What can facial mechanoreceptors tell the mouse brain about whisking?

Haptic perception synthesizes touch with proprioception, or sense of body position. Humans and mice alike experience rich active touch of the face. Because most facial muscles lack proprioceptor endings, the sensory basis of facial proprioception remains unsolved. Facial proprioception may instead rely on mechanoreceptors that encode both touch and self-motion. In rodents, whisker mechanoreceptors provide a signal that informs the brain about whisker position. Whisking involves coordinated orofacial movements, so mechanoreceptors innervating facial regions other than whiskers could also provide information about whisking. To define all sources of sensory information about whisking available to the brain, we recorded spikes from mechanoreceptors innervating diverse parts of the face. Whisker motion was encoded best by whisker mechanoreceptors, but also by those innervating whisker pad hairy skin and supraorbital vibrissae. Redundant self-motion responses may provide the brain with a stable proprioceptive signal despite mechanical perturbations such as whisker growth and active touch.

In rodents, whisker mechanoreceptors provide a signal that informs the brain about whisker 23 position. Whisking involves coordinated orofacial movements, so mechanoreceptors innervating 24 facial regions other than whiskers could also provide information about whisking. To define all 25 sources of sensory information about whisking available to the brain, we recorded spikes from 26 mechanoreceptors innervating diverse parts of the face. Whisker motion was encoded best by 27 whisker mechanoreceptors, but also by those innervating whisker pad hairy skin and 28 supraorbital vibrissae. Redundant self-motion responses may provide the brain with a stable 29

Introduction 35
Proprioception is the sense of where the body or its parts are in space. To interpret touch, it is 36 critical that the brain also knows where in space the touched body part was at the time of 37 contact. Thus, touch and proprioception are intimately linked during normal sensory-motor 38 function. Touch begins with the activation of low-threshold mechanoreceptors (LTMRs) in the 39 skin. Information about body position can come from efference copy signals that report the 40 motor commands ultimately used to control muscles. However, the nervous system contains 41 dedicated mechanoreceptive proprioceptor endings to provide feedback about actual, rather 42 than intended, position. Classical proprioceptors include muscle spindle and Golgi tendon organ 43

afferents. 44
Many rodents use rapid motions of their mystacial vibrissae (whiskers) to explore the 45 tactile world (Carvell and Simons, 1990;Welker, 1964;Wineski, 1983). Curiously, the muscles 46 controlling these "whisking" motions, as with other facial muscles, lack classical proprioceptor 47 endings (Moore et al., 2015). Therefore, feedback about whisker position must occur via self-48 motion-triggered ("reafferent") activity of peripheral mechanoreceptors other than classical 49 muscle proprioceptors, such as the cutaneous LTMRs responsible for sensing touch. Neurons 50 throughout the whisker somatosensory system respond to whisker self-motion in a manner that 51 depends on the relative position of the whisker within the current whisk cycle, or whisk "phase" 52 (Campagner et al., 2016;Crochet and Petersen, 2006;Curtis and Kleinfeld, 2009;Fee et al., 53 sensitive" units, defined in Glossary) and strongly modulated by phase, preferring to fire at a 113 particular phase of the whisk cycle ( Figure 2B,C). This sharp phase tuning largely reflects 114 sensitivity to inertial stresses (Severson et al., 2017). We used information theory analyses to 115 quantify how well spiking of single mechanoreceptors encoded phase and other variables 116 related to whisker kinematics. Specifically, we calculated the mutual information (MI; Cover and 117 Thomas, 2006), a measure of association between two random variables derived from their joint 118 probability distribution ( Figure 2D), between (1) spike counts obtained during 2 ms video 119 frames, and (2) binned values (Materials and Methods) of kinematic variables extracted from the 120 video frames, including θ, θ', θ'', θ amp , θ mid , and Φ. Mutual information between phase and spike 121 count for whisker afferents, expressed as a rate via multiplying by the 500 Hz sampling 122 frequency, was 9.1 ± 23.8 bits/s (median ± interquartile range [IQR]; n = 42 whisking-sensitive 123 units). To determine which kinematic variable best accounted for the spiking of whisker 124 afferents, we calculated a "normalized mutual information" by dividing MI by the spike count 125 entropy (Jamali et al., 2016). This quantity gives the fraction of spike count uncertainty 126 accounted for by a given kinematic variable. Whisker afferent spike counts were better 127 explained by phase ( Figure 2E; normalized MI = 0.096 ± 0.121; median ± IQR) than by θ (0.034 128 ± 0.064), θ' (0.046 ± 0.086), θ'' (0.030 ± 0.076), θ amp (0.012 ± 0.024), or θ mid (0.0089 ± 0.013; p < 129 0.0031 for all 5 comparisons, two-tailed K-S tests). 130 131

