mainly about learning, liking, or wanting — remains a live, unresolved question, not one any side has closed.

The receptors reading dopamine's signal multiplied along the way. A whole-genome duplication — an event in which an organism's entire genetic library is accidentally copied twice over, doubling every gene at once — occurred near the root of the vertebrate lineage roughly 450 million years ago, in the Ordovician, splitting one ancestral pair of dopamine-receptor genes into the four D1-like and several D2-like receptors carried by fish, birds, and mammals today, mapped in a 2013 study in Molecular Biology and Evolution. Not every lineage kept every copy afterward — mammals later lost one of the four, called D1C, and a different one, D1E, disappeared independently in several other branches. That the surviving system still functions the same way across that whole 450-million-year gap is not only a genetic inference: zebrafish dosed with morphine develop a place preference blocked by the opioid-blocker naloxone, and the same fish show conditioned preference for nicotine and ethanol persisting after three weeks of abstinence, reported in a 2013 study in PLOS ONE. Zebrafish and humans do not look alike, do not share a common ancestor for 450 million years, and do not build their reward systems the same way anatomically. Whatever computes reward in each of them still runs on the same borrowed alphabet.

From a jawless fish with no cortex at all to a small striped fish choosing the tank chamber where a drug once sat, the wager is the same wager: predict, act, and let one ancient molecule keep the score. The molecule stayed this constant across every one of these animals; what changed, in two lineages more than any other, was the size and shape of the brain built around it.

Overturned if: the TH/DDC dopamine-synthesis genes found in Daphnia turn out to be non-homologous to the vertebrate versions — independently evolved rather than inherited from a shared ancestor — converting what reads as one deep inheritance into two separate inventions of the same two-step chemistry.

Sources

• 2021 · El-Shehawy, R. et al., "Microbiota-Dependent and -Independent Production of l-Dopa in the Gut of Daphnia magna" — PMC8577283
• 2016 · "Uncertainty in the Timing of Origin of Animals and the Limits of Precision in Molecular Timescales" — PMC4651906
• 1997 · Schultz, W., Dayan, P., Montague, P.R., "A Neural Substrate of Prediction and Reward" — Science 275:1593
• 1998 · Berridge, K.C., Robinson, T.E., "What is the role of dopamine in reward: hedonic impact, reward learning, or incentive salience?" — Brain Research Reviews 28:309
• 2016 · "The Basal Ganglia Over 500 Million Years" — Current Biology30680-7.pdf)
• 2013 · "Evolution of Dopamine Receptor Genes of the D1 Class in Vertebrates" — PMC3603308