Information procession requires a dynamic coordination of activity of large groups of neurons in different brain structures. Brain rhythms organize neuronal activity in sensory information processing and cognitive tasks. The coordination of neurons is necessary both at the level of information transmission between brain structures and the neuronal ensembles within one brain area (Fries, 2015).

Rhythmic activity in the brain does not disappear during rest, when attention is not focused on external stimuli. But does it carry an information load or is it an “offline” state for neuronal computations? What are the mechanisms of transition between different rhythmic modes during active behavior and the resting state? We will consider these questions in regard to the hippocampus, since a lot of information about rhythms was obtained on this brain structure. The case of patient HM demonstrated the impossibility of acquisition of new memories without the participation of the hippocampus (Corkin, 2002; Buzsáki and Moser, 2013). The discovery of place cells, hippocampal neurons encoding the space, made the hippocampus a model object of memory research at the neuronal level (O’Keefe, 1976). The investigation of attention shows that the hippocampus is a central link in detecting the novelty of stimuli. It is probably the hippocampus that determines which information is new and requires memorization (Vinogradova, 2001; Numan, 2015).

There are two main functional states or modes of the hippocampal activity, theta and non-theta states (Buzsáki, 2002; Schultheiss et al., 2020). They have different behavioral correlates and a clearly different spectral content of local field potentials (LFPs) and neuronal spiking. The hippocampal theta state manifests itself during active exploratory behavior, locomotion, cognitive situations requiring attention, and rapid eye movement (REM) sleep. The slow-wave sleep (SWS) and quiet wakefulness (immobility, eating, grooming) represent the non-theta hippocampal state (Vanderwolf, 1969; Buzsáki, 2002; Young and McNaughton, 2009; Colgin, 2013). In the theta state, hippocampal LFPs exhibit strong regular theta (4–12 Hz) and gamma (30–120 Hz) oscillations (Bragin et al., 1995; Csicsvari et al., 2003; Tort et al., 2009; Belluscio et al., 2012). In the non-theta state, hippocampal activity is dominated by slower and often irregular oscillations in the delta (0.5–4 Hz) frequency range, interrupted by highly synchronous neuronal activity, sharp wave-ripple complexes (SWRs; Buzsáki, 2015). Theta and non-theta hippocampal states are traditionally characterized as two opposite and mutually exclusive states, the “online” and “offline” modes of information processing (Buzsáki, 2002).

However, “offline” does not mean that there is no active information processing. In the non-theta state, it was shown that memory consolidation takes place (Carr et al., 2011; Vandecasteele et al., 2014; Buzsáki, 2015; Mizuseki and Miyawaki, 2017). Moreover, recent studies demonstrated that in this mode, information is transferred from the hippocampus to other brain structures. For example, it was found that the activation of neuronal ensembles duri SWS in the medial prefrontal cortex (PFC) correlates with the preceding hippocampal ripples (Todorova and Zugaro, 2020).

The research on hippocampal rhythms is mainly devoted to the theta state. From numerous studies and reviews (Vinogradova, 1995; Buzsáki, 2002; Colgin, 2013; Tsanov, 2015), we know about the theta-gamma coupling (Belluscio et al., 2012), phase precession of place cells (O’Keefe and Recce, 1993), theta and gamma coherence between hippocampal regions during cognitive tasks, etc. (Mysin and Shubina, 2022). The non-theta state is investigated principally in the context of ripple oscillations (Buzsáki, 2015). Other aspects, such as delta and gamma rhythms, the functional significance of the non-theta mode and the transition between operational modes and its possible mechanisms, are poorly known and have not yet been discussed in reviews. We will try to correct this “bias” and systematize the available experimental data concerning the hippocampal non-theta state. We propose possible mechanisms underlying the transition between hippocampal theta and non-theta states and review the currently available data on the functional significance of the non-theta mode. We focus mainly on studies of natural waking and sleep states and studies on anesthetized animals are specifically mentioned.

Hippocampal network activity is spatially and temporally tuned in a behavior-dependent manner. During quiet wakefulness, consummatory behavior and grooming non-theta activity replaces “theta state” patterns (Figure 1). This activity includes low-frequency oscillations of different amplitudes, SWRs, and SWRs-associated gamma rhythms (Buzsáki et al., 1983; Buzsáki, 1986; Bragin et al., 1992, 1995). It should be noted that similar electrophysiological activity could be registered in the hippocampus during drowsiness and SWS (Jarosiewicz et al., 2002; Buzsáki, 2015).