Whisker motion coding by mechanoreceptors innervating hairy skin 132
While whisker mechanoreceptors showed strong phase coding, our goal was to put this coding 133 into context by comparing the information provided by these whisker afferents to that of any 134 other types of mechanoreceptor we could find that responded during whisking in air. We began 135 by recording from TG units with touch receptive fields on hairy skin (n = 85) rather than a 136 vibrissa ( Figure 3A). Afferents responded to manual deflections of all small hairs or a small 137 number of guard hairs within the mapped receptive field, were rapidly adapting, and responded 138 to touch in all directions (not shown). Remarkably, activity of a large number of facial hairy skin 139 afferents was modulated during whisking in air ( Figure 3B, 58 of 85 were whisking-sensitive). 140 Consistent with the lack of direction-selectivity, many facial hairy skin afferents fired at multiple 141 phases (e.g. both protraction and retraction phases) of the whisk cycle ( Figure 3C). Phase 142 coding by hairy skin afferents varied by receptive field location, such that units with receptive 143 fields closer to the whisker pad tended to encode phase more strongly than those distant from 144 the pad ( Figure 3D; overall MI rate = 1.3 ± 4.5 bits/s; median ± IQR). 145 We next grouped the hairy skin receptive fields into six different "zones" of the face 146 ( Figure S1), including the pad, cheek, snout, eye, lip, and jaw. Units with receptive fields on the 147 whisker pad (n = 14) were particularly modulated by phase (13 of 14 were whisking-sensitive). 148 We found several receptive fields comprised of small hairs surrounding whisker follicles, in 149 between whisker arcs or rows, or flanking the outer whiskers. Receptive fields on the pad were 150 smaller in area than other regions of the face ( Figure 3D). Pad hairy skin afferent encoding of 151 phase (MI rate = 9.26 ± 8.53 bits/s, median ± IQR) was comparable to whisker afferents and 152 significantly higher than afferents innervating hairy skin on the cheek (0.73 ± 2.10 bits/s), eye 153 (1.11 ± 3.57 bits/s), lip (0.75 ± 1.76 bits/s), snout (1.63 ± 1.23 bits/s), and jaw (0.08 ± 0.09 bits/s; 154 Figure 3E, p < 0.032 for all 5 two-tailed K-S tests). Across all facial hairy skin afferents, 155 normalized MI ( Figure 3F) was significantly higher for phase (0.0165 ± 0.0348) compared to θ' 156 (0.0062 ± 0.0098), θ'' (0.0056 ± 0.0098), θamp (0.0085 ± 0.012), and θ mid (0.0094 ± 0.011; 157 p<0.003 for all 4 two-tailed K-S tests), but similar to θ (0.014 ± 0.023; p = 0.77 two-tailed K-S 158 test). 159 Video capturing facial motion and whisker position suggested that widespread patterns 160 of skin strain likely occur in a manner correlated with whisking, with stronger correlations 161 between skin and whisker displacements occurring for facial regions on or near the whisker pad 162 (Supplementary Video 2; Figure S1). In addition to skin movements, we observed that the 163 whisking-sensitive), and microvibrissae (8 of 10 whisking-sensitive). Non-mystacial vibrissa 215 afferent spiking aligned with whisk phase ( Figure 4G). Similar to whisker afferents, we observed 216 examples of sharp phase tuning ( Figure 4H). 217 Supraorbital afferents encoded similar amounts of information about phase (MI rate = 218 16.2 ± 14.9 bits/s, median ± IQR) as genal afferents (5.2 ± 8.6 bits/s, median ± IQR; p = 0.23, 219 two-tailed K-S test), and more than microvibrissa afferents ( Figure 4I; 0.74 ± 1.60 bits/s, median 220 ± IQR; p = 0.0014, two-tailed K-S test). The spike counts of non-mystacial vibrissa afferents 221 overall were better explained by Φ ( Figure 4J; normalized MI = 0.062 ± 0.17, median ± IQR) and 222 θ (0.047 ± 0.042) compared to θ amp (0.019 ± 0.034), and θ mid (0.021 ± 0.033; p < 0.049 for all 4 223 two-tailed K-S tests). 224 225

Information encoded by jaw proprioceptors 226
