Single Neuron Recordings Research in Children: Ethical Considerations, Feasibility, Technical Aspects, and Scientific Opportunities

Abstract

Background:

Human intracranial recordings, and single neuron recordings in particular, have provided much knowledge on the mechanisms of human cognition and its impairment by disease. Improvements in recording technology, experimental design, and computational analysis methods have permit increasingly sophisticated understandings of uniquely human brain processes, including those underlying executive function, memory, and language. Despite the routine clinical use of intracranial recordings for invasive epilepsy monitoring in the pediatric population, there remains a significant gap between the associated research conducted in adult and pediatric neuroscientific investigation.

Summary:

Single neuron recordings in pediatric epilepsy patients are ethical, technically feasible, and safe. These data can provide mechanistic insights into the neurophysiology of the developing human brain.

Key Messages:

Routine use of invasive electrophysiological monitoring via SEEG studies in pediatric drug-resistant epilepsy offers opportunities to extend the utility of single neuron recordings to the pediatric population and advance our knowledge of the neuronal basis of behaviors in children.

1. Introduction

Intracranial electrophysiology research in humans has significantly advanced our understanding of human cognition and disease. The pioneering work by Wilder Penfield and Herbert Jasper on the functional anatomy of the human brain[1] famously initiated an effort to learn from the unique opportunities provided by neurosurgical intervention. Over the past 70 years, evolution of surgical technique and improvements in biomedical technology have facilitated an exponential increase in the number of intracranial recordings-based research studies (shown in Fig. 1), leading to, among many others, advances in our understanding of human language[2], memory[3,4], sensorimotor function[5–7], psychiatric disease[8,9], executive control[10–14], and brain-machine interfaces [15–18]. The corresponding increase in neurosurgical patient involvement has also fostered the development of bioethical initiatives ensuring the maintenance of integrity of clinical care and the voluntary aspect of research participation[19–23].

Fig. 1.

Evolution of intracranial neurophysiological research in adult and pediatric populations (1950–2024). PubMed database was searched to survey the number of publications for each recording technique from 1950 – 2024. A). The number of publications involving stereoelectroencephalography (SEEG) was obtained from the PubMed database and is shown using bars, with pediatric-focused SEEG studies indicated as a subset of the total. To compare the evolution of intracranial research with non-invasive neurophysiology techniques, the number of publications using electroencephalography (EEG, excluding SEEG and ECoG studies) and functional MRI (fMRI) are displayed as line plots. B). The number of Electrocorticography (ECoG)-related publications was gathered from the PubMed database and is shown as bar plots, with pediatric-specific ECoG publications highlighted within the total. EEG (excluding SEEG and ECoG studies) and fMRI publication trends are plotted as lines to provide a comparison of the evolution of intracranial research vs non-invasive techniques.

Altogether, human intracranial recordings-based research has become a well-established field of neuroscientific investigation. Large multidisciplinary teams of clinicians and non-clinician scientists collaborate to expand our understanding of the human brain; however, very few intracranial studies have focused on pediatric patients or the developing nervous system, despite the unique opportunities that routine invasive monitoring in pediatric patients provides. This paper reviews the current state of the field within pediatric intracranial recordings research, focusing on scientific advancements and ethical principles.

2. Stereoelectroencephalography (SEEG) in pediatric epilepsy

Approximately 1% of the global pediatric population is affected by epilepsy [24,25]. This includes 11 million children and adolescents worldwide, with an estimated 456,000 patients under 17 in the United States alone[26]. While most patients are successfully managed with anti-seizure medications (ASMs), 30–35% of them are characterized as drug resistant[27], defined by the International League Against Epilepsy (ILAE) as having failed two or more appropriately dosed and tolerated ASMs, either in tandem or individually[28]. In both adult and pediatric patients, those with a diagnosis of drug-resistant epilepsy (DRE) are often considered for surgical treatment options[29,30], as surgical intervention can offer dramatic reduction in seizure burden[30,31] with corresponding improvements in brain development[32,33], cognition[34,35], mortality risk[32,33,36,37], and quality of life[38,39]. Within the pediatric population, these benefits have led to the consensus that surgical intervention should be considered as early as possible to achieve optimal neurodevelopmental and survival outcomes[40,41].

