Theta- and gamma-band oscillatory uncoupling in the macaque hippocampus

Version of Record

Accepted for publication after peer review and revision.

Download
Cite
Share
CommentOpen annotations (there are currently 0 annotations on this page).

521 views

Version of Record published: May 24, 2023 (This version)
Accepted Manuscript published: May 4, 2023 (Go to version)
Accepted: March 13, 2023
Received: January 31, 2023
Preprint posted: January 1, 2022 (Go to version)

1. Of interest
Cholinergic and noradrenergic axonal activity contains a behavioral-state signal that is coordinated across the dorsal cortex

Lindsay Collins, John Francis ... David A McCormick

Research Article Updated Jun 2, 2023
Further reading

Article
Figures and data
Abstract
Editor's evaluation
Introduction
Results
Discussion
Methods
Data availability
References
Decision letter
Author response
Article and author information
Metrics

Abstract

Nested hippocampal oscillations in the rodent give rise to temporal dynamics that may underlie learning, memory, and decision making. Although theta/gamma coupling in rodent CA1 occurs during exploration and sharp-wave ripples emerge in quiescence, it is less clear that these oscillatory regimes extend to primates. We therefore sought to identify correspondences in frequency bands, nesting, and behavioral coupling of oscillations taken from macaque hippocampus. We found that, in contrast to rodent oscillations, theta and gamma frequency bands in macaque CA1 were segregated by behavioral states. In both stationary and freely moving designs, beta2/gamma (15–70 Hz) had greater power during visual search whereas the theta band (3–10 Hz; peak ~8 Hz) dominated during quiescence and early sleep. Moreover, theta-band amplitude was strongest when beta2/slow gamma (20–35 Hz) amplitude was weakest, instead occurring along with higher frequencies (60–150 Hz). Spike-field coherence was most frequently seen in these three bands (3–10 Hz, 20–35 Hz, and 60–150 Hz); however, the theta-band coherence was largely due to spurious coupling during sharp-wave ripples. Accordingly, no intrinsic theta spiking rhythmicity was apparent. These results support a role for beta2/slow gamma modulation in CA1 during active exploration in the primate that is decoupled from theta oscillations. The apparent difference to the rodent oscillatory canon calls for a shift in focus of frequency when considering the primate hippocampus.

Editor's evaluation

The rodent hippocampus is one of the main neuroscience models for memory, navigation, and plasticity, and research has suggested important roles for the theta-rhythmic modulation of firing activity and γ oscillations in these processes. This valuable study shows solid evidence of differences between a non-human primate and the rodent hippocampus, in that theta and γ frequencies are segregated by behavioral states, with theta dominating in quiescence and early sleep and β/γ during visual search.

https://doi.org/10.7554/eLife.86548.sa0

Introduction

Hippocampal oscillations are heralded as canonical examples of how oscillations support cognition by coordinating neural circuit dynamics (Buzsáki and Draguhn, 2004; Colgin, 2016; Hahn et al., 2019; Klausberger and Somogyi, 2008). In turn, behavioral states constrain and entrain specific neural oscillations. In rodents, locomotion and other exploratory movements elicit an ~8 Hz theta oscillation in hippocampal CA1 (Buzsáki, 2002; Kramis et al., 1975; Vanderwolf, 1969; Whishaw and Vanderwolf, 1973) and a faster gamma oscillation (25–100 Hz) that nests within theta (Bragin et al., 1995b; Colgin and Moser, 2010; Colgin, 2016; Csicsvari et al., 2003). In contrast, during quiescent states, theta and gamma oscillations are suppressed and sharp-wave ripple (SWR) complexes emerge, the latter consisting of large high-frequency oscillations (HFO, 150–250 Hz) in CA1 that occur within a slower (sharp-wave) deflection (Buzsáki, 2015; Ylinen et al., 1995). Although the occurrence of SWRs during quiescence is highly conserved across species (Buzsáki, 2015), its dichotomy with theta is questionable (Hussin et al., 2020; Leonard et al., 2015). This may stem from differences in how and when theta oscillations appear across phylogenetic order (Green and Arduini, 1954; Ulanovsky and Moss, 2007), particularly among primates, including humans (Courellis et al., 2019; Green and Arduini, 1954; Halgren et al., 1978; Herweg et al., 2020; Jacobs, 2014; Mao et al., 2021 Stewart and Fox, 1991; Talakoub et al., 2019; Tamura et al., 2013). Consequently, gamma coupling to hippocampal theta (Bragin et al., 1995a; Colgin, 2016; Lisman and Jensen, 2013), and the presence – as postulated – of sub-bands of gamma (Buzsáki and Wang, 2012; Colgin, 2016; Colgin et al., 2009; Csicsvari et al., 2003), could understandably be affected by the scarcity of theta oscillations in monkeys during species-relevant exploration (Courellis et al., 2019; Hoffman et al., 2013; Jutras et al., 2013; Leonard et al., 2015; Mao et al., 2021; Skaggs et al., 2007; Talakoub et al., 2019). In the present study, we therefore adopted a hypothesis-generating (data-driven) approach to identify (i) which oscillatory bands emerge in macaque hippocampal CA1 as a function of behavioral state; (ii) whether these oscillations coalesce or compete; and (iii) to what extent local single units are modulated at these rhythms.

Results

Spectral analysis of hippocampal LFP during active visual search and quiescence

We recorded 42 sessions (M1: 26 sessions, M2: 16 sessions) in the hippocampal CA1 subfield of two macaques (Figure 1; Figure 1—figure supplement 1) during active visual search and quiescence (henceforth: ‘rest’). As a control for the effects of the stationary animal, we recorded from one of the above animals (M2) and a third animal (M3) in freely moving and overnight sleep conditions (Figure 1—figure supplement 2). Consistent with previous reports (Leonard and Hoffman, 2017; Leonard et al., 2015), we observed bouts of roughly 20–30 Hz oscillations predominantly during search, and slower-frequency, larger-amplitude local field potentials (LFPs) during rest (Figure 1A and C). To visualize the relationship between spectral power across frequency bands, we sorted quantiles of ~1 s segments based on their average power in the 20–30 Hz frequency band, revealing an antagonistic relationship between 20–30 Hz and <10 Hz frequencies (Figure 1B), that is when power at 20–30 Hz was greatest, <10 Hz power was qualitatively weakest. In contrast, stronger power at <10 Hz was accompanied by stronger power at >80 Hz. To determine whether the spectrum varies with behavioral epoch, as it does in rodents (Buzsáki, 1996; Whishaw and Vanderwolf, 1973), we calculated the power spectrum for each behavioral state (Figure 1D). To identify power beyond the 1/f background, we used the aperiodic-adjusted power spectrum (Demanuele et al., 2007; Donoghue et al., 2020). We found stronger 7–10 Hz power during rest compared to active search (Figure 1D, middle). In contrast, power in the higher frequencies from 15 Hz to 70 Hz was stronger during active search compared to rest, with a peak in the 20–30 Hz range (both, p<0.05, Wilcoxon signed rank test with FDR correction). In most non-rodent species examined, including humans, hippocampal theta-band oscillations during alert wakefulness are described as occurring intermittently, in short-lived bouts, unlike the protracted and predictable trains of theta oscillations seen in the rodent hippocampus (Green and Arduini, 1954; Jacobs, 2014; Jutras et al., 2013; M Aghajan et al., 2017; Talakoub et al., 2019; Ulanovsky and Moss, 2007; Watrous et al., 2013). To identify oscillations that are rare and short lived, and to allow for more direct comparison to conventions used in human iEEG/macroelectrode studies, we used the BOSC method (Caplan et al., 2001; Hughes et al., 2012; M Aghajan et al., 2017). This method quantifies the fraction of time that band-limited power exceeds amplitude and duration thresholds. The amplitude threshold is set after fitting signal to a log-log linear regression to account for the spectral tilt (1/f^x) of the distribution under consideration and accepting only residuals with at least 3 cycles exceeding 95% of the χ² distribution. This classifies the graded power spectral measure into discrete ‘hits’ and ‘misses’ across time. Consistent with continuous power spectral results, we found that theta bouts (‘hits’) were more prevalent during rest/sleep compared to online active behavioral states. The pattern was the opposite for beta2/gamma frequencies, which were more prevalent during active states (Figure 1—figure supplement 2, p<0.05, Wilcoxon signed rank test with FDR correction).

