Most communication between neurons runs on two amino acids: glutamate, which excites, and GABA, which inhibits. Between them, they handle the vast majority of signalling in the brain---every perception, movement, and thought is carried by glutamate and GABA firing across neural pathways. They are the workhorses. The neurotransmitters that get all the attention---dopamine, serotonin, norepinephrine---are not workhorses. They are modulators. They don't carry the signal; they tune the system's responsiveness to it. This distinction matters, because popular neuroscience collapses it constantly.

The brain doesn't just process signals---it tunes how it processes them. Modulatory neurotransmitters like dopamine, norepinephrine, serotonin, and acetylcholine alter the responsiveness of neural circuits, changing learning rates, adjusting the precision of predictions, and gating which pathways are active. These signals don't carry content like "this is a dog" or "move your hand"; they broadcast context: how much to weigh this signal, how quickly to update, whether to exploit familiar routes or explore alternatives.

Think of neuromodulation as the brain's gain control. Dopamine spikes when predictions are violated---signalling "this matters, update your model." Norepinephrine tracks unexpected uncertainty---"something changed, pay attention." Acetylcholine marks expected uncertainty---"the world is volatile here, stay flexible." These modulators don't tell you what to learn; they tell you when and how much.

This architecture explains why the same event can produce very different behavioural outcomes depending on state. High arousal (elevated norepinephrine and cortisol) tightens the aperture---you fall back on familiar neural pathways, filter inputs aggressively, and execute with bias towards what you already know. Low arousal broadens sampling---you're more plastic, more exploratory, more willing to update. Dopamine shapes which updates stick: dopamine during an event strengthens the pathways that were active, making them more likely to fire next time the context recurs. This is why anticipation of reward activates much of the same machinery as reward itself, and why context around a reward becomes predictive---the modulatory signal marks the whole episode for consolidation.

Practically, this means state management is learning management. Regulate arousal and you regulate the plasticity-stability trade-off. If you need to explore or update, calm down---lower arousal broadens the system. If you need to execute a trained routine under pressure, a bit of urgency sharpens focus and tightens the model. But push too far in either direction and performance degrades: too calm and you're unfocused; too aroused and you're rigid.

The taxonomy of these chemicals is itself a patchwork. Amino acid transmitters (glutamate, GABA) handle the heavy lifting. Monoamines (dopamine, serotonin, norepinephrine) modulate mood, arousal, and learning. Neuropeptides (endorphins, oxytocin) operate on slower timescales, shaping motivation and social bonding. Even gases like nitric oxide serve as messengers. The system wasn't designed---it evolved, and the categories we impose on it are conveniences, not natural joints. This matters because the pop-neuroscience habit of assigning one function to one molecule---"dopamine is the reward chemical," "serotonin is the happiness chemical"---is a confidence game. It sounds explanatory but has roughly the same informational content as "rewards feel rewarding." The real insight is that these molecules tune operating parameters, and any given parameter is tuned by multiple molecules acting in concert across different circuits. No single neurotransmitter explains any behaviour.