The electrode you choose is a research design decision, not a hardware spec

Every mobile EEG study eventually runs into the same problem: the electrode that produces the cleanest signal is rarely the one participants can comfortably wear, discreetly, and for as long as the study requires. Wet, dry, semi-dry, around-the-ear, and in-ear electrodes are not simply competing technologies, nor is one universally "better" than another. Each represents a different balance between signal quality, wearability, setup time, and spatial coverage. The right choice depends not on the hardware itself, but on the research question you are trying to answer.

Wet electrodes: still the reference standard

Gel-based electrodes remain the gold standard for EEG acquisition. Their low and stable contact impedance provides excellent signal quality across a wide frequency range, and they are widely used as the benchmark against which other electrode types are evaluated (Advancements in dry and semi-dry EEG electrodes, 2025).

Outside the laboratory, however, the same characteristics become practical limitations. Wet electrodes require skin preparation and gel application, are visually conspicuous, and generally need a trained operator to achieve reliable recordings (Hinrichs et al., 2020). Over extended sessions, the gel gradually dries, increasing impedance and reducing signal stability. For tightly controlled laboratory experiments, these drawbacks are often acceptable. For naturalistic, ambulatory, or longitudinal studies, they can become major obstacles.

Dry electrodes: faster setup, but performance depends on the paradigm

Dry electrodes eliminate the need for conductive gel, making them particularly attractive for repeated measurements, self-administered recordings, and studies outside the laboratory. Their primary limitation is higher and less stable contact impedance, which generally increases the recording noise floor.

Importantly, a higher noise floor does not automatically make dry electrodes unsuitable. Their performance depends largely on the neural features being measured. Resting-state alpha and beta power have been found not to differ statistically between dry and wet systems, and mean latencies and amplitudes of the P100 visual evoked potential and the P3 showed comparable values with similar spatial distributions across both (Hinrichs et al., 2020).

More subtle or lower-amplitude components, however, are often more sensitive to the increased impedance and motion-related artefacts associated with dry electrodes. In one direct comparison, dry EEG still reliably detected mismatch negativity, but underestimated its mean amplitude, peak latency, and theta power relative to wet EEG, and low-frequency measures showed the largest divergence (Comparison of dry and wet EEG, 2024). The most accurate conclusion is therefore not that dry electrodes are "comparable" or "inferior," but that their suitability depends on the physiological signals your study relies on.

Semi-dry electrodes: a promising middle ground

Semi-dry electrodes aim to bridge the gap between wet and dry technologies by using a small reservoir of saline or hydrogel instead of a conventional gel application. The goal is to combine the signal quality of wet electrodes with the convenience and rapid setup of dry systems.

Reported designs support that ambition. One ceramic-based semi-dry electrode showed impedance rising by only about 20 kΩ over eight hours of continuous use (Li et al., 2016), and a flexible multi-layer design reported temporal correlation with wet-electrode signals of roughly 94 percent under static and 91 percent under dynamic conditions at hairy sites (Li et al., 2019). Although the technology is still evolving and results vary considerably between designs, semi-dry electrodes represent one of the most promising compromises currently available for mobile and longitudinal work.

cEEGrid (around-the-ear): unobtrusive recording with targeted spatial coverage

Around-the-ear electrode arrays such as cEEGrid trade whole-head spatial coverage for something conventional scalp EEG cannot easily provide: unobtrusive recording. Positioned around the ear and often concealed beneath hair or the arms of a pair of glasses, cEEGrid enables multi-hour or even all-day recordings with far less social conspicuousness than a traditional EEG cap.

Naturally, this comes with reduced spatial coverage. Whether that limitation matters depends on the research question. In a direct comparison with high-density cap EEG during a visual Simon task, N1, P1 and P300 waveforms extracted from cEEGrid were strongly correlated with the corresponding cap waveforms, though with lower signal strength and lower signal-to-noise ratio (Bleichner et al., 2016). A broader review has since confirmed that the auditory N100, MMN, P300 and N400 can be recorded reliably from around-the-ear electrodes, with expected signal loss relative to standard scalp positions (Meiser and Bleichner, 2022). In a spatial auditory attention paradigm, single-trial classification accuracies of around 70 percent were achieved with cEEGrid, comparable to those from high-density cap EEG (Debener et al., 2017).

Paradigms dominated by central or motor cortical activity are considerably harder to capture reliably from around the ear, though not always out of reach: in the same Simon task comparison, event-related lateralizations recorded at posterior scalp sites were well reflected in middle cEEGrid pairs, and the effect size of the Simon correspondence effect was similar between the two systems (Bleichner et al., 2016). cEEGrid is therefore well suited to studies involving auditory processing, temporal cortical activity, sleep, and many everyday cognitive applications, and requires more careful validation when accurate measurement of central sensorimotor activity is required.

In-ear EEG: maximum concealment, minimal spatial coverage

In-ear EEG takes unobtrusive recording one step further. Integrated into custom ear molds or hearing-aid-like devices, these electrodes are almost invisible during everyday use and can comfortably support long-term ambulatory recordings.

The trade-off is a further reduction in spatial coverage and signal amplitude. Nevertheless, ear-centred configurations have been used to capture subcortical auditory measures including auditory brainstem responses and envelope-following responses (Bech Christensen et al., 2018), typically at lower amplitudes that require more extensive averaging and careful artifact handling. Their discreet form factor makes them particularly attractive for studies where participant comfort, compliance, and ecological validity matter more than comprehensive cortical coverage.

The better question is not which electrode is best, but which electrode best fits the scientific question, the recording environment, and the participants.

The actual decision tree

Rather than asking which electrode technology is "best," it is often more useful to work backwards from the scientific question and the practical constraints of the study.

Ask yourself:

What neural signal does the study depend on? Resting-state oscillations and large cognitive ERPs generally tolerate higher impedance and fewer electrodes much better than fine time-locked components or motor-related potentials.

How long will participants wear the system, and where? A 45-minute laboratory experiment can accommodate gel application and technician support. A multi-day home study or all-day workplace recording usually cannot.

Does the recording need to be socially unobtrusive? If participants are expected to go about their normal daily activities, visibility is more than a comfort issue. It influences behaviour itself. The more conspicuous the recording system, the less natural the behaviour you are attempting to measure.

How much spatial coverage does the research question actually require? Studies involving motor cortex or widespread cortical networks often require conventional scalp EEG. Auditory, temporal, sleep, and many affective or arousal-related paradigms can often be addressed with much more localized recordings than many researchers initially assume.

The familiar "wet versus dry" debate reduces electrode selection to a single trade-off between setup time and signal quality, while implicitly assuming that every study requires laboratory-grade signal-to-noise ratios and whole-head spatial coverage. Many real-world EEG studies simply do not. The better question is not Which electrode is best? but rather Which electrode best fits the scientific question, recording environment, and participants?

Ultimately, electrode selection is less a hardware decision than a research design decision.