Figure 1 with 4 supplements see all

Download asset Open asset

Oscillatory decoupling in CA1 field potentials.

(A) Example traces of broadband local field potential (LFP) in CA1 during search. Data segments were taken from epochs with characteristic high 20–30 Hz power, shown in B (traces were linearly detrended for visualization). (B) Spectral density sorted by 20–30 Hz power. Surface plot shows data segments sorted into quantiles according to 20–30 Hz power, revealing an apparent increase in 5–10 Hz power when 20–30 Hz power is weakest. See Methods for details. (C) Example traces of wideband LFP in CA1 during the rest epoch showing characteristic interactions between <10 Hz and >60 Hz oscillations. Conventions as in A. (D) *Top.* Mean power spectral density during search (red), and rest (blue). *Inset*: mean power for low frequencies of main plot, with shaded 95% bootstrap confidence interval (N=42 sessions). *Middle.* Power spectral density after fitting and subtracting the aperiodic 1/f component during search and rest, with shaded 95% bootstrap confidence intervals. Gray areas show significant differences in power across behavioral epochs (p<0.05, Wilcoxon signed rank test, FDR corrected) *Bottom.* Power difference between search and rest. (E) Average cross-frequency power comodulogram (N=42 sessions). Dark outline represents areas that were significant in at least 80% of samples (p<0.05, cluster-based permutation test corrected for multiple comparisons). (F) Average bicoherence of the CA1 LFP (N=42 sessions). The dark outline represents areas that were significant in at least 80% of sessions (p<0.05, Monte Carlo test corrected for multiple comparisons). The decoupling is preserved when applying analysis methods sensitive to transient oscillations (Figure 1—figure supplement 2), and when analyzing CA1 LFPs from an additional monkey who moved freely in a search task, and during night-time rest (Figure 1—figure supplement 3).

In rodents, the frequency and the amplitude of theta oscillation is related to the speed of locomotion (Buzsáki, 2002; Fuhrmann et al., 2015; McFarland et al., 1975; Sheremet et al., 2016) although theta activity is also observed during awake immobile states of alertness (Kramis et al., 1975; Sainsbury, 1998; Tai et al., 2012; Vanderwolf, 1969). Theta movement correlates raise the concern that the scarcity of theta during active behaviors in the present study might be attributed to the animals’ immobility. To address this concern, we recorded wirelessly from the hippocampal CA1 of the second monkey (M2) and a third adult female macaque (M3) during freely moving active states including immersive visual search and during overnight rest/sleep (M3: 15 sessions, M2: 3 sessions). The results match those of the previous experiment, thereby demonstrating that the decrease in theta-band power and the increase in gamma power during active behaviors generalize, that is they were not merely due to the immobile state of the animals (Figure 1—figure supplement 3, Video 1). Furthermore, the duration and consistency of theta bouts during early sleep indicate that these methods (recording sites, electrode signal, and states) are capable of detecting theta oscillations, but that they appear during different epochs than those for recordings in rats and mice (Figure 1—figure supplement 3, Video 1, Figure 1—figure supplement 4).

Video 1

Download asset

posterframe for video — Oscillatory dynamics in monkey CA1 during sleep and waking states.

(Top) Broadband local field potential (LFP) during sleep (blue), and free movement in an enclosure during a search task (red), the 4–8 Hz bandpass filtered LFP, and the 25–50 Hz LFP, shown top to bottom, respectively. Detected bouts in each frequency band are highlighted with blue (sleep) and red (waking). The black vertical line shows 3 SD above the mean. (Bottom) Left: Movement, expressed as the vector norm of angular velocity. The gray horizontal line shows the threshold for movement. Middle: The envelope peak of detected theta (4–8 Hz) bouts plotted as a function of the gamma (25–50 Hz) peaks. Right: 1/f corrected (fitted residual) power spectrum. The distribution of power shifts to higher frequencies during awake compared to sleep.

Hippocampal cross-frequency coupling

To assess the coupling of theta and gamma oscillations at a finer temporal scale, we computed the cross-frequency power correlations across the spectrum. Consistent with the qualitative pattern shown in Figure 1B, power at 3–8 Hz and 20–35 Hz were negatively correlated (Figure 1E, p<0.05, using a cluster-based permutation test corrected for multiple comparisons). In addition, power in the slower, 3–8 Hz band was positively correlated with that of a much faster, 80–150 Hz band. We next applied a complementary approach, comparing the amplitude envelopes of these two bands (theta and slow gamma) over time, to track the finer temporal structure of power correlations. The results supported the epoched data results (Figure 1—figure supplement 2).

To estimate phase-amplitude coupling in the LFP, we performed bicoherence analysis (Figure 1F; Giehl et al., 2021; Hyafil, 2015; Kovach et al., 2018), revealing a peak cluster around the 25 Hz frequency range which confirms an interaction between the activity at this frequency and its second harmonic. In addition, the 3–8 Hz band was coupled to high frequencies of 95–150 Hz (p<0.05, cluster-based Monte Carlo statistical test). This is consistent with our cross-frequency amplitude coupling results and indicates that the correlated envelopes in Figure 1E are driven by phase-specific coupling of the high frequencies (Figure 1B). Nevertheless, our bicoherence results showed no significant phase-amplitude coupling between theta and gamma frequency range.

Oscillatory modulation of spiking activity

Peaks in spectral power do not necessarily indicate the presence of oscillations in the underlying neural activity (Buzsáki and Wang, 2012; Pesaran et al., 2018; Herweg et al., 2020; Jones, 2016). If oscillations are present in the local neural population, regular comodulation between spikes and local field oscillation phases should occur. We measured the spike-field coherence for the whole duration of the sessions by calculating pairwise phase consistency (PPC) (Vinck et al., 2010) for well-isolated units (N=404). Individual cells phase locked to multiple frequencies (Figure 2A; p<0.05, permutation test and Rayleigh test, p<0.05), with the population showing the full range of spike preferred frequencies of modulation (Figure 2C).

Figure 2 with 2 supplements see all

Download asset Open asset

Spike-field coherence and its influence by sharp-wave ripples (SWRs).

(A) Spike-local field potential (LFP) pairwise phase consistency (PPC) spectra for an example unit. Shaded gray shows significant values (p<0.05, permutation test and Rayleigh test, p<0.05). (B) Spike-LFP coherence for spikes during detected SWRs (dark gray) and for spikes remaining after extracting the SWR epochs (‘residual’, red). Light gray shading shows significant difference at p<0.05 in FDR-corrected permutation test for the SWR group. Inset: Normalized histogram of the phase values at 8 Hz, obtained from the spike-LFP coherence analysis shown in A. (C) Probability distribution of observing significant PPC values across all frequencies (solid black line), and only for preferred frequency (frequency with maximum PPC value) before adjusting for SWRs (N=404 units). (D) Difference in the proportion of cells with greater spike-LFP coherence for SWR (gray) and SWR-removed residual (red), for six frequency bands (N=185 units). (E) SWR residual difference of mean spike-LFP coherence. Shading shows 95% bootstrapped confidence interval (N = 185). Positive values indicating greater PPC for residual than SWR groups are shown in red, negative values (SWR>residual) in dark gray.

One of the caveats of spike-field coherence measures is that they can be sensitive to large amplitude non-periodic deflections. SWRs have a non-oscillatory amplitude envelope within the frequency range of 0.1–10 Hz in rodents and non-human primates (Hussin et al., 2020; Leonard et al., 2015; Maier et al., 2003; Rex et al., 2009; Skaggs et al., 2007). Furthermore, we previously observed that SWRs occur in primates during active visual exploration (Leonard and Hoffman, 2017; Leonard et al., 2015) and that the probability of firing of cells increases during these events, for putative pyramidal and inhibitory cells alike (Hussin et al., 2020; Leonard et al., 2015; Skaggs et al., 2007). Given the limitations of spike-field coherence, the frequency characteristics of SWRs, spiking activity profile of neurons around these events, weaker power at 7–10 Hz during active search, and the strong coupling between bands matching the SWR events (3–8 Hz and >95 Hz), we hypothesized that spike-LFP coherence at low frequencies might be partly produced as a byproduct of the slow deflection of SWRs (Hussin et al., 2020) rather than via harmonic oscillations. To test this, we extracted peri-SWR spikes, computed PPC only for these spikes in each cell [PPC_swr], and then compared this to the PPC for spikes outside the SWR windows [PPC_residual]. Figure 2B shows an example cell that exhibits stronger spike-LFP coherence at 8 Hz during SWR than outside the SWR window (p<0.05, permutation test with FDR correction). At the population level, coherence at lower frequencies (2–10 Hz) was greater during SWR than in the SWR-removed distributions. In contrast, the higher frequencies (10–200 Hz) maintained a greater coherence outside of the SWR time window (Figure 2D). This led to weaker mean spike-field coherence restricted to the <10 Hz range, after removing the influence of SWRs, suggesting a contribution of the non-oscillatory slow deflections in the SWR complex to the apparent cross-frequency interactions.