The extensive preoperative workup for patients with DRE is intended to identify a seizure-onset zone (SOZ) that can be targeted for curative resection or ablation. This workup[42] includes the characterization of seizure semiology, noninvasive (scalp) electrophysiology, and structural and functional neuroimaging studies, including MRI, PET, and functional MRI (fMRI). In circumstances where an SOZ cannot be clearly identified with these techniques, or when there is discordance between a radiographic lesion and electrophysiology, invasive monitoring is indicated (see Fig. 2 for details). Historically, invasive monitoring (so called phase 2) studies involved a craniotomy with placement of subdural grid or strip electrodes (SDE); however, advances in surgical technique and biomedical technology have resulted in the widespread use of stereoelectroencephalography (SEEG), in both children and adults, given the superior safety and morbidity profiles of SEEG and the unique ability of SEEG to sample directly from deep structures such as the hippocampus, amygdala, and insula [43,44].

Fig. 2.

Surgical workflow for patients with drug-resistant epilepsy (DRE). Phase 2 invasive monitoring for DRE is indicated in cases with either a lack of concordance between structural imaging (such as MRI) and noninvasive (scalp) electrophysiology, or when there is evidence of seizure onset laterality/focality on EEG, but no identifiable focus on structural or functional neuroimaging. These phase 2 patients undergo intracranial electrode implantation for high resolution sampling of parenchymal activity, providing opportunity for research investigation during the associated hospital stay (see text for details). Abbreviations: SEEG stereoelectroencephalography; SDG subdural grid electrodes; RXN resection; RNS responsive neurostimulation.

SEEG involves stereotactic implantation of multi-contact cylindrical electrodes into brain parenchyma thus facilitating electrophysiological sampling across broad areas of the brain, such that a precise SOZ can be localized to inform therapy. Typically, these electrodes are implanted with a high degree of precision using MRI guided stereotactic frame or a robotic platform. SEEG is hypothesis-driven, meaning the implant strategy reflects a suspicion for the SOZ based on the non-invasive studies and seizure semiology. Patients are implanted in the operating room and are subsequently monitored in an epilepsy monitoring unit (EMU), where ASMs are tapered to enable capture of sufficient seizure activity to identify the SOZ. The duration of phase 2 monitoring depends on seizure frequency and can safely last up to 4 weeks, if needed.

As described above, SEEG has become the standard means of phase 2 invasive monitoring for the pediatric DRE population[43–46]. As of 2019, there are 198 epilepsy centers with NAEC level 4 accreditation status in the US [47], which requires the ability to perform invasive phase 2 monitoring[48]. A recent analysis of these centers revealed that 2,309 intracranial monitoring surgeries were performed by these centers in 2019 alone[47]. Notably, 68% of the NAEC level 4 centers treat pediatric patients and thus perform SEEG implantations in pediatric patients [49]. Given that there are approximately 47 pediatric epilepsy patients for every 290 epilepsy patients over the age of 18 (16%) [26], and that the incidence of DRE in pediatric epilepsy is the same as that for adults (30–35%), we estimate that there are upwards of 250 pediatric SEEG implantations in the US every year (0.68 × 0.16 × 2309).

The widespread adoption of SEEG, and the unique research opportunities it provides, has led to a significant increase in neuroscientific research involving SEEG recordings (see Fig. 1). These studies, however, are almost exclusively conducted with adult patients (notable exceptions discussed further below). The substantial underrepresentation of pediatric SEEG studies in the neuroscience literature warrants exploration.

