Métodos

Following our proposition that decision making processes occurs periodically during continuous motor control under visual guidance in order to scale the sub-movements and evaluate cursor deviation (i.e. evidence for cor- rection) after the sub-movement, it comes as no surprise that the brain uses an intermittent motor controller for such behaviors.

Intermittent motor control was originally proposed based on the observation of sub-movements during visuo- motor tracking (Craik, 1947; Navas and Stark, 1968; Miall et al., 1985), isometric force tasks (Slifkin et al., 2000; Vaillancourt et al., 2006) or drawing (Doeringer and Hogan, 1998). These observations are hard to ex- plain in the framework of continuous, optimal feedback control (Todorov, 2004). Rather, intermittent motor control theory posits that sub-movements are planned intermittently based on sensory feedback but then unfold in an open-loop manner (Gawthrop et al., 2014). This open-loop time interval could provide time for the plan- ning of the next sub-movement, a possibly serial processing step that could thus have a refractory period pre- venting continuous control (Vince, 1948; Neilson et al., 1988). Intermittent control has received support from computational models (Hanneton et al., 1997; Ben-Itzhak and Karniel, 2008; Bye and Neilson, 2010; Loram et al., 2012). However, there is still no agreement on whether the brain uses such a control mechanism (Karniel, 2011).

One argument against the existence of a refractory period in intermittent control comes from double step para- digms. In such paradigms, a target is shown (first step), triggering a reaching movement. During a subset of movements, the target is displaced (second step), triggering a corrective (sub-)movement. Reaction time to the second step were not longer than for the first step (Archambault et al., 2009). These results were interpreted as evidence that no central bottleneck delayed the processing of the second step and the authors argued against a refractory period in such tasks. However, when the same steps were applied repeatedly during visuo-motor track- ing, there was an increase in response times for the second of two steps when both occurred in less than 200 ms (van de Kamp et al., 2013).

We propose that there is no refractoriness in motor planning per se but in the decision-making process that transforms the current state of evidence for correction (i.e. performance monitoring) to an up-scaling of the next sub-movement. This hypothesis implies that the motor planning of the first movement in the double step paradigm does not show refractoriness since there is no error detection. Nonetheless, the error detection and subsequent corrective action generated in response to the second step could be refractory. However, reaching movements are fast enough that no further error detection is likely to happen in time for a second correction (that would be a triple step paradigm). This is not the case for sustained visuo-motor tracking for which error detection and correction occurs periodically, as suggested by our results and proposed in our model. This pro- cess could thus be serial and show refractoriness, similarly to other cognitive processes (Marois and Ivanoff, 2005).

In Chapter 3, we showed that sub-movements are coupled with an ERP that encodes cursor deviation. In other words, performance monitoring occurs repetitively after each sub-movement. Behaviorally, sub-movements were shown to be scaled by cursor deviation (Miall et al., 1986; Selen et al., 2006). Sub-movements were also shown to couple to cortical oscillations (Gross et al., 2002; Jerbi et al., 2007). Our findings suggest a link between these two parallel lines of research: sub-movements are coupled to a network of oscillations that serve to syn- chronize the motor system with an intermittent performance monitoring and correction process. In Chapter 4, we show that sub-movements are not only coupled to cortical oscillations -and thus possibly to a periodic modu- lation of neuronal excitability (Lakatos et al., 2005)-, but also to high-gamma activity reflecting spiking activity. Our findings also put forward the complexity of the planning of sub-movements and monitoring of feedback in the medial frontal cortex, with some electrodes showing high-gamma activity coupled to sub-movements and some other electrodes showing high-gamma activity coupled to theta oscillations. Finally, in Chapter 5, we ob- served a transient activation of the ipsilateral sensorimotor cortex after large corrective sub-movements. All three chapters thus uncovered correlates of neural processing that depend on the phase of the sub-movements, implying that sub-movements are not just a consequence of intrinsic muscular properties driven by a continuous neural signal but are processed in an intermittent way in the brain.

Further experimental evidence is needed to find which neural processes show refractoriness during visuo-motor tracking. Some evidence comes from providing intermittent visual feedback of the cursor position at different frequencies (Slifkin et al., 2000). The authors showed that behavioral performance increased with increasing feedback rate until 6.4 Hz, after which performance reached an asymptote. This suggests that performance is not impaired if the motor system runs in open-loop every 150 ms (i.e at 6.4 Hz). Using this paradigm, one could observe how sub-movements are scaled depending on the phase of the sub-movement at which the visual feed- back is provided. Such a paradigm could also answer interesting questions such as: do sub-movement couple to the intermittent feedback? And if not, do theta oscillations couple to the intermittent feedback or to the sub- movements?

From a evolutionary perspective, an additional argument in favor of intermittent control lies in the fact that it is implausible that we evolved any neural process specifically for continuous movements under visual feedback. Indeed, in a competitive world in which survival depends on fast actions that should not be predictable by preys or predators, we cannot think of any evolutionary advantages for slow visuo-motor tracking before humans started hunting with bows and arrow. With this in mind, it is logical that such behaviors result from the concat- enation of so-called motor primitives (i.e. sub-movements) (Thoroughman and Shadmehr, 2000) and show in- termittency.