To ensure that oscillations were local and to avoid the influence of aperiodic deflections, we generated spike autocorrelograms. Periodic peaks of the spike autocorrelograms demonstrate theta rhythmicity in rat and mouse CA1 (Cacucci et al., 2004; O’Keefe and Recce, 1993; Royer et al., 2010). To directly compare theta rhythmicity in our cell population with observations in the rodent, we computed the autocorrelogram for a complete population of well-isolated cells with at least 100 spikes for the whole session and for the complete cell population from homologous subregions of CA1 in rodents, also across the whole session. Figure 3A shows an autocorrelogram of the theta spike-field coupled cell in Figure 2A (black, monkey) overlaid onto a theta rhythmic cell from the homologous CA1 region of a rat (i.e. temporal CA1) shown in gray. To quantify theta autocorrelogram rhythmicity, we calculated the theta modulation index (Jacobs, 2014). Compared to modulations seen in the rat (Figure 3B in gray), monkey cells typically showed near zero index values, that is, they were not modulated (Figure 3B in black). This was also evident in the sorted-cell and mean population autocorrelogram (Figure 3C, inset shows rat distribution). For a given firing rate, higher frequencies are less likely to demonstrate cycle-by-cycle periodicity due to their shorter periods; nevertheless, we observed a few examples of >20 Hz spike modulation in the spike-triggered averages and autocorrelograms of several cells (Figure 4).

Figure 3

Download asset Open asset

Examining spiking periodicity for theta modulation in macaque hippocampus.

(A) Autocorrelogram of an example ‘theta’ unit from Figure 2A (black, N=945 total spikes) and a theta-modulated unit in rat (gray, N=5655 total spikes). (B) Distribution of the theta index in CA1 units from this study (black, N=240 units), and CA1 units of rat (gray, N=197 units). Dashed lines show mean values, color coded by species. (C) Sorted autocorrelograms of CA1 units, with mean, population ACG shown as a white trace. Inset: same as the main plot but for rats.

Figure 4

Download asset Open asset

Spike-local field potential (LFP) coherence of the slow gamma oscillation in macaque hippocampus.

(A) Average spike-LFP coherence in gamma frequency range. Shading shows 95% bootstrap confidence interval (N = 404). (B) Spike-LFP coherence for a representative gamma-locked unit. This unit had a significant peak at 30 Hz (p<0.05, permutation test and Rayleigh test, p<0.05). (C) Top: Spike-triggered average LFP of the unit in B. Bottom: Autocorrelogram of the same unit.

Discussion

In this study we evaluated the oscillatory dynamics of primate hippocampus during two general behavioral states: first, during ‘online’ states of awake active behavior and second, during offline states. The waking-behavior recordings included stationary monkeys who were engaged in hippocampal-dependent, memory-guided visual exploration (Chau et al., 2011; Dragan et al., 2017; Yoo et al., 2020), and freely moving monkeys exploring their environment. The offline-state recordings included post-task quiescent states in the darkened booth, and the early stages of overnight sleep, respectively. The most prominent oscillation in amplitude and prevalence during alert active behavior was in the beta2/slow gamma band (~20–35 Hz for stationary search and ~20–50 Hz in freely moving subjects). In contrast, theta-band (3–10 Hz, peak at ~8 Hz) amplitude and bout prevalence were greater during offline states of rest and sleep than during online active states. This bimodal oscillatory profile was also evident in finer temporal scales. Theta and beta2/gamma bands were decorrelated in the cross-frequency amplitude coupling, and in phase-amplitude bicoherence, quite unlike that observed in rodents (Zhou et al., 2019). Spike modulation by LFP frequency was seen in both bands; however, only the theta-band effect appeared to be partially due to non-oscillatory field events, such as the SWR. Despite longer and more prevalent theta oscillations in rest, overall, we found no clear evidence of theta-periodic modulation. This contrasts with the strong and prevalent modulation seen using the same analyses methods applied to signal from the rat in homologous regions of CA1. Across several measures we found consistent or compatible results in the coalescence among frequency bands, and between bands and behavioral states within this study. These patterns differ from well-established spectral-behavioral coupling in rodents.

Dormant theta

In rat and mouse hippocampal LFPs, theta activity (centered at 6–9 Hz) is most consistently present during activated cortical states, that is, during alert aroused behaviors and REM sleep (Buzsáki, 2002; Colgin, 2013; Nuñez and Buño, 2021; Vanderwolf, 1969). In contrast, theta gives way to SWRs and irregular slow activity during quiescence and NREM sleep (Buzsáki, 2015; Buzsáki, 1989). Further, theta ‘nests’ different frequencies of gamma in a phase-specific manner during these ‘active’ cortical states (Colgin and Moser, 2010; Colgin, 2016; Lasztóczi and Klausberger, 2014; Lisman and Jensen, 2013). And finally, most classes of cells in CA1 are modulated by the theta rhythm (Klausberger and Somogyi, 2008; Klausberger et al., 2003; Royer et al., 2010). Indeed, theta is so reliably present and so spectrally dominant in fields, spikes, and intracellular currents during the active states when gamma occurs, that studies of hippocampal gamma oscillations and spikes are commonly conditioned on them.

Monkey CA1 LFPs also show reliable, durable theta oscillations with peak frequencies at 6–9 Hz; however, where measured across states, they do not demonstrate the above characteristics of theta. Instead, they show divergent brain-behavior coupling to that seen in rodents. Much of the literature in humans and monkeys has focused on brief epochs of alert, goal-directed behaviors within stationary visual and visuospatial tasks and may therefore be suboptimal for detecting theta oscillations. Yet where rest or sleep have been measured, stronger, more durable theta field potential oscillations appear during quiescence and NREM sleep (or anesthetized states, Kleen et al., 2021) than during alert/task-on states or REM sleep (Bódizs et al., 2001; Cox et al., 2019; Moroni et al., 2007; Tamura et al., 2013; Uchida et al., 2001). Whereas these behavioral-state correlates of field potential oscillations apparently differ by species, local control of unit activity may nevertheless reveal strong underlying rhythmic motifs.

To better understand local modulation, we measured CA1 spiking activity as a function of LFP oscillatory phase. All tested frequencies had at least some units that were preferentially phase locked to them, but the most common preferred frequencies were roughly matching the 3–10 Hz, 20–30 Hz, and 60–150 Hz bands. The lowest band would be consistent with theta modulation of local spiking, which has been reported using a few different methods in human and macaques. Spike-field coherence in the theta frequency range was seen for microelectrode bundles including microwires in unspecified subregions of the human hippocampus (Rutishauser et al., 2010), and for recording sites estimated to be within the hippocampus proper of the freely moving macaque (Mao et al., 2021), though in the latter case, the higher gamma bands and delta bands were several times more likely to show SFC. Within the theta band, spurious coherence may arise in response to nonrhythmic deflections (Aru et al., 2015; de Cheveigné and Nelken, 2019; Jones, 2016; Vinck et al., 2022), such as the slow components of the SWR which is especially prominent in macaques and human (Hussin et al., 2020; Liu et al., 2022). Indeed, we found that SWRs (Figure 2—figure supplement 1) disproportionally influenced the 2–10 Hz band (Figure 2B, D and E), whereas the gamma frequency range was largely unaffected. Because SWRs occur infrequently, it was somewhat surprising that SWR removal affected the overall results; however, the population activity during ripples typically includes 2–10× firing rate increases in nearly all cell types (Buzsáki, 2015; Hussin et al., 2020; Leonard et al., 2015; Skaggs et al., 2007). As such, each ripple is likely contributing disproportionately to the spike-locked activity. In addition, we know that visual (Katz et al., 2020; Rey et al., 2014; Roux et al., 2022) and saccade-elicited (‘ERP-like’) responses in the hippocampus (Hoffman et al., 2013; Jutras et al., 2009; Katz et al., 2022) increase theta-band power without necessarily producing harmonic oscillations. This would presumably only affect the segments from search, not sleep. Future human and monkey hippocampal studies should factor out the contributions of SWRs, and to the same end stimulus and saccade-locked transients, when evaluating low-frequency (<10 Hz) oscillations for non-stationary signals and when using methods that do not discriminate oscillations from transient deflections, as with spike-field coherence. The present results found residual coherence, and in addition, the typical duration of sleep theta bouts would well exceed the SWR envelope (as shown in Video 1 and Figure 1—figure supplement 4), therefore we consider the observed spectral peaks during rest likely reflect true oscillations and not simply collections of evoked transients.

