We have re-evaluated the level of inter-hemispheric synchrony of ripple oscillations. To our knowledge, this is the first study to fully quantify rodent CA1 ripple synchronization along the ipsilateral and contralateral longitudinal axes while controlling for bilateral homotopic and heterotopic sites in freely moving animals. The present work was motivated by a recent study suggesting that ripple events mainly occur asynchronously between hemispheres (Villalobos et al., 2017), a finding that contrasts with earlier literature (Buzsáki, 1989; Suzuki and Smith, 1988a; Suzuki and Smith, 1987). To address this issue, we first ensured reliable ripple detection in shanks positioned in the left and right hippocampi (Figure 1). We found that ripple abundance, peak frequency, inter-ripple intervals, and the mean number of ripple cycles per event are equivalent between hemispheres (Figure 2). We next investigated if ripples occurring within the same hemisphere have different synchronization properties compared to ripples recorded in different hemispheres. Our results demonstrate: (1) high phase synchronization among ipsilateral ripples, contrasted with much lower phase-locking for contralateral ripples (Figure 3), (2) high amplitude correlations for ipsilateral and contralateral ripples, with slightly stronger effect for ripples within the same hemisphere (Figure 4), (3) larger phase-coupling differences between hemispheres compared to amplitude coupling (Figure 5), (4) ripple events are still highly synchronized at 5ms when detected in ipsilateral shanks, though much less so in contralateral ones; however, at broader windows of 50 and 100 ms, ripple events were highly synchronized in both ipsilateral and contralateral shanks (Figure 6), and, finally, (5) CA1 pyramidal neurons and interneurons exhibit local phase-locking and global time-locking to ripple events (Figure 7). Taken together, our results show that ripple events predominantly occur in synchrony both within and between hemispheres and not asynchronously. Our results align with previous reports (Chrobak and Buzsáki, 1996; Buzsáki, 1989; Patel et al., 2013; Suzuki and Smith, 1988a; Buzsáki et al., 2003) but differ from Villalobos et al., 2017.

The CA3 and CA1 regions form a highly interconnected structure (Li et al., 1994; Laurberg, 1979; Amaral and Witter, 1989; Witter, 2007). CA3 pyramidal neurons have highly collateralized axons that project to CA3 and CA1 via two main pathways: ipsilateral (associational) and contralateral (commissural) connection systems (Shinohara et al., 2012). In the associational pathway, CA3 distributes its branches along and parallel to the CA3 septo-temporal axis through the longitudinal association bundle, terminating within CA3 at the stratum radiatum and stratum oriens (Swanson et al., 1978; Hjorth-Simonsen, 1973; Ishizuka et al., 1990; Li et al., 1994). Additionally, the same CA3 pyramidal neurons give rise to another associational pathway to CA1 via the Schaffer collaterals, which target the CA1 stratum radiatum and stratum oriens (Hjorth-Simonsen, 1973; Laurberg and Sørensen, 1981; Swanson et al., 1978; Swanson et al., 1980; Ishizuka et al., 1990; Witter, 2007). The commissural system exhibits a similar organization: CA3 pyramidal neurons across the entire septo-temporal axis project to the contralateral CA3 and CA1, reaching the same target strata (radiatum and oriens) (Laurberg, 1979; Laurberg and Sørensen, 1981). Interestingly, some CA3 pyramidal neurons send their collaterals to both associational and commissural pathways (Laurberg and Sørensen, 1981). Finally, CA3 to CA1 excitatory projections seem unidirectional, that is, the CA1 does not project back to CA3 (Hjorth-Simonsen, 1973; but see Sik et al., 1994, for evidence of CA1 GABAergic neurons that project back to CA3).

The anatomical findings are also supported by physiological evidence. Several studies have highlighted the functional role of CA3–CA3 and CA3–CA1 excitatory connectivity. Population spikes evoked by CA3 stimulation spread across distant portions of CA1 bilaterally (Finnerty and Jefferys, 1993). Using electrical stimulation, Buzsa`ki and Eidelberg, 1982 demonstrated that both CA3 associational and commissural pathways activate apical and basal dendrites of ipsilateral and contralateral CA1 pyramidal neurons. Moreover, population spikes evoked by CA3 stimulation spread across distant portions of CA1 bilaterally (Finnerty and Jefferys, 1993). In turn, despite the CA1 main outputs targeting the subiculum and entorhinal cortex, its septal (dorsal) region also projects sparsely to ipsilateral and contralateral CA1 (Swanson et al., 1978; van Groen and Wyss, 1990).