The functional state of the brain modulates the firing rate and patterns of individual neurons in the hippocampus (Figures 1, 2). In 1987 Colom and Bland described, in urethane-anesthetized rats, the presence in the hippocampus of “theta-off” cells, which differed from “theta-on” classical “theta” cells. “Theta-off” cells generated non-complex-spikes and were inactive during theta activity, discharging preferentially during the non-theta state (Colom and Bland, 1987; Ford et al., 1989). Later, in addition to theta-related cells, theta-unrelated neurons were recorded (also in urethane-anesthetized animals), which discharged with simple or complex spikes or were even “silent”. Thus, hippocampal pyramidal neurons were functionally heterogeneous in the generation of current LFPs (Bland et al., 2005).

The difference between hippocampal operational modes (theta and non-theta states) is also manifested in pathological conditions such as seizure activity. It was shown in a rat model of temporal lobe epilepsy that the hippocampal functional state preceding seizures influenced spontaneous seizure activity and determined the success of therapeutic intervention. Optogenetic stimulation significantly curtails seizures that emerged from non-theta states (Ewell et al., 2015).

In this section we summarize data on hippocampal LFPs and neuronal activity in the non-theta state. We will start with less popular low-frequency oscillations and then move on to much more studied SWRs.

Transitions between different hippocampal functional states occur relatively fast, within a few seconds (Schultheiss et al., 2020). During behavioral state transitions, hippocampal oscillatory activity may drop in several frequency bands (Lapray et al., 2012). The synchrony in one frequency band is suddenly replaced by synchrony in another. Pronounced delta or theta rhythm usually occurs when synchrony in the other band is weak. However, it is worth noting that such transitions are not always “clean”: in some cases, delta and theta correlate positively over short time intervals (Schultheiss et al., 2020).

When the behavior of an animal changes from the theta to the non-theta state, the hippocampal network excitability also changes. The activity of local interneurons, as well as septal and entorhinal inputs has different temporal structure compared with theta oscillations. During transitions from the theta to the non-theta state in freely moving and sleeping rats, the firing rate of hippocampal theta cells (presumably interneurons) may be significantly reduced. In contrast, the activity of complex-spike cells (presumed pyramidal cells) may increase (Stewart, 1993). Recording hippocampal CA1 units in awake guinea pigs, Rivas and colleagues showed that, when the animals change from the standing position to walking, the majority of units (presumed pyramidal cells) increased their rhythmicity and/or firing frequency or phase-locking with theta rhythm (Rivas et al., 1996). More recently it was shown that these transitions between different behavioral states with overall low oscillatory activity are accompanied by a decrease in the firing of PV basket cells, which may facilitate the network reorganization (Lapray et al., 2012; Kocsis et al., 2022).

Cortical slow and delta oscillations are believed to play an important role in memory consolidation, especially by facilitating the information flow between the hippocampus and the neocortex (Sirota et al., 2003; Maingret et al., 2016; Todorova and Zugaro, 2019). The functional significance of hippocampal delta oscillations and non-theta activity in general is not as well characterized. It has been noted in several studies that learning during non-theta states is less effective than during theta states. In the absence of theta oscillations, the rate of acquisition of eyeblink conditioning in rabbits was significantly slower, and the percentage of conditioned responses was lower compared with animals trained during theta states (Seager et al., 2002; Griffin et al., 2004; Hoffmann and Berry, 2009; Cicchese and Berry, 2016).

The traditional view that theta and gamma represent the “online”, and delta reflects the “offline” mode of information processing, is opposed by the results of Furtunato and colleagues (Furtunato et al., 2020). Analyzing LFPs in the left and right rat hippocampi during consecutive short-term treadmill runs at the same speed, they found prominent delta and theta oscillations. Delta power and interhemispheric phase coherence in the delta band increased during consecutive treadmill runs. At the same time, theta and gamma power decreased with the trial number (Figure 6). Thus, consecutive treadmill runs enhanced delta power and inter-hemispheric phase coherence in the rat hippocampus (Furtunato et al., 2020).

In general, it is believed that, in rodents, theta and non-theta hippocampal modes reflect different network coupling and distinct behavior. Analyzing the cortical-cortical and cortical-hippocampal coherence in freely behaving rats, Young and McNaughton showed that the intensity of theta and delta oscillations clearly depended on animal behavior (Young and McNaughton, 2009). As opposed to theta activity, the power of the delta rhythm was maximal during grooming and resting, and minimal during locomotion, exploratory behavior, and rearing. During “non-cognitive” behavior, there was also an extensive delta coupling in the frontal cortex, which was not associated with the posterior areas and the hippocampus (Young and McNaughton, 2009). The authors believed that rhythmic activity of different frequencies may bind different neuronal ensembles into functional groups in different circumstances. In a recent study on freely behaving rats, this assumption received confirmation (Schultheiss et al., 2020). The periods of delta-dominated activity were shown to be prominent in the hippocampus when animals are motionless or moving slowly (non-theta state). Such delta-dominated modes were orthogonal to theta modes prevailing in running. Delta power and synchrony in the hippocampus were negatively related to the speed of an animal. The orthogonality of delta and theta modes was also manifested in the interaction of the hippocampus with medial PFC. While theta coherence between these structures increased during running, delta coherence increased during quiet wakefulness (Schultheiss et al., 2020).