The spike autocorrelogram is a more stringent measure of oscillatory modulation, not susceptible to LFP spectral analysis artifacts. The <10 Hz range including the theta band failed to show clear examples of theta modulation (Figure 2—figure supplement 2). This was not simply due to low firing rates: even high firing rate cells, which should include ‘theta-on’ interneurons, failed to show theta modulation. Indeed, the same analysis applied to units from a homologous region of rat CA1 (Royer et al., 2010) showed a predominance of theta modulation (even after pooling functional cell types). In light of our rest-related theta power and SWR-removed spike-field coherence, this was somewhat surprising, but consistent with the results in free-flying bats (Yartsev et al., 2011), monkeys (Courellis et al., 2019; Mao et al., 2021), and humans (Qasim et al., 2021). In such cases, to observe spike timing regularity to field potentials, the aperiodic – that is non oscillatory – slow fluctuations in the LFP must first be warped to ‘fit’ one another, to create a pseudo-oscillation (Bush and Burgess, 2020; Eliav et al., 2018; Mao et al., 2021; Qasim et al., 2021). Due to the different criteria used, compared to the theta rhythmic spiking reported in rodents, the theoretical or computational roles ascribed to the theta oscillations may need to be re-evaluated, or, more precisely defined in terms that do not depend on periodicity.

Slower oscillations, or irregular low-frequency components at 1–4 Hz ‘delta’ band, are more frequently correlated with spatial or associative memory effects (Goyal et al., 2020; Vivekananda et al., 2021; Watrous et al., 2013), leading to the supposition that rodent hippocampal theta is simply faster than in primates (Jacobs, 2014; Mao et al., 2021). If our present results were based on two distinct oscillations – the faster ‘offline’ theta and the slower ‘online/memory’ delta band – we would have predicted anti-correlations or segmentation in the power plots into sub-bands. We have yet to see such a segmentation, but we would stress that this remains a candidate for closer inspection. Finer task behaviors might help to uncover this possible alert-behavior correlate. Even if this proves to be the case, it would raise further questions about which bands become considered homologous across species. If 1–4 Hz is envisaged as relating to septal-cholinergic theta in rats, what is the equivalent of the classic 8 Hz theta band in rats? Compounding this problem is a separate 1–4 Hz oscillation that was observed in rat hippocampus (Jackson et al., 2014; Schultheiss et al., 2020). When trying to consider homology across species, the LFP signals are only proxies for the underlying circuit activity. Identifying the local circuit motif, including receptor-specific neuromodulation in primates (Stewart and Fox, 1991), may help differentiate among various low-frequency oscillations and their functional roles across species.

Decoupled gamma

Gamma oscillations in rat and mouse CA1 comprise a wide high-frequency band (initially 40–100 Hz but more recently 20 Hz or 30–100 Hz) known for tight coupling within the theta oscillation (Bragin et al., 1995a; Buzsáki et al., 1983; Colgin and Moser, 2010; Colgin et al., 2009; Tort et al., 2008). In line with this coupling, gamma shares behavioral state correlates with theta: waking exploratory states and REM sleep. We found that the slower part of the gamma range (<~70 Hz) is decoupled from theta. Consistent with this finding, gamma power is seen as a relatively strong oscillation in monkeys and humans during REM sleep (Cantero et al., 2003; Takeuchi et al., 2015; Tamura et al., 2013; Uchida et al., 2001), as in rodents. Gamma oscillations, when measured, are often associated with hippocampal processing (e.g. memory guided search, Leonard et al., 2015; Montefusco-Siegmund et al., 2017), and retrieval (Montefusco-Siegmund et al., 2017) or subsequent memory effects (Jutras et al., 2009) in monkeys, or spatial coding and memory in rodents. In addition, a band-limited 20–40 Hz oscillation is seen in rodent CA1 (and is synchronized with LEC, Igarashi et al., 2014) during discrete item cueing or retrieval such as in olfactory associative place and sequence learning (Allen et al., 2016; França et al., 2014; Lansink et al., 2016; Lopes-Dos-Santos et al., 2018), and when exploring novel environments (França et al., 2021; Trimper et al., 2017). This frequency band is associated with activation of the trisynaptic pathway (DG and CA3, Colgin, 2016; Fernández-Ruiz et al., 2021; Hsiao et al., 2016; Rangel et al., 2015), but is also a coherent oscillation between CA1 and LEC (Igarashi et al., 2014) implicating direct, temporoammonic pathways. Finally, beta2/gamma coherence is seen in hippocampal-retrosplenial and hippocampal-mFPC oscillations in monkeys during object-scene associative memory (Hussin et al., 2022). The range of implicated pathways and regions underscores the need to identify the underlying microcircuits that give rise to these oscillations (Fernandez-Ruiz et al., 2023). Determining which are conserved across species will be an important topic for future research.

Gamma and theta coupling

The results of the current study show that theta oscillations are not prominent during awake behavior and do not couple with gamma, and thus, are not a good candidate for structuring possible gamma sub-bands. Although most studies in humans focus on coupling via slower or faster bands outside this range, we note two exceptions that ostensibly find increases in hippocampal theta-gamma coupling associated with active processing in the theta band (Axmacher et al., 2010; Stangl et al., 2021) and a third that included 5–11 Hz, for which the low-frequency-granting signal may have included the alpha band (Roux et al., 2022). Both of the former studies use macro-electrode iEEG with >1 mm contacts and 3–10 mm spacing between contacts, along with MR/CT coregistration, which are estimated to be in some regions of the hippocampus proper in only a subset of participants. Thus, any apparent discrepancies to the present findings may involve the inclusion of extra-hippocampal signals, though this would still be interesting to understand. Signal localization notwithstanding, the first study of working memory reported 6–10 Hz medial temporal lobe theta phase that was coupled with gamma oscillations (Axmacher et al., 2010). Although modulation strength was not associated with performance, modulation width predicted reaction time. Pertinent to the state correlates of theta, they show that the intertrial interval (ITI) – that is a potentially offline state – shared a strong modulation, similar to that seen during the memory epoch, except at the onset of the WM epoch, where intertrial coherence was strongest. In the second study, recording MTL iEEG in ambulating patients, theta power did not increase with movement speed (if anything, it nominally decreased), and search during movement was associated with weaker theta; however, when stationary, some theta modulation was seen and in addition, general theta-gamma coupling was reported between 6 Hz and 10 Hz and high gamma during movement and between 6 Hz and 10 Hz and a slower gamma when stationary and observing others (Stangl et al., 2021). The effects of saccade-evoked responses on this theta were not described but saccades were associated wtih increases in reported theta power, thus, it remains unclear what effect saccadic responses may have had on the reported theta gamma coupling. In the third study, using microwires, the most applicable measures to our study produced generally consistent results, including decreased theta and increased gamma power associated with successful associative memory, and no difference in low-frequency spike-field coherence for locally measured fields and spikes (Roux et al., 2022). Here, unlike the first study (Axmacher et al., 2010), the modulation of gamma power by peak 5–11 Hz phase was greater with successful memory (hits vs. misses). This might suggest a theta or alpha oscillation is regulating gamma magnitude; however, the MI measure of phase-amplitude coupling does not require strong or periodic phase-granting signal to cluster gamma (Tort et al., 2010). Future studies may help to disentangle the role of periodic theta in contrast to saccadic or other evoked response waves, in clustering active gamma oscillations during active behaviors. At present, we suggest that gamma oscillations – across species – can nevertheless work as a standalone rhythmic activity to select among inputs by virtue of laminar specificity, including activity of other hippocampal fields (e.g. CA3-dentate gyrus) or extrahippocampal areas (e.g. entorhinal cortex).

