The architecture of the brain — translated into clear, beautiful explanations of how cognition emerges from biology.
In the spring of 2018, two research groups published findings on the same question, in the same species, within weeks of each other, and reached opposite conclusions. One reported that the formation of new neurons in the human hippocampus falls to undetectable levels after childhood. The other reported that it continues into the eighth decade of life. Neither group was careless, and neither result was quietly withdrawn.
The first part of this series examined the defence of adult human neurogenesis mounted by eighteen researchers in 2018. This part examines the reply, written by the group whose negative finding prompted the exchange. It is short, pointed, and argues that the defence made the wrong argument.
The first two parts of this series followed a dispute conducted almost entirely within one structure. Both camps examined the dentate gyrus of the hippocampus, disagreed about what was in it, and drew conclusions about the adult human brain from that single site.
The progressive deterioration of cognitive function associated with brain aging and neurodegenerative disorders is closely linked to structural and functional changes at the synaptic level. Synaptic dysfunction and synapse loss may precede extensive neuronal death, particularly in disorders such as Alzheimer’s disease. This observation has encouraged investigation into biological strategies capable of supporting the maintenance, remodeling, and formation of neuronal membranes.
BDNF is probably the most name-dropped molecule in popular writing about the brain. It appears as a kind of fertiliser — more of it is better, and various activities and substances are said to raise it. Almost nothing about that framing survives contact with the literature.
Part One established that BDNF is processed into forms with opposing actions. This part concerns a single-letter difference in the gene that changes how that processing works — and which a large share of the human population carries.
The previous two parts described BDNF inside the brain: a protein processed into forms with opposing actions, and a common variant that alters that processing. This part deals with the only version of BDNF anyone is likely to encounter in practice — a number from a blood sample.
Peptides have become a fashionable category in the wellness market, usually discussed as though the only question were which one to take. The research literature is preoccupied with a different question entirely: how to get any of them into the brain at all.
Acetylcholine is the neurotransmitter most often invoked when a supplement mentions memory. The mechanism behind that association is real, and it does not work the way the marketing implies.
Part One described acetylcholine in the hippocampus as a signal that sets the operating mode of a circuit rather than carrying content. This part follows the same system outward, into cortex.
Long-term potentiation is the most studied mechanism in neuroscience. Stimulate a pathway in the right pattern and the synapses along it become stronger, for hours or days. It was first described in 1973, and in 2024 the Royal Society held a discussion meeting under the title "Long-term potentiation: 50 years on".
If a synapse becomes stronger and stays stronger for days, something has to hold the change in place — through protein turnover that replaces the molecules involved many times over.
Parts One and Two described the synaptic theory of memory: learning strengthens connections, and a molecular machine holds the strengthening in place. It descends from Hebb and it fits a great deal of data.
In 2015, a group at MIT gave mice a drug that blocks memory consolidation, confirmed the mice had no memory of what happened to them, and then switched the memory back on with a pulse of blue light.
Part One showed that a memory blocked by a protein synthesis inhibitor is still physically present and can be switched on with light. That is a striking laboratory result about an artificial amnesia.
A neuron cannot store fuel. It runs a metabolic rate among the highest in the body with essentially no reserve, which means something else has to supply it continuously — and that something is the astrocyte.
Part One described the astrocyte as a metabolic partner. This part covers the harder claim: that astrocytes participate in signalling, release transmitters of their own, and can tell one neural pathway from another.
The brain does not only build connections. It removes them, deliberately, using cells that belong to the immune system — and it does this throughout life, not only during development.
Part One established that microglia remove synapses as a normal part of how circuits are built. That raises an obvious question the field could not answer for years: how does a microglial cell know which synapse to remove?
Two parts of this series have been about microglia. A 2022 paper in Nature Aging shows that for one class of synapse, in a mouse model of tauopathy, the cell doing most of the eating is an astrocyte.
Three parts of this series have described a mechanism established in mice and in cultured rodent cells. This part covers the only study of the four that used human material, and it is the reason the topic reaches beyond neuroscience.
The gut–brain axis is the most enthusiastically marketed idea in nutritional science and one of the better documented at the mechanistic level. Both are true, which is what makes it hard to read honestly.
Part One established five routes from gut to brain and argued that the useful lens is metabolic rather than taxonomic: what the microbiota produces matters more than which species are present.
Short-chain fatty acids are the strongest part of the gut–brain story: named molecules, named receptors, a documented epigenetic mechanism, and — unusually for this field — a small number of controlled human experiments.