2.1. Literature search methods informing Figure 1:

The following PubMed query was used to obtain publication counts: (Research Technique[Title/Abstract]) AND ((journal article[Publication Type]) OR (casereports[Filter] OR classicalarticle[Filter] OR clinicalstudy[Filter] OR clinicaltrial[Filter] OR researchsupportamericanrecoveryandreinvestmentact[Filter] OR researchsupportnihextramural[Filter] OR researchsupportnihintramural[Filter] OR researchsupportnonusgovt[Filter] OR researchsupportusgovtnonphs[Filter] OR researchsupportusgovtphs[Filter] OR researchsupportusgovernment[Filter] OR technicalreport[Filter])) AND (humans[Filter]) AND (1950/01/01:2024/12/31[pdat]) AND (english[Filter]). To isolate pediatric studies (child: birth – 18 years), the following filter was appended: AND (allchild[Filter]). (A) SEEG ((sEEG[Title/Abstract]) OR (Stereoelectroencephalography[Title/Abstract]) OR (Stereotactic electroencephalography[Title/Abstract]) OR (intracranial recording[Title/Abstract]). EEG: ((EEG[Title/Abstract]) OR (Electroencephalography[Title/Abstract])) NOT (sEEG[Title/Abstract]) NOT (Stereoelectroencephalography[Title/Abstract]) NOT (Stereotactic electroencephalography[Title/Abstract]) NOT (intracranial recording[Title/Abstract]) NOT (Electrocorticography[Title/Abstract]) NOT (Electrocorticography[Text Word]) NOT (Electrocorticography[Text Word]) NOT (Electrocorticography[Title/Abstract]) [AND (filters)]. fMRI: (fMRI[Title/Abstract]) OR (Functional magnetic resonance imaging[Title/Abstract]) [AND (filters)])(B) ECoG ((Electrocorticography[Title/Abstract]) OR (Electrocorticography[Text Word]) OR (Electrocorticography[Text Word]) OR (Electrocorticography[Title/Abstract])).

3. Ethical Considerations

The NIH BRAIN Initiative, European Union Human Brain Project, and China Brain Project (among others) ushered in a productive era of funding support geared toward an improved understanding of the human brain. With the corresponding increase in intracranial human recordings research has come the development of a body of work dedicated to the associated ethical nuances [20–23]. As described by Feinsinger et al.[19], there are three common elements to human intracranial recordings research that render it ethically unique. These are: 1) a goal of scientific understanding human brain function (involving direct study and manipulation of human cognition); 2) the absence of near-term therapeutic intent; and 3) the use of intracranial recordings and/or stimulation in human subjects requiring neurosurgical intervention. While some of this research involves patients undergoing a neurosurgical procedure as part of an investigational device trial [5], the majority of work with human intracranial recordings takes place in patients undergoing neurosurgical procedures as independently deemed necessary for their clinical care. We will focus here on the latter studies.

The two overarching ethical principles espoused by the Research Opportunities In Humans (ROH) Consortium of the NIH BRAIN initiative with respect to human intracranial recordings research arise from the Belmont Report principles of Beneficence and Respect for Persons[50]. These ethical pillars of the ROH are: 1) the maintenance of integrity of the clinical care space; and 2) ensuring the voluntariness of participation in intracranial research[19]. In short, recognition that patients undergoing invasive neurosurgical procedures are vulnerable to the possibility that any related research will impact the course of their clinical care, or that their willingness to participate will impact their care, is crucial. Extreme caution must therefore be taken when designing experiments and communicating the goals and procedures of research to patients as a part of the process when obtaining their informed consent. A detailed description of these considerations is provided by a recent ROH Consortium publication[19].

When it comes to intracranial recordings in pediatric patients, there is no substantial difference than when pursuing such recordings for adult patients; these principles also apply. Additional consideration, however, must be taken given the particular vulnerable nature of the pediatric population itself [51,52]. When undertaking research involving pediatric patients undergoing invasive neurosurgical procedures as a part of their clinical care, the following additional stringencies should be considered:

3.1. Safety

The Code of Federal Regulations (CFR) governing human subjects research in the United States outlines the ethical principles for research involving human subjects [53]. It is enforced by 20 federal agencies and includes a set of ethical standards and protections for child participants (45 CFR 46 subpart D [54]). Institutional Review Boards (IRBs) may only approve research with children that satisfies one or more of the conditions outlined in Table 1, which contextualize the research with respect to “minimal risk” and direct patient benefit.

Table 1.

Code of Federal Regulations (45 CF 46) Subpart D IRB Approval Criteria [53]

Section	Description
46.404	Research not involving greater than minimal risk.
46.405	Research involving greater than minimal risk but presenting the prospect of direct benefit to the individual subjects.
46.406	Research involving greater than minimal risk and no prospect of direct benefit to individual subjects, but likely to yield generalizable knowledge about the subject’s disorder or condition.
46.407	Research not otherwise approvable which presents an opportunity to understand, prevent, or alleviate a serious problem affecting the health or welfare of children.
46.408	Requirements for permission by parents or guardians and for assent by children.
46.409	Wards.

