Abstract
Active sensing involves the fusion of internally generated motor events with external sensation. For rodents, active somatosensation includes scanning the immediate environment with the mystacial vibrissae. In doing so, the vibrissae may touch an object at any angle in the whisk cycle. The representation of touch and vibrissa self-motion may in principle be encoded along separate pathways, or share a single pathway, from the periphery to cortex. Past studies established that the spike rates in neurons along the lemniscal pathway from receptors to cortex, which includes the principal trigeminal and ventral-posterior-medial thalamic nuclei, are substantially modulated by touch. In contrast, spike rates along the paralemniscal pathway, which includes the rostral spinal trigeminal interpolaris, posteromedial thalamic, and ventral zona incerta nuclei, are only weakly modulated by touch. Here we find that neurons along the lemniscal pathway robustly encode rhythmic whisking on a cycle-by-cycle basis, while encoding along the paralemniscal pathway is relatively poor. Thus, the representations of both touch and self-motion share one pathway. In fact, some individual neurons carry both signals, so that upstream neurons with a supralinear gain function could, in principle, demodulate these signals to recover the known decoding of touch as a function of vibrissa position in the whisk cycle. Animals interrogate the world around them with actively moving sensory organs, resulting in a blend of sensory inputs: one input is from the object under study, while the second is from self-generated movement of the sensor. The detection of an object thus depends on the ability of the animal to distinguish among internally versus externally generated sensations. Nervous systems employ various signaling mechanisms to reference inputs from a sensory organ relative to its position. A well-known example is proprioception, in which receptors in the limb muscles and joints are used to infer the position of tactile sensors in the hands. In this case, the signals of limb position are encoded in areas of the brain that are distinct from those encoding touch. Here, we investigate the analogous problem of encoding the position of the vibrissae, or whiskers—essential orofacial sensorimotor organs in rodents. In contrast to the case for limbs, we find that vibrissa position is encoded along the same neuroanatomical pathway as vibrissa touch. The seeming ambiguity that results from the mixed representation of position and touch can be resolved by a nonlinear neuronal input-output relation that demodulates touch with respect to vibrissa position. This scheme enables the rodent to determine where an object is located relative to its body axis.

This publication has 89 references indexed in Scilit: