Rodent whisker sensation occurs both actively, as whiskers move rhythmically across objects, and in a passive mode in which externally applied deflections are sensed by static, non-moving whiskers. Passive whisker stimuli are robustly encoded in the somatosensory (S1) cortex, and provide a potentially powerful means of studying cortical processing. However, whether S1 contributes to passive sensation is debated. We developed 2 new behavioral tasks to assay passive whisker sensation in freely moving rats: Detection of unilateral whisker deflections and discrimination of right versus left whisker deflections. Stimuli were simple, simultaneous multi-whisker deflections. Local muscimol inactivation of S1 reversibly and robustly abolished sensory performance on these tasks. Thus, S1 is required for the detection and discrimination of simple stimuli by passive whiskers, in addition to its known role in active whisker sensation.
The whisker region of rodent primary somatosensory cortex (S1) is a powerful system for studying neural coding of sensory stimuli. While most studies focus on S1 coding during active, rhythmic whisking, whisker sensation also occurs in a “passive” mode, defined as sensation of external deflections by static (non-moving) whiskers (Milani et al. 1989; Krupa et al. 2001; Stuttgen and Schwarz 2008). Passive whisker tasks are powerful because they enable the efficient study of cortical sensory coding with precisely controlled stimuli (e.g., Stuttgen and Schwarz 2008, 2010; Gerdjikov et al. 2010; Adibi et al. 2012). S1 neurons robustly encode passive whisker deflections in a manner that often correlates with discrimination behavior, consistent with a role in passive whisker sensation (Simons 1978; Stuttgen and Schwarz 2008, 2010; Lak et al. 2010). However, whether S1 is required for passive sensory behavior has not been clear.
This is an important question because sensory cortex is not required for all sensory behaviors, including many unconditioned reflexes (Sprague 1996; Koch 1999; Yeomans 2002), delay eyeblink conditioning (Weiss and Disterhoft 2011), and fear conditioning and other aversive behavioral responses to simple sensory stimuli (Hutson and Masterton 1986; Phelps and LeDoux 2005; Cohen and Castro-Alamancos 2007). In particular, classical lesion experiments in primates and rodents indicate that cortical involvement varies with sensory processing complexity: Discrimination of complex sensory features (speech sounds, tactile object shape, and texture) requires cortex, while discrimination of simpler features (e.g., tone frequency, tone duration) and basic sensory detection (e.g. of passive fingertip touch) do not (Randolph and Semmes 1974; LaMotte and Mountcastle 1979; Pickles 1982; Heffner and Heffner 1990; Zainos et al. 1997; Prusky and Douglas 2004; Porter et al. 2011).
In rodents, S1 is required for active whisker sensory behaviors such as gap crossing (Hutson and Masterton 1986), roughness discrimination (Guic-Robles et al. 1992), and object localization (O'Connor, Clack et al. 2010), and aperture width discrimination, in which whisking is absent but head motion relative to objects generates complex multi-whisker patterns (Krupa et al. 2001). These behaviors require discrimination of complex spatial and temporal patterns of whisker input. In contrast, it is unclear whether S1 is necessary for the detection and discrimination of simple stimuli applied to passive whiskers. Prior studies found that S1 is not required for the detection and discrimination of air-puff stimuli to non-moving whiskers, or for the detection of whisker pad electrical stimulation (Hutson and Masterton 1986; Cohen and Castro-Alamancos 2007). However, these studies used aversively conditioned behavior (conditioned place avoidance and suppression of licking) with long-duration stimuli (5–15 s) and reinforcement delivered immediately post-stimulus. This is unlike the studies of active sensation and aperture width discrimination, which used appetitively conditioned behavior (lick or jump for reward) with brief stimuli (0.2–1.5 s) and delayed reinforcement (Hutson and Masterton 1986; Guic-Robles et al. 1992; Krupa et al. 2001; O'Connor, Clack et al. 2010). Thus, it is unclear whether S1 is unnecessary for detection of all simple passive stimuli, or instead is differentially required depending on these other task variables. In addition, recent studies in the rodent auditory cortex found that unlike lesions, rapid pharmacological inactivation abolishes simple tone discrimination (Gerstein et al. 2002; Tai and Zador 2008). Thus, lesion studies may have underestimated the role of primary cortex in simple sensory processing, perhaps due to lesion-induced functional reorganization.