Thus, the hippocampal delta rhythm, along with theta, may serve as a means of communication between the hippocampus and the neocortex (Roy et al., 2017; Mofleh and Kocsis, 2021). Roy and coauthors observed decreased delta coherence between the hippocampus and the PFC in anesthetized animals when the influence of the nucleus reuniens of the thalamus was abolished (Roy et al., 2017). The authors suggested that rhythmic couplings in delta and theta bands may serve as parallel communication channels between the PFC and the hippocampus operating in opposite directions. Later, the coupling between the hippocampus and the PFC at the respiratory rate (2 Hz, delta-range) was also shown in freely behaving rats (Mofleh and Kocsis, 2021).

To provide a more complete description of cognitive processes occurring in the non-theta state, we will briefly review the functions of SWRs. The most discussed functions of SWRs are memory consolidation, route planning, and information transfer from the hippocampus to the neocortex. These functions are based on the phenomenon of reactivation of theta sequences, i.e., the activation of place cells in the same neuronal ensemble in the order in which the rat traverses place fields (Both et al., 2008; Buzsáki, 2015). The hypothesis about the role of SWRs in memory consolidation is based on the fact that the SWRs-related reactivation of place cell sequences is accelerated compared to the theta-state reactivation. The activation of place cells in the “correct” order with an interval of a few milliseconds should strengthen the connection between neurons according to the STDP (Bi and Poo, 1999). This hypothesis received direct experimental confirmation. The blockade of SWRs generation by commissural stimulation impairs memory consolidation (Girardeau et al., 2009). There is also a lot of indirect evidence. For example, it was shown that the stability of the reactivations of pyramidal cells positively correlates with the quality of memory consolidation (Carr et al., 2011).

The hypothesis about the “route planning” (retrieval of “episodic-like” memory) function of SWRs is based on observations that place cells are activated in the order in which the animals later will run through the maze (Pfeiffer and Foster, 2015; Wu et al., 2017; Pfeiffer, 2020). Studies in humans also show that SWRs play a role in episodic memory retrieval. Sakon and Kahana found that just before word recall from previously studied lists, rate and power of SWRs raised. Moreover, if the words in the sequence were semantically related (for example, fish and whale, cupcake and pie), then the effect was stronger (Sakon and Kahana, 2022).

In 1971, D. Marr proposed the idea of the hippocampus as a temporary storage of information (Marr, 1971; Willshaw and Buckingham, 1990). This hypothesis received many confirmations, and SWRs are considered today as a reflection of the information transmission from the hippocampus to the neocortex (Willshaw and Buckingham, 1990; Buzsáki, 2015). The main arguments in favor of this hypothesis are the data on the induction of neuronal and field activity in the neocortex after SWRs in the hippocampus during SWS. Hippocampal SWRs were shown to induce delta oscillations and sleep spindles in the PFC (Sirota et al., 2003; Maingret et al., 2016; Todorova and Zugaro, 2020; Skelin et al., 2021). Other studies demonstrate the reactivation of neuronal ensembles in the PFC after hippocampal SWRs (Peyrache et al., 2009), as well as the important role of the hippocampus in stabilizing neuronal patterns in the PFC after sleep (Euston et al., 2007). In the context of this hypothesis, it is important to understand how cognitive processes occurring during hippocampal delta oscillations are related to SWRs. We have already noted that the hippocampal delta rhythm may also serve to interact with the neocortex. However, it is not known whether there is any connection between cognitive processes occurring in the non-theta state “outside” and “inside” SWRs.

Whether all described functions “coexist” in one SWR event is unknown. Perhaps SWRs are heterogeneous, and some cognitive operations are performed in one SWR, and others in another (Buzsáki, 2015). But it is evident that SWRs are the most discussed aspect of the non-theta state. On the other hand, there is no scientific consensus about the cognitive load of the non-theta-non-SWR mode.

It is common practice that patients suffering from drug-resistant epilepsy undergo invasive electrophysiological diagnostics with implantation of intracranial depth electrodes to localize the focus of seizure onset. This provides a unique opportunity to conduct additional studies of the rhythmic activity of the hippocampus in humans.