Despite the anatomical symmetry of the bilateral hippocampus, evidence for lateralization is limited (Shinohara et al., 2008; Kawakami et al., 2003; Klur et al., 2009; Shipton et al., 2014; Villalobos et al., 2017), and its potential impact on ripple features remains uncertain (Villalobos et al., 2017). Our ripple detection method followed the approach of Villalobos et al., 2017, allowing us to replicate similar ripple features reported in the literature. Ripple mean frequency distributions ranged between 140 and 150 Hz (Figure 2C, D), consistent with findings by Nitzan et al., 2022 and Gan et al., 2017. Ripple events frequently occurred as doublets, reflected by an inter-ripple interval peak between 100 ms (Figure 2E, F), in agreement with Buzsáki et al., 2003. Typically, the number of cycles per ripple event ranges from three to nine cycles (Buzsáki, 2015), and we found that most ripple events comprised six to seven cycles (Figure 2G, H), similar to Villalobos et al., 2017, Ylinen et al., 1995, and Sullivan et al., 2011, but fewer than reported by Gan et al., 2017. The ripple event rate (ripple abundance) was approximately 0.5 events per second (Figure 2A, B), which exceeds the rates reported by Villalobos et al., 2017 and Eschenko et al., 2008, but it was lower than those reported by Ego-Stengel and Wilson, 2010. Most importantly, we found no significant differences in ripple features between the left and right hemispheres. Furthermore, equivalence testing between hemispheres confirmed that ripple features were indeed equivalent within the specified bounds. Thus, while some evidence suggests CA1 lateralization, it does not appear to influence the ripple features examined in this study.

The study of ripple oscillation synchrony has a long history in rodent CA1 electrophysiology; however, we believe that ripple synchrony remains underexplored, particularly from a quantitative perspective. This conclusion is based on the lack of extensive studies identified through key term searches, citation reviews, and book chapters on SWRs (Draguhn et al., 2000; Andersen, 2007; Buzsáki, 2015; Liu et al., 2022; Morris et al., 2024). While some authors describe SWRs as highly synchronous or among the most synchronous patterns in the mammalian brain (Liu et al., 2022; Sullivan et al., 2011; Buzsáki, 2015), these claims often lack clarification and are used in varying contexts, including: (1) SWRs as the result of synchronous firing from CA3 and/or CA1 neurons (Suzuki and Smith, 1988a; Sullivan et al., 2011), (2) SWRs as synchronous field events spanning the hippocampal and parahippocampal structures (Chrobak and Buzsáki, 1994) and (3) SWRs as simultaneous events along the hippocampal septo-temporal axis and/or bilaterally (Suzuki and Smith, 1987).

Another source of confusion arises from the fact that not all studies detect sharp waves and ripples simultaneously, despite referring to them collectively as SWRs. Much of the literature has focused on either sharp waves originating from the CA1 stratum radiatum or ripples from the CA1 stratum pyramidale, rather than both. Nevertheless, it is generally assumed that most sharp waves are accompanied by ripples, and vice versa. To address this ambiguity, we explicitly clarify whether a study considered only sharp waves or only ripples, as we have done in our analysis.

