When is a lick not a lick?
The neural control of movement is so fiendishly complex in part due to the many distinct systems that contribute—neocortex, cerebellum, striatum, midbrain and brainstem nuclei, along with a tangle of spinal circuits (see Chapter 8 of Principles of Neurobiology). Moreover, these circuits cannot instantaneously communicate with our motor neurons and movement sensing organs. Instead, a range of “sensorimotor loop delays” characterize the round-trip journey from sensation transduced at the periphery, up to the neural control circuits, and back down to motor neurons. For human limb movements, these latencies can be as brief as a ~tens of milliseconds (ms) for segmental reflexes and up to as much as ~150 ms for higher neocortical responses. Researchers have thus traditionally divided movement control systems into two coarse types, with different brain regions potentially recruited into each: “feedforward”—when we execute a pre-planned trajectory; and “feedback”—when we respond to new sensory information to generate corrective action. In fact, it is thought that certain movements may be “purely” feedforward, without appreciable feedback corrective components. With the passage of time, however, the bright line dividing feedforward and feedback has become a bit blurrier.
A holdout in the purely feedforward movement category has been licking movements of the tongue in rodents. In the classical imagination, a mouse plans a tongue movement, executes, and retracts the tongue to initiate another discrete lick. This made intuitive sense: mouse can lick as rapidly as ~10 extensions and retractions of their tongue per second. Is there time for corrective action, and if so, under what circumstances, and how does the brain pull it off? Bollu and colleagues explore this question in a study published this year in Nature (Bollu et al., 2021).
The authors trained mice to lick toward a target to obtain a water reward, in response to a water-predicting auditory cue. At the same time, they high-speed videorecorded the tongue (1,000 video frames per second). They then computationally inferred from these two-dimensional images the actual time-varying three-dimensional shape of the tongue (Figure 1, left). This allowed them to determine with unprecedented detail certain key features of movement: when did the tongue leave the mouth, when did it reach its maximal protrusion, when did it start to retract into the mouth and, most importantly, did anything else happen in between? Unexpectedly, the authors found that licks came in two general kinds: classical licks where the tongue protruded out of the mouth and then returned; but also, licks where, after protrusion, the mouse made corrective movements before retraction (Figure 1, right). Why? As it turned out, corrective movements were most likely when mice were most uncertain about spout location—for example, when it had been a while since they had last localized the spout with their tongue.
Where did these corrective licks come from? A known major site of tongue control in the mouse neocortex is the “anterolateral motor” (ALM) region. When the authors optogenetically disrupted ALM, corrective licks were mostly abolished, and animals generally failed to contact the spout. By contrast, “easy” licking targets close to the mouth that normally lacked corrective movements were less severely affected. Alternatively, when the authors electrophysiologically recorded from ALM, they found widespread signals correlated with the occurrence of corrective tongue movements—even those that had yet to happen, perhaps indicating that when an animal is more uncertain of its tongue movements, ALM is preemptively recruited to ready for upcoming corrections. ALM similarly contributed to corrective motions when the authors directly induced errors by unexpectedly moving the target.
These findings clearly show that feedback control is a key part even of one of the most rapid motor systems, and that it critically involves neocortex. Much remains to be unraveled: what other brain sites contribute to feedback tongue control, and how does this system differ from or mirror some of the more “canonical” feedback control systems, such as forelimb reaching movements? Finally, given the ubiquity of licking-based assays in neuroscience, in the future, greater care may be needed to disaggregate licks by type to understand their neural underpinnings.