Distinct Spiking Patterns of Excitatory and Inhibitory Neurons and LFP Oscillations in Prefrontal Cortex during Sensory Discrimination

Prefrontal cortex (PFC) spike activity and local field potential (LFP) oscillation dynamics are broadly linked to various aspects of behavior. PFC neurons can encode the identity of sensory stimuli and related behavioral outcome in a range of sensory discrimination tasks. However, it remains largely unclear how different neuron subtypes and related LFP oscillation features are modulated in mice during sensory discrimination. To understand how excitatory and inhibitory neurons in PFC are selectively engaged during sensory discrimination and how they relate to LFPs oscillations, we used tetrode devices to probe well isolated individual PFC neurons, and LFP oscillations, in mice performing a three-choice auditory discrimination task. We found that a majority of the PFC neurons, 78% of a total of 711 individual neurons, exhibited sensory evoked responses that are context and task-progression dependent. Using spike waveforms, we classified these responsive neurons into excitatory and inhibitory neurons, and found that both neuron subtypes were transiently modulated, with individual neurons’ responses peaking throughout the entire task duration. While the number of responsive excitatory neurons remain largely constant throughout the task, an increasing fraction of inhibitory neurons were gradually recruited as trial progressed. Further examination of the coherences between individual neurons and LFPs revealed that inhibitory neurons in general exhibit higher spike-field coherence with LFP oscillations than excitatory neurons, first at higher gamma frequencies at the beginning of the task, and then at theta frequencies during the task, and finally across theta, beta and gamma frequencies at task completion. Together, our results demonstrate that while PFC excitatory neurons are continuously engaged during sensory discrimination, PFC inhibitory neurons are preferentially engaged as task progresses and selectively coordinated with distinct LFP oscillations. These results demonstrate increasing involvement of inhibitory neurons in shaping the overall PFC network dynamics as sensory discrimination progressed towards completion.


33
The prefrontal cortex (PFC) is known to be critically involved in decision making, and damage to the PFC 34 leads to deficits in various cognitive performance [1][2][3][4][5][6][7]. Goal-orientated decision making involves a 35 number of cognitive aspects, including detecting sensory stimuli, applying learned rules, and executing 36 an outcome. PFC activities, both individual neurons' spiking patterns and population local field potential 37 (LFP) oscillation dynamics, have been correlated with many aspects of the decision making process, such 38 as attention, sensory processing, rule utilization, working memory, task progression tracking, and result 39 anticipation. Recent studies using optogenetics to manipulate the activity of genetically defined cell 40 types have showed that different PFC cell types are associated with distinct aspects of cognitive tasks [8-41 11]. In general, excitatory neurons participate in various aspects of a task, whereas different subtypes of 42 inhibitory neurons seem to be preferentially recruited during different stages of a task. Calcium imaging 43 of PFC neurons in a go/no-go task further revealed that excitatory neurons exhibit heterogeneous 44 responses, while inhibitory neurons tend to be more correlated within their subtypes [12, 13] 45 presumably due to gap junction coupling [14]. Parvalbumin-expressing (PV) neurons were shown to 46 respond to various aspect of a task [12,15], especially to reward [10,11], whereas somatostatin-47 expressing (SST) neurons tend to be more selective and respond primarily to sensory stimuli and motor 48 activity [10,12]. 49 LFP oscillations in PFC have been associated with a range of cognitive functions, and related to 50 oscillations in other cortical and subcortical areas. For example, theta oscillations (~5-10Hz) are closely 51 linked to working memory [16], and are thought to coordinate long range connections between PFC and 52 the hippocampus [17]. PFC Beta oscillations (~15-30Hz), largely associated with status-quo and rule 53 application [18], are often synchronized between PFC and other cortical areas. Higher frequency gamma 54 oscillations (~35-100Hz) are found to be mainly local within the PFC, and are thought to be primarily 55 involved in working memory [19] and attention [20]. 