Different phase delays of peripheral input to primate motor cortex and spinal cord promote cancellation at physiological tremor frequencies

Saša Koželj,Stuart N Baker

doi:10.1152/jn.00935.2012

Abstract

Neurons in the spinal cord and motor cortex (M1) are partially phase-locked to cycles of physiological tremor, but with opposite phases. Convergence of spinal and cortical activity onto motoneurons may thus produce phase cancellation and a reduction in tremor amplitude. The mechanisms underlying this phase difference are unknown. We investigated coherence between spinal and M1 activity with sensory input. In two anesthetized monkeys, we electrically stimulated the medial, ulnar, deep radial, and superficial radial nerves; stimuli were timed as independent Poisson processes (rate 10 Hz). Single units were recorded from M1 (147 cells) or cervical spinal cord (61 cells). Ninety M1 cells were antidromically identified as pyramidal tract neurons (PTNs); M1 neurons were additionally classified according to M1 subdivision (rostral/caudal, M1r/c). Spike-stimulus coherence analysis revealed significant coupling over a broad range of frequencies, with the strongest coherence at <50 Hz. Delays implied by the slope of the coherence phase-frequency relationship were greater than the response onset latency, reflecting the importance of late response components for the transmission of oscillatory inputs. The spike-stimulus coherence phase over the 6–13 Hz physiological tremor band differed significantly between M1 and spinal cells (phase differences relative to the cord of 2.72 ± 0.29 and 1.72 ± 0.37 radians for PTNs from M1c and M1r, respectively). We conclude that different phases of the response to peripheral input could partially underlie antiphase M1 and spinal cord activity during motor behavior. The coordinated action of spinal and cortical feedback will act to reduce tremulous oscillations, possibly improving the overall stability and precision of motor control.

Highlights

EFFECTIVE CONTROL OF MOVEMENT requires integration of sensory input to ensure successful performance of the task goal (Todorov and Jordan 2002)
Slower central oscillations at ϳ10 Hz are responsible for the dominant component of physiological tremor, which can be a major limitation to the precision of motor output
We recently investigated the relation of neural activity to tremor around 10 Hz in various key motor centers (Williams et al 2010) in monkeys performing a slow finger movement that generates especially high peripheral oscillations in this band

Summary

METHODS

Experiments were carried out on two adult female Macaca mulatta monkeys (monkey T, weight 7.7 kg; monkey Z, weight 5.8 kg). It was important to correct for either of these errors, as they represent a correlation between spike and stimulus timing that would bias subsequent estimates of coherence phase This correction was carried out as follows. For situations with an artifactual suppression (C Ͻ B), we need to add in a spike within the artifact window in B Ϫ C of the N sweeps where there is no spike to restore the counts to the expected value B In this case, sweep i of the unedited spike train has a single interval ai ϩ bi, and the data likelihood is P(ai ϩ bi). To compile this figure, we first generated stimulus event times as a Poisson process (mean rate 10 Hz). This procedure was repeated 1,000 times; if the phase difference found from the original data exceeded 950/1,000 of the differences found after shuffling, the phases were assumed significantly different (P Ͻ 0.05)

RESULTS

DISCUSSION

DISCLOSURES