The trigeminal mesencephalic nucleus (MeV) resides in the brainstem and contains 227 mechanoreceptor neurons that innervate the masseter muscles involved in mastication. 228 Recently, it has been suggested that MeV neurons respond to aspects of whisker motion 229 (Mameli et al., 2017;Mameli et al., 2010;Mameli et al., 2014), which necessitates their inclusion 230 in a full account of possible sources of peripheral information about whisker motion available to 231 the brain (Bosman et al., 2011). We thus recorded the activity of single neurons using 32-232 channel tetrode microdrives implanted in MeV ( Figure 5A). 233 As with TG recordings, head-fixed mice were placed on a treadmill to elicit running and 234 whisking. Mice also licked at a lickport for water rewards. We used this preparation to identify 235 jaw muscle proprioceptors, as their activity was strongly modulated by the licking associated 236 with reward consumption ( Figure 5B,C). For analysis we considered both these putative jaw 237 muscle proprioceptors (n = 23 units), plus units that were recorded on the same tetrode as a 238 putative proprioceptor (n = 20 units) and therefore also presumably in MeV. We did not observe 239 obvious phasic modulation of MeV activity during whisking ( Figure 5B; periods of licking 240 excluded from this analysis). MeV units (n = 33 whisking-sensitive) were not tuned to whisk 241 phase ( Figure 5D) and thus did not encode much information about phase (0.04 ± 0.04 bits/s, 242 median ± IQR). 243 However, we did observe a correlation between MeV activity and whisking midpoint 244 ( Figure 5D,E). Several units increased or decreased spiking with increasing θ mid . In addition, the 245 kinematic variables that associated best with MeV spike counts were midpoint, amplitude, and 246 position ( Figure 5F). However, MI values between spike count and these quantities were low 247 (e.g. θ mid : 0.17 ± 0.38 bits/s, median ± IQR). Thus, MeV activity is not correlated with whisk 248 phase and appears only weakly correlated with whisking midpoint. We speculate that this weak 249 correlation may be explained by slight changes in jaw position associated with whisking around 250 more or less protracted midpoints. 251 252

Discussion 293
Here we surveyed primary mechanoreceptive afferents that innervate multiple regions of the 294 face to quantify correlations between spiking activity of these mechanoreceptors and whisker 295 motion. Our specific goal was to provide a comprehensive account of the possible sources of 296 reafferent information sent to the brain about whisking. This quantitative survey provides 297 important context to interpret the encoding of whisker motion-and in particular, whisk phase-298 previously observed among whisker afferents (Campagner et al., 2016;Severson et al., 2017;299 Wallach et al., 2016), and more generally to investigate the hypothesis that facial proprioception 300 relies on the reafferent activity of cutaneous LTMRs. We found that whisker afferents as a group 301 encoded whisk phase best, together with supraorbital and genal vibrissae afferents. Thus, our 302 results support the hypothesis that the strong phase coding observed in prior work with whisker 303 afferents (Campagner et al., 2016;Severson et al., 2017;Wallach et al., 2016) could serve as a 304 basis for whisker proprioception. 305 We found that a large number of mechanoreceptors with receptive fields on the hairy 306 skin of the face responded in a phasic manner during whisking. While passive or active touch of 307 the whiskers did not strongly activate facial hairy skin mechanoreceptors, passive stretch of the 308 skin within or near their receptive fields was sufficient to cause spiking (not shown). This 309 suggests that skin strain occurring within the receptive field, and in a pattern correlated with 310 whisker motion, likely underlies the self-motion responses of these afferents. Activity of 311 cutaneous afferents has also been reported during jaw movements in rabbits, with activity of 312 non-direction selective, hairy skin afferents responding to self-motion in a manner proportional 313 to movement speed (Appenteng et al., 1982). In humans, microneurography studies have 314 reported activity in cutaneous afferents related to active movement of the face (Johansson et 315 al., 1988), ankle (Aimonetti et al., 2007, knee (Edin, 2001), and finger (Edin and Abbs, 1991;316 Hulliger et al., 1979). Thus, "cutaneous" (reafferent) signals of potential use for proprioception 317 occur across a wide variety of body parts and animal species. 318 Using high-speed videography, we found correlated motions of the non-mystacial 319 vibrissae and the mystacial whiskers. In rodents, major aspects of the structure and innervation 320 (Fundin et al., 1995;Wineski, 1985) of the supraorbital and genal vibrissae closely resemble 321 those of mystacial vibrissae (Fundin et al., 1994). Motions of these non-mystacial vibrissae were 322 assessed in the golden hamster, and they were found to be relatively immobile (Wineski, 1983). 