3.11. Minimal Risk

The language used in CFR Subpart D centers on the notion of “minimal risk,” which is defined by 45 CFR 46 as “the probability and magnitude of harm or discomfort anticipated in research is not greater in and of itself than ordinarily encountered in the daily life of the routine physical or psychological examination or test” [53]. This definition, more importantly, is also open to IRB interpretation and hence is not wholly objective. With respect to intracranial recordings research, we can separate investigations into two groups: 1) those in which standard SEEG or SDE are used as needed for clinical care and 2) those in which special research electrodes are implanted explicitly for research purposes.

3.111. Clinical electrodes:

Invasive monitoring performed in the pediatric population utilizes the same intracranial electrodes that are used in adults. These electrodes are implanted explicitly for clinical use. Estimates from the SEEG literature indicate a very low rate of hemorrhagic (1%), infectious (0.8%), and mortality (0.3%) complications[55], while that from the SDE literature demonstrates a higher overall complication rate (4% hemorrhagic, 2.3% infectious, 2.4% intracranial pressure elevation, 12.1% cerebrospinal fluid leak, with 3.5% of patients requiring additional surgery)[56]. Research that capitalizes on electrode recordings from clinically implanted electrodes poses no more than minimal risk, as defined by CFR Subpart D.

3.112. Research electrodes:

The “hybrid” macro/micro electrodes used at our institution (Behnke-Fried; Ad-Tech Medical[57]) are 1.1 mm in diameter compared to 0.8 mm standard SEEG electrodes (Spencer^® SEEG electrodes; Ad-Tech Medical). Although slightly larger in diameter, it is important to note that all intracranial electrodes used for deep brain stimulation (DBS) and responsive neural stimulation (RNS), in pediatric and adult patients, are between 1.27 and 1.41 mm in diameter[58,59], depending on the brand. DBS and RNS electrode safety profiles are similar to those for SEEG electrodes [60–62], and thus the size difference between research and clinical electrodes in this setting does not have any expected increases in risk [63–65].

Following implantation of the hybrid depth electrode in standard fashion [57,66,67], the microbundle is advanced through the hollow core, such that the microwires protrude 4–6 mm from the electrode tip into the gray matter (see Fig. 3). The microwires consist of 1 non-insulated reference wire and 8 insulated recording wires, each 40 microns in diameter[68]. The standard anchor bolt cap is then tightened to secure the electrode assembly.

Fig. 3. Behnke-Fried recording electrode.

(A) The Behnke-Fried electrode is a hybrid electrode consisting of a hollow outer probe with macro-contacts (top probe) through which a bundle of microwires can be inserted (bottom probe). (Inset) Magnified view of the assembled electrode with the micro-wires splaying out of the tip of the macro-electrode probe. (B) Illustration of the loaded Behnke-Fried electrode. (from left-to-right): splay of microwires that record single neuron activity, black bars denote macro-contacts in the brain, large gray bars denote output of the macro-signals which connect to the clinical and research systems, small gray bars denote output of the microwires which connect to the research system. (C) A Post-operative CT fused with a pre-operative MRI depicts two Behnke-Fried electrodes targeting the anterior cingulate. (D) Illustrative depiction of recording locations displayed on an Atlas brain in MNI152-space [114]. yellow=pre-Supplementary Motor Area, blue=dorsal Anterior Cingulate Cortex, purple=Orbitofrontal Cortex)

These research electrodes have been demonstrated in the literature as equally safe and effective for clinical monitoring when compared to standard SEEG electrodes[63,64]. Although these electrodes are only utilized if the patient consents to participate in research, they are implanted exclusively into clinically necessary regions that would otherwise house standard SEEG electrodes. Their equivalent safety profile thus renders the corresponding research as no more than minimal risk. Of note, the existing data on safety of hybrid electrodes is limited to adult patients, chiefly due to lack of research investigations in pediatric patients; however, we find no reason to suspect a difference in pediatric patients, especially given the similar complication rates for standard SEEG between children and adults [43,55,69,70]. That said, it is important that in the future safety studies focused on experience in children are conducted once enough experience has been accumulated.

