GloNeuro

Brain networks

Network neuroscience studies how regions connect and communicate, and how patterns of connection give rise to thought, emotion and behaviour.

SubjectConnections between neurons and between regions
ScaleFrom local circuits to the whole brain
Built fromImaging, electrophysiology and microscopy data

Methods by scale

Microscopy
about 1 nm to 1 cm
Neuroimaging
about 0.5 mm to the whole brain
Brain networks
single neurons to the whole brain
Electrophysiology
about 10 nm to the whole brain
1 nm10 nm100 nm1 µm10 µm100 µm1 mm1 cm10 cm1 m
  1. Ion channelabout 10 nm
  2. Synaptic vesicleabout 40 nm
  3. Neuron cell bodyabout 10–30 µm
  4. Thickness of cortexabout 2–4 mm
  5. Human brainabout 15 cm long
Spatial scale, logarithmic: each step along the axis is ten times larger than the last. Ranges and sizes are approximate.
Construction of a brain network. The brain is divided into regions, which become nodes. The connection between every pair is measured, from white-matter tracts (structural) or from correlated activity (functional). The resulting network is analysed for modules of tightly linked regions and for hubs with many connections.

A network is described by nodes and the links between them. In the brain, nodes may be individual neurons or whole regions, and links may be physical wiring (structural connectivity) or activity that rises and falls together (functional connectivity).

The complete map of these connections is called a connectome. Brain networks share recurring features: highly connected hub regions, groups of regions that work closely together, known as modules, and short paths that let information travel efficiently across the brain.

Many neurological and psychiatric conditions, including depression, schizophrenia and Alzheimer’s disease, are associated with changes in network organisation. Network measures are therefore being studied as possible markers of disease and of response to treatment.

Key ideas

Structural connectivity

The physical wiring diagram of the brain.

Structural connections are the axons and synapses that physically link neurons and regions. In animals they are mapped with tracers that travel along axons, and with electron microscopy. In living humans, diffusion MRI and tractography estimate the large white-matter pathways between regions.

Applications include understanding how anatomy constrains activity, studying how connections develop and decline and assessing damage after injury.

Structural maps show the possible routes for communication; they do not show when those routes are used.

Functional connectivity

Regions whose activity rises and falls together.

Functional connectivity is the statistical similarity, often a correlation, between the activity of different regions over time. Recordings from fMRI, EEG or MEG are compared pair by pair, producing a matrix of how strongly each pair is coupled. Regions can be functionally connected without a direct physical link, through intermediate regions.

Applications include identifying large-scale networks at rest, comparing brain organisation between groups and tracking change with learning or treatment.

A correlation does not reveal direction or cause; methods for “effective connectivity” attempt to estimate these.

Hubs and the rich club

A small set of highly connected regions forms a communication backbone.

Some regions, including parts of the precuneus, posterior cingulate and superior frontal cortex, connect to many others. These hubs also tend to be densely connected with one another, a pattern known as a rich club. Much of the communication between distant parts of the brain is thought to pass through this backbone.

Applications include explaining efficient brain-wide integration, understanding why damage to hubs has widespread effects and studying vulnerability in neurodegenerative disease.

Hubs are metabolically costly, and several studies report that they are disproportionately affected in Alzheimer’s disease.

Modules

Communities of regions that work closely together.

Network analysis can divide the brain into modules: groups of regions with many connections inside the group and fewer to other groups. Modules often correspond to functional systems, such as visual, sensorimotor or attention networks. Connector hubs link one module to another.

Applications include describing specialised processing, studying how modules reorganise with tasks and age and comparing networks across individuals.

A modular design lets parts of the system adapt without disrupting the whole.

Small-world organisation

Local clustering combined with short long-range paths.

In a small-world network most connections are local, so neighbouring nodes form tight clusters, but a few long-range links act as shortcuts. Any two nodes are then only a few steps apart. Brain networks at many scales show this pattern, balancing specialised local processing with efficient global integration while keeping wiring costs low.

Applications include comparing network efficiency between groups, modelling how information flows and understanding the cost of long-range wiring.

The idea was formalised by Watts and Strogatz in 1998, using examples that included the nervous system of the worm C. elegans.

Large-scale networks

Recurring systems linked to particular kinds of mental activity.

Resting-state fMRI consistently reveals a handful of large-scale networks. The default mode network is most active during rest, recollection and imagining the future; the salience network detects important events; the frontoparietal control network supports goal-directed attention. Shifts between these networks are thought to help organise attention and thought.

Applications include studying attention, memory and self-referential thought, investigating depression, schizophrenia and dementia and assessing consciousness after brain injury.

Network names and boundaries vary somewhat between brain atlases.

Large-scale brain networks

NetworkMain regionsAssociated with
Default modeMedial prefrontal cortex, posterior cingulate, angular gyrusSelf-referential thought, memory and imagining the future
SalienceAnterior insula, anterior cingulate cortexDetecting important events and switching between networks
Frontoparietal controlLateral prefrontal and posterior parietal cortexGoal-directed attention and working memory
SensorimotorPrecentral and postcentral gyriMovement and bodily sensation

Network names and boundaries differ somewhat between research groups.

History of the field

  • Discovery
  • Technique
  • Application

  1. 1986Discovery

    The first complete connectome

    White and colleagues map every connection of the 302 neurons of the worm C. elegans.

  2. 1998Discovery

    Small-world networks

    Watts and Strogatz describe networks that combine local clustering with short global paths.

  3. 2001Discovery

    The default mode network

    Raichle and colleagues describe a set of regions that is most active when the brain is at rest.

  4. 2005Technique

    The connectome

    The term is proposed for a comprehensive map of the brain’s connections.

  5. 2010Application

    The Human Connectome Project

    A large consortium begins mapping the connections of the healthy adult human brain with MRI.

  6. 2011Discovery

    The rich club

    Hub regions of the human brain are shown to be densely interconnected with one another.

  7. 2016Application

    A new map of the cortex

    Multimodal MRI divides each human cerebral hemisphere into 180 areas.

  8. 2020Application

    The fly hemibrain

    Electron microscopy maps the connections of a large part of the fruit-fly central brain.

  9. 2022Application

    Network-guided brain stimulation

    An accelerated, fMRI-guided magnetic stimulation protocol for treatment-resistant depression is cleared in the United States.

  10. 2023Application

    A complete insect brain

    The connectome of the larval fruit-fly brain, with about 3,000 neurons, is published.

  11. 2024Application

    The adult fly brain

    The FlyWire connectome maps roughly 140,000 neurons and their connections.

  12. 2025Application

    Connectomics with function

    The MICrONS project relates the wiring of mouse visual cortex to the recorded activity of the same neurons.

Strengths and limitations

StrengthsLimitations
Links brain structure to function and behaviourResults depend on how nodes and edges are defined
Combines data from many methods and scalesFunctional connectivity does not show cause and effect
Offers new ways to describe how disease affects the brainHuman connectomes cannot yet be mapped at the level of single neurons

Last updated