Microstimulation of human somatosensory cortex evokes task-dependent, spatially patterned responses in motor cortex

Abstract

The primary motor (M1) and somatosensory (S1) cortices play critical roles in motor control but the signaling between these structures is poorly understood. To fill this gap, we recorded - in three participants in an ongoing human clinical trial (NCT01894802) for people with paralyzed hands - the responses evoked in the hand and arm representations of M1 during intracortical microstimulation (ICMS) in the hand representation of S1. We found that ICMS of S1 activated some M1 neurons at short, fixed latencies consistent with monosynaptic activation. Additionally, most of the ICMS-evoked responses in M1 were more variable in time, suggesting indirect effects of stimulation. The spatial pattern of M1 activation varied systematically: S1 electrodes that elicited percepts in a finger preferentially activated M1 neurons excited during that finger's movement. Moreover, the indirect effects of S1 ICMS on M1 were context dependent, such that the magnitude and even sign relative to baseline varied across tasks. We tested the implications of these effects for brain-control of a virtual hand, in which ICMS conveyed tactile feedback. While ICMS-evoked activation of M1 disrupted decoder performance, this disruption was minimized using biomimetic stimulation, which emphasizes contact transients at the onset and offset of grasp, and reduces sustained stimulation.

Fig. 1. Array placements and interactions.

a Four NeuroPort electrode arrays (Blackrock Neurotech, Inc.) were implanted in the hand and arm representations of motor cortex (M1) and the hand representation of somatosensory cortex (Brodmann’s area 1, S1). Here, the implant locations are shown for participant C1. The implant locations for the other two participants are shown in Supplementary Fig. 1. Black lines indicate the posterior-medial corner of each array, which is used as a reference in later figures. b M1 responses to ICMS trains delivered to S1. Responses of three example motor channels (spike rasters above and averaged, smoothed firing rates below) that were excited by ICMS (left) and three that were inhibited by ICMS (right). Black horizontal lines indicate the period of ICMS. The green rasters are from participant P3, the rest are from participant C1.

Fig. 2. Prevalence of ICMS-evoked activity in motor cortex.

a Proportion of stimulating channels that significantly modulated each motor channel on the lateral motor array of each participant (range: 0–0.7). In P2, gray squares indicate channels that are not wired. The majority of M1 channels could be modulated by ICMS through at least one S1 channel. The green square indicates the posterior-medial corner of the array (see Fig. 1a and Supplementary Fig. 1). b ICMS-driven modulation of activity in individual M1 channels, averaged across stimulating channels. Modulation is the ICMS-driven change in the response, normalized by baseline activity.

Fig. 3. Short-latency, pulse-locked responses in M1.

a Pulse-triggered average of the responses of three motor channels to ICMS at 100 Hz. On a subset of channels, such as these, responses were tightly locked to each pulse with millisecond or even sub-millisecond jitter across pulses. Blue line denotes the response during stimulation, black line the response during baseline (sham stimulation), gray box indicates blanked recording time to eliminate the stimulation artifact. Error bars represent bootstrapped standard error. Scale bar indicates a 10% probability of a spike occurring in a 0.5-ms bin. b Cumulative distribution of the latency of the peak pulse-locked, direct, response. Latencies tended to be shorter than 6 ms. c Pulse-triggered average of the response of a motor channel whose activity increases with stimulation but is not pulse locked. Error bars represent bootstrapped standard error.

Fig. 4. Shared somatotopy between movement-evoked and ICMS-evoked activity in participant C1.

a Rendering of the extrema of thumb and ring flexion in virtual reality. b Z-scored difference in firing rate during attempted flexion of the thumb (left) and ring finger (right) vs. the mean activation during attempted flexion of each of the 5 digits. The green square indicates the posterior-medial corner of the array (Fig. 1a). c The green regions on the hand diagrams denote the projected fields reported by participant C1 when stimulated through one channel in the lateral and medial sensory array, respectively (indicated by a black dot in the array maps). Channels shaded in green on the array diagram denote electrodes with projected fields on the thumb and ring finger, respectively. Channels shaded in gray denote unwired electrodes. Pink and orange squares in the top right indicate the posterior and medial corner of the medial and lateral sensory array, respectively (Fig. 1a). d Average M1 activity evoked by stimulation through S1 channels with projected fields on the thumb (left) and the ring finger (right). Motor channels that respond strongly to attempted thumb or ring finger movements tend to also be strongly activated by stimulation of electrodes with projected fields on the thumb or ring finger, respectively. Green squares indicate the posterior and medial corner of the array (Fig. 1a).

Fig. 5. M1 is somatotopically linked to S1.

a In participants C1 and P3, M1 electrodes are more susceptible to ICMS delivered through S1 electrodes whose projected fields match the movement fields. b When the dominant movement field matches the dominant digit in the projected field, the susceptibility is strongest; when the second most dominant movement field matches the dominant digit projected field, the susceptibility is weaker; etc. Lines denote the mean, error bars the standard error of the mean, n = 96 channels.

Fig. 6. ICMS-evoked activity depends on behavior.

a Top: Squeeze, grasp, and transport in the VR environment. Bottom: Three example motor channels from Participant C1 exhibit different responses to four levels of ICMS across three motor conditions (squeeze, grasp, transport). Traces denote the firing rate evoked by stimulation at the four levels after subtraction of the mean across conditions. b Stimulation amplitude classifier performance. Classifiers were trained from M1 activity on one of the three conditions and tested on activity in each condition (cross-validated within condition). c Task dependence—gauged by the strength of the condition/amplitude interaction divided by the strength of the main effect of amplitude—is nearly zero for the pulse-locked responses (direct) but varies widely for the non-pulse locked (indirect) ones. The units with direct input from somatosensory cortex respond the same way to ICMS across behavioral conditions.

Fig. 7. Decoder performance with and without sensory feedback (from participant C1).

a Failure rate for the three conditions. Rates collected during a single session are connected by a dotted line. b Path length during the transport phase with different stimulation conditions. Linear stimulation caused the path length of the transport phase to be significantly longer than without stimulation (p = 10⁻¹², K-S Test, two-sided). In contrast, biomimetic stimulation was significantly more efficient than its linear counterpart (p = 10⁻¹⁰⁾ and not significantly different from the no-stimulation condition (p = 0.4).