Our findings offer a hypothesis-generating framework for future analysis in (human and non-human) primate hippocampal physiology. The present results suggest that theta oscillations were not prevalent during search in primates and did not consistently modulate single unit activity, but rather form the strongest oscillatory marker of offline or quiescent states. Instead, beta2/slow gamma oscillations constitute the chief, self-contained oscillation that arises during active exploration in the primate hippocampus and stands as the most likely oscillation for organizing local dynamics during exploration. Aside from understanding the nature of these cross-species differences, future work may focus on the better-conserved aspects of CA1 activity, including gamma synchrony during exploration, locking of spiking to exploratory movements, and aperiodic spike timing measures that don’t require autocoherent oscillations. Despite several surface differences in hippocampal-behavioral coupling across phylogenetic orders, the underlying neural circuit activity giving rise to these oscillations may yet reveal fundamentally conserved mechanisms.

Methods

Subjects and task

Two adult female macaques (Macaca mulatta, ‘M1, M2’) were used in the main, visuospatial search experiments whose results are shown in all figures except those of the control task/recordings from Figure 1—figure supplement 3. Data from M1 and M2 have been reported previously (Hussin et al., 2020; Leonard and Hoffman, 2017; Leonard et al., 2015). The apparatus, training procedure, and task have been described previously (Leonard and Hoffman, 2017; Leonard et al., 2015), and are summarized briefly here. During search, animals performed a hippocampally dependent visual target detection task. In the task, seated, head-fixed monkeys were placed in front of a monitor and were required to identify a target object from nontargets in unique visual scenes presented on a monitor positioned in front of them, and report their selection of scene-unique target objects by holding their gaze on the target region for a prolonged (≥800 ms) duration. Target objects were defined as a changing item in a natural scene image, where the original and changed images were presented in alternation, each lasting 500 ms, with a brief gray screen (50 ms) shown between image presentations. An ITI of 2–20 s followed each trial. The daily sessions began and ended with a period of at least 10 min when no stimulus was presented within the darkened booth and animals could sleep or sit quietly (‘rest’). All procedures associated with this task were conducted with approval from the local ethics and animal care authorities (Animal Care Committee, Canadian Council on Animal Care). Two adult female monkeys (M. mulatta, ‘M2’ and ‘M3’) were used in a separate experiment to extend the analysis to include freely behaving conditions. The search epochs of M2 were observational (identified post hoc) and of M3 were experimentally controlled. M2 was placed in an enriched environment where she could actively forage, play with toys (manipulanda), walk, climb, self-groom, and groom another animal. Blind raters denoted the times of foraging, walking, and exploratory ‘search’ behaviors. M3 was placed in a testing enclosure equipped with multiple touchscreens around the periphery that presented spatially distributed arrays of objects. To obtain fluid reward the monkey was required to locate and select (touch) designated objects in a global spatial sequence across the enclosure, thereby requiring visual search, reaches, and walking/climbing during a trial. For M3 search data analysis, we extracted the trial sequence duration+2 s beyond the first/last touches of the sequence which included goal-directed walking, but excluded the ITI containing reward consumption or idle time prior to the animal’s approach and trial initiation. For both M2 and M3, rest epochs consisted of 40 min of recordings in the housing area during the start of the ‘night’ cycle of the room’s automatic lighting system. A total of 18 sessions (M3: 15, M2: 3) were analyzed for both task and rest epochs. All procedures for M3 were conducted in accordance with the approved protocols and authorized procedures under the local animal care authorities (Institutional Animal Care and Use Committee).

Electrophysiological and movement recordings

For monkeys M1 and M2, indwelling bundles of movable platinum/tungsten multicore tetrodes (96 μm outer diameter; Thomas Recordings) were implanted into the anterior half of hippocampus and lowered into CA1. In M3 we recorded from an indwelling active multichannel probe on an adjustable microdrive (‘Deep Array’ probe, beta-test design, Diagnostic Biochips, Inc). Recording sites were verified with postoperative CT coregistered to pre-operative MRI and using functional landmarks that changed with lowering depth, including the emergence of depth-specific SWRs in a unit-dense layer, as described in the previous studies (Figure 1—figure supplement 1, Figure 2—figure supplement 1). Post-explant MRI verified the electrode locations in M1 (Leonard et al., 2015). For the current study, we detected channels within the pyramidal layer based on the strongest amplitude of ripples during SWRs and single unit activity, and only used these channels for all analyses. LFPs were digitally sampled at 32 kHz using a Digital Lynx acquisition system (Neuralynx, Inc) and filtered between 0.5 Hz and 2 kHz for M1 and M2. LFPs for M2 and M3 that were used for results shown in Figure 1—figure supplement 2 were sampled at 30 kHz using the active headstage and recorded using a Cube/Freelynx wireless recording system (Neuralynx, Inc) to SD card (for rest) or using wireless transmission to the Cheetah acquisition system (for task). Single-unit activity was filtered between 600 Hz and 6 kHz, recording the waveform for 1 ms around a threshold-triggered spike events. Spike sorting was performed semi-automatically using KlustaKwik based on wave shape, principal components, energy, and peak/valley across channels. This was followed by manual curation of clusters in MClust (A.D. Redish). We used 3D accelerometer in Freelynx data acquisition system to record the movement of freely behaving animal subjects during active search and rest. Angular velocity (AV) traces were recorded at 3 kHz and samples were synchronized with the neural recording. For movement detection, we first calculated the vector norm of the three AV axes using vecnorm function in MATLAB. Then, we thresholded the vector norm to find periods of time when animal was moving. Threshold was calculated based on the vector norm of AV recordings when Freelynx was placed statically on a flat surface. In addition to the accelerometer data, we used recorded videos to detect animal behavioral state (e.g. sleep).

SWR detection

SWR detection was performed on the tetrode channels of M1 and M2 with the most visibly apparent ripple activity using the previously described method (Leonard et al., 2015). Raw LFPs recorded from the tetrode channel were filtered between 100 Hz and 250 Hz. To determine the SWR envelope, filtered LFPs were transformed into z-scores and rectified and subjected to a secondary bandpass filter between 1 Hz and 20 Hz. Events with a minimum amplitude exceeding 3 SDs above the mean with a minimum duration of 50 ms, beginning and ending at 1 SD were designated as potential ripples. High-frequency energy is present for non-SWR events such as EMG and other non-biological noise, though these artifacts are distinct from ripples because the latter are restricted to the regions near the pyramidal layer. For artifact (non-ripple-event) rejection, a distant tetrode channel was selected as a ‘noise detecting’ channel (Talakoub et al., 2016). Events that were concurrently detected on the noise channel and the ‘ripple-layer’ channel were removed from the ripple pool. High gamma (80–120 Hz) and HFO (110–160 Hz) events were similarly identified, but with the filter criterion set at 80–120 Hz and 110–160 Hz, respectively, and in both cases identifying peaks as those >1 SD. Duplicate high gamma/HFO and SWR events were labeled as SWRs. Events with a repetition rate <125 ms were considered a single event.

Power spectral parametrization and fitting

To compare the spectral content during search and rest, we selected successfully completed trials lasting longer than 1 s. For rest segments in the stationary (tethered) recordings M1 and M2, we extracted the LFP signals that were recorded before the start of the task and after the end of the task when the animal was in a dark environment in a quiescent or inactive state. For rest segments of wireless recordings in M2 and M3, we extracted LFP signals recorded during the evening after the task, during the dark cycle of the housing area. We used Welch’s method with a 50%-overlapping 1024-sample sliding Hanning window to estimate power spectra for the frequency range of 1–150 Hz with a frequency resolution of 0.25 Hz.

To identify spectral peaks and compare between search and rest states, we parameterized power spectra using the method described by Donoghue et al., 2020. This method models power spectra as a combination of the 1/f frequency components (aperiodic) in addition to a series of Gaussians that capture the presence of peaks (periodic components). The model was fit to a frequency range between 1 Hz and 200 Hz with a frequency resolution of 0.5 Hz. Settings for the algorithm were set as: peak width limits: (0.5, 12); max number of peaks: infinite; minimum peak height: 0; peak threshold: 2.0; and aperiodic mode: ‘Fixed’.

To assess statistical significance for the difference in parametrized spectra at each frequency, we used Wilcoxon signed rank test at p<0.05 with FDR correction for multiple comparisons.