3.2. Patient participation: informed consent, assent, and motivation

Obtaining informed consent is the primary means by which researchers demonstrate respect for autonomous decision-making in research subjects (Principle of Respect for Persons) [50]. As delineated in the Belmont Report, informed consent requires attention to three key elements: i) disclosure of adequate information; ii) voluntariness of participation; and iii) a capacity to understand the relevant information [50]. The first two elements are the focus of the ethical considerations for intracranial human recordings via the ROH Consortium [19]. Research involving pediatric patients, however, further requires attention to the vulnerability of children, which is characterized by their lack of capacity to understand informed consent information; their dependence upon, and close bonds with, their parents or guardians, such that they generally lack the ability to refuse participation; and that even resistance to participation may be dismissed due to their age and lack of understanding. Indeed, in recognition of their special vulnerability, for all research in children, 45 CFR 46.408 requires assent by the child in addition to permission by parents or guardians [54].

Specifically, according to 45 CFR 46.408, “the IRB shall determine that adequate provisions are made for soliciting the assent of the children, when in the judgment of the IRB the children are capable of providing assent. In determining whether children are capable of assenting, the IRB shall take into account the ages, maturity, and psychological state of the children involved. This judgment may be made for all children to be involved in research under a particular protocol, or for each child, as the IRB deems appropriate” [54]. In our practice, consent capacity is determined both by the clinical team and research team. Parental consent and child assent to participate in research during their stay and have research electrodes implanted is obtained prior to surgery by the surgeon-investigator. Likewise, for each research-related behavioral task/recording session, parental consent and patient assent are obtained by the research team, with an emphasis on voluntariness of participation and independence of their decision on clinical care [19].

One particularly unique challenge to working with pediatric patients is motivation. While patients may assent to participation in a given setting, disinterest or boredom can easily distract pediatric patients and lead to a premature termination of their research session. Adult patients will typically “indulge” the research team once they “commit” to a study via consent. Children, on the other hand, are generally not constrained by such social etiquette; they thus often have no problem saying “no” or avoiding participation altogether. That said, children will often follow the lead of their parents/guardians even if they would rather not do whatever is being proposed. The prospect of either raises the importance of paying particular attention to behavioral aspects of children, and when possible, providing them with opportunities to express themselves independently of their parents/guardians.

Navigating these issues and maintaining a balance between informed consent/assent, voluntariness of participation, and successfully engaging patients, we have developed several approaches that have proven fruitful. First, we obtain permission from the parents and assent from the patient separately, such that the child has ample opportunity to refuse or agree without pressure from the parents one way or another. In the epilepsy monitoring unit, we specifically ask parents outside of the room for consent to engage in the behavioral task; in the case of approval, we then approach the patient in the room without the parents to obtain assent. Second, to help the patient stay engaged, the behavioral task is “gamified” as much as possible to make it captivating and entertaining. A simple “change signal task [71]” for example, is transformed into a cartoonish game that is both intuitive/easily understandable and fun but at the same time still examines a scientific question rigorously. We further include a point system, whereby the subject can accrue points for successful trials within a session. They periodically see their “score” in comparison to their previous scores and those of other participants, which has effectively motivated them to keep playing the game/task. Lastly, our incentive model transforms the points they accrue into tickets they can exchange for small prizes (akin to an arcade). These techniques have resulted in robust participation from numerous patients aged 10 and up, with one patient asking to play the game each time he comes to the hospital for routine medical assessments unrelated to his invasive monitoring.

3.3. Justice

A final ethical consideration for intracranial recording research in pediatric patients centers on the Principle of Justice, which demands a fair distribution of the benefits and burdens of research [50]; by excluding children from intracranial recordings research, this principle is not observed in the case of children. Human intracranial recordings research in adults has yielded significant advancements in our understanding of neurophysiological phenomena, ranging from language production [72–74] to motor control [7,14]; however, we lack a similarly sophisticated understanding of the development of these phenomena or how they occur in the brains of developmentally immature humans. The brain changes dramatically from early life to adulthood, and thus we cannot assume that results from the adult population are translatable to children.

As described above, pediatric patients are routinely implanted with intracranial electrodes as a part of the standard surgical workup for DRE. We thus have an opportunity, and indeed an ethical obligation, to develop a more robust understanding of the immature brain, and hence to extend the benefits of intracranial recordings research as they translate into possible therapeutic advancements for the pediatric population.

4. Single neuron recordings in human neuroscience

Single-neuron recordings, or single-unit recordings, refer to the measurement of action potentials from individual neurons. In clinical settings, such recordings are extracellular, meaning the tip of the recording electrode is close to a neuron but does not penetrate it. This method allows investigators to monitor the spiking behavior of single neurons in vivo with millisecond temporal precision [75]. Typically acquired using fine-tipped microelectrodes inserted into cortical or subcortical structures, single-neuron recordings provide a direct window into the fundamental computational units of the nervous system [76], which are action potentials.