We re-examined whether S1 is required for the detection and discrimination of simple deflections applied to passive whiskers, using a behavioral task with brief whisker stimuli and appetitive conditioning, similar to tasks used previously to assay active sensation. Using local reversible inactivation, we found that S1 was clearly required for passive sensory detection and discrimination in these tasks. Thus, S1 is involved in both passive and active whisker sensory behaviors.
The procedures were approved by the UC Berkeley Animal Care and Use Committee. Eighteen female Long-Evans rats (174–202 g at start of training) were trained on 1 of 2 whisker stimulus detection tasks. Training was performed in 30–45 min sessions once or twice per day, 5–6 days per week. Rats were water restricted 22 h before each training session, and received water reward for correct performance, as well as ad libitum water during a 60-min free drinking period after each training session and all day on non-training days. Rats gained weight normally and displayed normal behavior throughout the entire training period (up to ∼6 months). Of these rats, 11 performed the tasks to criterion accuracy, and 9 of which were shown to perform the tasks by detecting panel movement, not piezo sounds (see below). Data from these 9 animals were analyzed in this study.
Training took place in a custom-built, computer-controlled operant conditioning chamber (Fig. 1A,B). The chamber measured 24 × 32 × 41 cm (H × W × D) and was constructed from rigid plastic with a removable plexiglass lid. One wall contained a central nose port (CNP) for trial initiation, flanked by 2 whisker stimulus panels of lightweight plastic (2.0 × 2.0 cm). These panels were attached to piezoelectric actuators (see below for details). The angle between the panels and the CNP was adjusted for each rat so that the long whiskers (macrovibrissae) would rest passively against the right and left stimulus panels while the rat's nose was in the CNP. Lateral to the whisker stimulus panels were a right- and a left-side drink port (DP) that contained an infrared (IR)-LED beam sensor to detect nose entry and that delivered calibrated water rewards via solenoid valve in response to the correct choice. The chamber floor was plexiglass. An IR-LED array below the CNP area provided illumination for video monitoring from above. In some sessions, high-speed IR CCD video images (Prosilica GC660, 119 frames/s) were taken through a macro lens (Vivitar 28 mm f2.8) to monitor whisker movement during behavior.
For the right–left-whisker discrimination task, rats self-initiated each trial by nose poking in the CNP, which brought the whiskers to rest against the right- and left stimulus panels. Nose poke in the CNP was detected via IR beam break and triggered delivery of whisker stimulus panel deflections. On each trial, impulses were delivered via either the right or the left stimulus panel, with a random choice of stimulus on each trial. Rats were rewarded (0.05–0.1 mL) for choosing the DP on the side of the whisker stimulus. Incorrect DP choice triggered a time out tone (4–6 s) and no reward. For the right-whisker detection task, nose poke in the CNP triggered either delivery of impulses on the right stimulus panel or no impulse delivery. (The left panel never moved in this task.) Rats were rewarded for choosing the right DP when a stimulus was presented, and the left DP when no stimulus was presented. Both these tasks are standard “yes–no” tasks in which only 1 stimulus is presented on each trial, unlike 2-alternative forced-choice tasks in which 2 stimuli are presented simultaneously for direct comparison (Blough and Blough 1977; Schwarz et al. 2010).
