Brain Development and How We Study the Brain

The arc of brain development

The human brain is not assembled the way a machine is, part by part to a fixed blueprint. It is grown, and growth is a sequence of processes that overlap, overshoot, and correct themselves. The sequence begins astonishingly early. By the third week after conception, before most people know a pregnancy exists, a flat sheet of cells called the neural plate is already folding into a tube. That tube is the whole future central nervous system: everything from the spinal cord to the cerebral cortex is derived from it.

What follows is a set of overlapping stages. Cells divide at enormous rates in the proliferative zones lining the tube's cavity. Newly born neurons then migrate outward to their destinations, and in the cortex they do so in a striking pattern: the earliest arrivals settle in the deepest layers, and each later wave climbs past the cells already in place, so the cortex is built from the inside out. Having arrived, neurons differentiate, extend axons and dendrites, and find their targets, guided by molecular cues that their growth cones read as they travel.

Then comes the part that most surprises people meeting it for the first time. The developing brain makes far more synapses than it will keep. Connection numbers rise steeply after birth, peak in infancy and early childhood depending on the region, and are then cut back substantially over the following years. This is not damage. Pruning is how experience shapes circuits: connections that are used are stabilised, and connections that are not are removed. Microglia, the brain's resident immune cells, are among the machinery that does the removing.

Insulation comes last and takes longest. Myelination of the connecting fibres proceeds roughly from the back of the brain forward and from primary sensory and motor areas toward association cortex, and it does not finish in childhood. White matter continues to increase into the twenties, with the prefrontal cortex among the last regions to reach maturity. That fact is the honest kernel of the "adolescent brain" story, and it is worth stating carefully: it does not mean teenagers are incapable of reasoning, and it does not license the tabloid claim that the adolescent brain is simply broken. It means the systems supporting flexible self-regulation are still consolidating while other systems are already fully online.

Development does not stop at maturity. Grey matter volume declines gradually across adulthood, white matter changes in its own pattern, and cognitive ageing is decidedly not uniform. Processing speed and some aspects of memory decline on average; vocabulary and accumulated knowledge often hold up and can continue to improve well into later life. The single sweeping curve of "decline" is a poor description of what actually happens.

Full detail: Brain Development.

The contested question of new neurons

For most of the twentieth century, the textbook position was that the adult brain makes no new neurons. Santiago Ramón y Cajal, the founder of modern neuroanatomy, put the doctrine in memorably bleak terms: in the adult centres, he wrote, the nerve paths are fixed and immutable, everything may die, nothing may be regenerated. For decades that was simply the received view.

It did not survive intact. Rodent work from the 1960s onward, and later human studies using birth-dating markers and carbon-14 dating, showed that at least some parts of the adult mammalian brain do generate neurons. But the story did not end in a tidy reversal. In 2018 two prominent papers reached opposite conclusions about the adult human hippocampus in the same year: one found essentially undetectable neurogenesis in adults, the other found it persisting into old age. The disagreement turns on tissue handling and marker choice, and it has not been resolved.

This is a case where the honest answer is a mixed one, and this reference gives it. Robust adult neurogenesis in the rodent dentate gyrus is not in doubt. Whether, and how much, it occurs in adult humans is genuinely open. And the popular claims built on top of it, that this or that food, supplement, or exercise regime "grows new brain cells" in you, run far ahead of the human evidence.

Full detail, including the evidence on each side: Neurogenesis.

The toolkit: how we look inside

Every claim in this reference rests on a method, and no method is neutral. Each one buys certain kinds of information at the cost of others, and understanding the trade-offs is the difference between reading neuroscience and being taken in by it.

The pattern is a familiar trade-off. Methods with excellent temporal resolution tend to have poor spatial resolution, and vice versa. Methods that establish causation are invasive or coarse. Methods that are safe and non-invasive are indirect. Serious neuroscience triangulates: a claim believed on the strength of one modality alone should be held loosely.

Full detail: Brain Imaging Methods.

What the methods cannot tell you

Popular writing about the brain leans heavily on one image: a scan with a coloured blob, captioned as a region "lighting up". Almost everything about that framing is misleading, and it is worth being blunt about why.

Two specific traps deserve naming. The first is reverse inference: seeing a region active and concluding that the person was in a particular mental state. Because most regions participate in many functions, this reasoning is unsound unless the region is highly selective, a point Russell Poldrack set out in 2006 and which has been widely accepted since. The second is the multiple-comparisons problem, memorably illustrated by Craig Bennett and colleagues, who ran a standard uncorrected analysis on a dead Atlantic salmon and duly found significant "activation" in its brain. The salmon was a joke with a serious point: with tens of thousands of voxels, uncorrected statistics will find something.

To these should be added the field's ongoing reckoning with statistical power. Many influential neuroimaging findings were built on small samples, and large-scale replication efforts have shown that brain-behaviour correlations estimated from small samples are unstable. This is not a reason to dismiss neuroimaging. It is a reason to prefer large, pre-registered, replicated work and to be sceptical of a single striking result.

What powers it all

The brain is a metabolically extravagant organ. It is roughly 2 per cent of adult body mass and consumes around 20 per cent of the body's energy at rest, and in young children the share is higher still. It runs almost entirely on glucose under normal conditions, it stores essentially no fuel of its own, and it therefore depends on a continuous blood supply that takes something like 15 per cent of cardiac output.

Where does it all go? Overwhelmingly, into signalling itself, and specifically into undoing signalling. Every action potential and every synaptic event lets ions move down their gradients, and the sodium-potassium pump then has to push them back. Detailed budgets, most influentially by Attwell and Laughlin in 2001, put the great majority of the cortex's energy cost on this housekeeping. And note the corollary that makes functional imaging possible at all: because activity costs energy and energy demands blood, local activity raises local blood flow. That coupling is exactly what fMRI measures.

Full detail: Brain Energy and Metabolism.

Explore this section

Related topics