When recording hippocampal activity in 10 patients, Watrous and colleagues attempted to determine the behavioral correlates of human delta and theta rhythms. They showed that the power of delta and theta activity increased with increasing speed of navigation in virtual reality on a significantly larger number of electrodes than would be expected if it happened randomly. However, delta and theta power were modulated more uniformly by the “spatial view” when subjects looked at “spatial landmarks” in the virtual environment (Watrous et al., 2011). Later, the same group of researchers demonstrated that low-frequency oscillations (delta and theta) in the human hippocampus are not associated with sensorimotor processing but are involved in encoding the spatial information, for example, distance-related (Vass et al., 2016). This was shown with the help of short- and long-distance teleporters in virtual reality. Navigating in a cross-shaped environment, patients could use short-distance teleporters, placed about half way along the plus maze arms, or long-distance teleporters, placed near the ends of the maze arms. By the locations of teleporters, patients could predict the distance they would travel (short or long). The use of teleporters made it possible to avoid sensorimotor processing during navigation (visual cues were absent). During long-distance teleportation (the teleportation time did not depend on the distance), the periods of low-frequency oscillations were also longer. Thus, the pattern of oscillations in the hippocampus during teleportation made it possible to determine what the teleported distance was, long or short (Vass et al., 2016). It is worth to note, that, studying neuronal activity in humans during virtual navigation, Jacobs and colleagues show that most hippocampal neurons are modulated by delta (1–4 Hz) rather than theta rhythm (4–8 Hz; Jacobs et al., 2007).

Human hippocampal delta and theta rhythms (as well as gamma) may also be involved in encoding the environmental novelty. During the search for objects and memorizing their location in virtual reality, hippocampal delta and theta oscillations in the first block of tasks were smaller than in the subsequent ones, in contrast to the gamma rhythm, which was higher in the new environment (the first block of tasks; Park et al., 2014; Figure 7).

The synchrony of the rhythmic activity of the hippocampus and the entorhinal cortex was analyzed using recordings from the non-epileptic brain hemisphere of 22 patients with unilateral hippocampal sclerosis (Mormann et al., 2008). The analysis of phase coherence throughout different frequency ranges revealed a lower interstructural (entorhinal-hippocampal) synchrony in comparison with the levels of intrastructural synchrony. However, the effect reached statistical significance only in the delta and theta ranges. The authors suggested the presence of independent delta/theta rhythms in the entorhinal cortex and the hippocampus in humans (Mormann et al., 2008). It is interesting to mention here an earlier work in which it was shown that in patients with temporal lobe epilepsy theta coherence between the hippocampus and the perirhinal/entorhinal cortex increased with successful word memorization (Fell et al., 2003). Thus, the formation of declarative memory probably requires an active synchronization of independent delta/theta oscillations in the hippocampus and the entorhinal cortex.

In contrast to rodents, where the theta and delta modes probably reflect different brain states, in humans, a similar pattern of changes in hippocampal theta and delta oscillations is observed during navigation. Moreover, in humans, correlated change in the delta and theta rhythms is observed not only in the hippocampus, but also in the neocortex in virtual navigation and word memorization tasks (Kahana et al., 1999; Serruya et al., 2014). It can be assumed that the division of human hippocampal low-frequency activity into delta and theta bands is probably not always justified. Moreover, canonical frequency bands were determined on the basis of human scalp EEG studies (Niedermeyer, 1999) and may not accurately reflect the dynamics of oscillations in the hippocampus. It is suggested in the literature that the human hippocampal theta range may extend from 1 to 8 Hz (Jacobs et al., 2007; Herweg and Kahana, 2018).

Non-theta and especially non-theta-non-SWRs hippocampal modes have been little studied. In most works, these states are defined as the absence of the theta rhythm, and are not based on the analysis of specific parameters. Although the frequency patterns of the non-theta state and its behavioral correlates (SWS, quiet wakefulness) are known, many parameters of hippocampal neuronal activity, as well as causal relationships between neuronal and field activities, remain to be clarified. Based on the reviewed data, we provide a list of the most important unresolved questions about the non-theta state:

More extensive experimental and theoretical studies are needed to clarify these issues. The answers to these questions will bring us closer to understanding what the brain does when it rests and whether it rests at all.

IM has made general structure of the article, abstract, introduction, conclusions and section “2. Hippocampal activity in the non-theta state”. LS has made figures and the following sections: “3. Transition between hippocampal theta and non-theta states”, “4. Functional significance of hippocampal non-theta states”, “5. Peculiarities of non-theta states in humans”. All authors contributed to the article and approved the submitted version.