323 Here, we show that in mice, supraorbital and genal vibrissae are indeed mobile and whisk in 324 phase with the whiskers. The observation of tight coupling of whisker and non-mystacial vibrissa 325 movements adds to our understanding of the exquisitely coordinated orofacial motor actions in 326 rodents (Kurnikova et al., 2017;Welker, 1964) and suggests that their premotor circuits are 327 linked (Deschenes et al., 2016;Kleinfeld et al., 2014;McElvain et al., 2018;Moore et al., 2013). 328 Afferents with receptive fields on these structures, especially the supraorbital and genal 329 vibrissae, displayed strong phase tuning and carried information about phase comparable to 330 that of whisker afferents. While these afferents encode the phase of the whiskers in the whisk 331 cycle, the supraorbital and genal vibrissae are unlikely to contact objects that are in reach of the 332 whiskers. Thus, an interesting possibility is that afferents with these non-whisker vibrissae 333 receptive fields could provide the brain with a phase signal that is, unlike that of the whisker 334 afferents we report here and in past work (Severson et al., 2017), unperturbed by contacts 335 between whiskers and objects in the world. Alternatively, whisker afferents that respond to 336 whisking in air but not touch have also been found and could serve this role (Szwed et al., 337 2003). Neural circuits that separate touch and self-motion signals arising from the same 338 licking. We found that MeV units did not encode whisk phase nor other rapid aspects of whisker 346 motion. MeV units did encode the midpoint of whisking, albeit very modestly relative to other 347 afferent classes. MeV houses the muscle spindles of jaw muscles, which spike during jaw 348 movements (Goodwin and Luschei, 1975). We therefore speculate that these weak correlations 349 with midpoint occur due to coordinated motion of the jaw and whisker pad, perhaps with subtle 350 jaw muscle changes occurring at more protracted whisking midpoints (which occur at higher 351 locomotion speeds; Sofroniew et al., 2014). However, we identified MeV units in our 352 extracellular recordings based on responses to licking (presumably jaw-motion-correlated), or 353 based on a unit being recorded on the same tetrode (nearby location) as a licking-correlated 354 unit. It is possible that MeV houses neurons that we did not sample and that encode other 355 aspects of whisking. 356 Together, our results provide a quantitative survey of how much information 357 mechanoreceptors in the face can provide the mouse brain about whisking. Our data reveal that 358 non-mystacial vibrissae can whisk in phase with the whiskers, and that mechanoreceptors 359 innervating these non-mystacial vibrissae, as well as a subset of mechanoreceptors innervating 360 facial hairy skin, can provide the brain with information about whisker motion comparable to 361 mechanoreceptors that innervate the whiskers. Whisker mechanoreceptors provided the best, 362 but not the only, source of information about whisking for the brain to use in whisker 363 proprioception. We conclude that the coding of whisker self-motion occurs via a multitude of 364 sensory signals arising from distinct classes of facial mechanoreceptors.  Raw video (slowed 20-fold) showing mouse whiskers during whisking. The tracked γ whisker is 508 indicated with the black overlay, and its angular position (θ) is shown in the trace at bottom. 509 Spike times from a simultaneously recorded afferent responsive to the C2 whisker (third from 510 right) are indicated as black ticks above the θ trace. Audio is the playback of the spike waveform 511 at the corresponding spike time.

Materials and Methods 539
All procedures were in accordance with protocols approved by the Johns Hopkins University 540 Animal Care and Use Committee. 541 542 Mice. Mixed background mice were housed singly in a vivarium with reverse light-dark cycle (12 543 hours each phase). Behavior experiments were conducted during the dark (active) cycle. The 544 sex and line of each mouse used for recordings is detailed in Table S1. 545 546 Surgical preparation -TG recordings. Adult mice (6-18 weeks old) were implanted with 547 titanium headcaps (Yang et al., 2016). Prior to electrophysiological recordings, two small 548 openings (0.5 mm anterior-posterior, 2 mm medial-lateral) in the skull were made centered at 0 549 and 1.0 mm anterior and 1.5 mm lateral to Bregma, with dura left intact. Craniotomies were 550 covered acutely with hemostatic gelatin sponge (VetSpon, Ferrosan Medical Devices) or 551 chronically with silicone elastomer (Kwik-Cast, WPI) followed by a layer of dental acrylic (Jet 552 Repair Acrylic).