Training was performed in stages. In Stage 1, rats were habituated to handling and to the behavioral chamber (10–15 min per day, ∼5 days). In Stage 2, water restriction began, and rats were trained in darkness to nose poke in the CNP to obtain water from a nearby DP. The DP was gradually moved to its final position over 4–5 days, and then training was repeated for the opposite DP (2 days). In Stage 3, full stimulus-reward contingency was implemented, initially using large (10° amplitude), sustained (200–400 ms) panel vibrations. Rats were required to maintain CNP nose poke for 100–300 ms per trial, with early withdrawal triggering a time out (0.5–1 s). Some animals were initially trained in Stage 3 using stimulus blocks (5–20 sequential stimuli presented on one side), while other animals were initially trained with random stimulus selection. Once they achieved learning criterion (>70% accuracy), all rats were transitioned to random stimulus presentation and to smaller standardized panel vibrations (3 impulses, 50-ms interval, 700–800°/s velocity, 3.8° amplitude).
The behavioral apparatus, including whisker stimulus delivery, was controlled and recorded by custom programs in IgorPro (Wavemetrics).
Whisker Deflection Stimuli
Whisker deflections were delivered via the whisker stimulus panels, which were mounted on a lightweight fiberglass rod (45 mm length) glued to a piezoelectric actuator (Piezo Systems Inc., Woburn, MA, United States of America; PSI-5H4E bender actuator; 12.7 × 31.8 mm). For all experiments, macrovibrissae were trimmed to 15 mm in length. Video analysis showed that cut whisker tips rested on the whisker stimulus panels.
Whisker stimuli were trains of up-down deflection impulses. Command voltage waveforms were synthesized in Igor and presented as analog waveforms (10 kHz sample rate) using ×20-gain piezo amplifiers (Piezo Systems Inc.). The standardized stimulus train consisted of 3 ramp-hold-return impulses (0.7–1 mm amplitude, 4–12 ms rise–fall time, 6–11 ms hold duration, 50-ms interval between impulse start times). Stimulus parameters reported in the text represent actual panel movement, measured using a custom laser-photodetector system. Amplitude and velocity were expressed in degrees of whisker angle, based on the 15 mm whisker length. After each impulse, a modest amount of ringing occurred (typically <10% of pulse amplitude), lasting 100–150 ms after impulse offset (e.g., Fig. 1D).
The piezos produced soft click sounds. To prevent rats from using piezo clicks to solve the task, 2 steps were taken. First, an additional “dummy piezo” was mounted behind, but not attached to, each whisker panel (1 cm below the panel actuator piezo, see Fig. 1B). For the left–right discrimination task, whenever a whisker panel stimulus was not delivered on one side, the dummy piezo on that side was actuated, so that piezo clicks were generated on both right and left sides during every trial. For the right detection task, on all trials either a whisker panel stimulus or dummy piezo stimulus was delivered on the right side, so a click was always generated on the right side for every trial. Secondly, we recorded the piezo click sounds generated by all stimulus trains (200 kHz sampling rate, ¼ inch Type 4939 free-field microphone, Bruel & Kjaer North America Inc., Norcross, GA, United States of America), and generated a masking noise stimulus composed of these recorded clicks, played at random intervals and a range of amplitudes, combined with white noise. This masking noise stimulus was played from speakers located ∼10 cm above the whisker panels during training. These precautions appeared to be effective, because fixed-panel control experiments showed that of 11 rats that performed the discrimination and detection tasks to criterion, 9 used panel movement, not piezo clicks, to perform the tasks (see Fig. 3).
Whisker Movement Tracking
Whisker movements were recorded by a high-speed CCD video camera (Prosilica GC660, 119 frames/s) during nose poke epochs (3 rats, 10 trials, 19 whiskers), and in separate rats during free whisking in air (2 rats, 4 trials, 8 whiskers). Whiskers were manually traced frame-by-frame over a 300–400-ms period, and the angle between each whisker and the cheek was calculated in each frame using ImageJ (http://rsb.info.nih.gov/ij/; National Institute of Health, United States of America). The most visible whiskers were tracked: for CNP periods, D2 and D3; for free whisking, D1 and D2. (Whiskers move largely coherently during whisking.) Whisking frequency, peak-to-peak amplitude in 100-ms bins, and angular velocity were calculated.