To our knowledge, only a few studies have investigated SWRs synchronicity between hemispheres (Buzsáki, 1986; Buzsáki, 1989; Buzsáki et al., 2003; Chrobak and Buzsáki, 1996; Guan et al., 2021; Suzuki and Smith, 1987; Suzuki and Smith, 1988a; Suzuki and Smith, 1988b; Valeeva et al., 2019; Villalobos et al., 2017), whereas intra-hemispheric ripple dynamics along both the longitudinal and transverse axes have been more extensively analyzed (Buzsáki, 1989; Buzsáki et al., 1992; Chrobak and Buzsáki, 1996; Sullivan et al., 2011; Ylinen et al., 1995; Nitzan et al., 2022; Patel et al., 2013; Valeeva et al., 2020; Csicsvari et al., 2000; Villalobos et al., 2017). Research on intra-hemispheric SWRs synchrony has demonstrated that sharp-wave and ripple events are highly phase- and/or time-locked along the longitudinal axis within the same hemisphere of rats. For example, Buzsáki, 1989 reported the simultaneous occurrence of sharp waves and population spike bursts across distances of up to 3 mm in the rat CA1 stratum radiatum. Building on this, Buzsáki et al., 1992 demonstrated ripple phase-locking across shanks spaced up to 1.8 mm along the septo-temporal axis. However, their Figure 2B reveals variability in the averaged ripple signals, without clarifying whether the diminished averaged traces are due to inter-electrode distance. Ylinen et al., 1995 observed ripple phase-locking along the CA1 septo-temporal axis with shanks spaced 300 μm, spanning up to 1.5mm, while Chrobak and Buzsáki, 1996 reported ipsilateral phase-locked ripples over distances of 4mn to 5mm. Afterward, Patel et al., 2013 found that approximately 38% of ripple events were time-locked within 6 mm, 37% propagated bidirectionally along the septo-temporal axis at speeds of 0.33 to 0.4 mm/ms, and 21% were localized within 1 mm. Notably, our dataset spans 1.2 mm, and Villalobos et al., 2017 analyzed intervals smaller than 200 μm. This suggests that propagating ripples would be separated by less than 5 ms, a time window matching the lower limit of our ripple coincidence and correlation analyses (Figure 6).

Sharp-wave events have been observed to occur simultaneously on ipsilateral electrodes spanning 1.2–2.4 mm in the septal and intermediate CA1 regions of neonatal rats (Valeeva et al., 2020). Furthermore, Nitzan et al., 2022 reported that ripple features such as amplitude, frequency, and duration remain consistent along the longitudinal axis of CA1, with ripple events time-locked across the entire hippocampal axis, spanning from the CA1 septal regions to posterior areas and the subiculum (3 mm in mice). In contrast, Csicsvari et al., 2000 and Sullivan et al., 2011 investigated ripple synchrony along the mediolateral axis. Sullivan et al., 2011 found high ripple amplitude correlations along the CA1 transverse axis in rats using shanks spaced 300 μm, while Csicsvari et al., 2000 reported a high ripple coincidence probability within 50 ms bins using electrodes with similar spacing. Collectively, these studies demonstrate that sharp-wave events and ripple oscillations are highly phase- and/or time-locked along both the septo-temporal and transverse axes within the same CA1 hemisphere.

Conversely, inter-hemispheric ripple synchrony appears less phase-locked but still exhibits temporal coordination in the rat CA1. Buzsáki, 1986 demonstrated simultaneous sharp waves in contralateral CA1 stratum radiatum spaced 5 mm apart, with a time lag of less than 5ms and similar amplitudes triggered by homotopic regions. Similarly, Suzuki and Smith, 1987 presented a bilateral recording example of synchronous sharp waves between electrodes spaced 5 mm (homotopic sites). They also noted that sharp waves occurred simultaneously in both homotopic and heterotopic CA1 sites, with spatial shifts of 1–2.5 mm; however, this early study lacked a thorough quantification of synchronization. In a subsequent study, Suzuki and Smith, 1988a adopted a more quantitative approach, showing several superimposed traces of bilateral sharp waves spaced 5mm (homotopic sites). They suggested that contralateral ripples are not strongly phase-locked but may appear either in phase or out of phase. Buzsáki, 1989 provided an example of simultaneous bilateral sharp waves in rat CA1, also spaced approximately 5mm (based on our estimates). Chrobak and Buzsáki, 1996 reported increased ripple event coincidences between contralateral electrodes spaced 3mm, although phase coupling remained restricted to ipsilateral recordings. Buzsáki et al., 2003 observed a diminished ripple-filtered average response when using a contralateral ripple cycle as a trigger (similar to our Figure 3E, F). Nevertheless, ripple amplitudes were correlated, indicating contralateral synchrony in timing but not in phase (also similar to our Figure 4). Studies on neonatal rats by Valeeva et al., 2019 demonstrated a strong amplitude correlation and high coincidence of sharp-wave events between homotopic contralateral CA1 sites spaced up to approximately 2.4 mm. Interestingly, Guan et al., 2021 showed that silencing CA3 reduces inter-hemispheric ripple event cross-correlation, suggesting that CA3 provides a common bilateral input governing CA1 ripple timing. Collectively, these findings, along with ours, suggest that global ripple synchronization depends more on event timing than on phase coupling. In ipsilateral recordings, distance plays a critical role in phase-locking, with phase coupling strength exhibiting a linear decline as electrode spacing increases along the septo-temporal axis of CA1.