Twenty to 30 Hz sorted spectral density map

We estimated the power spectral density using Welch’s method described in the previous section to obtain (frequency * PSD segments) matrix. We then sorted the PSD segments based on the mean power in 20–30 Hz frequency range and normalized each segment by dividing it by its median. We clustered all sorted segments into 50 total segments of equal size (frequency * 50 segments) by averaging original PSD sorted segments. We repeated this procedure across sessions and animals separately.

Cross-frequency power correlation

On continuous LFP time-series data, Welch’s method with a 50%-overlapping 1024-sample sliding Hanning window was used to estimate the spectrogram for the frequency range of 1–150 Hz with a frequency resolution of 0.25 Hz.

We computed the pairwise correlation between cross-frequency power using the following formula (Masimore et al., 2004):

c o r r_{i j} = \frac{\sum_{k} (S_{k} (f_{i}) - \bar{S} (f_{i})) (S_{k} (f_{j}) - \bar{S} (f_{j}))}{σ_{i} σ_{j}}

where $S_{k} (f_{i})$ is the PSD at the frequency $f_{i}$ in time window $k$ , $\bar{S} (f_{i})$ the averaged PSD at the frequency $f_{i}$ over all sliding windows, $σ_{i}$ the standard deviation of the PSD at the frequency $f_{i}$ , and $k$ ranges over all sliding windows. In the wireless recordings in M2 and M3 this procedure was applied to the data segments extracted during task performance.

To test the null hypothesis that the power spectral time series of two different frequencies, $f_{i}$ and $f_{j}$ , are not coupled, we performed a non-parametric surrogate data method with cluster-based multiple comparison correction (Thammasan and Miyakoshi, 2020). This method preserves the original data’s statistical properties while generating time series that are randomized such that any possible nonlinear coupling is removed. In this method, we randomized time window $k$ differently for each frequency bin to build surrogate time-frequency time series and computed the surrogate cross-frequency power correlation. This process was repeated 5000 times to produce distributions for the dataset in which the null hypothesis holds. The original, non-permuted data are then compared to the surrogate distribution to obtain uncorrected p-values. The significance threshold was selected to be 0.05. For cluster-based multiple comparison correction, all samples were selected whose p-value was smaller than 0.05. Selected samples were then clustered in connected sets based on their adjacency and the cluster size was calculated. This procedure was performed 5000 times to produce the distribution of cluster sizes. If cluster sizes in the original correlation matrix were larger than the cluster threshold at 95th quantile, they were reported as significant. We performed this statistical procedure at the level of single channels per animal. We consider as robust significance those areas that were significant for at least 80% of all samples.

Hilbert amplitude envelope correlation

We applied a third-order Butterworth filter at each frequency of the LFP with a 2 Hz bandwidth. We then Hilbert transformed the bandpass-filtered LFP and estimated the continuous amplitude envelope. We computed the pairwise correlation between cross-frequency amplitude envelope using previously described correlation method.

Bicoherence

For bicoherence, we used the HOSA toolbox. Bicoherence was estimated for frequencies $f_{1}$ (1–75 Hz) and $f_{2}$ (1–150 Hz) in steps of 1 Hz according to the following formula (Bullock et al., 1997):

B (f_{1}, f_{2}) = \frac{| {⟨ F_{t} (f_{1}) F_{t} (f_{2}) F_{t}^{*} (f_{1} + f_{2}) ⟩}_{t} |}{{⟨ | F_{t} (f_{1}) F_{t} (f_{2}) F_{t}^{*} (f_{1} + f_{2}) | ⟩}_{t}}

where $F_{t} (f)$ is the signal’s time-frequency transformation at time t, || represents the absolute value, and $⟨ ⟩$ is the average over time. We set the segment length to 1024 samples for this analysis.

Bicoherence has a higher spectral resolution for disentangling harmonic from non-harmonic cross-frequency coupling. Additionally, bicoherence relaxes the artificial spectral constraints introduced by conventional PAC, corrects for its poor biases, and accounts for asymmetry in the rhythms (Giehl et al., 2021; Kovach et al., 2018; Sheremet et al., 2016).

Theoretically, the bispectrum is statistically zero for linear systems with mutually independent Fourier coefficients. For nonlinear systems, the bispectrum will exhibit peaks at triads ( $f_{n}, f_{m}, f_{n + m}$ ) that are phase correlated, measuring the degree of three-wave coupling (Sheremet et al., 2016). In practice, however, bicoherence has a positive bias. The background activity of LFP signals can be estimated by properties of red noise which can then be used for significance testing (Bédard and Destexhe, 2009; Torrence and Compo, 1998). To calculate the statistical significance of the local autobicoherence, we generated red noise with the same length of our original signals and computed bicoherence for the red noise sample. We repeated this procedure 5000 times to obtain the null distribution. We then compared the original data to the null distribution to obtain uncorrected p-values, thresholded for significance at 0.05. We then performed cluster-based multiple comparison correction as described in the Cross-frequency power correlation section.

Detection and prevalence of transient oscillatory events

We used the BOSC algorithm (Caplan et al., 2001; Hughes et al., 2012) to detect transient bouts of heightened frequency-specific power using a joint amplitude and duration thresholding procedure. A specific concern we wished to address was that the 1 s windowed power spectral method may result in false negatives (i.e. missing brief theta epochs), therefore we set a maximally permissive duration threshold of 3 cycles, and an amplitude threshold using sixth-order wavelets passing the 95th percentile of model fit distributions. From these detected events, we computed occupancy rate using the formula:

O c c u p a n c y r a t e (%) = \frac{T o t a l d u r a t i o n o f d e t e c t e d e v e n t s}{D u r a t i o n o f t h e o r i g i n a l s i g n a l} \times 100

Occupancy rate is a measure of prevalence, showing the percentage of time spent in an oscillatory event of a specific frequency, also referred to as P_episode(f).

BOSC algorithm provides logical matrix of the form [number of frequencies * timepoint] where values are 1 when an event was detected and 0 otherwise. For bout duration distributions we summed values across frequency ranges of interest and then performed the logical operation larger than 1, thus in MATLAB it will be, where detected is the BOSC output:

thresholded = sum(detected(ThetaRange,:)) ≥ 1;

We found start (thresholded switches from 0 to 1) and stop (thresholded switches from 1 to 0) of events and removed events that were incomplete, only had start or stop, and computed duration. We then fitted a kernel probability distribution to the duration values using fitdist function in MATLAB. We used bandwidth of 50 and 10 for theta and gamma oscillations, respectively.

Spike-field synchronization

To quantify spike-field synchronization, we used fieldtrip toolbox (MATLAB) to compute PPC which is unbiased by the number of spikes (Vinck et al., 2010). Raw continuous recordings were resampled with a 1000 Hz sampling rate. The spectral content was estimated with a frequency-dependent Hanning window with 5 cycles per frequency and frequency resolution of 1 Hz. All detected spikes of a unit during the session were included. To assess the statistical significance of spike-field synchronization, we first used a non-parametric permutation test with minimal assumptions. In this procedure, the distribution of PPC values was estimated from 1000 iterations of shuffled spike times of each cell. We used the PPC distribution of shuffles to compute the PPC threshold for significance at each frequency. We applied a threshold of uncorrected p<0.05 to determine the significant synchronization at each frequency. Only PPC values that exceeded the statistical threshold and had a Rayleigh test p<0.05 and a minimum peak and peak prominence of 0.005 were reported as significant. To obtain the probability distribution of observing significant PPC values at a frequency, we fitted a kernel probability distribution to significant frequency values using fitdist function in MATLAB. We used a bandwidth of 4.

To compare spike-field synchronization during SWRs, we extracted spikes inside a 600 ms window centered around the SWR events and computed PPC [PPC_SWR]. These spikes were then excluded from the unit spike timestamps and PPC was calculated for the remaining ‘residual’ spikes [PPC_residual]. PPC_SWR was then compared with PPC_residual. Only cells with at least 20 spikes during ripple time windows were included in this analysis (N = 185).

To test the significance of differences in spike-field coupling within SWR epochs or excluding them, on a per-unit basis, spikes were randomly selected and assigned to SWR and residual conditions. In this random selection, spike counts were controlled to correspond to the original condition. We performed the random selection 1000 times and measured the difference between PPC in each iteration to obtain the null distribution. Then, we grouped frequencies into six bands 2–3, 4–10, 11–20, 21–40, 41–100, 101–200 Hz. In each frequency band, we found the peak frequency at which the absolute PPC difference was largest and only tested these for significance. If the p-value of PPC difference was less than 0.05 (two-tailed) after FDR correction, it was labeled as significant.

