Here I outline current theories regarding the functional role of sensorimotor beta
oscillations in motor control to determine whether the hypothesis that this activity could represent sensory uncertainty is plausible. It has been known for a long time that beta oscillations over sensorimotor cortex desynchronise prior to and during movement and resynchronise following a movement (Gastaut, 1952; Hari and Salmelin, 1997; Jasper and Penfield, 1949; Jasper and Andrews, 1936; Pfurtscheller, 1981). However, the functional
37
role of this activity remains controversial. Motor control theories suggest that this modulation of beta oscillatory activity actively controls movement rather than being an epiphenomenon of movement. Evidence for this comes from patients with PD outlined above. However, the beta event-related desynchronization (ERD) during movement is not modulated by different types of movement execution, such as the speed of movement (Stancák and Pfurtscheller, 1995), index vs fourth-finger flexion (Salmelin et al., 1995) or ballistic vs sustained wrist movements (Alegre et al., 2003) and there is no difference in the ERD depending on whether participants are focusing on speed rather than accuracy in a reaching task (Pastötter et al., 2012). Moreover, sensorimotor beta power can be decreased by motor imagery (McFarland et al., 2000; Nakagawa et al., 2011), the
observation of movement (Babiloni et al., 2002; Koelewijn et al., 2008), passive movement (Keinrath et al., 2006) and tactile stimulation (Cheyne et al., 2003; Gaetz and Cheyne, 2006), which suggests that the beta ERD is not purely associated with motor execution.
On the other hand, the pre-movement decrease in beta power does appear to be related to movement preparation (J. Kilner et al., 2005): predictive warning cues revealing which hand to respond within a RT task leads to a greater decrease in beta power suggesting a role for beta oscillations in response selection (Doyle et al., 2005; van Wijk et al., 2009).
However, other studies suggest this beta decrease is modulated more readily by
experimental conditions in a task rather than specific movement parameters (Alegre et al., 2003; Cassim et al., 2000; Sanes and Donoghue, 1993; Stancák and Pfurtscheller, 1995).
The literature on modulations of this pre-movement beta decrease is inconsistent and variable; however the consensus is that beta oscillations are not solely involved in motor execution.
It has been suggested that beta oscillations over sensorimotor cortex act to maintain the status quo of the system (Engel and Fries, 2010). Following the termination of a
movement there is an increase in beta power due to a post-movement beta
synchronisation (PMBS). This is thought to recalibrate the motor system and prevent the generation of any new movements. This is supported by the finding that corticospinal excitability is reduced during this period (Chen et al., 1998) and GABA levels in the motor cortex correlate with the magnitude of the PMBS (Gaetz et al., 2011). Indeed, beta
oscillations appear to have an active akinetic process as spontaneous increases in beta power have been shown to slow movements (Gilbertson et al., 2005) and cortical
stimulation of sensorimotor cortex in the beta frequency band has been shown to reduce motor output (Joundi et al., 2012; Pogosyan et al., 2009). Further evidence that the PMBS may encode proprioceptive error feedback (Tan et al., 2014a, 2014b) supports the idea that beta oscillations monitor and maintain the status of the sensorimotor system. In
38
addition, beta power increases during static postural maintenance and motor unit activity in the periphery is phase-locked to sensorimotor beta oscillations during postural
maintenance supporting the idea that beta oscillations aim to maintain the state of the sensorimotor system by encoding cortical reafferance (Baker et al., 1997). This theory closely resembles predictions about how modulations in sensory precision would affect motor output: the precision-weighting of prediction errors determines their effect on current processing in sensorimotor cortex and therefore this precision-weighting acts to maintain the status of the system. This supports that sensorimotor beta power may correlate with this uncertainty estimate.
One assumption of this hypothesis is that sensorimotor beta oscillations have one distinct role that can be generalised across all changes in beta power. However, there are multiple distinct beta components prior to, during and following a movement, which suggests this assumption may not be correct. In warned RT tasks, there is often an increase in beta power prior to the cue onset, which appears to be separate from the PMBS following the previous trial. (Saleh et al., 2010) show a modulation in this power between two tasks with differing complexity (colour association vs simpler spatial cueing task), but the same timing and movement requirements. This suggests that here beta oscillations may act as part of a large-scale visuomotor attentional network to upregulate sensorimotor
processing beyond somatosensation. This could be a distinct role from the beta ERD during movement. In addition, several studies have highlighted a decrease in beta power after the onset of a warning cue that is distinct from the preparatory decrease in beta power prior to movement (Alegre et al., 2006; Tzagarakis et al., 2010; van Wijk et al., 2009). This has been shown to modulate with uncertainty in the direction of the
upcoming movement, such that there is a greater decrease in beta power on trials with the least uncertainty (Tzagarakis et al., 2015, 2010) and this is also reflected in a subthalamic beta power decrease which is greater following predictive warning cues which are likely to have less uncertainty compared to non-predictive warning cues (Williams et al., 2003).
There is no confirmed hypothesis for what this post-warning-cue beta decrease is functionally involved in, however I hypothesise that this may be involved in the
attentional modulation of synaptic gain shown previously to encode uncertainty. It will be important to determine if beta power both prior to and after a movement correlates with changes in sensory precision as predicted by a single unifying hypothesis or if this hypothesis only holds true for certain beta components.
39