Standardized EEG Protocol for Comparing Cortical Responses to Mechanical and Electrical Tactile Stimuli

Summary

This proof-of-concept protocol details EEG recording of somatosensory evoked potentials elicited by electrical and mechanical stimulation using oddball and roving paradigms.

Abstract

The development of realistic tactile feedback in haptic interfaces is contingent on understanding the neural processing of electrical versus mechanical stimuli. A protocol is presented for directly comparing the two modalities using electroencephalography (EEG). The protocol details procedures to achieve precise EEG-stimulator synchronization, calibrate subject-specific electrical and mechanical stimuli, and implement both oddball and roving paradigms to compare early event-related potential (ERPs) and the mismatch negativity (MMN) component. The proof-of-concept protocol was validated on four healthy participants, each assigned to a distinct experimental condition: mechanical or electrical stimulation of the right index fingertip under roving or oddball paradigms, manipulating duration (100 ms vs. 145 ms). Additionally, a 5^th participant underwent a roving paradigm with alternating modalities with all stimuli at 100 ms. Under unimodal stimulation, the roving paradigm proved more sensitive than the oddball in detecting somatosensory ERPs for both modalities and a discernible MMN-like component for electrical stimulation. The cross-modal comparison using the roving paradigm similarly revealed a clear MMN-like component but only for the electrical stimulus. These results demonstrate that the current protocol is suitable for correlating neural activity with perception and for directly comparing the cortical processing of artificial and natural sensations. This framework is a necessary step toward developing electrical stimulation protocols that elicit cortical responses indistinguishable from mechanical touch.

Introduction

A fundamental challenge in haptic technology is recreating natural tactile sensations through skin-machine interfaces. This requires replicating the precise spatiotemporal patterns of neural activity that natural stimuli evoke. While mechanical actuators can replicate physical interactions, epidermal electrical stimulation of mechanoreceptors presents a compact and versatile alternative¹. However, electrical stimulation often elicits unnatural or unpleasant paresthesia. A direct comparison of neural responses to electrical and mechanical stimulation represents a novel area of interest^2,3, raising the question: can refined electrical protocols evoke cortical responses that closely mimic natural mechanical touch?

Electroencephalography (EEG) provides an ideal tool to address this question by capturing cortical responses to somatosensory stimuli with high temporal resolution⁴. Critical metrics include early event-related potentials (ERPs), which reflect the initial cortical processing of stimulus features⁵, and the mismatch negativity (MMN) component^6,7. The MMN is a pre-attentive, change-detection response, typically elicited in an oddball paradigm, that signals a violation of sensory prediction when a sequence of repetitive standard stimuli is interrupted by a rare deviant stimulus⁸. This response is a robust electrophysiological correlation of sensory discrimination and memory. A complementary approach is the roving paradigm, which efficiently generates MMN by presenting short trains of a repeating stimulus; the first stimulus of a new train acts as a deviant due to its novelty, providing a more balanced number of standards and deviants than the classical oddball paradigm⁹.

Although MMN has been extensively characterized in auditory and visual domains, a standardized framework for comparing MMN across somatosensory sub-modalities is absent. Previous work has separately established reliable MMN responses to mechanical^6,7 and electrical stimuli^10,11,12. However, the lack of a direct, within-subject comparison using temporally precise and perceptually matched stimuli prevents meaningful conclusions. This methodological gap fundamentally limits the ability to assess whether artificially evoked electrical responses engage the same pre-attentive change-detection mechanisms as natural mechanical touch.

To address this methodological limitation, a standardized EEG protocol was developed to enable direct neural comparison of electrical and mechanical stimulation. The current method ensures: precise EEG-stimulator synchronization, participant-specific electrical calibration to a salient yet non-aversive level, application of a standardized mechanical stimulus, and implementation of both oddball and roving paradigms to compare ERPs and MMN across modalities.

