Low and high beta rhythms have different motor cortical sources and distinct roles in movement control and spatiotemporal attention

Introduction

A link between the beta rhythm in human sensorimotor cortex and voluntary movements was established 75 years ago [1]. Yet, the functional role of sensorimotor beta remains elusive. Beta was associated with many aspects of motor behavior, ranging from motor cortical idling or postural maintenance [2–11] to sensorimotor integration or temporal predictions [12–22]. As beta rhythms were observed in many cortical and subcortical regions and in many different behavioral contexts, they might serve multiple roles [23–26].

Most studies have treated the broader beta frequency range (approximately 13 to 35 Hz) as one common motor cortical rhythm. This has hindered the association of specific beta frequencies to specific aspects of sensorimotor behavior and prohibited building a unified theory regarding the role(s) of sensorimotor beta. A few studies divided this broad band into low beta (below 20 Hz) and high beta (above 20 Hz). In a published dataset [27,28], we observed concurrent and distinct low and high beta bands during visuomotor behavior in macaque motor cortical local field potentials (LFP). However, the low and high bands modulated similarly in power and peak frequency in that behavioral task. Also Stoll and colleagues [29] observed 2 distinct low and high beta bands in macaque frontal cortical electrocorticography (ECoG), in a trial-and-error task comprising search and repetition phases. They found only the high band to be systematically sensitive to attentional effort and cognitive control. Chandrasekaran and colleagues [30] correlated behavioral reaction time (RT) with dorsal premotor cortex (PMd) beta power, in an RT-task. Their data contained a single band peaking at 25 Hz, which shifted slightly towards higher peak frequency in the pre-stimulus period for shorter RT, resulting in positive correlations below 20 Hz and negative correlations above 20 Hz. In comparison, Zhang and colleagues [31] found positive correlations with RT for sensorimotor pre-stimulus alpha/beta power covering 8 to 33 Hz in an RT task.

These studies remain far from conclusive in determining potentially distinct correlations between behavior and motor cortical low and high beta bands. We therefore designed a new visuomotor behavioral task to maximize at the same time spatiotemporal attention and the required motor control, with the aim of disentangling which of the different task variables, and task-instructed and uninstructed (spontaneous) behavioral factors [32,33] affect the 2 beta bands. We hypothesized that low beta might be related to dynamic postural control and movement preparation. This band was shown to be more affected in human Parkinson’s disease patients than high beta [34,35], and more attenuated by stimulation in the subthalamic nucleus [36] and levodopa administration [37], 2 therapies that improve motor control. Furthermore, low beta is attenuated after kinematic errors [38]. In contrast, motor cortical high beta might be more closely associated with attention, working memory, decision-making or reward [29,38–40], and originate anterior to low beta [41]. Consistent with these predictions, we found low beta to be dominant in primary motor cortex (M1) and correlate positively with behavioral RT in 2 macaques. Low beta also correlated negatively with spontaneous hand postural micro-movements that were frequent during the maintenance of stable central hold during delays. High beta, on the other hand, was dominant in PMd, and was unrelated to RT and hand postural micro-movements. Instead, it modulated selectively during anticipation and processing of visual cues. This modulation was enhanced by focal overt attention, when the animal oriented the gaze towards the work area. We conclude that motor cortex contains multiple, independent beta rhythms operating simultaneously, with band-specific behavioral correlates.

Results

We studied LFP low and high beta band rhythms recorded in the motor cortex (M1 and PMd) of 2 macaque monkeys engaged in a complex visuomotor reaching task (Fig 1A). We determined the motor cortical regions in which each band dominated, and we quantified their relationship to task conditions and performance, and to spontaneous hand and eye movements.