554
Surgical preparation -MeV recordings. Custom microdrives with eight tetrodes (Cohen et al., 555 2012) were built to make extracellular recordings from MeV neurons. Each tetrode comprised 556 four recording wires (100-300 kΩ). A ~1 mm diameter craniotomy was made (centered at -5.4 557 mm caudal to bregma, 0.9 mm lateral to midline) for implanting the microdrive to a depth of 2 558 mm, ~0.5 mm dorsal to MeV. Adult mice (9-18 weeks old) were implanted with a titanium 559 headcap for head-fixation. The microdrive was advanced in steps of ~100 µm each day until 560 reaching MeV, identified by the presence of clear high-frequency firing responses to jaw 561 opening and/or closing. Putative MeV jaw proprioceptors were identified post hoc by clear 562 modulations of spike rate aligned to lick times ( Figure 5C). 563 564 Behavioral training and apparatus. Mice received 1 ml water per day for ≥ 7 days prior to 565 training. Mice were head-fixed and placed on a linear treadmill to promote whisking, as mice 566 whisk during running. Voluntary bouts of running were encouraged by providing subsequent 567 water rewards via a custom lickport. On training days (2-10 days total), mice were weighed 568 before and after each session to determine the volume of water consumed. If mice consumed < 569 1 ml, additional water was given to achieve 1 ml total. During recordings, treadmill position was 570 tracked with a custom optical rotary encoder comprised of a 3D printed encoder disk (2 cm 571 diameter, 20 holes) and a commercial photointerrupter (1A51HR, Sharp). 572 573 Whisker and other hair trimming. One day prior to electrophysiological recording, non-574 mystacial hairs on the left side of the face were trimmed short with fine forceps and 575 microdissection scissors (Fine Science Tools), during isoflurane (1.5%) anesthesia. For TG 576 recordings, all whiskers and microvibrissae were trimmed short except β, γ, δ, B1-4, C1-4, and 577 D1-4. For improved tracking of whiskers, we minimized obstruction of the field of view by hairs 578 that were not whiskers intended to be tracked. We did not use chemical hair remover. Fur 579 between the whiskers was manually removed by plucking or trimming. Non-whisker hairs were 580 maintained at this short length by repeating this procedure as necessary. Receptive fields on 581 facial hairy skin were always on fur cut <1 mm by trimming. Whisker and non-mystacial vibrissa 582 afferents were recorded while the vibrissa in the receptive field was at or near its intact length. 583 584 Trigeminal ganglion electrophysiology. Recordings from TG afferents were performed as 585 described (Severson et al., 2017). Briefly, awake mice were head-fixed and allowed to run on 586 the treadmill. The craniotomy was exposed and covered with PBS. A single tungsten recording 587 electrode (2 MΩ nominal, Parylene coated; WPI) was lowered ~5.5 mm until it reached the TG. 588 The tissue was allowed to relax at least 10 min to stabilize recordings. An identical reference 589 electrode was lowered to a similar depth or placed outside the craniotomy in the PBS. The 590 differential electrophysiological signal between recording and reference electrodes was 591 amplified 10,000x, bandpass filtered between 300 Hz and 3,000 Hz (DAM80, WPI), and 592 acquired at 20 kHz in 5 second sweeps. Electrophysiology, high-speed video, and other 593 measurements were synchronized by Ephus (Suter et al., 2010) or WaveSurfer 594 (http://wavesurfer.janelia.org) software. A micromanipulator (Sutter Instruments) advanced the 595 recording electrode until a well-isolated unit responsive to manual touch stimulation was 596 encountered. The unit's receptive field, response type (RA or SA), and direction selectivity were 597 manually classified. All whiskers except the row containing the whisker-of-interest and/or 598 surrogate tracking whisker were trimmed short. Small manual movements of the treadmill 599 encouraged the mouse to run and whisk. After recordings, the craniotomy was covered with 600 silicone elastomer and a thin layer of dental acrylic. Spike waveforms were obtained by 601 thresholding high-pass filtered (500 Hz) traces and clustered using MClust-4. were bandpass filtered online between 0.1 Hz and 10 kHz, highpass filtered offline below 500 610 Hz, and spikes were detected using a threshold of 4-6 standard deviations of the filtered signal. 611 The timestamp of the peak of each detected spike, as well as a 1-ms waveform centered at the 612 peak, were extracted from each channel of the tetrode for spike sorting, and clustered using 613 MClust (AD Redish et al.).

