The neural representation of time distributed across multiple brain regions differs between implicit and explicit time demands

Animals appear to possess an internal timer during action, based on the passage of time. However, the neural underpinnings of the perception of time, ranging from seconds to minutes, remain unclear. Herein, we considered the neural representation of time based on mounting evidence on the neural correlates of time perception. The passage of time in the brain is represented by two types of neural encoding as follows: (i) the modulation of firing rates in single neurons and (ii) the sequential activity in neural ensembles. Time-dependent neural activity reflects the relative time rather than the absolute time, similar to a clock. They emerge in multiple regions, including the hippocampus, medial and lateral entorhinal cortices, medial prefrontal cortex, and dorsal striatum. Moreover, they involve different brain regions, depending on an implicit or explicit event duration. Thus, the two types of internal timers distributed across multiple brain regions simultaneously engage in time perception, in response to implicit or explicit time demands.


Introduction
Time perception, the awareness of the continuous connection of past, present, and future, allows individuals to optimize behaviors ranging from foraging to social interactions. Humans can voluntarily or involuntarily act at the appropriate time using the flow of time and time intervals; therefore, we likely possess a timekeeping function that can be regarded as an "internal timer." For example, we can estimate the time required by the elevator used every day to reach the desired floor, or we can easily establish a 10-s waiting period by explicitly counting the time.
Patients with Parkinson's disease show impairments in accurate timing of elapsed times of tens of seconds and regularly rhythmic timing behavior (Jones & Jahanshahi, 2014;Malapani et al., 1998Malapani et al., , 2002. Moreover, patients with Huntington's disease show impairments in discrimination of stimulus times (Rowe et al., 2010). These clinical findings suggest that neuronal activity is involved in the internal clock in several brain regions.
Researchers have postulated several theories on perceiving time and determining behavior (Coull, Cheng, & Meck, 2011;Meck, Penney, & Pouthas, 2008;Merchant, Harrington, & Meck, 2013). Unlike circadian rhythms, which are constructed on a 24-h cycle (Dunlap, 1999), and millisecond-scale time perception, which entails the fine-tuning of motor timing (Mauk & Buonomano, 2004), perceiving time in the seconds to minutes range is essential for several adaptive behaviors in animals and can be acquired through learning. Thus, it is useful to explore the behavioral and physiological underpinnings of time perception (Buhusi & Meck, 2005). In this study, we focused on time perception ranging from few seconds to few minutes. We seek to add to the recent reviews (Lee, Thavabalasingam, Alushaj, Çavdaroglu, & Ito, 2020;Paton & Buonomano, 2018;Tallot & Doyère, 2020) that explored the neural basis of timing behavior, by providing a comprehensive review of the distinctions and classifications of time perception and the neural representations of time perception across brain regions.
Investigators have studied the neural mechanisms underlying an animal's biological clock from several perspectives (Tallot & Doyère, 2020). For example, in timing behavior studies, timing is distinguished between explicit timing, which refers to monitoring the current passage of time sequentially and acting on this information, and implicit timing, which occurs for the duration of a passive event that does not require explicit timing behavior. However, previous studies on temporal information processing mechanisms have focused on explicit representations of time (Cook et al., 2022;Heys, Wu, Allegra Mascaro, & Dombeck, Abbreviations: dStr, Dorsal striatum; EC, Entorhinal cortex; HPC, Hippocampus; LEC, Lateral entorhinal cortex; MEC, Medial entorhinal cortex; mPFC, Medial prefrontal cortex; PFC, Prefrontal cortex; SNc, Substantia nigra pars compacta. 2020), whereas implicit representations of time for the duration of events (Gill, Mizumori, & Smith, 2011;Pastalkova, Itskov, Amarasingham, & Buzsáki, 2008) were not emphasized. Furthermore, multiple candidate brain regions may possess temporal information; nonetheless, each region has been independently studied and there is limited information about the network-level interactions between them.