56 It has been suggested that LFP oscillations may organize neurons into functional ensembles [21,22]. For 57 example, coherence between the PFC spikes and the sensory cortex LFPs was increased during covert 58 attention [20], and spike-field coherence within the sensory cortex was found to be correlated with 59 behavioral performance [23]. Recent optogenetic experiments showed that abnormal activity of 60 inhibitory neurons can disrupt gamma oscillations in PFC and lead to cognitive deficiency [24]. While 61 much of our knowledge of the PFC has been obtained in humans, monkeys, and rats, much less is known 62 about PFC function in mice, a model organism with advanced genetic tools that allow detailed 63 examination of the functional significance of distinct neuron subtypes. Here, we recorded both spikes 64 and LFPs simultaneously in the PFC while mice were performing a 3-choice auditory discrimination task 65 in an open field. With tetrode devices, we identified well-isolated individual PFC neurons and distinguish 66 excitatory neurons from inhibitory neurons using spike waveform features. We found that a large 67 fraction of mouse PFC neurons and LFP oscillations were dynamic modulated during the sensory 68 discrimination task, as observed in other animal models and humans. Inhibitory neurons were 69 increasingly recruited towards task completion, suggesting that the inhibitory neural network is 70 particularly engaged as sensory discrimination progressed. In addition, we found that inhibitory neurons 71 overall showed higher coherence with LFPs compared to excitatory neurons, consistent with the idea 72 that the activity of inhibitory neurons are more significantly coupled with LFP oscillations than excitatory 73 neurons. 74 75

76
A majority of individual PFC neurons exhibited task related spiking activities during a three-choice 77 sensory discrimination task 78 To understand how distinct cell types are recruited during sensory discrimination, we designed a 3-79 choice auditory discrimination task, which required freely moving mice to associate a specific auditory 80 stimulus with a predefined "reward" location ( Figure 1A). Briefly, mice self-initiated each trial by 81 stepping into the "initiation" location to trigger one of the three auditory cues (10 kHz sine wave, 25 82 click/second, and 100 click/second), which was presented throughout the trial. After trial initiation, mice 83 were given 5 seconds to reach the "reward" location on the other end of the arena to receive a reward 84 (correct trial). If mice reached the other two incorrect "reward" locations (incorrect trial) or failed to 85 reach any "reward" location within 5 seconds (incomplete trial, excluded in this study), they were 86 presented with a 5-second timeout, with bright light illuminating all "reward" locations. After training, 87 all mice maintained a performance of >60% correct rate over the recording period ( Figure 1B). Most  88  mice were able to complete each trial within 3 seconds, with an average reaction time of 1.31±0.57  89 seconds across all completed trials in 6 mice ( Figure 1C). 90 We performed a total of 251 recording sessions in 6 mice, in PFC bilaterally, and identified 711 single 91 neurons based on simultaneously recorded waveforms from four closely positioned tetrode wires (Fig.  92 1D). Among these 711 neurons, 552 (78%) showed significant changes in their firing rates during the task 93 when compared to the inter-trial interval (ITI) (p<0.05, Wilcoxon rank sum test). To understand how 94 different PFC neurons are selectively modulated during the sensory discrimination task, all subsequent 95 analysis was performed on responsive neurons only. 96 We first examined the timing of PFC spiking relative to the stage of a trial. When aligned to trial start, 97 some neurons showed an immediate increase in firing rate at trial start with the firing rate gradually 98 decaying towards the end of the trial ( Figure 1D), some exhibited delayed increase after trial start and 99 mainly fired in the middle of the trial ( Figure 1E), and some exhibited small increase at trial start but 100 sharp rise towards the end of the trial ( Figure 1F). Given the variable duration for a mouse to complete 101 each trial, we normalized the firing rate to task duration. We found that individual neuron firing rates 102 were dynamically modulated at different stages of the task. However, as a population, PFC neurons 103 response covered the entire task duration, suggesting that the PFC neural network is engaged 104 throughout the entire sensory discrimination task period ( Figure 1D3, 1E3, and 1F3). 105 To rule out the possibility that PFC was solely driven by bottom up auditory stimuli, we designed a 106 "passive listening" block, during which mice received the same auditory stimuli but without performing 107 the discrimination task. Mice were placed in the same arena with a floor that covered the sensors at the 108 "initiation" and "reward" locations. The "passive listening" block contained 200-250 trials. During each 109 trial, one of the same three auditory stimuli was randomly presented for 1.5-3 seconds, followed by a 110 1.5-3 seconds long inter-trial interval (ITI). To compare the activity of the same neuron during 111 discrimination task versus during passive listening, the "passive listening" block was performed either 112 before or after the "discrimination task" block in the same recording session. Of the 422 task-modulated 113 neurons that were also tested with the "passive listening" condition, only 12 (3%) neurons showed 114 significant changes in firing rate (representative neurons shown in Figure 1D, E and F). Together, these 115 results demonstrate that a large fraction of PFC neurons are selectively modulated during sensory 116 discrimination. Even though individual PFC neurons are transiently activated during different phases of 117 the task, the overall PFC neural network is engaged throughout the entire task period. 118 119 Inhibitory neurons were preferentially modulated towards trial completion. 120 To explore the difference in excitatory and inhibitory neural responses, we sorted the recorded 121 individual neurons based on their spike waveforms. Of the 552 task-relevant neurons, the width of their 122 spike waveforms followed a bi-normal distribution, with one peak centered at 0.45 ms, consistent with 123 the general observation of putative excitatory neurons, and another peak centered at 0.25 ms, 124 consistent with the general observation of putative inhibitory neurons (Figure 2A  To examine how excitatory and inhibitory neurons are differentially modulated during the 130 discrimination task, we sorted neurons according to the relative timing of their peak firing rates. We 131 found that both excitatory and inhibitory neurons were transiently modulated throughout the task, with 132 different neurons exhibit peak firing rate at different phases of the task ( Figure 2C1). Excitatory neurons 133 tended to be active throughout the entire task period, whereas increasing number of inhibitory neurons 134 were recruited towards the end of the task ( Figure 2D; p<0.01, 2 test, between the distribution of 135 excitatory and inhibitory neurons). On the other hand, these task responsive PFC neurons failed to 136 produce any responses during the "passive listening" condition, confirming that PFC neurons exhibit 137 task-specific modulation, rather than simply responding to bottom-up auditory stimuli alone ( Figure  138 2C2). While PFC neurons are known to respond to auditory stimuli during passive conditions [29,30], in 139 our study, the three auditory stimuli at ~70dB delivered over an ambient environment of ~60dB were 140 not sufficient to evoke significant passive responses in the PFC neurons. We further compared excitatory 141 versus inhibitory neuron firing rates during correct versus incorrect trials, and found that both neuron 142 subtypes showed clear sequential activity during the correct trials ( Figure S1A1), but not during the 143 incorrect trials ( Figure S1A2), confirming that both excitatory and inhibitory neurons are modulated by 144 task outcomes. Together, these results demonstrate that both excitatory and inhibitory neurons in the 145 PFC encode task stage specific information that correlate with behavioral outcome. The fact that an 146 increasing fraction of inhibitory neurons are recruited towards trial completion suggests that inhibitory 147 network exhibits greater influence over the overall PFC dynamics towards the completion of the 148 auditory discrimination as tested here. 149 150 PFC neurons encode auditory stimulus identify, and inhibitory neuron populations exhibit increasing 151 discrimination ability towards the completion of the task. 152 While PFC neurons are broadly tuned to auditory stimuli, they are known to discriminate sensory stimuli 153 and to categorize sensory inputs [5, 31, 32]. To examine whether mouse PFC neurons are selectively 154 modulated by different auditory cues, we compared their responses to the three auditory cues used in 155 the task ( Figure 3A and B). We observed that PFC neurons exhibited highly heterogeneous responses to 156 different cues. Some PFC neurons responded to only one auditory cue but not the other two ( Figure 3A), 157 whereas some responded to two cues but not the third one. This demonstrates that individual PFC 158 neurons can encode cue identity with highly heterogeneous response profiles, highlighting that PFC 159 networks could utilize the heterogeneity of individual neurons to expand the coding capacity of a large 160 variety of cues among a population of individual neuron with various response amplitude and temporal 161 kinetics. Consistent with the observation that PFC spiking responses are dynamic during the task, tone 162 specific responses are often restricted to certain stages of the task. 163 To quantify the temporal auditory cue selectivity of each PFC neuron, we calculated the discrimination 164 score, defined as 1 minus the p-value from One-way ANOVA test, between the firing rates upon the 165 presentation of the three auditory cues at the different stages of the trial. A larger selectivity score 166 indicates that the firing rates of a neuron exhibit greater difference in response to different sounds. We 167 then plotted the selectivity scores of each neuron over the entire trial, and calculated the fraction of 168 neurons that exhibit significant discriminatory activity during different task stages ( Figure 3C1 and D). 169 Overall, we found that an increasing number of excitatory and inhibitory neurons becomes sound 170 discriminative as task progressed ( Figure 3D). However, a significantly larger fraction of inhibitory 171 neurons can distinguish tone identity toward the end of the task, compared to the excitatory neuron 172 population ( Figure 3D, excitatory: dark blue, inhibitory: dark red, 2 test, p<0.01). During control 173 passive listening condition, these same PFC neurons failed to discriminate sound identity ( Figure 3C2), 174 and the percentage of modulated neurons stayed low and constant throughout the auditory stimuli 175 presentation ( Figure 3D, dot lines). Together, these results demonstrate that both excitatory and 176 inhibitory PFC neurons spiking activity evolves to encode the identity of sensory stimuli during sensory 177 discrimination, consistent with the idea that accumulating information converges onto PFC to facilitate 178 the identification of sensory stimuli. In addition to recruiting a greater fraction of inhibitory neurons as 179 sensory discrimination task progressed ( Figure 2D), the proportion of inhibitory neurons exhibits sensory 180 discrimination also increases more than excitatory neurons ( Figure 3D) the reduction in oscillation powers was significant for all frequencies analyzed ( Figure 4C; comparison 192 between the averaged powers during the 500 ms time windows pre and post trial start: N=5 mice, 193 paired t-test, p<0.05). At trial end, the decrease in oscillation power was also broad brand across all 194 frequencies analyzed ( Figure 4C; comparison between the averaged powers during the 500 ms time 195 windows pre and post trial end: N=5 mice, paired t-test, p<0.05). Conversely, during passive listening, 196 LFP powers seemed constant throughout the trial and progressed into the ITI ( Figure 4B). However, 197 similar analysis on the averaged power during pass listening revealed a small but significant increase 198 across all frequencies at trial start (N=5 mice, paired t-test, p<0.05), in contrast to the reduction in 199 oscillation powers at this stage during discrimination task (N=5 mice, paired t-test, p<0.05). 200 Even though LFP oscillations changes are broad band, inhibitory neurons are known to be related to 201 specific oscillations [24]. To investigate the relationship between spiking activities of excitatory versus 202 inhibitory neurons with LFPs, we calculated spike-field coherence between individual neurons and the 203 LFPs recorded from the adjacent electrode within the ipsilateral hemisphere ( Figure 4D). Interestingly, 204 while LFP power is reduced across all frequencies at trial start, the reductions did not impact the 205 coherence of excitatory and inhibitory neurons equally at all frequency bands. Inhibitory neurons 206 exhibited significantly higher coherence only at gamma frequencies, but not at theta and beta 207 frequencies ( Figure 4D). This result is similar to the finding that inhibitory neurons are important for PFC 208 gamma oscillations during rule shifting behavior [24], highlighting the general coupling of inhibitory 209 neurons and PFC network dynamics during cognitive tasks. The fact that gamma frequency oscillation 210 power is reduced at this stage, but yet inhibitory neurons are still more coherent with gamma 211 frequencies, highlights that the PFC inhibitory network is preferentially engaged upon the initiation of 212 the sensory discrimination task. 213 As trial progresses towards completion, spike-field coherence at theta frequency increased in both 214 neuron types, with inhibitory neurons showing a higher coherence than excitatory neurons, which 215 sustained beyond trial completion ( Figure 4D1). At beta and gamma frequencies, the divergence of 216 spike-field coherence between excitatory and inhibitory neurons only occurred after the trial end, where 217 inhibitory neurons again showed stronger coherence than excitatory neurons ( Figure 4D2 and 4D3). In 218 summary, these results demonstrate that inhibitory neurons showed a stronger coherence with LFPs 219 than excitatory neurons across multiple frequencies during the entire behavioral task, suggesting a more 220 coordinated inhibitory neuron network that emerges with distinct LFP oscillations. It is possible that 221 different inhibitory neuron subtypes are preferentially recruited as task progressed, which could 222 account for the observation of the spike-field coupling with distinct frequency components as previously 223 suggested [12][13][14]. 224 225 PFC LFP oscillation changes are outcome-dependent 226 To further understand the role of LFP oscillations in task performance, we investigated whether LFP 227 oscillations were differently modulated by task outcome. When aligned to trial start, LFP oscillation 228 power showed similar trends during the correct and incorrect trials ( Figure 5A left vs 5B left, and 5C). 229 We further quantified the dynamics of LFP power at trial start, by comparing the average power at 230 theta, beta and gamma frequencies power during the 500 ms time windows before trial start versus the 231 500 ms window after trial start. We found that while the power is reduced on both correct and incorrect 232 trials, the reduction is greater on correct trials than incorrect trials, across all frequencies ( Figure 5D, 233 N=5 mice, paired t-test, *: p<0.05). 234 Interestingly, at trial end, changes in oscillatory power diverged based on the trial outcome. On correct 235 trials, oscillation powers continued to decrease and remained low over a prolonged period into the ITI, 236 across all frequencies ( Figure 5E, blue line). For incorrect trials, oscillation powers first decreased, similar 237 to those in correct trials, but then rose across all frequency bands ( Figure 5E, red line). As a result, the 238 power changes at the trial end were bifurcated and significantly different, where the averaged powers 239 showed reductions in correct trial, but exhibited an opposite relationship in incorrect trials ( Figure 5F, 240 N=5 mice, paired t-test, *: p<0.05). Together, these results demonstrate that LFP oscillation dynamics 241 are linked to task performance, and the prolonged difference in LFP oscillations after each trial may 242 serve as a feedback signal for the correct/reward and incorrect/timeout. 243 244

245
Understanding the different dynamics of excitatory versus inhibitory neural networks in PFC is of great 246 interest in studying PFC involvement in cognitive functions. Here we designed a three-choice auditory 247 discrimination task, and performed tetrode recordings from definitive single PFC neurons, in task 248 performing mice. We distinguished well isolated excitatory neurons and inhibitory neurons based on 249 spike waveforms, and found that inhibitory neurons were preferentially recruited at later stages, 250 whereas excitatory neurons were active throughout the trial, but only during the discrimination task but 251 not during passive listening. These results demonstrate that excitatory neurons and inhibitory neurons 252 in PFC were recruited in a context-dependent manner at different stages of the auditory discrimination 253 task. Not only an increasing faction of inhibitory neurons was recruited as task progressed, inhibitory 254 neuron populations exhibited increasing discriminatory ability of the three auditory stimuli, and also 255 selectively interacted with LFP oscillation at specific frequency bands and task stages. can be sequentially recruited to process task-relevant information. The 3-choice auditory discrimination 265 task allows us to examine more complex features of the PFC encoding ability. We found that while some 266 neurons increased their firing rate specifically to only one auditory cue, others could be modulated by a 267 group of cues. The ability of individual neuron responding to multiple cues could expand the PFC's 268 coding capacity, as each specific sensory information could be collectively represented from the 269 combined selectivity of an ensemble of neurons. 270 Although mice were presented with the same auditory cues in both the "discrimination task" and the 271 "passive-listening" blocks, PFC was only modulated during the "discrimination task" block, indicating 272 that PFC neurons' activities are context-dependent. Such context-dependent modulation has been 273 shown when animals were exposed to different environments field coherence, suggest that inhibitory neural network increasingly impact overall PFC network 293 dynamics as discrimination task progressed. 294 While we found that LFP oscillatory powers were altered similarly across multiple frequency bands, the 295 coherences between LFP oscillation and spike activity were neuron type specific. Inhibitory neurons 296 showed stronger spike-field coherence than excitatory neurons with LFP oscillations at higher 297 frequencies (gamma) at task start, which then switched to lower frequencies (theta) during the task. The 298 fact that inhibitory neurons exhibited a higher degree of coherence across all frequency bands is 299 suggestive of inhibitory neural networks in supporting PFC oscillation dynamics, which may play a crucial 300 role in organizing cell assemblies within the PFC in a context-dependent manner as postulated for the 301 general functional significance of LFP oscillations. 302 LFP oscillations at specific frequencies have been related to different aspects of behavioral tasks and 303 states [13, 21, 60], and LFP synchrony within the PFC and between the PFC and other areas has been 304 observed in many tasks [61]. In addition to single neuron responses in the PFC, we also observed wide 305 spread changes in LFP oscillation patterns across all frequencies. At first glance, the oscillations across 306 multiple frequency band seems to have similar power dynamics during the task, but a more detailed 307 examination revealed that their coherences with the excitatory and inhibitory neurons differ, depending 308 on the task stage and the frequency bands. Moreover, the dynamics of LFP power also reflected in task 309 performance. As the LFP represents the collective activities from populations of neurons, a more 310 phenomenal changes in oscillation could require a collaboration involving more PFC neurons and may be 311 crucial in successfully performing the task. 312 313 Committee. Female C57BL/6 mice (Taconic, Hudson, NY), were water-restricted during behavioral 327 testing, and were closely monitored to ensure that they maintained at least 85% of their pre-experiment 328 body weight. Adult female mice (2-3 months old at the start of the experiments) were trained to 329 perform a 3-choice auditory discrimination task in following steps: 330

Materials and methods
Step 1: Obtain water from reward port. At this step, only the middle reward port was accessible to the 331 animal. It delivered a water reward whenever an animal reached the reward location and triggered the 332 IR beam sensor in front of the water port. 333 Step 2: Detection of the first auditory cue. A white LED at the start location was illuminated to indicate 334 trial initiation. Mice learned through trial and error to initiate a trial by reaching the start location, which 335 triggered the IR beam sensor and the first auditory cue was presented. Mice were allowed to reach the 336 reward location to obtain the reward without any time limit. 337 Step 3: Detection of the second auditory cue. After the animal learned initiating the task and responding 338 to the first cue, we made the reward location associated with the second cue accessible, while blocking 339 the reward locations for the other two cues. At this step, when animals initiated a trial, only the second 340 cue was presented, and animals were required to reach the corresponding location for reward with no 341 time limit. 342 Step 4: Two-choice auditory discrimination task. At this step, the reward locations for both the first and 343 the second cues were accessible. When animals initiated the trial, one of two auditory cues would be 344 presented randomly, and mice were required to reach the corresponding reward location for reward 345 with no time limit. When animals reached the incorrect reward location, a 5 second timeout occurred, 346 indicated by white LED lights around the reward locations. 347 Step 5: Detection of the third auditory cue: After the animal learned the 2-choice discrimination task, we 348 repeated Step 3 to introduce the third auditory cue. At this step, only the third reward location was 349 accessible and the other two were blocked. When an animal initiated the trial, only the third auditory 350 cue was played, and the animal was required to reach the third reward location to complete the trial 351 and receive the reward with no time limit. 352 Step 6: Three-choice auditory discrimination task. All three reward ports were accessible. When animals 353 initiated the trial, one of the three auditory cues was presented, and mice were required to reach the 354 corresponding reward location to obtain reward. At this stage, a reaction time limit of 5 seconds was 355 introduced. Failure of reaching the reward location within 5 seconds would cause a timeout, indicated 356 by white LED lights around the reward locations. To obtain a balanced number of trials with each cue, 357 auditory cues were presented in a group of three, and within each group, each auditory cue was 358 presented once in a random order. 359 During training, each mouse was trained 20 minutes per day. Once well trained, defined as performing 360 over 60% correct rate per day over 3 consecutive training days and capable of completing a minimum of 361 100 trials per day, animals were provided free water access in their home cage for a week and then 362 underwent tetrode implantation. After tetrode implantation and recovery from surgery, animals were 363 briefly re-trained using procedures described in Steps 2-6 until their performance reached 60%, and 364 then recordings were performed. 365

Electrophysiology 366
Custom tetrode devices (16 channels) were assembled in house, which contained four tetrode bundles, 367 two bundles targeting each hemisphere. were implanted with the center positioned at the midline (AP: 2, ML: 0), so that the tetrode bundles 372 targeted the PFC (AP: 2+/-0.2, ML: +/-0.2). We advanced tetrode bundles gradually during the re-373 training period, so that the tip of the tetrode bundles reached the PFC at the recording stage (AP: 1.0 -374 1.9). All recordings were performed in freely-moving mice. During recording, the tetrode device was 375 connected to a commutator (ACO32, Tucker-Davis Technologies, Alachua, FL) to ensure free movement 376 in the behavior arena. Data was acquired with a Plexon OmniPlex system (Plexon Inc, Dallas, TX). Spike 377 waveforms and local field potentials were sampled at 40 kHz and at 1 kHz respectively. The Plexon 378 OmniPlex system also received time stamps from the NiDAQ board to record timing of behavioral 379 events. 380 Mice underwent one recording session per day. Each recording session constituted one "discrimination 381 task" block and one "passive listening" block in random order. Animals could move freely in the 382 behavioral arena during the entire recording session. The "discrimination task" block lasted about 20 383 minutes, and animals were allowed to initiate the task as many times as they desired. In general, mice 384 performed 100-200 trials within 20 minutes. During the "passive listening" block, mice were placed in 385 the same behavioral arena with a plastic floor positioned above all IR beam sensors and LEDs, so that 386 mice had no access to any sensors or water ports. The same three auditory cues (~70db) used in the 387 "discrimination task" were played in pseudo-random order for 200-250 trials. The durations of the 388 auditory cues were randomized from 1.5 to 3 seconds with random 1.5 -3 seconds inter-trial intervals. 389

Data analysis 390
Spike: Spikes were sorted with Offline Sorter (Plexon Inc, Dallas, TX) and then imported into Matlab 391 (2014, MathWorks, Natick, MA) for further analysis. Spike width was defined as the duration between 392 the valley and the peak of a spike waveform. Due to the difference in lengths of each trial, to calculate 393 the firing rate throughout trial progression, we first normalized each trial based on its duration, so that 394 trial start and trial end were aligned at 0% and at 100% of trial progression, respectively. We then 395 calculated the firing rate from -20% to +120% of trial progression using a 1% moving window, and 396 smoothed the results by averaging each data point with its two adjacent data points. When presenting 397 the population data, we further normalized the firing rate between -20% to 120% of trial progression by 398 calculating the z-scores for each neuron with its own mean and standard deviation. 399 LFP: LFPs were imported into Matlab with the Matlab custom script provided by Plexon, and then 400 analyzed with the Chronux toolbox (chronux.org). The power spectrogram of each LFP trace was 401 calculated with mtspecgramc function (moving window size: 500 ms, moving window step: 5ms, tapers: 402 [3 5]) in Chronux. 403 In a few occasions, animal movement caused artifacts in our tetrode recording, such as when the 404 tetrode devices bumped the arena walls, which create large voltage deflections in our recordings. To 405 eliminate the impact of such movement artifact in our analysis, we identified these artifact periods as 406 having the 5% maximum amplitude, either positive or negative, of the whole recording session. Trials 407 that contained these artifact periods were excluded from all analysis involving LFPs. 408 The spectrogram examples were log-normalized (10*log(power/max)), with the maximum power of the 409 examined time window (2 second, centered at either trial start or end). To compare the powers at 410 specific frequency bands across mice, the powers within the examined time window (2 second, centered 411 at either trial start or end) of each trial were first normalized by converting to their z-scores and then 412 averaged within a given frequency band to obtain the power of each trial. The powers of all trials from 413 the same animal were then averaged as the representative power of each animal. In our analysis, we 414 examined three frequency bands, and the ranges of each frequency bands were defined as follows: 415 theta (5-8 Hz), beta (15-30 Hz), and gamma (30-50 Hz). 416 Spike-field coherence was calculated with cohgramcpt function (moving window size: 500 ms, moving 417 window step: 5ms, tapers: [3 5]) in Chronux. For each trial, we first calculated the spike-field coherence 418 for each neuron with LFPs at a specific frequency range, and then averaged across trials. 419

Statistical testing 420
For spike rate modulation, we used one-way ANOVA to compare the firing rates at the same trial 421 progression period for different auditory cues. For LFP power spectrum, we used a paired t-test to 422 compare the power during the 500 ms before and after either trial start or trial end. For spike-field 423 coherence, we used a non-paired t-test to compare the coherence at the same trial progression period 424 between the "discrimination task" blocks and "passive listening" blocks. All analyses were performed in 425 Matlab. Details for each test are presented throughout the Results section. but from a representative excitatory neuron with elevated firing rate in the middle of the task. (F) 608 Similar to (D), but from a representative inhibitory neuron with elevated firing rate at the trial end. 609 spike waveforms, with a clear separation at 0.4 milliseconds between the two peaks, which was used as 613 a threshold to identify excitatory (blue) and inhibitory (red) neurons. (C) Normalized population firing 614 rate during the discrimination task (C1) and during passive listening (C2, sorted in the same order as in 615 C1. 130 neurons were recorded only in auditory discrimination task without corresponding passive 616 listening block were filled with dark blue). Neurons were grouped by type (excitatory and inhibitory) and 617 sorted based on the timing of their peak firing rates. (D) Distribution of excitatory (blue bars) and 618 inhibitory neurons (red bars), based on the timing of their peak firing rate during the task (p<0.01, 2 619 test). 620