Theta modulation index estimation

We used the method described by Royer et al., 2010, to quantify the degree of theta modulation in single units. For all units, we first computed the autocorrelogram of the cell, in 10 ms bins from –500 ms to +500 ms, normalized to the maximum value between 100 ms and 150 ms (corresponding to theta modulation), and clipped all values above 1. We only included autocorrelograms with at least 100 counts for further steps (N=240 units). We then fit each autocorrelogram with the following function:

y (t) = [a (\sin (ω t) + 1) + b] * e^{- | t | / τ_{1}} + c * e^{- t^{2} / τ_{2}^{2}}

where $t$ is the autocorrelogram time lag from –700 ms to 700 ms, and $a - c$ , $ω$ , and $τ_{1 - 2}$ were fit using the $f m i n s e a r c h$ optimization function in MATLAB. The theta indexes were defined as the ratio of the fit parameters $a / b$ . For best-fitting performance, we restricted possible values for $ω$ to (4, 10), for $a$ and $b$ to non-negative values, for $c$ to (0, 0.2), and for $τ_{2}$ to (0, 0.05).

Additional single-unit datasets

To generate example plots of theta rhythmic cells (Figure 3), recordings from the Buzsáki laboratory were included (https://buzsakilab.nyumc.org/datasets/).

Data availability

The code used to process these data are available at https://github.com/hoffman-lab/Manuscripts/tree/main/AbbaspoorHussinHoffman2023 (copy archived at Abbaspoor et al., 2023). Data structures can be downloaded at https://zenodo.org/record/7757458. Previous reports from the stationary data are Hussin et al., 2020; Leonard and Hoffman, 2017; Leonard et al., 2015.

The following data sets were generated

1. Abbaspoor S
(2023) Zenodo
Theta-and gamma-band oscillatory uncoupling in the macaque hippocampus.
https://doi.org/10.5281/zenodo.7757458

References

Software
(2023) Manuscripts, version swh:1:rev:1a7deaa154172b248cb063b4e9a789df0b3f63a8
Software Heritage.

