Human neuromarkers of tactile perception: state of the art in methods and findings

Abstract

Tactile perception is a multifaceted sense with complicated convergent/divergent peripheral pathways. Its neuromarkers remain poorly understood, due to the sense’s inherent complexity and the confounding factor of intricate motor, cognitive and affective correlates. This gap hinders research evaluating interventions to restore touch in artificial hands. We inventorize state-of-the-art and recent innovations in control systems with soft and hard robotics that are poised to unlock more targeted non-invasive stimulations. We review neuromarkers observed for pressure, vibration, brushing, texture discrimination, pain, heat and cold, complemented with the covariates from movement, attention, working memory, multisensory and sensorimotor integration or competition (audition, vision) and affect. We analyze neural oscillations during sensory and (peripheral and central) electro-magnetic stimulation. This review matures a framework of reverse prediction, in which non-invasive observation of neural activity robustly and unobtrusively quantifies tactile perception.

I. Introduction

Several research groups, including our own, are working toward the development of artificial hands that restore the sense of touch, in an effort to facilitate dexterity and fluidity of daily manual actions for amputees and limb absent users of prosthetic arms, and to support remote manual control in teleoperation. The development of technologies to restore afferent pathways calls for tools to monitor tactile perception continuously, unobtrusively, and non-invasively. A preferred site to achieve this monitoring lies with the brain: it is the destination of afferent tactile signals, with slightly better spatial separation of somatosensory and motor activities than what is achievable noninvasively in the peripheral nervous system, where both afferent (sensory) and efferent (motor) pathways are blended in shared nerves. To monitor tactile perception unobtrusively via non-invasive recording of brain activity (e.g. electro- and magneto-encephalography, EEG and MEG [10]), it is necessary to understand which neurophysiological activities correlate with qualitative and quantitative variations of tactile sensations. The non-invasive neuroscience of touch, however, remains less well understood than other senses, for a plurality of reasons. First, instrumentation to stimulate the various tactile experiences with temporal, spatial and functional precision remains a work-in-progress. Second, the sense of touch possesses a large and diverse array of receptors working synergistically. Those receptors’ complicated convergent and divergent connectivity to the brain compounds the usual difficulties with teasing apart perception from its cognitive, behavioral, and environmental modulators. And finally, naming of electrophysiological activities confuses syntheses (viz. identical names to activities from distinct brain networks, and distinct names for activities that pertain together). In the following presentation, we aim to accomplish four goals: (1) review technological developments supporting the disentanglement of tactile perception’s multiple facets, leading to breakthrough neurophysiological studies; (2) identify neuromarkers of somatosensation and (3) their cognitive, affective, motor and environmental covariates; and (4) list bottlenecks to resolve in order to establish robust neuromarkers of tactile perception.

II. Technology review: tactile stimulation paradigms

A variety of techniques has been used to apply tactile stimulations and quantify associated brain responses. Studies span the simplest approaches (e.g. brush strokes applied onto a subject’s arm) into engineered control systems that leverage mechanical, thermal, optical, acoustic, or electrical stimulations to improve temporal, spatial and functional precision. All sites along the pathways from sensory receptor to brain have been targeted: stimuli applied to the skin to recruit tactile receptors; electrical or ultrasonic stimulation applied to afferent nerves to induce peripheral activities mimicking the response to natural stimuli (directly or with co-optation of muscles); electrical or magnetic stimulations directly aimed at somatosensory brain areas to confirm elicitation of tactile experiences.

III. Neuromarkers of somatosensation

With the aforementioned tools, considerable evidence implicates 6 oscillatory activities of the brain in somatosensation (Fig.1). Mu (also called Rolandic alpha, a 8–12Hz activity over somatosensory/motor cortex) and Beta (13–30Hz, a slightly more anterior activity toward motor cortex and putatively involved in sensorimotor integration) undergo suppression and rebound during and after somatosensory stimulation and the direct electrical stimulation of peripheral nerve. Their sources were tied to somatosensory cortices, and reciprocally, stimulation of somatosensory areas with Mu frequencies elicited tactile percept [5]. Prestimulus Mu also modulated perception of faint tactile stimuli [6].] Theta (4–7Hz, fronto-central) and Gamma (30–100Hz, central regions of the brain) increased during somatosensory events [8], [9], and electrical stimulation of the somatosensory cortex at Gamma frequency elicited tactile sensations [5]. Finally, transient ultra-fast activities, Sigma (450–750Hz) and Kappa bursts (850–1200Hz), were discovered with Median Nerve Stimulation and seemed specific of afferent tactile information [3], [4], though few studies aid their functional understanding due to the great technical challenge of their recording.

Fig. 1.

Overview of EEG/MEG neuromarkers examined in this work.

IV. Cognitive, Behavioral, Affective and Environmental Covariates

To reverse inferences and unobtrusively monitor the quality of artificially-restored somatosensory feedback with neuromarkers, a solid understanding of somatosensory covariates is needed. Beta and Theta activities were modulated by affective contexts and pain [8], [11]. Attention modulated the dynamics of Mu and Beta [2], [6]. Increased Gamma activity in the somatosensory cortex was also tied to attention [2]. Lastly, a tremendous overlap exists between somatosensory and motor functions that intimately coevolved. Movement modulates most of the same activities as somatosensation, namely Mu (and its variant Sigma), Beta, Theta and Gamma [7]. Current theoretical models suggest that Gamma might be related to spatial attention, sensory gating and inhibition, whereas theta has been proposed for sensorimotor integration, novelty and change detection.

V. Discussion

We asked if the neuroscience of touch had achieved maturity to quantitatively assess somatosensory experience from non-invasive monitoring of the brain. The answer to this question is not yet. There are six candidate neuromarkers of tactile perception, but four of them are confounded with other mental activities as well as environmental modulation [12]. The other two [3], [4], albeit challenging to record, bear some promise but will need increased vetting of their modulators. We also raised the functional heterogeneity of beta (an endogenous rhythm of sensorimotor cortex possibly distinct from entrained rhythm with vibrotactile stimulation). It would further be desirable to disentangle the different facets of somatosensory experience with respect to specialized receptors pathways, (e.g., high frequency vibration signaling a slipping object or sustained modulation of pressure signaling force control). Except for beta entrainment, the current literature on oscillations does not offer unambiguous specificity of its oscillations with respect to different facets of somatosensation (the timing of evoked potentials might help, since receptor types are wired with sharply distinct conduction velocities, but this lies beyond the scope of the present work). Insight into specific receptor pathways will require new well-integrated electrophysiological studies generalizing the approach undertaken in [8], complemented with better stimulation devices leveraging progress in soft and hard robotics [1]. Those advances in tactile neuroscience, combined with the application of bidirectional prosthetics, will open the path to much-needed neurorehabilitation studies comparing behavioral and neurophysiological characteristics of somatosensation in traumatic or congenital amputees and limb-intact populations using haptically-augmented prosthetic hands.