1/N How do neural dynamics in motor cortex interact with those in subcortical networks to flexibly control movement? I’m beyond thrilled to share our work on this problem, led by Eric Kirk @videtide1 with help from Kangjia Cai!

2/N To control discrete, voluntary movements like reaching, motor cortex generates time-varying patterns of activity and sends commands to the spinal cord. Thus, muscle activity is often modeled as a function of cortical firing rates.

3/N This approach doesn’t always work. Consider breathing, which is controlled by a medullary CPG, and is usually involuntary. During speech, voluntary commands must modulate the motor pattern. Respiratory muscle activity, then, is a function of both cortical and CPG states.

4/N Similarly, the basic locomotor pattern in mammals is generated by a spinal CPG. Animals with cortical lesions can walk on a flat surface. Motor cortex becomes critical, however, when animals must voluntarily modify their gait to clear obstacles.

5/N We studied cortical dynamics in a “hurdling” task for mice, which combines a spinally-generated pattern with voluntary modifications. In this task, motor cortex must transform information about obstacle proximity and locomotor phase into commands which sculpt motor output.

6/N We performed large-scale recordings in the motor cortex of freely-moving mice during the task. The 5120-contact, 384-channel Neuropixels 2 probes were really a game-changer here, as they allowed us to record from ~250 neurons simultaneously and resolve single-trial dynamics.

7/N Single-neuron responses were highly diverse. Some cells were phase-locked to the locomotor rhythm. Some had transient responses just before or during obstacle traversals, and others exhibited superpositions of rhythmic and transient patterns.

8/N The transient could be synced with the motion of the contralateral limb (limb-dependent response), or with the first limb to cross the barrier (limb-independent). Surprisingly, the latter was more common. Overall, most cells were modulated by at least one task variable.

9/N At the population level, we identified a large limb-independent transient, a large representation of locomotor phase, and a small “readout” signal with the properties of a motor command.

10/N The limb-independent factor was synchronized more tightly with obstacle proximity than movement onset in the leading limb. Does it reflect the propagation of a sensory (e.g., vibrissal) impulse across the cortical network, or is it an abstract preparatory signal?

11/N The signal was robust to optogenetic silencing of barrel cortex, bilateral whisker trimming, and removal of visual input. We conclude it is a motor preparatory signal abstracted from sensory information.

12/N Are rhythmic dynamics driven by sensory feedback, or an efference copy from the CPG? The latter appears to be the case: cortical dynamics were organized into distinct subspaces for active and passive movements.

13/N How does motor cortex transform representations of motor preparation and the locomotor rhythm into appropriate descending commands? We propose a phase-dependent gating mechanism involving a nonlinear interaction between the prep and rhythmic factors.

14/N Overall, we think these results highlight the need to generalize models of cortical control to encompass interactions between voluntary cortical commands and motor programs orchestrated by largely independent subcortical centers.

15/15 Thanks again to Eric Kirk for leading the charge on this project, Kangjia for contributions to analysis and infrastructure, our funders, and @CWRUSOM for supporting our work!

@videtide1 figures are so beautiful!

@videtide1 Thanks! We try to live by Tufte’s rules.

@videtide1 Ha, good to know. Which one of his books do you think is the most relevant?

@videtide1 The Visual Display of Quantitative Information is most useful as a polemical style guide, but the examples in Envisioning Information are arguably better. Also, William Cleveland's Visualizing Data is a must-have!

@videtide1 Great, thank you, Britton!

@videtide1 This was such a cool thread!

@videtide1 Thanks! :)

@videtide1 This research looks so cool! 👏 I'm wondering if similar changes in motor cortex rhythms can be observed in human studies?

@videtide1 Hi Britton can I ask you a question about how possible it is to build a memory erasing machine?