615
Mapping facial hairy skin receptive fields. The touch receptive fields of TG units were 616 identified with a hand-held probe, while monitoring activity using an audio monitor (Model 3300, 617 A-M Systems). When a whisker receptive field could not be found, the receptive field could often 618 be located after probing hairy skin on the entire face. In these cases, before recording began, 619 the extent of the receptive field was mapped by determining the region of hair and skin in which 620 gentle touch with fine forceps (Dumont AA, tip dimensions 0.4 mm x 0.2 mm; FST, #11210-10) 621 evoked spikes and marked with a fine, water-based color marker (0.3 mm tip, Micro-Line, 622 Platinum Art Supplies). Following the recording, the mouse's head with marked receptive fields 623 and a micro-ruler (Electron Microscopy Sciences, #62096-08) were photographed (13 624 megapixel camera, LG Stylo 2) from the side, above, and/or below. The receptive fields were 625 then compiled on a template "face map". The template image was drawn by outlining the profile 626 and fiducial marks (e.g. eye, whisker follicles, nostrils) of a side view image of a mouse's face in 627 Adobe Illustrator CS 6 (Adobe Systems). The approximate shape, location, and relative size of 628 each imaged receptive field were mapped onto the template by: outlining the receptive field, 629 locating nearby fiducial marks in the original image, applying a fixed scaling to match receptive 630 field and template image dimensions, and translating to align to fiducial marks in the template 631 image. Using the SVG Interactivity Panel in Illustrator, receptive fields were tagged with unique 632 identifier text and their coordinates exported to a text file subsequently read into MATLAB. 633 Borders of each zone of the face (e.g. pad, cheek) were drawn by outlining and connecting 634 fiducial marks ( Figure S1). Receptive fields were designated to the zone in which the center of 635 mass was located.

637
High-speed videography. Video frames (640 pixels x 480 pixels, 32 µm/pixel) were acquired at 638 500 Hz using a PhotonFocus DR1-D1312-200-G2-8 camera (90 µs exposure time) and 639 Streampix 5 software (Norpix). Light from a 940 nm LED (Roithner Laser) was passed through a 640 condenser lens (Thorlabs), through the whisker field, reflected off a mirror (Thorlabs), and 641 directed into a 0.25X telecentric lens (Edmund Optics). Ephus or WaveSurfer triggered 642 individual camera frames (5 seconds, 2,500 frames per sweep) synchronized with 643 electrophysiological recordings. To record microvibrissa movement, whiskers were trimmed, 644 except for the D-row whiskers used for tracking whisker movement. The LED was rotated 30° to 645 capture an oblique view of the profile of the mouse's face, thus maximizing the apparent length 646 of the microvibrissae to enable tracking. To record facial and supraorbital vibrissa movements, 647 the mouse's fur was trimmed to <1 mm, as described above. Whiskers and microvibrissae were 648 trimmed to the base, except for the A-row whiskers used for tracking whisker movement. An 649 additional mirror was placed in the light path to capture a side view of the mouse's face. e.g. whisker pad, and outside of the expected length range were excluded. Whisker identity was 662 then determined based on its follicle X-coordinate in either ascending or descending order. 663 Finally, a number of events could render individual videos ineligible for further processing. 664 These events included objects placed in or entering the video frame or grooming behavior. 665 Using a custom GUI, every sweep was inspected to determine if an exclusion event had 666 occurred. 667 668 Processing kinematics. We used the Hilbert transform to quantify the instantaneous phase 669 (Φ), amplitude (θamp) and midpoint (θ mid ) of bandpass (8-30 Hz, Butterworth) filtered θ (Hill et al., 670 2011). Instantaneous whisking frequency (f whisk ) was calculated by taking the time derivative of 671 the unwrapped Φ signal. We first smoothed θ with a Savitzky-Golay filter (3 rd order, span of 9 672 frames) and interpolated missing frames when possible. Angular velocity, θ', the time derivative 673 of θ, was calculated using central differences and smoothed with the same Savitzky-Golay filter.

674
Sweeps with more than 2% of frames having missing θ data were excluded. For θ, θ', θ amp , f whisk , 675 θ mid , observations outside of the 0.25 and 99.75 percentiles were excluded. No outlier removal 676 was performed on Φ. We calculated cross-correlation values (MATLAB 'xcorr' with 'coeff' 677 option) on pairs of traces for whiskers and non-mystacial vibrissae ( Figure 4E) after converting 678 the sampling intervals from equally spaced time intervals to equally spaced phase intervals, 679 using linear interpolation separately for each whisk cycle. Whisk cycles containing any non-680 whisking frames were removed. For cross-correlation analysis, we included between 79 and 681 333 videos for each session, including 195-591 seconds of whisking data.

