A seizure does not spread through the brain like ink diffusing evenly through water. Propagation follows anatomy. It depends on which neurons are connected, how strongly they are connected, the direction of those connections, the state of inhibitory circuits, the availability of long-range pathways and the excitability of the regions receiving the activity.
This is why two seizures that begin in similar regions can spread differently, why some remain localised, why others recruit distant networks quickly, and why the symptoms of a seizure can change as different brain systems become involved.
Propagation is not the same as onset
The site where abnormal activity is first detected is not automatically identical to every region that later becomes involved. A focal seizure can begin in a relatively restricted network and then recruit surrounding cortex, neighbouring lobes, the opposite hemisphere or deeper structures.
This distinction is clinically important. The first symptom may reflect the earliest recruited functional network rather than the exact point of origin, and scalp EEG may display propagated activity more clearly than the smaller or deeper source that produced it.
Local spread through cortex
Within cortex, seizure activity can recruit neighbouring tissue through local excitatory connections. Pyramidal neurons make extensive horizontal and recurrent connections, allowing activity in one cortical area to influence nearby populations.
Local propagation is not unlimited. Inhibitory interneurons surrounding the developing seizure can initially restrain recruitment. Experimental work describes a transition between tissue showing strong seizure-related field potentials and tissue in which local neurons are fully recruited into ictal firing. That advancing boundary is often described as an ictal wavefront.
The seizure core and the surrounding penumbra
Microelectrode recordings in human focal epilepsy have helped distinguish a densely recruited seizure core from surrounding tissue that may show large rhythmic electrical signals without the same degree of local neuronal firing. This surrounding region is sometimes called the ictal penumbra.
The concept explains an important paradox: an electrode can record dramatic seizure activity even when the nearby neurons are not yet fully participating. Strong synaptic currents generated elsewhere can produce large field potentials. Propagation therefore has to be understood through both local neuronal recruitment and the electrical influence of connected regions.
Why inhibition can slow the wavefront
Inhibitory interneurons at the edge of an ictal region can fire intensely and delay the recruitment of neighbouring principal neurons. This creates a moving contest between seizure-driving excitation and inhibitory restraint.
As seizure activity continues, inhibition can become less effective through several mechanisms, including activity-dependent chloride accumulation, synaptic fatigue and changes in extracellular ions. Once local restraint is overcome, previously protected tissue may join the ictal network.
White-matter pathways shape long-range spread
Seizures can also recruit distant regions through long-range axonal pathways. White-matter tracts connect cortical and subcortical areas that may be centimetres apart. Activity can therefore appear in a distant but strongly connected region without spreading continuously through every piece of tissue in between.
Relevant pathways vary with seizure origin. Temporal-limbic seizures may propagate through connections involving the hippocampus, amygdala, entorhinal cortex, fornix and cingulate networks. Frontal seizures can recruit bilateral frontal and motor networks rapidly through dense interhemispheric and subcortical connections.
The corpus callosum and interhemispheric spread
The corpus callosum is the largest white-matter connection between the cerebral hemispheres. It provides a major route through which activity in one hemisphere can influence homologous or functionally connected regions in the other.
Its importance is illustrated by corpus callosotomy, a palliative epilepsy surgery that disconnects much of the callosal pathway. The procedure does not remove the seizure-generating tissue, but it can reduce the rapid bilateral propagation responsible for some dangerous drop attacks and generalised motor events.
Subcortical structures can amplify or coordinate spread
The thalamus, basal ganglia, brainstem and other subcortical systems are not passive bystanders. The thalamus has extensive reciprocal connections with cortex and can participate in the synchronisation and distribution of seizure activity. Different thalamic nuclei connect to different cortical and limbic networks.
This network role is one reason thalamic stimulation is being used or studied in some drug-resistant epilepsies. It is also why a seizure originating in one cortical region can eventually produce widespread bilateral activity through recurrent cortico-subcortical loops.
Hub regions may matter more than distance
Brain networks contain highly connected regions sometimes described as hubs. A seizure that reaches a strategically connected hub may gain access to many other network nodes more easily than a seizure moving into equally nearby but weakly connected tissue.
This is why physical distance alone is a poor predictor of propagation. Two regions separated by several centimetres can be functionally and anatomically more connected than neighbouring areas separated by only millimetres.
Structural connectivity and functional connectivity are not the same
Structural connectivity refers to physical pathways such as axonal tracts. Functional connectivity describes statistical relationships between activity in different regions. Two areas can show strongly correlated activity even when their anatomical connection is indirect.
Epilepsy network research uses both concepts. Diffusion MRI can estimate major white-matter pathways, while EEG, MEG, fMRI and intracranial recordings can estimate functional interactions. Neither method by itself provides a perfect map of seizure propagation.
Why some seizures remain focal
Many focal seizures never become bilateral. Several factors can restrict spread: strong local inhibition, poor access to long-range pathways, low excitability in connected regions, network geometry, medication effects and the state of the brain at that moment.
A seizure can therefore stop after involving a relatively small network. That may produce subtle sensory, cognitive, autonomic or experiential symptoms without widespread motor activity or loss of awareness.
Why other seizures spread rapidly
Some regions are positioned within highly interconnected networks. Frontal lobe seizures, for example, can propagate very rapidly because frontal areas have dense intrahemispheric, callosal and subcortical connections. Rapid propagation can make scalp EEG and clinical semiology difficult to interpret because the earliest local signal may be brief or hidden before bilateral activity appears.
Deep sources can create a similar problem. Activity arising from mesial temporal, insular or other deep structures may need to reach superficial cortex before scalp EEG records it clearly. By then, the visible electrical pattern may reflect propagation rather than onset.