The ability of each paradigm to elicit clear ERPs and MMN in response to mechanical and electrical stimuli of varying durations was first assessed. The oddball paradigm generated discernible ERPs only for electrical stimulation and failed to elicit MMN for either modality. In contrast, the roving paradigm successfully elicited ERPs for both modalities and a notable MMN-like component for electrical stimulation. Leveraging the sensitivity of the roving paradigm, a direct cross-modal comparison was conducted using alternating trains of mechanical and electrical stimuli. Clear ERPs were observed for both modalities, whereas a clear MMN-like component was elicited only by electrical stimulation. Although these initial results are based on a single participant per condition, they demonstrate the potential of this standardized framework to guide the development of electrical stimulation protocols that can eventually elicit cortical responses indistinguishable from those of natural mechanical touch.

Only recently have MMN responses to electrical and mechanical stimulation been compared directly, and so far, only using magnetoencephalography^12,13. The proposed protocol is, therefore, the first to provide a unified and detailed EEG framework for recording somatosensory MMN – one elicited by both electrical and mechanical stimulation and implemented as standard and deviant stimuli within the same experimental design. The key novelty of this protocol lies not in the EEG preparation itself, but in the integration of stimulation, recording, synchronization, and analysis into a comprehensive, end-to-end methodology that enables direct cross-modal comparison of somatosensory processing. The EEG preparation procedures follow established best practices and are based on previously published protocols¹⁴. While aspects of the EEG setup have also been demonstrated in video-based formats¹⁵, these approaches have not been combined with precisely controlled electrical and mechanical tactile stimulation, nor applied to MMN paradigms. By integrating these elements, the present protocol extends existing methodologies and provides a transferable framework for the mechanistic investigation of somatosensory prediction and deviance detection.

Protocol

The current experimental protocol was conducted in accordance with the ethical guidelines of the Sirius University Human Investigation Committee (Approval Date: 10.06.2025). Written informed consent is obtained from each participant prior to data collection. Exclusion criteria included the presence of implanted cardiac pacemakers or other electronic devices, a history of epilepsy or seizure disorders, pregnancy, or dermatological conditions at the stimulation site.

1. Multimodal stimulation system

1. Use a custom-built multimodal stimulator developed to deliver precisely controlled and synchronized somatosensory stimuli. The system consists of a control unit and a stimulation module, engineered for high-fidelity neurophysiological research.
   NOTE: While custom-built for this study, similar experimental requirements could also be met by commercial systems, such as the mini-shaker type 4810⁶ and Digitimer DS7A¹⁶.

2. Experimental set-up

1. Record EEG signals using a 64-channel active EEG system. Position the ground electrode at Fz (10–10 system) and place the reference electrode on the tip of the nose. Sample data at 10 kHz and apply a 1–40 Hz band-pass IIR filter. Synchronize stimulation and EEG systems via TTL triggers, with trigger durations set individually for each stimulus type (20-100 ms, in 20 ms increments). Perform offline data analysis using custom-written scripts implemented in MATLAB.
2. Perform a control task, in which mechanical stimuli were sent to the stimulator while the subject's hand remained outside the stimulating surface, with four subjects, to ensure that ERPs did not result from auditory noise produced by the vibrating actuator.

3. EEG preparation

1. Measure the participant's head circumference and identify anatomical landmarks (nasion, inion, and left/right preauricular points). Select the EEG cap size accordingly.
2. Position the cap so that the central electrode (Cz) is located at the vertex. Verify proper alignment by measuring the distances from Cz to the nasion, inion, and both preauricular points to ensure symmetry. Use the international 10-10 electrode placement system.
3. Connect all electrode leads to the EEG amplifier according to the manufacturer's specifications.
4. Prepare each electrode site by parting the hair, gently abrading the skin with abrasive gel, and cleaning with alcohol. Fill each electrode cavity with conductive gel. Apply the ground and reference electrodes first to ensure a stable connection.
5. Confirm that electrode signal quality meets manufacturer-recommended criteria, ensuring low noise levels and reliable recording of physiological activity, with impedance values (when available) within the acceptable range (usually below 10 kΩ).
6. Arrange electrode cables to minimize movement artifacts and secure them to the participant's clothing to reduce strain on connectors.
7. Seat the participant comfortably in front of a monitor. To support stable alertness during prolonged recordings, present a silent, neutral video.
8. Inspect the raw EEG signal to ensure consistent baseline noise across channels. Verify that physiological artifacts (e.g., eye blinks, muscle activity) are clearly identifiable.