683
Tracking facial movement. We acquired epochs of facial movement with high-speed video 684 (500 Hz, 480 pixels x 640 pixels, 32 µm/pixel) to analyze correlations between facial skin 685 movement and whisker kinematics ( Figure S1). Two mirrors were placed in the light path to 686 capture a side view of the mouse's face. Facial hair and whiskers were trimmed short except 687 two A-row whiskers for tracking whisker movement. Displacement of each pixel for each frame 688 was estimated by applying an image registration algorithm (MATLAB "imregdemons" with 689 pyramid level iterations 32, 16, 8, and 4) that aligns each "moving" frame with a "fixed" template 690 frame. First, fixed and moving frames were resized by half on each dimension (to 320 pixels x 691 240 pixels; MATLAB "imresize" with bicubic smoothing) to reduce compute time and file size. 692 Next, pixel values outside of the face and in the eye were set to zero. Image registration was 693 then applied to every video frame in the session. We then calculated Pearson's correlation 694 coefficients between the time series of x-dimension pixel displacement values (Δx) and whisker 695 position (θ) time series. Y-displacement values were not used for calculating correlations 696 because they could not be estimated as accurately from 2D images, due to substantial out-of-697 image-plane curvature of the mouse face that varies along the y-dimension. Mean Pearson's r 698 values for each facial region ( Figure S1E) were obtained by averaging r values across all pixels 699 within each facial region. These regions were determined for the fixed template image using 700 fiducial marks as described above. To obtain the "MI rate", we multiplied MI by the sampling frequency, which was always 500 Hz. 720 To calculate "normalized MI" for each recording, we first calculated the entropy of the spike 721 count distribution:

735
"Non-whisking" periods: Frames with < 1° for the tracked whisker. 736 "Whisking-sensitive": Applies to a unit with 95% confidence interval (CI) on mean spike rate 737 during whisking in air non-overlapping with 95% CI for mean spike rate during non-whisking and 738 with mean spike rate > 1 Hz during whisking. 739 "Whisker afferents" or "whisker mechanoreceptors": LTMRs with single-whisker receptive 740 fields, presumably which innervate the whisker follicle. To track whisking kinematics, the angle (θ) of a whisker (black trace) relative to the mediolateral axis (dotted line) was measured. (C) Top, example one second trace of whisker position (θ, black) and its midpoint position (θmid, gray). Middle, whisk amplitude (θamp) is the half-width, in degrees, of the whisk cycle. Lower bound of scale bar indicates 0°. Bottom, whisk phase (Φ, black), computed for the same trace, is the relative position of the whisker within a whisk cycle. Times when the whisker is in its fully retracted position (Φ = π) are indicated by gray lines. (D) Schematic illustrating different types of afferents (open circles with dotted lines) recorded in these experiments, grouped by type of receptive field. These include four populations: trigeminal ganglion (TG, beige) low threshold mechanoreceptors (LTMRs) with receptive fields localized to either (1) a mystacial whisker follicle (filled black dots; e.g. black whisker), (2) hairy skin (e.g. gray patch on cheek), or (3) a non-mystacial vibrissa (red dots; e.g. red supraorbital vibrissa), or trigeminal mesencephalic nucleus (MeV) proprioceptors innervating facial muscles. (E) Table indicating electrode location (either TG or MeV), number of recorded units, and number of whisking-sensitive units (non-overlapping 95% CI for mean spike rate during whisking vs. non-whisking and >1 Hz mean spike rate during whisking) recorded for each mechanoreceptor group.   (top, supraorbital, blue; middle, genal, red; bottom, microvibrissa, green) tracked simultaneously (gray lines: fully retracted whisker positions). (C) Scatter plot of whisk phase for whisker vs. non-mystacial vibrissae (n = 1000 randomly chosen frames; top, SO, blue; middle, G, red; bottom, μ, green; dashed black lines: unity). (D) Trajectories of whisker and non-mystacial vibrissa angles through each whisk cycle, normalized to the whisker angle (n = 500 randomly chosen cycles; top, A1 and SO angle; middle, C1 and G angle; bottom, D2 and μ angle). (E) Peak cross-correlation (Pearson's r) and phase lag (open circles) for θ of tracked whisker and either adjacent whisker (W, gray; n = 14 whisker pairs from 12 