Dopamine¶

A neuromodulator. Not the pleasure chemical. Not a unitary "amount" sloshing around the brain. Not something you can sense introspectively as "high" or "low."

What it actually is: a molecule released by specific neurons in the midbrain (VTA, substantia nigra) onto specific pathways, modulating the activity patterns of the circuits those pathways project into. Each pathway has a different computational job.

The four pathways¶

Same molecule, different circuit, different effect:

Mesolimbic (VTA → nucleus accumbens, ventral striatum) — motivation, salience, reward signaling.
Mesocortical (VTA → prefrontal cortex) — working memory, cognitive control, gating of what enters working memory.
Nigrostriatal (substantia nigra pars compacta → dorsal striatum) — action initiation, motor vigour. The pathway that degenerates in Parkinson's disease.
Tuberoinfundibular (hypothalamus → pituitary) — prolactin regulation. Endocrine, not cognitive — but worth knowing exists, since antipsychotics that block dopamine cause hyperprolactinaemia as a side effect.

Receptor families¶

Dopamine acts via five receptor subtypes grouped into two families, both G-protein-coupled:

D1-like family — D1 and D5. Gs-coupled → activate adenylyl cyclase → ↑ cAMP. Mostly post-synaptic, generally excitatory in their net effect on the target neuron's response to other inputs.
D2-like family — D2, D3, D4. Gi/Go-coupled → inhibit adenylyl cyclase → ↓ cAMP. Includes autoreceptors (D2 in particular) that sit on the presynaptic dopamine neuron itself and inhibit further dopamine release when dopamine is in the synapse — a built-in negative feedback loop.

The split was introduced by Kebabian & Calne (Nature, 1979) based on cAMP coupling; the five subtypes were cloned in the late 1980s–early 1990s.

Why the receptor split matters clinically:

Typical antipsychotics (chlorpromazine, haloperidol) block D2 — effective on positive symptoms of schizophrenia, but heavy motor side effects (extrapyramidal symptoms) because they also block D2 in the nigrostriatal pathway.
Atypical antipsychotics (clozapine, olanzapine, risperidone) block D2 + 5-HT2A — comparable positive-symptom efficacy with reduced motor side effects; some efficacy on negative symptoms.
Stimulants used clinically (methylphenidate, amphetamines) increase synaptic dopamine by blocking or reversing the dopamine transporter (DAT); the resulting effect on D1 in the prefrontal cortex underlies cognitive- enhancement effects in ADHD.

The receptor-level detail is what lets the same molecule ("dopamine") have such different functional consequences across its four pathways — which receptors are dominant in which target region determines whether the modulation is excitatory or inhibitory, and on what timescale.

What dopamine release actually correlates with¶

Dopamine release patterns do have well-mapped correlates — but they are computational, not phenomenal.

Phasic dopamine in mesolimbic VTA neurons → reward prediction error (Schultz, Dayan & Montague, 1997). When a reward is better than expected, phasic firing increases; when worse, it dips below baseline; when exactly as expected, no signal. Robust, quantitatively precise, the foundation of modern reinforcement-learning theory of the brain.
Phasic dopamine in mesocortical projections → gating of working memory updates. Tells the prefrontal cortex when to swap out the contents of working memory.
Phasic dopamine in nigrostriatal pathway → action initiation and vigour. Loss of this signal is why Parkinson's patients have difficulty initiating movement.

These are computational roles in specific circuits, not unitary subjective states. Reward prediction error is a number, not a feeling.

Why "increasing dopamine = feeling better" is wrong¶

Direct empirical evidence:

L-DOPA in Parkinson's restores dopamine signaling and treats motor symptoms — but does not produce euphoria. If dopamine were the pleasure chemical, L-DOPA would be a recreational drug. It isn't.
Anhedonia (the loss of pleasure, central to depression) doesn't reliably correlate with low dopamine. The monoamine-deficiency model of depression has been substantially weakened.
Cocaine and amphetamines produce euphoria via massive dopamine release — but also massive noradrenaline and serotonin release. The euphoria doesn't track simply with dopamine elevation, and the experience is not the same as what you get with L-DOPA at therapeutic doses.
Wanting vs liking (Kent Berridge & Terry Robinson, 1998+) — the canonical demonstration that even within rewarded behaviour, dopamine drives wanting (motivational pull, incentive salience) but not liking (hedonic pleasure). Hedonic pleasure lives in opioid hotspots in the nucleus accumbens shell. You can pharmacologically separate them in rats: knock out the dopamine system and rats still show liking responses to sucrose, they just don't want it enough to pursue it.

"Feeling better" isn't a unitary process¶

Even if dopamine did have a phenomenal correlate, "feeling better" wouldn't be a clean target. It's a bundle of:

Mood (sustained affective tone)
Motivation (drive to act)
Anhedonia / hedonia (capacity for pleasure)
Energy / fatigue
Attention / engagement
Interest / curiosity

Each of these is supported by its own constellation of circuits, and dopamine modulates some of them differently than others. There can't be a single neurochemical correlate even in principle, because there isn't a single thing being correlated.

Where the felt-states actually live¶

The mental-state correlates live in activity patterns — spatiotemporal dynamics across distributed networks — not in the level of any single neurotransmitter. This is the modern view from:

The neural correlates of consciousness (NCC) research programme: find what patterns of activity change when conscious content changes.
Predictive processing: mental states emerge from prediction-error / precision dynamics across hierarchical circuits.
Integrated information theory (IIT, Φ): consciousness is the irreducible integrated information structure of a system's causal dynamics.

Dopamine modulates these patterns — sets the gain, biases which states the system explores — but doesn't constitute the felt state. Saying "dopamine = pleasure" is like saying "the volume knob on the radio is the music."

The cleaner formulation¶

The version to keep in mind across the KB:

Dopamine has well-defined computational roles (reward prediction error, salience, gating, vigour), each in a specific pathway. It has no unitary phenomenal correlate. Felt states like "feeling better" are correlates of distributed activity patterns across many circuits — patterns that dopamine modulates but doesn't constitute.

Implications for the KB¶

Treat any "low dopamine / high dopamine" claim as folk vocabulary that needs translation before it can carry scientific weight.
When discussing reward, motivation, addiction: use wanting vs liking, reward prediction error, incentive salience — not "dopamine hit."
When discussing depression: the monoamine hypothesis is weakened; treatments that work probably do so via neuroplasticity over weeks, not acute monoamine elevation (Moncrieff et al., Mol. Psychiatry, 2022 umbrella review on serotonin and depression).
When discussing psychedelics: the relevant target is 5-HT2A, not dopamine — see the serotonin row in the glossary.
When discussing "modulation of activity patterns by neurotransmitters": this is exactly the right framing. Dopamine is one example; serotonin, ACh, NE, and the neuropeptides all do similar circuit-mode-setting jobs in their own pathways.