The gamma-aminobutyric acid A (GABA-A) receptor agonist muscimol was used for reversible inactivation of S1 cortex. After training on the behavioral task, a craniotomy was made either bilaterally over S1 (for the left–right discrimination task) or unilaterally over the left S1 (for right-whisker detection task). For craniotomy surgery, anesthesia was induced with ketamine/xylazine (90 and 10 mg/kg, i.p.) and maintained with isofluorane (0.5–2%). The skull was exposed and covered with a thin layer of cyanoacrylate adhesive (Vetbond) and dental cement, and a small screw was attached for head fixation. A 2–3-mm craniotomy was made over S1, centered at 2.5 mm caudal, 5.5 mm lateral from bregma. For the detection task, an additional craniotomy was made over left secondary visual cortex (V2) at 5.2 mm caudal, 4.5 mm lateral from bregma. S1 craniotomy location was confirmed by recording whisker-evoked spikes. The craniotomies were filled with silicone elastomer (Kwik-Sil, World Precision Instruments, Sarasota, FL, United States of America) and sealed with dental cement. Rats received buprenorphine (0.05 mg/kg, subcutaneously, 8-h interval) for post-operative analgesia.
Behavioral measurements began ≥5 days after surgery. Prior to each behavioral session, rats were anesthetized by isofluorane (4% induction and 2% maintenance), head-fixed, and the craniotomy was opened. Drugs were injected at a depth of 500 μm by a glass pipette (30–40 μm tip diameter), and the craniotomy was re-sealed. For the right–left discrimination task, muscimol (Sigma-Aldrich, G019) was dissolved at 5 μg/μL in sterile saline, and 1 μL was injected bilaterally into each hemisphere. For the right whisker detection task, muscimol was dissolved at 1 μg/μL and 1 μL was injected unilaterally in the left hemisphere. Control experiments involved either saline injection, no injection, or muscimol injection into V2. Drug injection was performed 90 min prior to the start of the behavioral session.
Passive Whisker Sensation Task
Rats were trained to perform either a right–left discrimination task or a right-side detection task in which they sensed mechanical impulses applied to non-moving whiskers. In both tasks, freely moving rats initiated trials by nose poke into a CNP in a computer-controlled operant behavior chamber. The whiskers naturally rested on right- and left-whisker stimulation panels that flanked the CNP. Typically whiskers D2, D3, D4, C2, C3, C4, and E3, E4 contacted each panel. These panels delivered calibrated up-down whisker deflections via attachment to a piezoelectric actuator (piezo). Rats maintained nose poke for 100–375 ms, attended to whisker panel stimuli presented during the nose poke interval, and then exited the CNP to retrieve a reward at either a right or left DP (Fig. 1A–D). In the right–left discrimination task, a 150-ms series of impulses was delivered to either the right- or left-side stimulus panel on each trial, and the rat was rewarded for choosing the DP on the side of whisker stimulation. In the right-side detection task, the left-whisker stimulus panel never moved. The right-side panel delivered either the 150-ms train of deflections, or no deflections, on each trial. The rat was rewarded for choosing the right DP on right panel deflection trials, and the left DP on no-deflection trials. In both tasks, incorrect DP choice triggered a time out. Training took place in darkness with a masking noise and other precautions that prevented rats from solving the tasks using the auditory detection of piezo sounds (see Methods and below).
