The hippocampus is a curved structure buried deep in the medial temporal lobe. It is best known for its role in memory, but in epilepsy it is also important because it contains densely interconnected excitatory and inhibitory circuits that can support seizure generation and propagation.
The hippocampus is not a single uniform block of tissue. It contains several specialised regions — including the dentate gyrus, CA3, CA2, CA1 and the subiculum — connected in organised pathways. Their normal job is to process and relay information. In epilepsy, those same recurrent connections can become part of a seizure-generating network.
Where the hippocampus sits
There is one hippocampus in each hemisphere, extending through the medial temporal lobe. It lies close to the amygdala, entorhinal cortex, parahippocampal cortex and other limbic structures. These regions communicate extensively with one another and with widespread cortical areas.
Because of this position, the hippocampus can participate in networks involving memory, emotion, spatial orientation, autonomic function and conscious experience. A seizure affecting hippocampal circuitry can therefore produce much more than a simple memory symptom.
The hippocampus is central to forming new memories
The hippocampus is particularly important for forming and consolidating new declarative memories — memories for facts and events that can later be consciously recalled. It also helps organise spatial and contextual information.
It is not the permanent storage site for every memory. Over time, many memories become increasingly represented across distributed cortical networks. The hippocampus is better thought of as a system that helps bind together the elements of an experience and supports their later retrieval.
The hippocampal circuit
A simplified version of the classic hippocampal circuit begins with input from the entorhinal cortex to the dentate gyrus. Granule cells then project to CA3 pyramidal neurons through mossy fibres. CA3 projects strongly to CA1 through Schaffer collateral pathways, and CA1 communicates with the subiculum and entorhinal cortex.
This description is intentionally simplified. Real hippocampal connectivity includes numerous recurrent, inhibitory and extrahippocampal pathways. Those additional connections are important because epilepsy often involves a broader temporal-limbic network rather than one isolated hippocampal loop.
Why CA3 is especially interesting in seizure research
CA3 pyramidal neurons form extensive recurrent excitatory connections with one another. Under normal conditions, those connections contribute to memory processing and pattern completion. In experimental epilepsy models, however, recurrent excitation can also support highly synchronous population activity.
This does not mean CA3 is the universal source of temporal lobe seizures. Human temporal lobe epilepsy is heterogeneous, and seizure-generating networks can involve the hippocampus, subiculum, amygdala, entorhinal cortex and other temporal structures in different combinations.
The dentate gyrus as a gate
The dentate gyrus is often described as a gate controlling the flow of excitatory information into the hippocampal network. Its granule cells are normally relatively difficult to activate compared with many other neurons, and strong inhibitory control helps prevent excessive excitation from propagating through the hippocampus.
In some forms of temporal lobe epilepsy, structural and functional reorganisation can weaken this filtering role. Mossy-fibre sprouting, altered inhibition and changes in intrinsic excitability have all been studied as mechanisms that may reduce the effectiveness of the dentate gate.
The subiculum is more than an output relay
The subiculum sits between the hippocampus and several connected cortical and subcortical regions. It was once treated mainly as a passive output station, but modern research shows that it has its own organised microcircuits and intrinsic bursting properties.
Human tissue studies and experimental work suggest that the subiculum can participate actively in focal epileptic activity, particularly when hippocampal sclerosis has altered neighbouring networks. It may therefore help transmit or generate abnormal activity rather than simply passing it onwards.
Why the hippocampus can be seizure-prone
Several features make hippocampal circuits biologically capable of sustaining epileptic activity: recurrent excitatory connections, strong synchronising interneuron networks, plastic synapses, sensitivity to changes in extracellular potassium and chloride, and close communication with other limbic structures.
Those same properties are essential for normal learning and memory. The hippocampus is therefore a good example of how a circuit designed for flexibility and synchronisation can become vulnerable when inhibition, connectivity or cellular properties are altered.
Interictal activity and the hippocampus
Between seizures, epileptic hippocampal networks can produce interictal epileptiform discharges. At the cellular level, researchers have long studied paroxysmal depolarising shifts and related population events as mechanisms underlying interictal spikes.
Interictal activity is not identical to a clinical seizure. A network can generate spikes between seizures without progressing into a sustained ictal state. Why some abnormal events stop while others recruit a full seizure remains an active area of research.
What can a hippocampal seizure feel like?
Seizures involving medial temporal networks can produce experiential symptoms such as déjà vu, jamais vu, sudden familiarity or unfamiliarity, a rising epigastric sensation, fear, altered perception, memory fragments or changes in awareness.
These symptoms are not uniquely hippocampal. The amygdala, insula, temporal neocortex and other connected regions can produce overlapping phenomena. Semiology can suggest network involvement, but it cannot identify a seizure focus with perfect precision.
Why memory can fail during a seizure
Forming a memory requires coordinated communication between the hippocampus and distributed cortical networks. During a seizure, that coordination can be disrupted. A person may remain partly responsive yet fail to encode what is happening into a durable memory.
This helps explain why some people can perform actions or speak during a seizure but later remember little or nothing about the event. The absence of memory afterwards does not prove that awareness was completely absent at the time.
Why memory problems can continue between seizures
Memory difficulties in temporal lobe epilepsy are not caused by one mechanism. Structural pathology, recurrent seizures, interictal discharges, sleep disruption, antiseizure medicines, mood, attention and the underlying epilepsy network can all contribute.
Frequent interictal activity can transiently interfere with cognitive processing, particularly when it occurs during memory encoding or retrieval. Repeated seizures can also disrupt sleep and recovery, which indirectly affects learning and recall.