4.1. Neuroscientific advantages of human single neuron recordings

One major benefit of single-neuron recordings is their ability to capture the activity of individual neurons with high precision during behavior, enabling the study of brain networks, encoding schemes, and neural computations at the cellular scale. This resolution allows researchers to determine what sensory and cognitive variables are represented in a given brain area, and if so, in what format. Specifically, we can ascertain whether this representation is sparse, invariant, conjunctive, or abstract. By contrast, local field potentials (LFPs) measure slower voltage fluctuations resulting from of summed post-synaptic potentials within a local neuronal ensemble, generally in the range of a few hundred micrometers around the recording site with likely some contribution from volume-conducted signals up to several millimeters away [77]. Intracranial EEG (iEEG), including electrocorticography (ECoG), records field potentials at a coarser scale, integrating activity from thousands to millions of neurons. Importantly, LFPs are recorded from high impedance micro-wires, while iEEG is measured via low-impedance and comparatively large “macro” contacts on intracranial electrodes (SEEG or SDE) [78]. While LFPs and iEEG are invaluable for detecting population-level rhythms, synchrony, and oscillatory coordination, they lack the temporal and spatial resolution to resolve how neural microcircuitry contribute to these phenomena.

Single-neuron recordings have been central to the discovery of several foundational principles in cognitive neuroscience. In the following paragraphs, we briefly highlight a few select studies to illustrate the breath and significance of research that such recordings have enabled.

In memory research, they have revealed that neurons in the MTL are tuned to stimulus novelty and familiarity, often responding within a few hundred milliseconds of stimulus presentation [3,79]. Distinct populations of memory-selective neurons code for recognition strength and successful associative retrieval, and their activity correlates with subjective confidence during recall tasks [80]. Another key discovery on the neural substrate of human memory is that neurons in the medial temporal lobe (MTL) respond selectively and invariantly to abstract concepts, regardless of modality or specific visual presentation [81]. These “concept cells” provide a highly selective and sparse neural substrate for semantic memories and are some of the most direct evidence we presently have for the physical substrate of human memory. Similarly, neurons in the hippocampus and amygdala have been shown to signal stimulus novelty, memory strength, and episodic recall with trial-level specificity [80,82]. Recent work has also demonstrated the capacity of single neurons to learn and encode latent structural relationships.

Tacikowski et al. (2024) found that hippocampal and entorhinal neurons implicitly acquired the topology of a stimulus transition graph, adapting their firing to reflect relational proximity within an abstract temporal space—even in the absence of explicit instruction [83]. Such findings suggest that the hippocampal–entorhinal system integrates “what” and “when” information into predictive models of experience. Moreover, the format of single-unit representations can be geometrically analyzed to assess abstraction, such as whether task variables are encoded in disentangled subspaces that support generalization. For example, Courellis et al. (2024) demonstrated that hippocampal neurons encode orthogonal representations of context, stimulus identity, and outcome, a structure that enables flexible inference across latent task states [84]. In perception, researchers recorded hundreds of neurons across the depth of auditory cortex and found that different layers are tuned to distinct spectro-temporal features of speech [2]. Neurons responded selectively to groups of phonemes with shared articulatory or acoustic features, such as nasals or fricatives, and exhibited consistent tuning across repeated sentences.

Single neuron recordings have also been successfully employed in pursuit of greater pathophysiological understanding of epilepsy and its effect on cognition. Reed et al. (2020) investigated the effects of interictal epileptiform discharges (IEDs) on single neuron activity during declarative memory retrieval [85]. While IEDs have been previously linked with memory impairment in patients living with epilepsy [86–88], the mechanisms by which they impact the neural substrate of memory requires high resolution electrophysiological investigation that can only be provided by in vivo single neuron recordings. This work revealed that memory deficits occur as single neuron activity corresponding to declarative memory recall (of familiar images only) is directly interrupted by a passing IED. They further identified a more pronounced effect on putative inhibitory, rather than excitatory, neurons.

