Whisking in air: encoding of kinematics by trigeminal ganglion neurons in awake rats

Abstract

Active sensing requires the brain to distinguish signals produced by external inputs from those generated by the animal's own movements. Because the rodent whisker musculature lacks proprioceptors, we asked whether trigeminal ganglion neurons encode the kinematics of the rat's own whisker movements in air. By examining the role of kinematics, we have extended previous findings showing that many neurons that respond during such movements do not do so consistently. Nevertheless, the majority ( approximately 70%) of trigeminal ganglion neurons display significant correlations between firing rate and a kinematic parameter, and a subset, approximately 30%, represent kinematics with high reliability. Preferential firing to movement direction was observed but was strongly modulated by movement amplitude and speed. However, in contrast to the precise time-locking that occurs in response to active whisker contacts, whisker movements in air generate temporally dispersed responses that are not time-locked to the onset of either protractions or retractions.

FIG. 1.

Trigeminal ganglion activity during whisker movements in air. Each panel shows the spikes (red ticks) generated by a trigeminal ganglion neuron during a 2-s episode of whisking. Right: spike waveforms. The arrangement of the panels from top to bottom reflects increasing correlations between movement kinematics and spiking probability. Top: a ganglion neuron that fired consistently in response to passive deflections but rarely fired during self-generated whisker movements. Even movements as large as 40° failed to evoke a spike in this neuron. Middle: a neuron that responded during some but not all whisker movements. Bottom: a neuron that responded consistently throughout the episode.

FIG. 2.

Spikes of trigeminal ganglion (TG) neurons are not reliably triggered by the onset of protraction or retraction. For each of the 4 cells, 100 whisker movements are aligned to protraction onset, retraction onset, and end. Spike rasters are shown above the movements.

FIG. 3.

Comparison of spiking activity for stationary and moving whiskers. Relative to nonwhisking, 10 of 21 (∼48%) TG neurons produce higher firing rates (P < 0.05) during protractions (A) and 7 of 21 (∼33%) do so during retractions (B). •, significant differences.

FIG. 4.

Correlations between kinematic parameters and firing rates of TG neurons. A: a typical significant (P < 0.05) relationship between firing rate and a kinematic parameter for cell 2 from D. B: an example of a less common, strong relationship (cell 18 from D). Significant correlations for amplitude (C) and speed (D) in our sample of TG cells. Absence of a bar indicates that none of the correlations were significant. For a given parameter, correlations could be significant in 1 direction but not in the other. 14 of 21 neurons displayed significant correlations with ≥1 movement parameters.

FIG. 5.

Receiver operating characteristic (ROC) analysis for each movement kinematic. For amplitude, movements were sorted into 3 classes: small (S), intermediate (I), and large (L) movements. For speed, the classes were slow (S), intermediate (I), and fast (F). We plot those neurons for which ROC curves had areas ≥0.7 and thus could reliably encode movement kinematics.

FIG. 6.

Evaluating the ability of kinematic-responsive neurons to differentiate protractions from retractions. A: directional response difference (Hz) as a function of relative speed (protraction minus retraction) for cell 2 in B–D. Directional response difference is computed for each whisk with protraction firing rate minus retraction firing rate. Similarly, relative speed for each whisk, is protraction speed minus retraction speed. B: mean difference in firing rate for protractions vs. retractions is plotted for all movements. Bars are only shown for those cells (10/14, ∼70%) for which the difference is significant (P < 0.05). C: R² for the extent to which relative amplitudes and speeds modulate direction preferences. Speed has a stronger effect than amplitude in modulating the amount of directional tuning. D: the presence of a significant y intercept indicates a directional preference persisting after contributions from amplitude or speed are removed. Eight of the 10 cells displayed preferences after eliminating the influence of amplitude differences, and 5 of them responded more during protractions. Six of the 10 originally direction-selective cells continued to differentiate directions after accounting for speed, and they all preferred protractions.