Left and right hippocampi are not simply ‘verbal’ and ‘visual’
Clinical neuropsychology often finds stronger associations between the language-dominant medial temporal lobe and verbal learning, and between the non-dominant side and some visual-spatial memory functions. But the division is not absolute.
Memory is distributed and individual language lateralisation varies. A person’s cognitive pattern depends on seizure onset, pathology, age at onset, education, functional reorganisation and the health of the opposite hippocampus.
Bilateral hippocampal networks matter
The two hippocampi communicate indirectly through commissural and cortical pathways. A seizure beginning in one temporal lobe can recruit the opposite temporal network, while long-standing epilepsy can produce functional changes beyond the original focus.
This is one reason presurgical assessment examines both sides. Removing or disconnecting tissue that is important for memory carries different risks depending on how well the opposite medial temporal lobe can support function.
The hippocampus and awareness
The hippocampus alone is not an ‘awareness centre’. Conscious awareness depends on distributed cortical and subcortical systems. However, temporal-limbic seizures can disrupt those wider networks, especially when activity spreads beyond the hippocampus.
This helps explain why some medial temporal seizures preserve interaction initially and then progress to impaired responsiveness as additional networks are recruited.
Hippocampal sclerosis
Hippocampal sclerosis is a recognised pathological pattern associated with many cases of temporal lobe epilepsy. It involves characteristic neuronal loss and gliosis, with different patterns of damage recognised histologically.
This page does not treat hippocampal sclerosis in depth because it deserves its own structural epilepsy article. The future dedicated page will cover pathology, MRI appearance, ILAE histological classification, relationship with temporal lobe epilepsy, surgical relevance and the question of cause versus consequence.
The hippocampus after prolonged seizures
Prolonged seizures and status epilepticus can produce major metabolic and ionic stress in hippocampal networks. Experimental models show cell loss, inflammation, receptor changes, altered neurogenesis and synaptic reorganisation after severe prolonged seizure activity.
However, it is not accurate to assume that every brief seizure damages the hippocampus. The biological consequences depend on seizure type, duration, frequency, age, underlying pathology and many other factors.
The hippocampus and sleep
Normal hippocampal memory processing continues during sleep. Hippocampal sharp-wave ripples participate in memory consolidation, while sleep also changes the synchronisation of epileptic networks.
Epileptic activity during sleep can therefore intersect with systems that are normally involved in consolidating memories. This is one possible contributor to learning difficulties in some people with frequent nocturnal epileptiform activity, although the relationship varies between epilepsy syndromes.
High-frequency oscillations
High-frequency oscillations can be recorded from hippocampal and other brain networks. Some are normal physiological events, including hippocampal ripples involved in memory processing. Others, particularly certain pathological high-frequency oscillations, are being studied as biomarkers of epileptogenic tissue.
The challenge is distinguishing pathological activity from normal high-frequency physiology. Frequency alone does not provide a perfect answer.
Why hippocampal epilepsy can be difficult to see on scalp EEG
The hippocampus lies deep within the temporal lobe. Small or highly localised discharges may not generate a strong scalp signal. Activity may need to spread into more superficial temporal cortex before it becomes clearly visible on scalp EEG.
This is one reason a normal routine scalp EEG cannot exclude temporal lobe epilepsy and why prolonged video-EEG, specialised electrodes or intracranial monitoring may sometimes be required in presurgical assessment.
Why MRI matters
Dedicated epilepsy MRI can identify hippocampal atrophy, signal abnormalities, malformations and other structural changes that may be relevant to temporal lobe epilepsy. But not every epileptogenic hippocampus looks abnormal on routine MRI.
Imaging therefore has to be interpreted alongside seizure history, EEG, neuropsychology and other functional evidence.
The hippocampus in epilepsy surgery
In carefully selected drug-resistant temporal lobe epilepsy, surgery may involve removing or ablating part of the medial temporal seizure-generating network. The potential for seizure control has to be balanced against memory and language risks.
Presurgical assessment therefore asks two separate questions: which tissue is necessary for the seizures, and which tissue is necessary for the person’s remaining cognitive function?
Why the same hippocampal lesion can affect people differently
Two people with similar-appearing structural abnormalities can have different seizure patterns and different memory profiles. Functional reserve, age at onset, seizure frequency, bilateral network health and cortical reorganisation all influence outcome.
This is another reason epilepsy cannot be understood from MRI appearance alone.
What research still does not know
Major questions remain about why some hippocampal abnormalities generate seizures while others do not, which microcircuits initiate human temporal seizures, how pathological and physiological high-frequency activity can best be distinguished, and how to preserve memory while disrupting epileptic networks.
Much of the most detailed human evidence comes from people with drug-resistant temporal lobe epilepsy undergoing surgery or intracranial monitoring. Those findings are powerful but should not automatically be generalised to every epilepsy involving memory symptoms.
What this page does — and does not — explain
This page explains the hippocampus as a normal memory structure and as part of epileptic networks. It does not duplicate the broader Memory & Cognition page, the Temporal Lobe Epilepsy clinical overview, or the future dedicated article on Hippocampal Sclerosis.
Sources
Jasper’s Basic Mechanisms of the Epilepsies, 5th ed. (2024)
Houser — Hippocampal Sclerosis in Temporal Lobe Epilepsy: New Views and Challenges
Evstratova et al. — Role of the Subiculum in Focal Epilepsy
Prince & Avoli — The Paroxysmal Depolarizing Shift
Evidence checked: September 2026.