Reward system (the mesolimbic system)

The set of brain structures that decide whether a stimulus is worth approaching. Its job, stated in Lewis et al. 2021, is narrower than the popular version: the system "determines the valence of a stimulus and signals whether it is to be avoided or approached, as well as assigning the priority of one stimulus over another." Valence and priority — not pleasure.

The circuit

VTA → everywhere. Midbrain dopamine neurons in the ventral tegmental area project to the striatum, prefrontal cortex, amygdala, hippocampus and other limbic structures. A rewarding stimulus activates them and dopamine is released into those targets.

The striatum splits by job. The nucleus accumbens, in the ventral striatum, is the structure most associated with reward, evaluation and incentive-based learning. The dorsal striatum does action selection, decision initiation and habit. Both are involved; they are not doing the same thing.

The striatum splits again by receptor. Medium spiny neurons (GABAergic) are 90–95% of striatal neurons and the striatum's only output. They form two pathways:

Responding properly to a reward needs both signalled together. The remaining 5–10% of striatal neurons are interneurons — mostly inhibitory GABAergic ones, plus a 1–2% population of tonically active cholinergic interneurons that punch well above that share and whose activation tracks the salience of an event rather than its value.

Dopamine is central, not sole

The chapter opens by saying reward "requires the coordinated release of heterogenous neurotransmitters" and only then that dopamine holds "a central position." The order matters here, because this spoke holds the case where the qualification got dropped for a general audience and had to be retracted — see dopamine-nation and anna-lembke.

What hijacks it

Natural rewards — eating, drinking, mating — exist because they keep an organism and a species going. Drugs of abuse, licit and illicit, offer the reward without the biological function, and the chapter's account of what follows is that the pleasure of early use is lost to abuse while the seeking is not. Palatable food does a milder version of the same thing: repeated high-fat, high-sugar exposure produces compulsive consumption and poor intake control, which is the reward system overriding the homeostatic one.

Two consequences of that override are measured rather than asserted. Striatal D2R expression falls in mice fed a palatable diet, and striatal D2R availability is lower in obese humans than in controls (Wang et al. 2001). Whether the receptors fell because of the eating or the eating followed the receptors is not settled in the chapter.

It is also a clock

The finding this corpus did not previously hold: the reward system is wired to the suprachiasmatic nucleus, and the traffic runs both ways. Rewarding activity releases dopamine, and dopamine shifts circadian phase — so feeding, sex, drugs and social contact act as time-givers. See circadian-rhythm.

It runs differently at 3am (2026-09-19)

Tubbs et al. 2022 add a time axis to the circuit above. Their summary of the nocturnal state is a reward system showing elevated dopamine anticipation and a blunted response on receipt, with greater ventral striatum activation under sleep deprivation and reduced anterior insula activity alongside it — more appetite for the reward, less of the signal that would price its cost.

Two things follow for this page.

The D2 side weakens specifically. Sleep deprivation reduces available D2 (inhibitory) receptors in the ventral striatum. The table above says responding properly to a reward needs D1 and D2 signalled together; this is that requirement failing on a schedule, and it lands on the same receptor the obesity evidence here already turns on.

Anticipation and receipt come apart. The corpus's largest open gap is that it holds no theory of what dopamine computes — no reward prediction error, no wanting/liking. An "elevated anticipation, blunted receipt" state is the shape of Berridge's wanting-vs-liking split, arriving from a paper that never names it. That is a hint about which theory would do the work here, not evidence for it. See synthesis.

What this page does not cover

The founding source is a chronobiology chapter, so the standard reward-system topics it never touches are absent here too: reward prediction error and the Schultz dopamine-as-teaching-signal account, and Berridge's separation of wanting from liking. Both are central to the field and neither is in this corpus. Listed as a gap in synthesis.

Related

dopamine · circadian-rhythm · lewis-borrelli-reward-2021 · dopamine-nation · mind-after-midnight-hypothesis · tubbs-mind-after-midnight-2022 · robert-sapolsky · free-will · synthesis