Brain Development

The neural tube

The story begins in the third week after conception, which is to say before most pregnancies have been noticed. A strip of cells running down the back of the embryonic disc thickens into a structure called the neural plate. Its edges rise into folds, the folds lean toward each other, and they fuse along the midline to form a hollow tube. Closure begins in the middle and zips outward in both directions, and the process is complete by roughly the end of the fourth week.

That tube is everything. Its walls become the brain and spinal cord; its central cavity becomes the ventricular system that later carries cerebrospinal fluid. Nothing else in the body will produce a neuron of the central nervous system.

This is the reason folate matters so much. Randomised trial evidence, most influentially the UK Medical Research Council's Vitamin Study published in 1991, showed that folic acid supplementation around the time of conception substantially reduces the recurrence of neural tube defects. It is one of the clearest, most actionable findings in the whole of developmental neuroscience, and it is why folic acid supplementation is recommended before conception and in early pregnancy, and why many countries fortify flour with it.

From three vesicles to a brain

Almost immediately after closure, the head end of the tube begins to swell into three bulges, the primary vesicles. Everything in the adult brain descends from one of them.

The three-then-five vesicle scheme is more than an anatomical curiosity. It explains why the adult brain is organised the way it is: the divisions you can name on a dissected brain, described in detail on the brainstem and cerebellum and cerebrum pages, are the grown-up forms of these embryonic swellings. The cavity inside the tube likewise persists, becoming the ventricular system.

Proliferation: making the cells

Cells are manufactured in the proliferative zones lining the tube's cavity, chiefly the ventricular zone and, later, the subventricular zone. Progenitor cells there divide, and their daughters either divide again or exit the cycle and become neurons.

The rate is difficult to convey without sounding hyperbolic, but the arithmetic is unavoidable: an adult cortex contains billions of neurons, and essentially all of them are produced in a matter of months of prenatal life. Averaged over that period, the developing brain generates neurons at a rate on the order of a quarter of a million per minute. Peak production overlaps with the middle of gestation, and the great majority of cortical neurons are in existence before birth.

Whether the progenitor cell divides symmetrically (making two more progenitors, expanding the pool) or asymmetrically (making one progenitor and one neuron, spending the pool) is one of the key levers of brain size. Small shifts in the balance, sustained over many divisions, produce large differences in the final number of neurons, which is one reason evolution has been able to scale cortices up and down so effectively.

Migration and the inside-out cortex

Neurons are born in the wrong place. They are generated deep, next to the ventricle, and they belong in the cortical sheet at the surface. Getting there requires migration, and the principal mechanism is beautifully simple: the young neuron climbs.

Radial glial cells span the full thickness of the developing wall, from the ventricle to the outer surface. They are both progenitors, generating neurons, and scaffolding, providing the fibres along which those neurons crawl outward. This is radial migration, and it is how the cortex is populated. (Some populations, notably many inhibitory interneurons, arrive instead by tangential migration, travelling sideways from their birthplace in the ventral forebrain.)

The striking part is the order. The earliest-born neurons stop and settle in the deepest cortical layers. Each subsequent wave migrates past the neurons already in place, and settles more superficially. The cortex is therefore built from the inside out: layer six first, layer two last. A neuron's birthday determines its layer, and its layer largely determines what it will connect to.

Differentiation and wiring

Having arrived, a neuron must become the right kind of neuron and connect to the right partners. Differentiation gives it its shape, its transmitter, and its receptor complement. Then it must wire itself in.

The tip of a growing axon carries a motile structure called a growth cone, which extends fine probes into the tissue ahead of it and reads the chemical environment. Guidance cues do the steering: some molecules attract, others repel, and some act at long range while others must be contacted directly. Netrins, slits, semaphorins, and ephrins are the best-characterised families. The growth cone integrates these signals and turns accordingly, and in this way axons navigate over what are, at cellular scale, enormous distances to reach targets they have never seen.

When the axon arrives, it forms synapses. The initial pattern of connections is approximately right rather than exactly right, and refining it is the job of the next stage.

Overproduction and pruning

Here development does something that looks, at first sight, wasteful. It builds far too much and then throws a great deal of it away.

Synapse density in the cortex rises steeply after birth, peaks during infancy and early childhood at a level substantially above the adult figure, and then declines over childhood and adolescence to the adult level. The peak comes at different times in different regions: sensory areas peak and settle earlier, and association cortex, including the prefrontal cortex, peaks later and prunes for longer. Classic quantitative work on this developmental profile was carried out by Peter Huttenlocher and colleagues on human post-mortem tissue.

