Perception of microstimulation frequency in human somatosensory cortex

Christopher Hughes, Sharlene N. Flesher, Jeffrey M. Weiss, Michael L. Boninger, Jennifer L. Collinger, Robert A. GaunteLife 2021 · DOI

TL;DR

This study reveals that the perceived intensity and quality of artificial tactile sensations evoked by electrical stimulation in the human somatosensory cortex are highly dependent on the stimulation frequency, and this relationship is specific to individual electrode locations. These findings are crucial for developing more natural and effective sensory feedback in brain-computer interfaces for individuals with paralysis.

What Problem Does This Paper Address?

Restoring sensory feedback is a critical component of developing effective bidirectional brain-computer interfaces (BCIs) for individuals with paralysis. These BCIs aim to allow users to control prosthetic limbs or other devices using their thoughts and, in turn, receive sensory information from the device back into their nervous system. Intracortical microstimulation (ICMS) in the somatosensory cortex is a promising method for delivering this sensory feedback, allowing for the artificial evocation of tactile sensations. However, a significant knowledge gap exists regarding how specific stimulation parameters, particularly frequency, influence the perceived intensity and quality of these artificial sensations. While previous research in animals and humans has shown that increasing stimulation amplitude generally increases perceived intensity, the effect of frequency has been less clear and potentially more complex. Understanding these relationships is essential for designing stimulation strategies that can deliver nuanced and naturalistic sensory experiences, thereby improving the functionality and user experience of BCIs.

Methods

The researchers implanted microelectrode arrays into the somatosensory cortex of two participants with cervical spinal cord injuries. They then delivered ICMS trains and systematically varied stimulation parameters:

  • Magnitude Estimation Task: Participants were asked to report the perceived intensity of the evoked sensations on a self-selected scale. This was performed by varying stimulus amplitude, pulse frequency, and train duration.
  • Detection Threshold Task: A two-alternative forced choice task was used to determine the minimum stimulus amplitude required to detect a sensation at a given frequency.
  • Surveys: Regular surveys were conducted to gather qualitative descriptions of the evoked percepts, including their location, quality (e.g., pressure, vibration, warmth), and naturalness.
  • K-means Clustering: Electrodes were categorized into groups based on their frequency-intensity relationships and reported perceptual qualities.
  • Spatial Analysis: The spatial distribution of these electrode categories was analyzed to determine if there was a relationship between electrode location and its frequency-dependent response.

Datasets

The study utilized data collected from two human participants (P2 and P3) with chronic implants. The data included:

  • Magnitude estimation data from multiple stimulation sessions.
  • Detection threshold data.
  • Qualitative survey data describing percept qualities.

No new benchmark datasets were introduced; the focus was on analyzing data from the participants' ongoing BCI experiments.

Key Insights and Ideas

  • Frequency is a powerful modulator of tactile perception: Unlike amplitude, which has a more linear effect on intensity, frequency has a complex, electrode-specific impact on both the intensity and quality of perceived sensations.
  • Electrode-specific "preferences": Electrodes could be broadly categorized into groups that preferred low, intermediate, or high frequencies for maximal perceived intensity. This suggests a degree of functional specialization at the electrode level.
  • Quality and intensity are linked: The categories defined by frequency-intensity relationships were remarkably similar to those defined by reported percept qualities, suggesting these are manifestations of the same underlying neural organization.
  • Spatial organization matters: The electrode categories were not randomly distributed but showed significant spatial clustering within the somatosensory cortex, hinting at a topographical organization of frequency encoding.
  • Human perception is key: The study highlights the limitations of animal models for understanding the qualitative aspects of sensory perception, underscoring the necessity of human studies for advancing BCI sensory feedback.

Key Results

  • Amplitude and train duration: Increasing stimulus amplitude and train duration consistently increased perceived intensity across all tested electrodes.
  • Frequency-intensity relationships:
    • For some electrodes, increasing frequency increased perceived intensity.
  • For over half of the electrodes, increasing frequency decreased perceived intensity.
  • Electrodes were categorized into three groups (LFP, IFP, HFP) based on their frequency-intensity profiles.
  • IFP electrodes generally showed the highest median intensity, while HFP electrodes showed the lowest.
  • Perceptual qualities: Different electrode categories evoked distinct percept qualities at different frequencies. For example, LFP and IFP electrodes evoked pressure and tapping at low frequencies, while HFP electrodes evoked buzzing and vibration at higher frequencies.
  • Detection: Higher frequencies improved the detectability of stimuli at perithreshold amplitudes, aligning with previous animal studies. However, at suprathreshold levels, this did not always translate to higher perceived intensity.
  • Spatial clustering: Electrode categories based on frequency preference and perceptual quality showed significant spatial clustering on the implanted arrays.

Limitations and Future Work

  • Single participant complexity: Most experiments were conducted with a single participant (P2) with a chronic implant, limiting generalizability. While findings were replicated in a second participant (P3), the scope was more limited.
  • Limited residual sensation: Participants had limited natural sensation, making it difficult to directly compare ICMS percepts with responses to natural tactile stimuli.
  • Cortical reorganization: Spinal cord injury can lead to cortical reorganization, which might influence responses.
  • Limited electrode coverage: The implanted arrays covered only a small region of the somatosensory cortex, limiting the creation of detailed cortical maps.
  • Categorical vs. continuous representation: The division of electrodes into discrete categories might oversimplify a potentially continuous spectrum of frequency preferences.
  • Layer-specific responses: The study did not differentiate responses based on cortical layers, which could influence percepts.

Future Work: * Investigate the neurophysiological mechanisms underlying these frequency-dependent responses, potentially through animal models. * Explore other stimulation parameters like pulse width modulation and multi-electrode stimulation. * Develop electrode selection strategies for BCIs based on identified feature preferences. * Utilize machine learning to more rapidly and accurately categorize electrode feature preferences. * Collect more data across additional participants and cortical regions to confirm and expand these findings.

Summary updated

Abstract

Microstimulation in the somatosensory cortex can evoke artificial tactile percepts and can be incorporated into bidirectional brain-computer interfaces (BCIs) to restore function after injury or disease. However, little is known about how stimulation parameters themselves affect perception. Here, we stimulated through microelectrode arrays implanted in the somatosensory cortex of two human participants with cervical spinal cord injury and varied the stimulus amplitude, frequency, and train duration. Increasing the amplitude and train duration increased the perceived intensity on all tested electrodes. Surprisingly, we found that increasing the frequency evoked more intense percepts on some electrodes but evoked less-intense percepts on other electrodes. These different frequency-intensity relationships were divided into three groups, which also evoked distinct percept qualities at different stimulus frequencies. Neighboring electrode sites were more likely to belong to the same group. These results support the idea that stimulation frequency directly controls tactile perception and that these different percepts may be related to the organization of somatosensory cortex, which will facilitate principled development of stimulation strategies for bidirectional BCIs.
BibTeX
@inproceedings{hughes2021perception,
  author = {Christopher Hughes and Sharlene N. Flesher and Jeffrey M. Weiss and Michael L. Boninger and Jennifer L. Collinger and Robert A. Gaunt},
  title = {{Perception of microstimulation frequency in human somatosensory cortex}},
  booktitle = {eLife},
  year = {2021},
  doi = {10.7554/elife.65128},
}
Figure