Similarly, Lee et al. (2021) recorded single neurons from the SOZ of patients with temporal lobe epilepsy and discovered that neurons within the mesial temporal lobe (MTL) of patients with right-sided SOZ demonstrated impaired activity during memory recall when compared to MTL neurons in patients with a left-sided SOZ [89]. These findings corresponded to behavioral performance, such that subjects with right-sided SOZ suffered impaired recollection during memory recall compared with those with left-sided SOZ. These studies shed light on the neural correlates of epilepsy-related cognitive impairment in a manner that cannot be discerned by noninvasive means.

4.2. Logistical considerations for successful single neuron recordings research

Single-neuron recordings in humans offer unmatched resolution but are constrained by technical and logistical challenges. First, electrode placement is dictated by clinical necessity rather than research design. Second, signal acquisition requires the use of high-impedance microelectrodes capable of isolating spikes from individual neurons [57,90], thereby limiting the number of probes available for semi-chronic recordings in the in-patient (EMU) setting. Several electrodes are available that offer single cell resolution, including the BF electrode (Behnke-Fried; Ad-Tech Medical[57] – see Fig. 3); the Dixi-Medical hybrid electrode (MICRODEEP^® SEEG Electrodes and MICRODEEP^® Micro-Macro Depth Electrodes; Dixi-Medical); Utah Array (Blackrock Neurotech [91]); and the Neuropixel probe [2,92,93]. The Ad-Tech BF electrode and; the Utah array and the Neuropixel probe are currently used for acute, intraoperative recordings, with the Utah array also utilized for chronic, long-term brain-computer interface technology[5–7,91], but are not practical for the EMU setting in most cases

Challenges associated with both acute and chronic single neuron recordings include motion artifacts from brain pulsation or patient movement, which can cause electrode drift, altering spike waveforms and complicating unit tracking [75]. Background electrical noise from clinical equipment, as well as epileptiform activity, can further degrade signal quality. Sparse firing or low responsiveness in some neurons may also reduce the yield of isolatable units. Technological advancements have helped address these limitations. Modern spike sorting algorithms like MountainSort [94] or OSort [95] accommodate waveform drift and overlapping signals, improving single-unit isolation. Real-time monitoring of spike quality and firing rates helps ensure recording fidelity across sessions. Although bedside task paradigms must remain brief and low-burden, large-scale datasets are feasible to acquire by pooling data across many patients into a pseudo-population, an approach extensively used in systems neuroscience broadly as well as in human intracranial work [96].

5. State of the art in intracranial recordings research in pediatric epilepsy

Despite the ubiquity of intracranial recordings in pediatric patients with DRE [43,44,97], there is a relative paucity of neuroscientific research that takes advantage of the associated opportunities when compared to adults (Fig. 1). A handful of studies, however, demonstrate the unique questions that can be addressed using intracranial recordings in the pediatric population. Table 2 provides references to the existing research topics that have been addressed via pediatric intracranial recordings.

Table 2.

Areas of inquiry well-suited for pediatric intracranial recordings research

Cognitive Phenomenon	Associated Pathology	Current iEEG Literature
Development of Attention	ADHD	[109]
Development of Memory	TBI, epilepsy, depression, anxiety	[98,99,145]
Development of Face Recognition	ASD, ADHD, epilepsy	[146]
Development of Language	ASD, TBI, epilepsy	[101–108,147–156]
Development of Social Interaction	ASD	[157]
Development of Cognitive control	Self-injurious behavior, OCD, TS	[9,158–160]

Abbreviations: ADHD: attention deficit hyperactivity disorder; ASD: autism spectrum disorder; OCD: obsessive-compulsive disorder; TS: Tourette’s syndrome; TBI: traumatic brain injury

A 2018 study by Johnson et al. utilized SDE recordings from the lateral prefrontal cortex in children aged 6–19 performing a memory encoding task to elucidate high resolution field potentials from iEEG governing memory formation in the developing brain [98]. Through spectral decomposition and trial-wise logistic regression, they showed that early and sustained high-frequency broadband activity (30–250 Hz) in prefrontal subregions predicted subsequent memory performance, with age-related increases in both spatial specificity and temporal precision. This finding illustrates the utility of intracranial recordings in tracking the maturation of task-relevant neural engagement with millisecond resolution.