A recent study by Villalobos et al., 2017 challenged the notion that hippocampal ripple oscillations occur simultaneously across both hemispheres. To test this, they employed an arrangement of eight tetrodes, four in each hemisphere. According to their diagram, ipsilateral electrodes were spaced less than 200 μm, while contralateral electrodes were separated by up to 4.5mm. Due to their random electrode selection, ipsilateral electrodes could be positioned along either the medio-lateral or septo-temporal axis. By detecting ripple events from ipsilateral and contralateral hemispheres, they found that most contralateral ripples occurred independently. Specifically, they reported an ipsilateral coincidence of 22% within 5ms time bins and 29% within 100 ms bins, compared to a contralateral coincidence of only 3% for 5ms bins and 10% for 100 ms bins. This result is striking not only because of the low coincidence of ripples between contralateral sites but also due to the relatively low coincidence of ripples between electrodes spaced ≈ 125 µm apart within the same hemisphere.

Finally, we observed an increase in overall firing rates during ripple events, with CA1 pyramidal neurons firing maximally just before the ripple trough and interneuron activity peaking just afterward. This finding aligns with previous studies (Buzsáki, 1986; Buzsáki et al., 1992; Chrobak and Buzsáki, 1996; Csicsvari et al., 1999a; Csicsvari et al., 1999b; Csicsvari et al., 2000; Buzsáki et al., 2003). Our data further demonstrate that pyramidal neurons exhibit local ripple phase coupling, with coupling strength decaying rapidly in both ipsilateral and contralateral directions. In contrast, interneurons maintain a diminished but significant phase coupling to contralateral ripples, likely supported by their connections to commissural and associational excitatory fibers (Deller et al., 1994). Supporting this, Stark et al., 2014 showed that localized CA1 pyramidal activation induces ripple events with multiple ipsilateral loci phase-locking ripple cycles. They also demonstrated that disrupting GABAA signaling or silencing PV+ basket cells reduces spatial and spiking coherence in the ripple band while increasing interneuron phase-locking. Similarly, Patel et al., 2013 reported that the phase-locking between ripples and multi-unit spiking activity diminishes with increasing distances from ripples detected at the most septal CA1 recording site but remains consistently time-locked to them. Additionally, Suzuki and Smith, 1988b found that high-frequency stimulation of anterior CA1 evokes ripple-like events bilaterally, likely mediated by CA3 bursts, consistent with our observation of a global increase in firing rates during ripples. Firing rate synchronization during ripples also occurs along the medio-lateral axis. Sullivan et al., 2011 and Csicsvari et al., 2000 showed that interneurons couple to distant recording sites along the transverse axis, whereas pyramidal cells remain locally coupled – a pattern we also observed along the longitudinal axis. One potential role of time-locked contralateral ripples could be to synchronize place cells across hemispheres, facilitating the formation of global cell assemblies (Pfeiffer and Foster, 2015).

Conclusions and future directions

In summary, despite the long history of studies on CA1 ripple oscillations, our work provides one of the few detailed quantitative analyses of their inter-hemispheric synchronization. Notably, our findings challenge the conclusions of Villalobos et al., 2017 by demonstrating that ripple events, while primarily phase-locked within the same hemisphere, exhibit robust time-locking across hemispheres. This indicates that ripple synchronization is not purely local but involves global temporal coordination, likely driven by common bilateral inputs from CA3.

Our results further reveal that phase and amplitude coupling metrics capture distinct aspects of ripple synchrony. Phase coupling is highly localized, rapidly decaying with distance along the septo-temporal axis, whereas amplitude coupling remains consistent over broader spatial scales, highlighting the global nature of ripple time-locking. These findings suggest that the hippocampus balances localized processing with global coordination, a mechanism that may support memory consolidation and the formation of coherent cell assemblies across hemispheres.

Although our results robustly demonstrate the time-locked co-occurrence of ripples across hemispheres, they do not directly address whether the information encoded by each hemisphere is correlated. One promising avenue would be to investigate neuronal assembly dynamics during exposure to novelty, given that awake SWR-associated replay can express both forward and reverse reactivation of behavioral sequences, even in unfamiliar environments (Buhry et al., 2011). While such analyses lie beyond the scope of the present study, they represent a clear and compelling direction for future work.