High-speed video monitoring (119 frames/s) showed that rats did not whisk during the nose poke interval, when the whiskers were resting on the stimulus panel (Fig. 2A), but that whisking did occur during entry and exit from the CNP (Fig. 2B, Supplementary Movie). The absence of whisking during the nose poke period was confirmed by frame-by-frame analysis of whisker angle relative to the face on a small subset of trials (3 rats, 10 trials, 19 whiskers, 270–375 ms trial duration). When whiskers rested on the stimulus panel, no rhythmic whisking was detected, and whiskers moved only 5.8 ± 0.5° standard error of the mean (SEM) amplitude in each 100-ms time bin, with 62.8 ± 3.5°s−1 average speed, and 174.6 ± 11.2°s−1 peak velocity. In contrast, during free whisking in air (measured in 2 separate rats, 3 bouts of whisking, 8 whiskers, 380–430 ms per trial), rhythmic whisking occurred at 7.3 ± 1.0 Hz frequency, with whiskers moving 54.8 ± 7.2° amplitude in each 100-ms epoch, with 555 ± 70°s−1 average speed, and 1464 ± 188°s−1 peak velocity (n = 8 whiskers, 4 rats, 380–430 ms, 3 bouts of whisking). Thus, movement amplitude and velocity during nose poke were only ∼11% of that during free whisking in air (P < 0.001, t-test), and no rhythmic whisking was detected (Fig. 2C,D). Cessation of whisking in the CNP did not have to be trained, but occurred naturally, similar to a prior study of rats entering a narrow aperture (Krupa et al. 2001). Thus, panel deflections constitute passive whisker stimuli to largely static whiskers. Piezo stimulus delivery in the nose poke therefore enables reproducible whisker stimulation across trials, with no need for head fixation.
Rats use Whiskers, not Auditory Cues, to Solve the Task
Rats learned the right–left discrimination task in 40–70 training sessions (n = 6 rats) and the right-whisker detection task in 11–21 sessions (n = 3 rats; criterion for learning was >80% correct in the best 50 consecutive trials). Rats performed 100–200 trials per day. Once the task was learned, daily performance was stable (Fig. 3A). To ensure that rats used tactile, not auditory cues to solve the task, we secured the whisker stimulus panel to the wall, rather than the piezo, for one behavioral session (fixed-panel session). The piezo itself (unattached to the panel) continued to move as normal, but these movements were not conveyed to the whiskers. All 9 rats in this study performed the 2 basic tasks at high accuracy (average accuracy 76.2%, range [59.4–90.6%]) and were reduced to chance performance during fixed-panel sessions (average 50.1% [39.1–56.7%]). After each fixed-panel session (4–5 h), rats received a second session of normal training, and performed at high accuracy (average 75.9% [64.5–86.8%]). Accuracy in every fixed-panel session was indistinguishable from chance (P > 0.16, binomial test). Thus, these rats used panel movement, not piezo-generated auditory cues, to solve the task (Fig. 3B,C). (Two additional rats showed some residual performance during fixed-panel sessions, and were eliminated from this study.)
To test whether rats used the long whiskers (macrovibrissae) to sense panel movement, 3 rats that were trained on the left–right discrimination task (A06, A08, and A15) had their whiskers trimmed bilaterally. Before trimming, these rats discriminated at 83.5%, 76.9%, and 75.9% accuracy in the first 10 min of each training session (n = 4 training sessions; 59–84, 13–30, and 45–52 trials in these 10 min; with 15 whiskers). Performance for all sessions was significantly greater than chance (A06 P < 0.001, A08 P < 0.05, A13 P < 0.01, binomial test). After bilateral whisker trim, performance was reduced to 57.2%, 48%, and 51.7%, respectively (n = 42–50, 19–35, and 44–53 trials in 10 min). Not a single behavioral session after whisker trim showed accuracy greater than chance (A06 P > 0.07, A08 P > 0.2, A13 P > 0.09, binomial test). The decrease in accuracy relative to pre-trim sessions was statistically significant for each animal (P < 0.01, binomial test) (Fig. 3D). Thus, rats sense panel movement with the long whiskers, not with non-whisker tactile or visual cues.