The present article reviews the brain regions that represent temporal information on a scale ranging from seconds to minutes, focusing on studies with rodents, which can be examined at the level of neuronal activity. In the hippocampus-entorhinal cortex (EC), particularly activity features collectively encode the event durations on a scale of few seconds, and we will describe these features and their relationship with the behavior. In addition, we will present studies of brain regions, such as the striatum, which may contribute to explicit time perception and behavior, and discuss the possibility of variations in the neural circuits involved in time perception depending on the purpose and context. Finally, we will discuss future research required to uncover the neural network-level interactions between the firing activities and regions that enable an adaptive time-based behavior.

The temporal representation of neural activity
The dependence of time perception on the generation of an internal clock in a specific brain region or the presence of multiple internal clocks in various regions warrant investigation. Researchers have described numerous brain regions that may represent temporal information; however, it is unclear if they independently contribute to the encoding of time, function in a coordinated manner, or merely reflect the temporal information processing of a specific region.
Furthermore, it is necessary to consider the possibility of neural representations being influenced by differences in the context of the task that an animal must perform (Lee et al., 2020;Paton & Buonomano, 2018;Tallot & Doyère, 2020). First, the tasks performed by animals can be broadly classified as explicit or implicit (Fig. 1). Implicit tasks do not require direct monitoring of the passage of time, but refer to the length of an event experienced, such as the duration of a presented stimulus, the delay of a memory task, or the trace period of trace conditioning. Lee et al. (2020) classified explicit tasks based on prospective or retrospective temporal information processing. In other words, they distinguished between prospective time estimation, in which the animal acts on the current elapsed time information, and retrospective time estimation, in which the animal refers to and makes decisions based on duration memory of experienced events. Evidence from previous behavioral studies suggest the possible presence of brain regions corresponding to these classifications (Jacobs, Allen, Nguyen, & Fortin, 2013;Meck, 1996).
Tasks related to temporal information processing are broadly classified according to an explicit or implicit perception of the passage of time and are further classified according to the type of task required.

The hippocampus implicitly encodes the duration of events
In the hippocampus, neurons that fire in a sequential manner during the duration of an event have been identified in implicit timing tasks on a timescale ranging from seconds to minutes (Gill et al., 2011;Kraus, Robinson, White, Eichenbaum, & Hasselmo, 2013;MacDonald et al., 2011MacDonald et al., , 2013Modi, Dhawale, & Bhalla, 2014;Pastalkova et al., 2008;Salz et al., 2016). Pastalkova et al. (2008) trained rats to select the left or right arm of the figure-eight maze after running on a wheel for a 10 s to 20 s delay period. During the performance, they monitored single-unit activity in the hippocampal CA1 region and identified neurons that fired at specific moments during wheel running, which were later termed as "time cells" (Fig. 2). This population of neurons, observed in  the hippocampus in an unchanging position, expresses the passage of time even while running at varying distances on a variable-speed treadmill; moreover, they are observed during immobility, thus distinguishing them from place cells, which fire when passing through a specific position (Kraus et al., 2013;MacDonald, Carrow, Place, & Eichenbaum, 2013).
(a) Top: The duration of the event. Bottom: The sequential activity pattern of five cells. Vertical bars indicate a single spike. In the HPC, MEC, mPFC, and dStr, neurons with specific time fields fire sequentially to denote they possibly cover the duration of the event. The neurons that form the later part of the sequence tend to fire for a longer time. (b) With an elongated duration of events, in the HPC and dStr, the time field of each neuron displayed in (a) extends, thus indicating they represent the relative time rather than the absolute time.
The time-dependent firing patterns do not merely represent the passage of time but also distinguish between contexts. The delayed alternation task conducted by Gill et al. (2011) using a plus maze involved selecting arms in a specific direction (east and west), rather than left and right, by turning on each trial. The beginning position at the end of the delay differed across trials; nonetheless, the ensemble of time cells that filled the delay formed an ensemble, thereby distinguishing between eastern and western selections. Hence, hippocampal time cells do not merely represent the passage of time during the delay time, but are related to the processing of the episode, including the context of events prior to the delay or those to be selected in the future. Disrupted time-cell firing sequences were associated with error attempts, thus suggesting they support the task success by conveying information beyond the delay time (MacDonald, Lepage, Eden, & Eichenbaum, 2011. Therefore, the activity of time cells contributes to maintaining contextual information across time and associating with temporally distant events. Supporting a theory that the hippocampus is required to relate information across temporal gaps, neurotoxic lesions and the optogenetic inactivation of the rat hippocampus disrupted the trace conditioning learning with a temporal gap from hundreds of milliseconds to tens of seconds. Thus, the learning of trace conditioning with temporal displacement is dependent on the hippocampus (Bangasser, Waxler, Santollo, & Shors, 2006;Raybuck & Matthew Lattal, 2011;Sellami, Al Abed, Brayda-Bruno, Etchamendy, Valério, Oulé, Pantaléon, Lamothe, Potier, Bernard, Jabourian, Herry, Mons, Piazza, Eichenbaum, & Marighetto, 2017; Tracey J. Shors et al., 2001;Yoon & Otto, 2007).
Hippocampal time-cells have been primarily observed during the processing of implicit temporal information for experienced events. The hippocampus represents the elapsed time of events during delay periods in memory tasks. However, the animals do not require the online detection of elapsed times for task performance. Hippocampal neurons do not represent an absolute unidirectional flow of time from the past to the future, rather they represent relative elapsed time. The time cells that fire in the late stage of the delay period display a longer firing duration than those in the early stage (Kraus et al., 2013;Salz et al., 2016). Moreover, some cells extend their firing timing to match the time scale (Shimbo, Izawa, & Fujisawa, 2021), whereas others randomly change their firing timing (MacDonald et al., 2011). In the discrimination task between forced treadmill running times, a group of neurons in CA1 represented two running durations relative to each other (Shimbo et al., 2021). In other words, the neural activity in the hippocampus does not reflect an absolute flow of time, such as a clock, rather a gradual representation of the duration of an event that is subjectively experienced from the beginning to the end, encoding it relative to the durations of similar events.
As described above, the neural activity associated with implicit timing in the hippocampus is well understood. However, the contribution of the hippocampus to explicit timing is controversial. In explicit retrospective timing tasks, hippocampal time cells fire to fill the stimulus duration, but it is unclear whether these signals are used to guide behavior (Shimbo et al., 2021). The hippocampus is thought to play a primary role in episodic memory consolidation. Whether these firing activities during explicit timing tasks are signals related to explicit timing abilities or signals related to the "when" component of episodic memory is a question that should be explored. This question could be resolved by lesion or optogenetics studies. Several lesion studies report that hippocampal lesions have no effect on explicit timing behavior (De Corte, Farley, Heslin, Parker, & Freeman, 2022;Dietrich & Allen, 1998;Jacobs et al., 2013). On the other hand, other studies have reported that hippocampal lesions impair behavioral performance during explicit timing tasks (Meck, Church, Wenk, & Olton, 1987;Meck, 1988;Meck, Church, & Olton, 1984). However, it is noteworthy that the impairment of explicit timing due to hippocampal impairment is temporary and not complete (Meck, 1988).
Based on these studies, it is possible that the role of the hippocampus in explicit timing is in the memory aspect of the task, such as duration memory or context memory. For example, patients with damage to the temporal lobe, including the hippocampus, perform explicit prospective timing tasks normally (Shaw & Aggleton, 1994), but explicit retrospective timing, which requires duration memory information, is impaired (Melgire et al., 2005; but see Jacobs et al., 2013).
The generation of the hippocampal representation of time within the hippocampus and its dependence on external inputs remains questionable (Eichenbaum, 2014). Time cells have been principally identified in CA1 of the hippocampus; however, later studies have identified them in CA3 (Salz et al., 2016). Moreover, their relationship to temporal information processing in the upstream EC and cortical areas warrants investigation.

The medial entorhinal cortex encodes both implicit and explicit time
The EC is a major input/output source of the hippocampus and may contribute to the formation of time cells (Eichenbaum, 2017). It is classified into the medial entorhinal cortex (MEC) and lateral entorhinal cortex (LEC) based on its connections with other brain regions and differences in the neuronal structure and function (Witter, Doan, Jacobsen, Nilssen, & Ohara, 2017), thus leading to the hypothesis that functional differences exist in temporal information processing.
In the MEC, researchers have identified time cells similar to those recorded in the hippocampus (Fig. 2). Kraus et al. (2015) implied that the grid cells in rat MECs, which fire in a location-dependent manner (Hafting, Fyhn, Molden, Moser, & Moser, 2005), form fields for the passage of time and running distance. Lesion studies of the MEC have demonstrated that it impairs trace conditioning, thus suggesting it may be an organized activity for passing information across time, similar to the hippocampus (Esclassan, Coutureau, Di Scala, & Marchand, 2009;Morrissey, Maal-Bared, Brady, & Takehara-Nishiuchi, 2012;Otto & Eichenbaum, 1992).
The function of the MEC may not be limited to bridging temporal gaps; it may also be involved in explicit time perception. Two studies using calcium imaging and virtual reality-generated linear tracks reported that the MEC generates time-dependent activity sequences during active waiting, thereby suggesting its involvement in learning to generate target waiting times Heys & Dombeck, 2018). Using an explicit prospective timing task where mice waited in a fixed position for 6-s, Heys and Dombeck (2018) found systematic formation of cells in the MEC that fired at specific moments within the interval. Interestingly, the time cells recorded in this study were also present during spontaneous pauses in the task and the moments of exposure to novel environments, thus suggesting the presence of an activity that informs the hippocampal temporal coding formed by learning. In addition, Heys et al. (2020) examined the causal relationship between the waiting behavior and neural activity in the MEC by manipulating it using optogenetics. MEC inactivation during the learning phase of the waiting period prevented the mice from learning the waiting behavior, whereas its inactivation during the waiting behavior following learning did not affect the generation of the accurate waiting time . In other words, the MEC is involved in learning the duration for waiting but not for the detection of the online passage of time. However, considering the timescales may possibly change the responsibility for temporal information processing (Kraus et al., 2013), the 6-s waiting time was relatively short for an interval timing task. Moreover, the results may possibly change for the online detection of longer durations. For example, in a recent study using an explicit retrospective timing task with a 10-and 20-s interval, neurotoxic lesions in rat MEC significantly impaired discrimination performance during 20-s trials (Vo et al., 2021).

The lateral entorhinal cortex represents implicit time by ramping activity of single neurons
The temporal representation of the LEC displayed characteristics different from those recorded in the hippocampus and MEC. Tsao et al. (2018) recorded neural activity from the LEC of rats freely exploring food scattered in a square-shaped open field. In trials with a 150-s interval, the color of the wall randomly switched between white and black every 250 s. An analysis of the individual neurons revealed a decrease and increase in the firing rate at various time scales. For example, some cells displayed a gradual increase in the firing rate towards the end of each trial, whereas others displayed a gradual decrease in the firing rate throughout the session. Unlike the temporal coding features of hippocampal-ECs, the LEC represents event durations by gradually increasing or decreasing the firing rate of a single neuron, termed ramping activity (Fig. 3a). Ramping activity is categorized as one of the activity patterns commonly observed in cortical areas (Narayanan, 2016;Parker, Chen, Kingyon, Cavanagh, & Narayanan, 2014).
Time-dependent single neurons in the mPFC and LEC tend to modulate their firing monotonically or non-monotonically along the temporal axis during an event. (a) Cells 1 and 2 demonstrate monotonically modulated ramping activity, which gradually increases or decreases the firing rate from the beginning of an event. (b) Cell 3 displays an increase in their firing rate at the beginning and end of an event, whereas Cell 4 displays a gradual increase from the beginning of the event and a decrease towards the end.
Firing features that provide long durations in the LEC contain information on both the duration of each event and the temporal distance relationship between events, thus suggesting they may contribute to the formation of episodic memory with integrated where-when-what information represented in the hippocampus (Bright et al., 2020;Rolls & Mills, 2019;Tsao et al., 2018). The LEC contributes only to implicit time perception and episodic memory formation on a several-minute scale. However, no studies have reported on the activity in the several-second timescale or studies related to explicit temporal information processing. Lisman proposed the hypothesis that EC subregions utilize different frameworks to provide spatial information to the hippocampus for resolving the task (Lisman, 2007). Differences in the firing activity of EC subregions are similar for spatial and temporal information processing (Issa, Tocker, Hasselmo, Heys, & Dombeck, 2020;Save & Sargolini, 2017); therefore, this hypothesis may be applicable to temporal information-processing mechanisms.

The prefrontal cortex represents time in terms of the firing modulation of single neurons and sequential activity by populations
Neurons in the prefrontal cortex (PFC) exhibit different timedependent activities at both single neuron and neuronal population levels. At the single-cell level, each neuron represents the duration of an event by modulating its firing frequency (Fig. 3). Narayanan and Laubach (2009) trained rats to maintain a lever press for a specific time. Neurons in the PFC gradually increased their firing rate from the beginning to the end of lever pressing. This may attribute to the use of a temporal encoding strategy similar to the ramping activity (Fig. 3a) reported in the LEC (Tsao et al., 2018). Emmons et al. (2017) showed that monotonic ramping activity in the medial prefrontal cortex (mPFC) is scaled in response to changes in the time interval of the fixed interval task. Such flexible scaling for time intervals could be used to resolve explicit timing tasks (Emmons et al., 2017;Xu et al., 2014). Moreover, impairment of the mPFC disrupted the performance of explicit retrospective timing tasks (Dietrich & Allen, 1998;Emmons et al., 2017).
Furthermore, explicit timing behavior has been suggested to be achieved by nonlinear modulation of single cells. Xu et al. (2014) reported that single neurons in the mPFC of rats exhibited diverse modulation patterns when timing a 2.5-s sound stimulus. During sound stimulation, showed peak firing rates at the onset and offset of the sound stimulus, whereas others increased their firing rate at the onset of the sound and continued to fire continuously. Subsequently, they increased the rate at the end of the sound (Fig. 3b). An analysis of the firing rate of time-dependent cells in the task indicated that 65% of the cells exhibited monotonic firing rate modulation, either gradually increasing or decreasing the firing rate, whereas the remaining 35% exhibited nonmonotonic firing rate modulation. Such non-monotonic firing activity differs from monotonic ramping activity in the LEC (Tsao et al., 2018) and posterior parietal lobes (Leon & Shadlen, 2003). In explicit timing tasks where subjects generate target time intervals, neurons in medial frontal cortex were also reported to show complex response profiles that are heterogeneous and nonlinear (Wang, Narain, Hosseini, & Jazayeri, 2018). These timing-related signals were also scaled for different target time intervals. These studies suggest a contribution of linear/nonlinear modulation of single neurons to explicit prospective timing, but population firing activity is considered a signal for implicit timing.
Researchers have reported the existence of population-level activity, where neurons sequentially fire during intervals (Fig. 2) (Kim et al., 2013;Tiganj et al., 2017). This neuronal population shares some features with hippocampal time cells (Gill et al., 2011;MacDonald et al., 2011), such as the large number of cells that fire at the beginning of the sequence (Eichenbaum, 2014;Salz et al., 2016), longer firing times of cells with time fields later in the sequence, and lower accuracy with an increased elapsed time (Tiganj et al., 2017).
The firing modulation at the single neuron level and sequential activity at the population level communicate comparable amounts of temporal information; however, encoding schemes may differ owing to the differences in purpose (Tiganj et al., 2017). For example, Simen, Balci, deSouza, Cohen, and Holmes (2011) considered the possibility that population activity is used to construct memories of events, and that single neuron modulation contributes to decision making. Howard et al. (2014) proposed the need of ramping activity for the formation of sequential activity.

The striatum represents explicit temporal information in population activity
The basal ganglia process temporal information (Buhusi & Meck, 2005;Yin, 2014). Particularly, the striatum is a major input region of the basal ganglia (Hunnicutt et al., 2016;Yelnik, 2002), and its subregion, the dorsal striatum, supports the ability to make behavioral decisions based on explicit time perception, as evidenced by impaired lesioninduced performance on timing tasks (Meck, 2006).
In the dorsal striatum, neurons reportedly fire at different times during timing tasks. Mello et al. (2015) recorded neural activity in the dorsal striatum of rats during explicit prospective timing task that was rewarded by lever pressing following a fixed interval. Neurons in the dorsal striatum fired at different moments of the fixed interval duration, similar to the time cells observed in the hippocampus (Pastalkova et al., 2008). Interestingly, the firing activity of time-coding neurons in the dorsal striatum scaled their time field relative to the length of the interval upon switching to a different fixed interval (12 s-24 s-36 s-46 s-60 s). Gouvê et al. (2015) studied dynamic population activities of the dorsal striatum in an explicit retrospective timing task, in which rats explicitly discriminated whether the interval between two shortduration sound stimuli presented in succession was shorter or longer than 1.5 s. This population activity predicted the rats' judgments of the stimulus duration. Furthermore, inhibition of the striatal activity by muscimol impaired task performance. Toso, Reinartz, Pulecchi, and Diamond (2021) recorded neural activity in the dorsolateral striatum receiving input from the vibrissal somatosensory cortex while the rats performed an implicit timing task of discriminating the intensity of vibratory stimulus to their whiskers. Neurons in the dorsolateral striatum encoded the duration of the stimulus, similar to the aforementioned dorsolateral striatum study (Gouvê et al., 2015;Mello et al., 2015). Thus, the dorsolateral striatum implicitly encodes the time even when the monitoring of the passage of time is not required. In addition, the temporal coding pattern of the dorsolateral striatum did not change, despite incorrectly performing the temporal discrimination task of two successively presented vibratory stimuli (Toso et al., 2021).
Therefore, a subregion of the striatum may have a clock function that constantly monitors the time regardless of an explicit or implicit time.

Dopamine neurons control the subjective length of time
Midbrain dopamine neurons have been implicated in several cognitive functions, such as motivation, attention, and emotion (Bromberg-Martin, Matsumoto, & Hikosaka, 2010;Cools, 2008;Schultz, 1998). Numerous dopamine neurons in the substantia nigra pars compacta (SNc) project to the dorsal striatum (Meck, 2006), and their signals may influence the perception of time. Soares, Atallah, and Paton (2016) examined the role of dopamine activity during a task that required mice to judge whether the interval between two sound stimuli was shorter or longer than 1.5 s. Using fiber photometry to record calcium activity in the SNc during the explicit timing task, they observed that the mice were more likely to judge the duration of a sound stimulus as short during high dopamine activity; conversely, they were more likely to judge the duration as long during low activity. Furthermore, optogenetic activation and inactivation of dopamine neurons demonstrated a causal relationship between their activity and behavioral performance (Soares et al., 2016). Thus, the activity levels of SNc dopamine neurons may modulate the subjective perception of the passage of time, based on the internal factors in response to events. For example, the feeling of a rapidly passing pleasant event could be explained by the sensation that the activation of dopamine neurons promotes an underestimation of time and decelerates the internal clock. The aforementioned dopamine neuron activity may be an adaptive function of displaying the elapsed time of a beneficial event shorter for oneself, such that we experience that event for longer.

Discussion
In this study, we reviewed articles on the major brain regions involved in time-related neural activity. The hippocampus, MEC, PFC, and striatum displayed population activity that sequentially fired in response to the elapsed time of events as a shared neural ensemble representation. By contrast, the monotonic and nonmonotonic modulation of the firing rate of single neurons in the PFC and LEC encode the elapsed time. Their time-dependent activity reflects the relative time of subjectively experienced events, rather than generating an absolute time, such as that by a clock. Thus, the time-dependent firing activity may reflect the experience of underestimating or overestimating the passage of time owing to internal factors. In addition, different neural circuits may be prioritized for dissimilar purposes in time-informed brain regions, and the characteristics of diverse experiments, such as explicit timing, specific duration events, and implicit delays during working memory tasks, suggest possible modification of the primary temporal calculation regions and neural circuits.
Hippocampal time cells that reflect the implicit temporal information may be a part of the encoding of episodes that include the location and context. In MEC, the time cell population encodes time not only during the perception of implicit temporal information but also during the perception of explicit time. The LEC has a larger timescale and implicitly encodes event durations by ramping activity at the singleneuron level. The PFC encodes elapsed time by both single neuronlevel linear or non-linear activities and population-level dynamic activity, each of which may be used for different purposes. The striatum encodes explicit temporal information and may support the ability to act on the passage of time; the dopamine neurons of the SNc may be involved in the explicit judgments of the subjective passage of time.
This review focused on neural correlates, whereas many lesion studies have provided important insights into the causal relationship between timing behavior and neural activity. Hippocampal lesions impaired implicit timing but did not significantly impair explicit timing (Dietrich & Allen, 1998;Jacobs et al., 2013). Inactivation of the MEC impaired explicit prospective timing learning; however, it did not impair behavioral performance of learned explicit prospective timing tasks . Lesions of the PFC impaired explicit prospective timing behavior (Dietrich & Allen, 1998;Emmons et al., 2017), and inactivation of the dorsal striatum impaired many explicit prospective and explicit retrospective timing behaviors (Gouvê et al., 2015;Mello et al., 2015). Inactivation of dopamine neurons in the midbrain distorted subjective perception of passage of time (Soares et al., 2016).
We focused on individual brain regions that may function as part of the internal clock; however, further research is required to understand the functional interaction among these brain regions. The neural basis for temporal information requires simultaneous recordings of the neural activity from multiple regions while performing a single task because of varying context, task demands, and timescale across studies (Paton & Buonomano, 2018).
This finding may allow us to understand the network-level differences between explicit and non-explicit timing and differences by the purpose/context and timescale (Jacobs et al., 2013), thereby classifying the temporal information processing mechanisms (Fig. 4). PFC is separately involved in both explicit and implicit timing during different activities, which may enable us to distinguish between circuits and subregions in detail. Many explicit timing studies have reported neural activity in the mPFC with strong contacts with the dorsal striatum (Gabbott, Warner, Jays, Salway, & Busby, 2005;Wall, De La Parra, Callaway, & Kreitzer, 2013). On the other hand, implicit timing studies have reported neural correlates in the orbitofrontal cortex in addition to the mPFC (Bakhurin et al., 2017;Otto & Eichenbaum, 1992). The orbitofrontal cortex has strong interconnections with the hippocampus through direct and indirect channels (Wikenheiser & Schoenbaum, 2016). Therefore, temporal information processing processes in the PFC could be separated into circuits and subregions.
Simultaneous recordings of multiple regions would provide important insights to understand the dynamic timing processes at the neural network level. For instance, Bakhurin et al. (2017) recorded neural activity in the striatum and prefrontal cortex of mice using a silicon probe and observed sequential activity in both regions. By comparing the accuracy of time coding between both regions, they observed that the striatum encoded time more accurately. Moreover, the connections between regions can be assessed by analyzing neural oscillations (Buhusi & Meck, 2005;Buzsáki & Draguhn, 2004). Researchers have implicated synchronous theta oscillations in the PFC and hippocampus in memory and learning (Fujisawa & Buzsáki, 2011), and synchronous activity is possibly used to transfer temporal information during the temporal assessment tasks. This necessitates clarifying the temporal coding strategy for each brain region for different purposes and understanding the representation of temporal information at the network level in the future.
(a) Based on the reviewed neurophysiological evidence, a schematic representation is provided for the relationships between brain regions that may contribute to the success of an explicit timing task. The dStr is considered to integrate sensory information from the cortical area and signals related to memory and action planning from the HPC and mPFC to generate a subjective passage of time that is reflected in behavior. In addition, dopamine input from the SNc accelerates the pace of the internal clock in the dStr, consistent with the model devised principally based on behavioral findings (Matell, Meck, & Nicolelis, 2003;Matell & Meck, 2004). (b) Based on the neurophysiological evidence reviewed in the present paper, a schematic representation of the relationships between the brain regions associated with implicit temporal encoding is depicted. The HPC encodes the time of events as a region that integrates spatiotemporal information to generate episodic memories. The   Fig. 4. Neural circuitry responsible for explicit or implicit time-perception tasks. seconds-time scale population coding in MEC and the minutes-time scale single neuron coding in the LEC may project signals to the HPC to accurately reflect the temporal location of events. The activity of timerelated neurons in the mPFC reflects the onset and termination of delay and stimulus events and may share with HPC the information necessary for non-temporal task performance. The brain regions are depicted in the sagittal sections. The black arrows indicate the direction of the time-related signal flow. mPFC: medial prefrontal cortex, dStr: dorsal striatum, HPC: hippocampus, SNc: substantia nigra pars compacta, MEC: medial entorhinal cortex, and LEC: lateral entorhinal cortex.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability
No data was used for the research described in the article.