Representation of Mechanical Variables in the Rat Vibrissal-Trigeminal System

For more than one hundred years, the rat vibrissal-trigeminal pathway has been used as a model system to study tactile perception during active sensing behaviors. To date, however, it is unknown how movements and deflections of the whisker are represented in neural activity, especially during the natural behavior that is an important part of both sensation and perception. This thesis describes experiments and simulations that begin to correlate the mechanical variables transmitted by the whisker with neural activity at two different stages of the trigeminal pathway. We approach the problem from the outside of the system moving inwards--beginning at the whisker, moving to the trigeminal ganglion (Vg), and then to simulations of trigeminal brainstem operations. First, we quantified whisker movements and deflections during naturalistic whisking by calculating the two-dimensional (2D) forces, moment, and kinematics at the base of the whisker. The activity of corresponding Vg neurons was simultaneously recorded. We then used the simplest possible linear models to determine the extent to which Vg responses could be directly predicted either from geometrical or mechanical variables. On average, models constructed with mechanical variables were able to explain slightly more of the Vg response than models constructed with geometrical variables. These results suggest Vg neurons may directly encode forces and moments applied to the whisker. A subsequent analysis of three-dimensional (3D) whisker movements showed extensive out-of-plane motion. The predictive ability of both models--based either on geometrical or mechanical variables--improved when an approximation to this motion was included. Results of an initial analysis that incorporated the complete 3D shape of the whisker even more strongly suggested that 3D forces and moments can explain a very high fraction of the Vg firing rate. Finally, we applied our new understanding of 2D Vg coding preferences to simulate possible brainstem operations. Given that trigeminal brainstem nuclei have multi-whisker receptive fields, we calculated moment, force, and kinematics resulting from deflections of multiple whiskers of a robotic whisker array during wall-following behavior. A simple model using signals from these whiskers was able to detect edges, and find and predict upcoming object contours.