A follow-up study extended these insights by recording from both medial temporal lobe (MTL) and prefrontal cortex (PFC) in a larger cohort [99]. Time-frequency analyses revealed dissociable slow and fast theta-theta coupling between MTL and PFC that predicted memory formation, with interregional synchronization evolving over development. Such fine-grained temporal dynamics—particularly cross-regional phase synchrony, which depends on the precise timing of neural oscillations at the millisecond scale—are only observable with intracranial data, as noninvasive methods like scalp EEG suffer from spatial smearing and signal contamination, and fMRI lacks the temporal resolution to track sub-second neural coordination [100].

Electrocorticography of language development has also been well-studied by the Wayne State group, revealing insights into activity patterns governing human speech as it matures from infancy into adulthood [101]. Gamma oscillation patterns within frontotemporal regions aligned to phoneme articulation [102,103], object naming [104], speech production [103,105–107], and infant cooing and babbling [108] have been identified using intracranial recordings from pediatric epilepsy patients. The developmental trajectories of these oscillations can be uniquely studied in the pediatric population, and this work serves as a foundation of the nascent field of ECoG of human language development.

In another study investigating the network-level dynamics of attention, Warsi et al. (2023) utilized LFP recordings from 13 children performing an attentional set-shifting task [109]. Using deep learning models trained on stimulus-locked power spectra, they identified beta and gamma oscillations in default mode and executive networks that were predictive of trial-level reaction time. Further analysis revealed anatomically precise localization of this activity to default mode network (DMN) subregions, thereby demonstrating the role of the DMN in attentional control. This work capitalizes on the superior spatial and temporal resolution offered by intracranial electrophysiology.

Investigating white matter physiology as a function of developmental stage, van Blooijs et al. employed cortico-cortical evoked potentials (CCEPs) to quantify the electrical transmission speed between distant and adjacent cortical regions via white matter tracts [110]. In 74 neurosurgical patients aged 4 to 51, they stimulated one cortical site and recorded evoked responses at distant sites using intracranial electrodes, allowing precise measurement of conduction delays based on the latency of early response components (e.g., N1 peaks). Their findings revealed that transmission speed across these fibers continues to improve into the third decade of life, with particularly protracted maturation in longer association pathways. These results demonstrate how direct electrophysiological recordings can chart the ongoing structural and functional refinement of inter-regional communication during human development—data that cannot be reliably inferred from functional MRI or diffusion imaging alone.

Notably, pediatric intracranial research has so far remained limited to standard (clinical) SEEG or SDE electrodes and the associated LFP and ECoG. While intraoperative microelectrode recordings (of individual neuron action potentials) that guide decision making during deep brain stimulation surgery have been described in pediatric patients [111,112], only one study at the time of this writing has analyzed electrophysiological features of single neurons for research purposes. Steinmetz et al. (2013) utilized intraoperative microwire recordings from hypothalamic hamartomas (HH) in children undergoing surgical resection to characterize firing rate profiles of single neurons within the lesion [113]. They identified two distinct neuronal populations based on firing rate and burst patterns and found that over half of all neuron pairs exhibited significant firing synchrony, suggesting that epileptogenic HH tissue harbors tightly coupled neuronal microcircuits. Their work leveraged neurosurgical access to a unique pediatric pathology to facilitate an improved understanding of epileptogenesis in the affected patient population.

These studies demonstrate how pediatric intracranial recordings-based neuroscience can offer unparalleled insights into multi-scale neurophysiological dynamics underlying unique components of the developing brain. However, there remains a tremendous opportunity for expansion of this nascent field

Conclusions and future directions

Human intracranial recordings research, and single neuron recordings in particular, offer a powerful means of interrogating human cognition. The rare opportunities provided by neurosurgical patients have been increasingly capitalized upon in the adult epilepsy and functional neurosurgery realm [23], yielding advances in our understanding of human language[2], memory[3,4], sensorimotor function[5–7], psychiatric disease[8,9], and executive control[10–14]. Although there are an estimated 250 pediatric patients who undergo phase 2 invasive monitoring for epilepsy evaluation every year in the US alone (see above), only a handful of research studies have attempted to answer neuroscientific questions unique to the pediatric population, which is a necessary step toward a better understanding and treatment of diseases of the immature and developing brain.

As we describe here, the ethical considerations for such work are well-established, and the biomedical Principle of Justice as described in the Belmont report should further motivate work in this field. Our successful implementation of a pediatric SEEG-based single neuron intracranial recordings research program, which is beyond the scope of this manuscript, represents a first step toward the expansion of the benefits of the associated research opportunities to pediatric patients.