Muscimol Inactivation of S1 Abolishes Detection and Discrimination Performance
We reversibly inactivated S1 using the GABA-A receptor agonist muscimol to determine whether S1 is required for passive whisker detection and discrimination tasks. A re-sealable craniotomy was made over S1. For the left–right discrimination task, we interleaved behavioral sessions in which either 1) muscimol or 2) vehicle (saline) was injected bilaterally into S1, 90 min prior to the behavioral session, or 3) animals underwent anesthesia and transient craniotomy opening, but no injection was made (sham injection). The experiment was performed in 2 rats (A06 and A08), with 15 whiskers intact. Discrimination accuracy was high during vehicle and sham-injection sessions, but it was reduced to near chance during interleaved muscimol sessions (example rat: Fig. 4A, group data: Fig. 4B). Overall, mean accuracy during sham-injection sessions was 72.9% (range 68.9–81.3%, 48–61 trials per session, n = 2 sessions per rat). All sham-injection sessions showed greater than chance performance (P < 0.003). During saline injection sessions, mean accuracy was 75.8% (range 71.2–80%, 40–73 trials per session, n = 2 sessions from each 2 animals). During muscimol sessions (1 μL of 5 μg/μL muscimol injected into each hemisphere), mean accuracy was 53.1% (range: 44–58.3%, 25–62 trials per session, n = 3 sessions per rat). Accuracy in every muscimol session was significantly less than the mean control performance (P < 0.02, binomial test), and none of the muscimol sessions showed accuracy above chance (P > 0.15, binomial test). Rats entered the nose poke normally during muscimol sessions and the number of trials was not reduced, indicating that navigation and motivation were not impaired. During muscimol sessions, accuracy was reduced for both right- and left-side stimuli, suggesting that bilateral muscimol injection caused a loss of whisker sensation bilaterally (Fig. 4C).
To test whether S1 was required for the right-whisker detection task, we interleaved behavioral sessions in which 1) muscimol and 2) vehicle injection were injected unilaterally in left S1, 90 min prior to each behavioral session, 3) sham injection was performed, and 4) muscimol was injected in left V2, 2.9 mm away from the S1 injection site. This experiment was performed in 2 rats (A15 and A19, with 15 whiskers intact), and with a lower dose of muscimol than in the left–right discrimination task (1 μL of 1 μg/μL per injection) to reduce the chance of inactivating distant cortical areas. Again, performance was high in the sham-injection and vehicle-injection sessions, and reduced to near chance after muscimol injection in S1 (example rat: Fig. 5A; group data: Fig. 5B). Overall, mean performance in sham-injection sessions was 72.1% (range: 61.4–79.2%, 53–83 trials per session, n = 2 sessions [A15] and 4 sessions [A19]). Mean performance in saline injection sessions was 76% (range: 69.4–80%, 62–90 trials per session, n = 1 session [A15] and 4 sessions [A19]). Mean performance in muscimol sessions was reduced to 53.9% (range: 46.7–62.3%, 42–91 trials per session, n = 3 sessions [A15] and 6 sessions [A19]). Considering each session separately, 10 of 11 control and saline sessions showed accuracy that was significantly greater than chance (P < 0.05, binomial test). In contrast, 7 of 9 muscimol sessions had accuracy that was indistinguishable from chance (P > 0.05, binomial test), and was significantly lower than mean control accuracy for that animal (P < 0.03). The remaining 2 muscimol sessions (one in each animal) showed modest, residual discrimination above chance (62.3% and 60.9% correct, P = 0.046 and 0.049), which may reflect incomplete S1 inactivation.
To ensure that muscimol effects on task performance were due to local effects in S1, rather than broad inactivation of distant cortical or subcortical structures, we tested the effects of muscimol injection into the secondary visual cortex (V2) in the left hemisphere, 2.9 mm from the location of S1 injection. V2 muscimol injection did not impair performance in the right-whisker detection task (mean accuracy: 76.4%; range 70.1–88.9%, 64–95 trials per session, 3 sessions from each rat; Fig. 5A,B). Muscimol did not impair general navigation or significantly reduce trial number in this experiment.