Focal to bilateral tonic-clonic seizures
When a focal seizure recruits sufficiently widespread bilateral networks, it can evolve into a focal-to-bilateral tonic-clonic seizure. This is not simply a stronger focal seizure. It represents a major change in network participation, with bilateral motor systems and other widespread circuits becoming involved.
The route to bilateral involvement varies. Some seizures cross quickly through interhemispheric connections; others propagate through subcortical structures and distributed cortical networks before generalised motor activity appears.
Generalised epilepsies also use networks
Generalised seizures should not be imagined as every neuron in the brain becoming active at exactly the same instant. Modern models emphasise distributed networks, particularly thalamocortical systems in absence and other generalised epilepsies.
The network may be bilaterally organised from the outset while still showing regional differences in timing and participation. ‘Generalised’ therefore describes the organisation of seizure onset and involvement at the network level, not indiscriminate activation of the whole brain.
Why symptoms change as a seizure spreads
As additional functional systems are recruited, the clinical expression can evolve. A focal sensory or experiential symptom may be followed by impaired responsiveness, automatisms, tonic posturing, head or eye version, bilateral motor activity or autonomic changes.
The sequence can carry localisation information because it reflects the order in which networks are recruited. However, propagation can blur the picture. A late dramatic sign may reflect a distant recruited network rather than the origin of the seizure.
Why scalp EEG can show the spread more clearly than the start
Scalp EEG detects electrical fields only after they are large and synchronised enough to reach the scalp. Small, deep or radially oriented sources may be invisible or poorly localised.
When activity propagates into a larger superficial cortical territory, the scalp signal can suddenly become much clearer. This can create the false impression that the seizure began where the scalp EEG first becomes obvious. Intracranial recordings sometimes reveal an earlier onset elsewhere.
This issue is explored in depth on the separate page When the Scalp EEG Sees the Spread, Not the Start, which focuses specifically on the diagnostic consequences rather than the general biology of propagation.
Stereo-EEG and propagation mapping
Stereo-EEG places electrodes in selected three-dimensional brain regions to sample suspected seizure networks directly. It can show the timing relationship between candidate onset zones and downstream regions, helping clinicians distinguish early involvement from later propagation.
However, stereo-EEG samples only the implanted contacts. It does not continuously record every brain region. Interpretation therefore combines intracranial timing with MRI, PET, SPECT, MEG, semiology and other evidence.
Diffusion MRI and tractography
Diffusion MRI can estimate the orientation of white-matter fibres, while tractography reconstructs probable pathways computationally. These methods are increasingly used in epilepsy research to study how structural networks may support seizure propagation.
Tractography has limitations. It infers pathways from water diffusion rather than tracing individual axons directly, and results depend on acquisition and modelling choices. It is therefore a network estimate, not a literal wiring diagram.
Network measures and the search for critical nodes
Computational network studies use measures such as centrality, synchronisability and effective connectivity to identify regions that may be disproportionately important for propagation. The long-term goal is to distinguish tissue that generates seizures from tissue that merely receives them, and to identify nodes where surgery or stimulation could disrupt the network.
These approaches are promising but are not yet a universal clinical test. Different methods can identify different ‘important’ nodes, and findings require validation against actual surgical and treatment outcomes.
Why propagation can differ from seizure to seizure
The same seizure focus does not always recruit exactly the same network. Sleep stage, recent seizures, medication levels, hormonal state, alertness and ongoing network activity can alter which pathways are available and how excitable downstream regions are.
This helps explain why one seizure may remain focal while another from the same person progresses further. Variation in spread does not necessarily mean that the seizure began in a different place.
Can propagation pathways change over years?
Chronic epilepsy can be associated with structural and functional network reorganisation. Recurrent seizures, underlying pathology, surgery, ageing and disease progression can all alter connectivity. In some people, the pattern of seizure spread therefore changes over time.
That does not mean every seizure progressively ‘damages’ the brain or inevitably spreads farther. Network changes are condition-specific and vary greatly between individuals.
Why propagation matters for treatment
Epilepsy surgery aims not only to remove or disconnect seizure-generating tissue but also to interrupt the network required for disabling seizures. In some cases, disconnective procedures can be effective even when not all abnormal tissue is removed.
Neuromodulation also takes a network approach. VNS influences widespread brainstem and cortical systems, DBS targets deep network nodes such as the anterior nucleus of the thalamus, and RNS records from selected sites and delivers stimulation when abnormal activity is detected.
What research still cannot predict perfectly
Current tests cannot predict the exact route of every seizure with certainty. Structural connectivity does not fully determine functional propagation, and the network state at seizure onset can change the result.
Much of the most detailed propagation research comes from people undergoing invasive monitoring for drug-resistant focal epilepsy. That evidence is crucial but cannot automatically be generalised to every generalised epilepsy, childhood syndrome or self-limited focal epilepsy.
What this page does — and does not — explain
This page is about the routes and mechanisms by which seizure activity recruits additional brain networks. It does not repeat the cellular generation of action potentials, the overall sequence of a seizure, the long-term development of epilepsy or the diagnostic limitations of scalp EEG. Those subjects have their own dedicated pages.
Sources
Jasper’s Basic Mechanisms of the Epilepsies, 5th ed. — mechanisms of focal seizure initiation, development and termination
Schevon et al. — Evidence of an inhibitory restraint of seizure activity in humans
Smith et al. — The ictal wavefront and propagation in human focal epilepsy
Jasper’s Basic Mechanisms of the Epilepsies, 5th ed. — thalamic mechanisms in epilepsy
Jasper’s Basic Mechanisms of the Epilepsies, 5th ed. — full reference work
Evidence checked: September 2026.