mice), supraorbital (SO, blue; n = 6 recordings from 6 mice), genal (G, red; n = 6 recordings from 6 mice), or microvibrissa (μ, green; n = 3 recordings from 3 mice). (F) Mean whisk amplitude for whisker (W, gray), supraorbital (SO, blue), genal (G, red), or microvibrissa (μ, green). Bars indicate mean ± SD across recordings. (G) Example one second periods with spike times from supraorbital unit (top, blue ticks), genal unit (middle, red ticks), and microvibrissa unit (bottom, green ticks) aligned with position of the tracked whisker (θ, black trace; gray lines: fully retracted whisker positions). (H) Phase tuning curves (mean ± SEM) for the same examples in (G): top, SO, blue; middle, G, red; bottom, μ, green. (I) Cumulative distributions of mutual information rate between spike count and phase of whisking-sensitive neurons with receptive fields on non-mystacial vibrissae, including SO (blue, n = 8), G (red, n = 3), and μ (green, n = 8). (J) Heatmap of normalized mutual information for all whisking-sensitive non-mystacial vibrissa units (n = 19), measured between spike count and each kinematic quantity (•, columns): phase (Φ), position (θ), angular velocity (θ'), angular acceleration (θ''), amplitude (θ amp ), and midpoint (θ mid ). Units (rows) are sorted by receptive field location (labeled at right) and within each non-mystacial whisker by increasing normalized MI averaged across the kinematic quantities.  ) for whisking periods (330,085 frames) in the example recording in (A). Ratios above 1.2 are not shown on plot for clarity (0.09% of frames). The peak with ratio near zero, indicated by a blue arrow, indicates a substantial fraction of missed whisks in this example recording. (C) Same as (B) but for A2 vs A1 whiskers. There is no histogram peak indicative of missed whisks (0.16% of values are above the axis limit of 1.2). (D) Example traces of θ C1 and θ C2 (black traces) and θ G (red; gray lines: fully retracted phase of C1). Missed whisks by the genal vibrissa are marked by red arrows. (E) Histogram of ratio of genal vibrissa amplitude (θ amp,G ) versus C1 whisker amplitude (θ amp,C1 ) for whisking periods (96,429 frames) in the example recording in (C). Ratios above 1.2 not shown (0.36% of frames). Missed whisks with amplitude ratio near zero are indicated by the red arrow. (F) Same as (E) but for C2 vs C1 whiskers. There is no histogram peak indicative of missed whisks (0.75% of values are above the axis limit of 1.2). (A) Joint distribution for spike count and whisk midpoint (θ mid ) comparing linearly spaced (left, same calculation reported throughout paper) and uniform count (percentile) binning (right) of θ mid for an example whisker afferent. Note that bin edges are not equally spaced for uniform count binning. Marginal distributions are plotted for θ mid (top) and spike count (right). Mutual information (MI) rate values calculated using each method of binning are reported at the top. (B) Joint distribution for spike count and whisk phase (Φ) comparing linearly spaced binning (left) and uniform count binning (right) of Φ for the same example unit. Marginal distributions are plotted on top and to the right. Note that the distribution of phase is almost uniform, except fewer bins are observed during retraction phases due to rapid whisker retraction. MI rate values calculated using each method of binning are reported at top. (C) Joint distribution for spike count and angular acceleration (θ'') comparing linearly spaced (left) and uniform count binning (right) of θ'' for the same example unit. Marginal distributions are plotted on top and right. Note that distribution of θ'' has longer tails than θ mid and Φ. MI rate values are reported at top. (D) Cumulative distributions of MI rate between spike count and θ mid , calculated using linearly spaced ("LS", solid lines) or uniform count ("UC", dotted lines) binning for the different afferent groups (linearly spaced values are repeated from Figure 6A). (E) Cumulative distributions of MI rate between spike count and Φ, calculated using linearly spaced (solid lines) or uniform count (dotted lines) binning for the different afferent groups (linearly spaced values are repeated from Figure 6B). (F) Cumulative distributions of MI rate between spike count and Φ, calculated using linearly spaced (solid lines) or uniform count (dotted lines) binning for the best afferent subgroups (linearly spaced values are repeated from Figure 6C).  1 Fig. 2 Fig. 3 Fig. 4 Fig. 5 Fig. 6 Fig. S1 Fig. S2 Fig. S3