To understand how muscimol impaired right-side stimulus detection, we assessed DP choice separately for right-side stimulus trials and no-stimulus trials (Fig. 5C). On right-side stimulus trials in the 3 control conditions (sham-injection, vehicle injection, V2 muscimol), rats chose the right DP on 81.4% of trials (range 65.8–92.7%). With muscimol in left S1, rats chose the right DP on only 34.7% of right-side stimulus trials (9 sessions, range 0–65.7%), significantly less than in control sessions (P < 0.001, t-test). Indeed, their choice on right-stimulus trials was significantly biased toward the left DP, beyond chance performance of 50% left-side choice: in 7 of 9 muscimol sessions, rats chose the left DP significantly more than the right DP for right-stimulus trials (P < 0.05 criterion, binomial test). The mean behavioral choice on right-side stimulus trials with muscimol (65% left DP choice, 35% right DP choice) was identical to the mean behavioral choice during no-stimulus trials in the 3 control conditions (68.1% left DP choice [range: 50–89%], 31.9% right DP choice). Thus, muscimol in left S1 caused rats on right-side stimulus trials to behave as if they felt no right-whisker stimulus. On no-stimulus trials with muscimol, rats chose the left DP on 72.4% of trials [range 50–97%]. This is not significantly different than no-stimulus trials in control sessions (P = 0.52, t-test), strongly suggesting that muscimol in left S1 numbed sensation of right-whisker stimuli, rather than adding noise to the behavioral choice.
These results show that S1 is acutely required for the detection and discrimination of simple whisker stimuli during a passive whisker sensory task. The passive nature of the task was confirmed by videography and whisker tracking, which showed no active whisking during the sensory stimulation period when the nose is in the CNP and the whiskers rest on the stimulus delivery panel (Fig. 2). Three lines of evidence suggest that S1 inactivation by muscimol impaired task performance by reducing passive sensation of panel deflections, rather than by interfering with active whisking-based navigation during CNP entry and exit. First, rats whisked normally during S1 muscimol sessions. Secondly, S1 inactivation did not significantly reduce trial number, suggesting there was no major navigation deficit (Figs 4B and 5B). Thirdly, in the unilateral detection task, S1 inactivation often caused rats to respond to right-side stimuli by preferentially choosing the left DP, which is the behavioral indication that a stimulus was not detected (Fig. 5C). This strongly suggests that muscimol numbed whisker sensation, rather than impairing navigation, motivation, or other general task requirements.
To confirm that muscimol affected whisker sensory behavior by specifically silencing S1 cortex, rather than diffusing to inactivate distant cortical or subcortical structures (including the superior colliculus, which has been implicated in passive whisker-cued place avoidance [Cohen and Castro-Alamancos 2007]), we made control injections of muscimol in V2. The V2 injection site was 2.9 mm from the S1 site and ∼1.7 mm closer to the superior colliculus. V2 muscimol did not impair performance in the unilateral detection task, while interleaved injection into S1 did (Fig. 5A–C). This strongly suggests that muscimol acted by silencing S1 cortex, rather than by broadly silencing cortex or other distant sites. Normal whisking during muscimol sessions also suggests that muscimol did not act by impairing motor or memory functions of M1 (Komiyama et al. 2010; Erlich et al. 2011).
