How We Study the Brain: An Overview of Methods

From patients and post-mortem slices to scanners, electrodes, light-controlled neurons and supercomputers: the toolkit of neuroscience and the trade-offs between seeing where, seeing when and seeing causes.

No single tool can capture the brain, which spans scales from nanometre synapses to whole-brain networks and from microsecond channel openings to lifelong learning. Neuroscientists therefore use a toolkit, each tool with strengths and blind spots.

Where versus when

A key trade-off is spatial resolution (how precisely you can say where) versus temporal resolution (how precisely you can say when). EEG and MEG capture changes in brain activity within milliseconds but blur where they come from. fMRI localises activity to within a few millimetres but tracks slow blood-flow changes over seconds. Electrodes inside the brain achieve both, but only in small regions, and require surgery.

Invasive versus non-invasive

Non-invasive methods — EEG, MEG, MRI, PET (which uses a small radiation dose), TMS — can be used in healthy human volunteers. Invasive methods — implanted electrodes, optogenetics, injecting tracers — are used mostly in animals, or in patients who already need surgery (for example, people with epilepsy undergoing monitoring).

Correlation versus causation

Watching activity shows correlation: region X is active during task Y. To show a region is necessary, you need to disrupt it — through lesion studies in patients, temporary disruption with TMS, drugs, or switching off neurons with optogenetics or chemogenetics. To show it is sufficient, you stimulate it and see whether the behaviour appears.

Looking at structure

Anatomy remains fundamental: histology and staining (from Golgi to modern antibodies), electron microscopy of synapses, tissue-clearing methods that make whole brains transparent, diffusion MRI to trace white matter in living people, and connectomics to map every connection in small volumes.

Genes, molecules and models

Genetics and single-cell sequencing identify cell types and disease genes. Animal models — from worms and flies to mice and monkeys — allow experiments impossible in humans. Brain organoids grown from stem cells model development. Computational models test whether proposed mechanisms can actually produce the observed behaviour. The richest discoveries usually come from combining several methods.