Cell death does something similar. Substantially more neurons are generated than survive, and a programmed cell death (apoptosis) removes the surplus, particularly those that fail to secure adequate trophic support from their targets.

This is not inefficiency. Overproduction followed by selective elimination is a design strategy: it lets experience, rather than the genome alone, determine which connections are kept. A genome does not have the storage capacity to specify each of the brain's synapses individually. What it can do is build too many and let use decide. Connections that are active and useful are stabilised; connections that are not are removed. Microglia, the brain's resident immune cells (see glial cells), physically engulf and eliminate synaptic material during this process, a discovery that reframed them from mere immune sentries into active participants in circuit construction.

Myelination and the long finish

The last major structural process is insulation. Oligodendrocytes wrap axons in myelin, which speeds conduction dramatically, as described on the neuron page. Myelination begins before birth in some pathways and then proceeds over an extraordinarily long timescale.

Two gradients describe the pattern. Myelination proceeds roughly from the back of the brain toward the front, and from primary sensory and motor regions toward the association cortex that integrates them. The consequence is that the frontal lobes, and the prefrontal cortex in particular, are among the last regions to become fully myelinated. Structural MRI studies show white matter volume continuing to increase through adolescence and into the twenties, well after the brain has reached adult size.

The adolescent brain, accurately

The claim that "the teenage brain is not finished until twenty-five" has become one of the most-repeated pieces of neuroscience in public life, and it is used to explain everything from car insurance premiums to criminal responsibility. The underlying finding is real. The interpretation is frequently abused, and it is worth separating the two.

What the evidence supports: structural maturation, particularly myelination and the refinement of frontal circuits, continues well past the end of adolescence, and the prefrontal cortex is among the last regions to reach its mature form. Meanwhile, subcortical systems concerned with reward and emotion (see dopamine and reward) are functionally mature earlier. That mismatch in timing is the neurobiological core of the adolescent picture, and it makes sense of a specific and well-replicated behavioural pattern: adolescents take more risks in emotionally arousing situations and in the presence of peers than they do when calm and alone.

What the evidence does not support: the idea that adolescents are incapable of rational thought, or that they should be treated as though they lack understanding. On cold, deliberate reasoning tasks, adolescents perform much like adults. Nor is twenty-five a magic number at which a switch flips; it is a rough marker on a gradual curve that differs greatly between individuals. The honest statement is narrow: a specific maturational asymmetry makes certain kinds of decision, under certain conditions, systematically harder in adolescence. That is worth knowing. It is not the same as saying the adolescent brain is broken.

Critical and sensitive periods

Some circuits are wired by experience during a window, and the window matters. The classic evidence comes from the visual system: work by David Hubel and Torsten Wiesel in the 1960s showed that depriving one eye of input during a specific early period in kittens permanently altered the organisation of visual cortex, whereas the same deprivation later had far less effect. The finding won a Nobel Prize and established the principle that timing, not just experience, shapes the brain.

The clinical parallel in humans is congenital cataract: if it is not corrected early, vision in the affected eye may never develop normally, even after the lens is replaced. Language shows a softer version of the same pattern, with first-language acquisition proceeding easily in early childhood and becoming markedly harder for those deprived of language input in those years.

The term critical period implies an absolute window; sensitive period is usually the more accurate term, because the boundaries are gradual and because some plasticity persists afterwards. The distinction matters for a practical reason: a strict "windows close" framing has been used to sell parents an urgency that the science does not warrant. The brain retains a great deal of capacity for change throughout life, as covered under neuroplasticity. Early experience matters enormously; it is not the case that everything is decided by the age of three.

Ageing: what changes and what does not

Development does not have a terminus. Structural change continues across the adult lifespan, and while the direction is broadly one of gradual decline, the picture is far more nuanced than a single downward line.

Grey matter volume declines gradually from early adulthood onward, with the frontal and temporal regions among the more affected. White matter follows a different course, increasing into the middle decades before declining, and its microstructural integrity tends to deteriorate later. Ventricles enlarge as tissue is lost around them. All of this is normal ageing rather than disease, and it is distinct from the pathology of Alzheimer's disease and other dementias, which are conditions, not simply an extension of the normal curve.

Cognitively, the crucial point is that decline is not uniform. Processing speed, working memory, and some aspects of episodic memory tend to decline on average with age. But crystallised abilities, vocabulary, accumulated knowledge, and expertise, typically hold up well and can continue to improve into later life. An older adult may be slower to learn an unfamiliar list and simultaneously have a larger and better-organised vocabulary than a younger one. Averages also conceal enormous individual variation: some people in their eighties perform in the range of healthy adults decades younger.

Common misconceptions

Related topics