4. Stimulation setup and threshold determination

1. Clean the stimulation site (e.g., the index finger of the dominant hand) with an alcohol swab to reduce skin impedance and ensure stable electrode-skin contact.
2. Position the finger on the stimulation pad. Ask the participant to maintain a relaxed and immobile hand posture throughout the experiment. Ensure that stable contact is achieved. If available, use the device's automatic baseline force registration to confirm steady pressure; otherwise, verify manually that the contact remains consistent.
3. Determine the individual sensory threshold for electrical stimulation using a standardized staircase procedure¹⁷.
   1. Begin stimulation at 1 mA and increase intensity in 0.5 mA increments until the participant first reports sensation.
   2. Decrease intensity by 0.5 mA, then increase in 0.1 mA steps to refine the threshold.
   3. Present 10 stimuli at each candidate intensity and define the threshold as the lowest intensity at which at least 5 out of 10 stimuli are detected.
   4. Repeat the staircase procedure 3 times and calculate the mean value as the sensory threshold.
4. Determine the discomfort threshold for electrical stimulation.
   1. Starting from the sensory threshold, increase intensity in 0.5 mA steps. Ask the participant to indicate when the stimulus becomes unpleasant or intolerable.
   2. Define the discomfort threshold as the last intensity tolerated below this level. Confirm tolerability by delivering 4-7 test pulses.
5. Select the final stimulation intensity for the experiment as a value between the sensory and discomfort thresholds, typically the arithmetic mean of the two.
6. Set the mechanical stimulation intensity to be clearly detectable without causing discomfort or visible finger movement (e.g., at twice the sensory threshold).

5. Preparation of stimulation sequences

1. General parameters
   1. Set the pulse width of both electrical and mechanical stimulation to 200 µs. Define the duration of standards as 100 ms and deviants as 145 ms, unless otherwise specified. Jitter the inter-stimulus interval (ISI) randomly between 3.0-5.5 s. Present each stimulus type at least 100 times.
2. Oddball and roving paradigms (electrical and mechanical stimulation)
   1. In the oddball paradigm, present sequences of 5-8 standards followed by a single deviant. Typical oddball blocks included approximately 100 deviants and 650 standards (750 stimuli in total).
   2. In the roving paradigm, present 5-8 consecutive standards followed by 5-8 deviants. Typical roving blocks included approximately 100 standard sequences and 100 deviant sequences (each sequence contains 5-8 stimuli, resulting in a total of approximately 1300 stimuli per block).
   3. Apply both paradigms with electrical and mechanical stimulation. Adjust the intensity of mechanical stimulation to be clearly perceivable while remaining below the discomfort threshold.
3. Combined electro-mechanical stimulation (roving paradigm)
   1. Define standards and deviants by modality rather than duration. Present both electrical and mechanical stimuli with identical durations of 100 ms to eliminate duration as a confounding factor.
   2. For this condition, utilize a roving sequence structure, with 5-8 consecutive standards followed by 5-8 deviants. In this design, the roles of standards and deviants alternate naturally within the same session, removing the need for additional counterbalancing across sessions.

6. EEG recording during stimulation

1. Explain the experimental procedure to the participant, noting that no behavioral response to the tactile stimuli is required. To help maintain a stable level of alertness, it is recommended to present a silent, neutral video; however, this is optional. Inform the participant that they may withdraw from the experiment at any time without consequences.
2. Instruct the participant to minimize blinking, facial muscle contractions, and head movement during recording.
3. Ask the participant to use earphones with active noise cancellation to prevent the brain responses from being affected by the noise of the vibrating actuator.
4. Provide short breaks every 10-15 min to reduce fatigue and maintain data quality. During these breaks, assess the participant's perception of stimulus intensity and adjust the intensity if significant changes in subjective perception are reported. Verify synchronization between the stimulator and the EEG acquisition system using TTL triggers.
5. Start EEG recording at a sampling rate of 10,000 Hz that is at least 2x the stimulation frequency.
6. Run the stimulation sequences (electrical, mechanical, or electro-mechanical) according to the assigned paradigm (Table 1).
7. Collect a minimum of 100 deviant trials per condition. Adjust the number of standard trials accordingly, maintaining a ratio of approximately 5-8 standards for each deviant.
8. In addition to the main experiment, run the control sequence with stimuli sent to the stimulator normally, but the hand of the subject is placed outside of the stimulating surface, showing conditions where no ERP/MMN is expected, confirming that the observed effects are protocol dependent.