In prior behavioral studies, rodent S1 was found to be required for sensory tasks involving active whisking (Hutson and Masterton 1986; Guic-Robles et al. 1992; O'Connor, Clack et al. 2010) and for an aperture width discrimination task in which whiskers are passive but complex cross-whisker comparisons are required (Krupa et al. 2001). These tasks involved appetitive operant conditioning (2-alternative forced choice, yes–no, or go/no-go behavior) with brief whisker stimuli (0.2–1.5 s) and reinforcement after a brief delay. In contrast, S1 was not required for detection or discrimination of simpler stimuli applied to passive whiskers in 2 different tasks; however, these tasks utilized aversive conditioning (whisker-cued suppression of licking, whisker-cued active avoidance) with prolonged stimuli (5–15 s) and reinforcement immediately post-stimulus (Hutson and Masterton 1986; Cohen and Castro-Alamancos 2007). We re-examined S1′s role in passive whisker sensation using simple, brief stimuli (∼ 0.15 s) and an appetitively conditioned yes–no behavioral task, chosen to be similar to previous studies of active sensation (e.g., Carvell and Simons 1990; Guic-Robles et al. 1992; Krupa et al. 2001; Morita et al. 2011). Our results show that S1 is required for the detection and discrimination of simple stimuli by passive whiskers under these conditions. This suggests that S1 involvement in whisker sensory tasks is not related to active versus passive sensation or the need to process simple versus complex stimuli, but may instead vary with behavioral response type, stimulus duration, reinforcement delay, task complexity, or other variables. Subcortical structures including superior colliculus (Cohen and Castro-Alamancos 2007) and perhaps amygdala (Phelps and LeDoux 2005) may mediate non-cortically dependent detection and discrimination behavior.
Our results build on other reversible inactivation studies to suggest a reexamination of the standard model that primary cortex is not involved in simple sensory detection and discrimination. In the somatosensory system, this model is largely derived from classical lesion experiments (e.g., Randolph and Semmes 1974; Lamotte and Mountcastle 1979; Zainos et al. 1997). However, lesion experiments can underestimate acute cortical involvement in sensory tasks due to rapid functional reorganization during post-lesion recovery. Recent studies using reversible pharmacological inactivation of rodent auditory cortex (A1) show that A1 inactivation acutely impairs simple tone discrimination, even though A1 lesion does not (Gerstein et al. 2002; Tai and Zador 2008). This suggests that A1 normally contributes to simple auditory processing, but that parallel non-A1 pathways can compensate after A1 lesion. Consistent with this view, S1 lesion in primates does not abolish basic tactile detection, but does cause a lasting increase in tactile detection threshold, as does stroke in humans (LaMotte and Mountcastle 1979; Roland 1987). This suggests that S1 somehow contributes to basic sensory detection, even though it is not absolutely required.
The specific role of rodent S1 in passive whisker sensation is not known. Because S1 is an early cortical processing area in which most neurons encode simple features of single-whisker motion, a likely possibility is that S1 performs basic feature detection and early sensory processing for cortically-mediated whisker sensation, including both passive and active sensation, and for simple and complex stimuli. In this view, S1 represents the beginning of a general cortical processing stream leading to perception of whisker stimuli. Another possibility is that S1 is specifically required for learned sensory-response associations or other cognitive components of the behavior (Zainos et al. 1997). A third possibility is that S1 does not specifically encode tactile information for simple sensory behavior, but instead provides a tonic, permissive signal that is required for whisker processing in another brain region, perhaps superior colliculus (Cohen and Castro-Alamancos 2007). Distinguishing these possibilities will require more precise activity manipulation or measurement of the trial-by-trial correlation between S1 spiking and sensory behavior.
We conclude that S1 contributes to passive whisker sensation of simple stimuli, in addition to its previously accepted role in active whisker sensation and processing of complex, cross-whisker stimuli. This view is consistent with the existence of robust stimulus-evoked spiking in rodent S1 during both passive (Simons 1978; Stuttgen and Schwarz 2008, 2010) and active sensation (Curtis and Kleinfeld 2009; Jadhav et al. 2009; O'Connor, Peron et al. 2010). Processing of passive whisker input must be the principal role of S1 in species with non-motile whiskers, and may be evolutionarily older than processing for active whisker sensation.
This work was supported by National Science Foundation (grant #SBE-0542013) to the Temporal Dynamics of Learning Center, National Institute of Health (R01 NS072416) and a long-term fellowship from the TOYOBO BIO foundation to T.M.
We thank Heejae Kang, Josh Chang, and Pascal Guevara for assistance with behavioral training. Conflict of Interest: None declared.