7. EEG data preprocessing and analysis

1. Remove artifacts by inspecting the continuous EEG and removing channels with persistent noise contamination. Exclude time segments affected by technical breaks or excessive muscle activity.
2. Correct baseline offsets channel by subtracting the mean value of each channel from every time point to correct for baseline offsets.
3. Apply filtering using a 1-40 Hz band-pass fourth-order IIR filter to the continuous EEG data.
4. Segment the data into epochs time-locked to stimulus onset (−1000 ms to +1000 ms).
   NOTE: In our experience, using a symmetric 1 s pre- and post-stimulus window allows for a more detailed assessment of data quality, facilitates the application of additional frequency filtering without affecting the critical baseline period, and enables the detection of stimulus-related synchronization or desynchronization of ongoing neural rhythms.
   1. For unimodal stimulation (electrical or mechanical oddball/roving paradigms), define standard and deviant epochs according to stimulus duration (100 ms vs. 145 ms).
   2. For multimodal stimulation (combined electro-mechanical roving paradigm), define standard and deviant epochs according to stimulus modality (electrical vs. mechanical, both with 100 ms duration).
5. Reject contaminated epochs by discarding those containing amplitudes exceeding +50 µV or falling below −50 µV within the −1000 to +1000 ms interval.
6. Adjust baseline of epochs by subtracting the mean value in the −200 ms to −50 ms pre-stimulus interval.
7. Average waveforms by computing mean traces across epochs of the same type (standard or deviant) within each condition.
8. Visualize ERPs by plotting ERPs for individual channels or regions of interest, focusing on the centro-temporal regions of interest within the 150-350 ms post-stimulus time window.
   Contralateral hemisphere: FC5, FT7, C3, C5, T7, CP3, CP5, CP7, P3, P5, P7.
   Ipsilateral hemisphere: FC6, FT8, C4, C6, T8, CP4, CP6, TP8, P6, P8, P4.

Results

Oddball paradigm
The presentation of rare deviant stimuli (145 ms) among frequent standard stimuli (100 ms) during mechanical stimulation failed to elicit clear and reliable ERPs. Consequently, no significant difference was observed between standard and deviant responses (Figure 1). In contrast, the electrical oddball paradigm generated discernible bilateral ERP waveforms. However, comparing standard and deviant stimuli still did not yield a significant or consistent MMN...

Discussion

This work establishes a standardized EEG protocol for the direct, within-subject neural comparison of artificial electrical and natural mechanical tactile stimulation. The core methodological contribution is a framework that ensures precise temporal control, participant-specific perceptual calibration, and the application of different experimental paradigms to assess cortical responses. The initial proof-of-concept results, while based on a limited sample, demonstrate the protocol's efficacy and highlight its potenti...

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
actiCHamp Plus Versatile all-in-one lab amplifier	Brain Products GmbH	S/N ACBM17120955	Primary EEG data acquisition system
actiCAP slim / snap	Brain Products GmbH	BP-185-2200	Active electrode cap with 64 channels
Digitimer DS7A	Digitimer Ltd	UDI-DI 5060490360003	Electrical stimulator
MatLab	The MathWorks, Inc., Natick, MA, USA
Mini-shaker type 4810	Brüel & Kjær	BP-0232	Mechanical stimulator
Photo Sensor	Brain Products GmbH	N/A	For optical detection of stimulus onset
Shielded-room Neiroiconica "Expert	Neiroiconica Assistive Ltd	S/N 20210119011	Electromagnetically shielded chamber
TriggerBox Plus	Brain Products GmbH	S/N TB-A100009-0624	Synchronization interface for TTL triggers