https://archive.softwareheritage.org/swh:1:dir:6ebff0e03b59ec0da1b72fbc7fb5d04f68e1c547;origin=https://github.com/hoffman-lab/Manuscripts;visit=swh:1:snp:017a890e3e3f56006474c797316c413a8e0889fd;anchor=swh:1:rev:1a7deaa154172b248cb063b4e9a789df0b3f63a8
1. Allen TA
2. Salz DM
3. McKenzie S
4. Fortin NJ
(2016) Nonspatial sequence coding in CA1 neurons
The Journal of Neuroscience 36:1547–1563.
https://doi.org/10.1523/JNEUROSCI.2874-15.2016
- PubMed
- Google Scholar
1. Aru J
2. Aru J
3. Priesemann V
4. Wibral M
5. Lana L
6. Pipa G
7. Singer W
8. Vicente R
(2015) Untangling cross-frequency coupling in neuroscience
Current Opinion in Neurobiology 31:51–61.
https://doi.org/10.1016/j.conb.2014.08.002
- PubMed
- Google Scholar
1. Axmacher N
2. Henseler MM
3. Jensen O
4. Weinreich I
5. Elger CE
6. Fell J
(2010) Cross-Frequency coupling supports multi-item working memory in the human hippocampus
PNAS 107:3228–3233.
https://doi.org/10.1073/pnas.0911531107
- PubMed
- Google Scholar
1. Bédard C
2. Destexhe A
(2009) Macroscopic models of local field potentials and the apparent 1/f noise in brain activity
Biophysical Journal 96:2589–2603.
https://doi.org/10.1016/j.bpj.2008.12.3951
- PubMed
- Google Scholar
1. Bódizs R
2. Kántor S
3. Szabó G
4. Szûcs A
5. Erõss L
6. Halász P
(2001) Rhythmic hippocampal slow oscillation characterizes REM sleep in humans
Hippocampus 11:747–753.
https://doi.org/10.1002/hipo.1090
- PubMed
- Google Scholar
1. Bragin A
2. Jandó G
3. Nádasdy Z
4. Hetke J
5. Wise K
6. Buzsáki G
(1995a) Gamma (40-100 Hz) oscillation in the hippocampus of the behaving rat
The Journal of Neuroscience 15:47–60.
https://doi.org/10.1523/JNEUROSCI.15-01-00047.1995
- PubMed
- Google Scholar
(1995b) Dentate EEG spikes and associated interneuronal population bursts in the hippocampal hilar region of the rat
Journal of Neurophysiology 73:1691–1705.
https://doi.org/10.1152/jn.1995.73.4.1691
- PubMed
- Google Scholar
(1997) Bicoherence of intracranial EEG in sleep, wakefulness and seizures
Electroencephalography and Clinical Neurophysiology 103:661–678.
https://doi.org/10.1016/s0013-4694(97)00087-4
- PubMed
- Google Scholar
1. Bush D
2. Burgess N
(2020) Advantages and detection of phase coding in the absence of rhythmicity
Hippocampus 30:745–762.
https://doi.org/10.1002/hipo.23199
- PubMed
- Google Scholar
(1983) Cellular bases of hippocampal EEG in the behaving rat
Brain Research 287:139–171.
https://doi.org/10.1016/0165-0173(83)90037-1
- PubMed
- Google Scholar
1. Buzsáki G
(1989) Two-Stage model of memory trace formation: a role for `` noisy'' brain states
Neuroscience 31:551–570.
https://doi.org/10.1016/0306-4522(89)90423-5
- PubMed
- Google Scholar
1. Buzsáki G
(1996) The hippocampo-neocortical dialogue
Cerebral Cortex 6:81–92.
https://doi.org/10.1093/cercor/6.2.81
- PubMed
- Google Scholar
1. Buzsáki G
(2002) Theta oscillations in the hippocampus
Neuron 33:325–340.
https://doi.org/10.1016/s0896-6273(02)00586-x
- PubMed
- Google Scholar
1. Buzsáki G
2. Draguhn A
(2004) Neuronal oscillations in cortical networks
Science 304:1926–1929.
https://doi.org/10.1126/science.1099745
- PubMed
- Google Scholar
1. Buzsáki G
2. Wang XJ
(2012) Mechanisms of gamma oscillations
Annual Review of Neuroscience 35:203–225.
https://doi.org/10.1146/annurev-neuro-062111-150444
- PubMed
- Google Scholar
1. Buzsáki G
(2015) Hippocampal sharp wave-ripple: a cognitive biomarker for episodic memory and planning
Hippocampus 25:1073–1188.
https://doi.org/10.1002/hipo.22488
- PubMed
- Google Scholar
1. Cacucci F
2. Lever C
3. Wills TJ
4. Burgess N
5. O’Keefe J
(2004) Theta-modulated place-by-direction cells in the hippocampal formation in the rat
The Journal of Neuroscience 24:8265–8277.
https://doi.org/10.1523/JNEUROSCI.2635-04.2004
- PubMed
- Google Scholar
1. Cantero JL
2. Atienza M
3. Stickgold R
4. Kahana MJ
5. Madsen JR
6. Kocsis B
(2003) Sleep-dependent θ oscillations in the human hippocampus and neocortex
The Journal of Neuroscience 23:10897–10903.
https://doi.org/10.1523/JNEUROSCI.23-34-10897.2003
- Google Scholar
(2001) Distinct patterns of brain oscillations underlie two basic parameters of human maze learning
Journal of Neurophysiology 86:368–380.
https://doi.org/10.1152/jn.2001.86.1.368
- Google Scholar
1. Chau VL
2. Murphy EF
3. Rosenbaum RS
4. Ryan JD
5. Hoffman KL
(2011) A flicker change detection task reveals object-in-scene memory across species
Frontiers in Behavioral Neuroscience 5:58.
https://doi.org/10.3389/fnbeh.2011.00058
- PubMed
- Google Scholar
1. Colgin LL
2. Denninger T
3. Fyhn M
4. Hafting T
5. Bonnevie T
6. Jensen O
7. Moser MB
8. Moser EI
(2009) Frequency of gamma oscillations routes flow of information in the hippocampus
Nature 462:353–357.
https://doi.org/10.1038/nature08573
- PubMed
- Google Scholar
1. Colgin LL
2. Moser EI
(2010) Gamma oscillations in the hippocampus
Physiology 25:319–329.
https://doi.org/10.1152/physiol.00021.2010
- PubMed
- Google Scholar
1. Colgin LL
(2013) Mechanisms and functions of theta rhythms
Annual Review of Neuroscience 36:295–312.
https://doi.org/10.1146/annurev-neuro-062012-170330
- PubMed
- Google Scholar
1. Colgin LL
(2016) Rhythms of the hippocampal network
Nature Reviews. Neuroscience 17:239–249.
https://doi.org/10.1038/nrn.2016.21
- PubMed
- Google Scholar
(2019) Spatial encoding in primate hippocampus during free navigation
PLOS Biology 17:e3000546.
https://doi.org/10.1371/journal.pbio.3000546
- PubMed
- Google Scholar
1. Cox R
2. Rüber T
3. Staresina BP
4. Fell J
(2019) Heterogeneous profiles of coupled sleep oscillations in human hippocampus
NeuroImage 202:116178.
https://doi.org/10.1016/j.neuroimage.2019.116178
- PubMed
- Google Scholar
(2003) Mechanisms of gamma oscillations in the hippocampus of the behaving rat
Neuron 37:311–322.
https://doi.org/10.1016/s0896-6273(02)01169-8
- PubMed
- Google Scholar
1. de Cheveigné A
2. Nelken I
(2019) Filters: when, why, and how (not) to use them
Neuron 102:280–293.
https://doi.org/10.1016/j.neuron.2019.02.039
- Google Scholar
(2007) Distinguishing low frequency oscillations within the 1/f spectral behaviour of electromagnetic brain signals
Behavioral and Brain Functions 3:62.
https://doi.org/10.1186/1744-9081-3-62
- PubMed
- Google Scholar
1. Donoghue T
2. Haller M
3. Peterson EJ
4. Varma P
5. Sebastian P
6. Gao R
7. Noto T
8. Lara AH
9. Wallis JD
10. Knight RT
11. Shestyuk A
12. Voytek B
(2020) Parameterizing neural power spectra into periodic and aperiodic components
Nature Neuroscience 23:1655–1665.
https://doi.org/10.1038/s41593-020-00744-x
- PubMed
- Google Scholar
1. Dragan MC
2. Leonard TK
3. Lozano AM
4. McAndrews MP
5. Ng K
6. Ryan JD
7. Tang-Wai DF
8. Wynn JS
9. Hoffman KL
(2017) Pupillary responses and memory-guided visual search reveal age-related and alzheimer’s-related memory decline
Behavioural Brain Research 322:351–361.
https://doi.org/10.1016/j.bbr.2016.09.014
- PubMed
- Google Scholar
1. Eliav T
2. Geva-Sagiv M
3. Yartsev MM
4. Finkelstein A
5. Rubin A
6. Las L
7. Ulanovsky N
(2018) Nonoscillatory phase coding and synchronization in the bat hippocampal formation
Cell 175:1119–1130.
https://doi.org/10.1016/j.cell.2018.09.017
- PubMed
- Google Scholar
(2021) Gamma rhythm communication between entorhinal cortex and dentate gyrus neuronal assemblies
Science 372:eabf3119.
https://doi.org/10.1126/science.abf3119
- PubMed
- Google Scholar
(2023) Over and above frequency: gamma oscillations as units of neural circuit operations
Neuron 111:936–953.
https://doi.org/10.1016/j.neuron.2023.02.026
- PubMed
- Google Scholar
(2014) Beta2 oscillations (23-30 Hz) in the mouse hippocampus during novel object recognition
The European Journal of Neuroscience 40:3693–3703.
https://doi.org/10.1111/ejn.12739
- PubMed
- Google Scholar
(2021) Beta2 oscillations in hippocampal-cortical circuits during novelty detection
Frontiers in Systems Neuroscience 15:617388.
https://doi.org/10.3389/fnsys.2021.617388
- PubMed
- Google Scholar
1. Fuhrmann F
2. Justus D
3. Sosulina L
4. Kaneko H
5. Beutel T
6. Friedrichs D
7. Schoch S
8. Schwarz MK
9. Fuhrmann M
10. Remy S
(2015) Locomotion, theta oscillations, and the speed-correlated firing of hippocampal neurons are controlled by a medial septal glutamatergic circuit
Neuron 86:1253–1264.
https://doi.org/10.1016/j.neuron.2015.05.001
- PubMed
- Google Scholar
1. Giehl J
2. Noury N
3. Siegel M
(2021) Dissociating harmonic and non-harmonic phase-amplitude coupling in the human brain
NeuroImage 227:117648.
https://doi.org/10.1016/j.neuroimage.2020.117648
- PubMed
- Google Scholar
1. Goyal A
2. Miller J
3. Qasim SE
4. Watrous AJ
5. Zhang H
6. Stein JM
7. Inman CS
8. Gross RE
9. Willie JT
10. Lega B
11. Lin JJ
12. Sharan A
13. Wu C
14. Sperling MR
15. Sheth SA
16. McKhann GM
17. Smith EH
18. Schevon C
19. Jacobs J
(2020) Functionally distinct high and low theta oscillations in the human hippocampus
Nature Communications 11:2469.
https://doi.org/10.1038/s41467-020-15670-6
- PubMed
- Google Scholar
1. Green JD
2. Arduini AA
(1954) Hippocampal electrical activity in arousal
Journal of Neurophysiology 17:533–557.
https://doi.org/10.1152/jn.1954.17.6.533
- PubMed
- Google Scholar
1. Hahn G
2. Ponce-Alvarez A
3. Deco G
4. Aertsen A
5. Kumar A
(2019) Portraits of communication in neuronal networks
Nature Reviews. Neuroscience 20:117–127.
https://doi.org/10.1038/s41583-018-0094-0
- PubMed
- Google Scholar
(1978) Human hippocampal formation EEG desynchronizes during attentiveness and movement
Electroencephalography and Clinical Neurophysiology 44:778–781.
https://doi.org/10.1016/0013-4694(78)90212-2
- PubMed
- Google Scholar
(2020) Theta oscillations in human memory
Trends in Cognitive Sciences 24:208–227.
https://doi.org/10.1016/j.tics.2019.12.006
- Google Scholar
(2013) Saccades during visual exploration align hippocampal 3-8 Hz rhythms in human and non-human primates
Frontiers in Systems Neuroscience 7:43.
https://doi.org/10.3389/fnsys.2013.00043
- PubMed
- Google Scholar
(2016) Slow gamma rhythms in CA3 are entrained by slow gamma activity in the dentate gyrus
Journal of Neurophysiology 116:2594–2603.
https://doi.org/10.1152/jn.00499.2016
- PubMed
- Google Scholar
(2012) BOSC: a better oscillation detection method, extracts both sustained and transient rhythms from rat hippocampal recordings
Hippocampus 22:1417–1428.
https://doi.org/10.1002/hipo.20979
- PubMed
- Google Scholar
(2020) Sharp-wave ripple features in macaques depend on behavioral state and cell-type specific firing
Hippocampus 30:50–59.
https://doi.org/10.1002/hipo.23046
- PubMed
- Google Scholar
(2022) Retrosplenial and hippocampal synchrony during retrieval of old memories in macaques
The Journal of Neuroscience 42:7947–7956.
https://doi.org/10.1523/JNEUROSCI.0001-22.2022
- Google Scholar
1. Hyafil A
(2015) Misidentifications of specific forms of cross-frequency coupling: three warnings
Frontiers in Neuroscience 9:370.
https://doi.org/10.3389/fnins.2015.00370
- PubMed
- Google Scholar
1. Igarashi KM
2. Lu L
3. Colgin LL
4. Moser MB
5. Moser EI
(2014) Coordination of entorhinal-hippocampal ensemble activity during associative learning
Nature 510:143–147.
https://doi.org/10.1038/nature13162
- PubMed
- Google Scholar
1. Jackson J
2. Amilhon B
3. Goutagny R
4. Bott JB
5. Manseau F
6. Kortleven C
7. Bressler SL
8. Williams S
(2014) Reversal of theta rhythm flow through intact hippocampal circuits
Nature Neuroscience 17:1362–1370.
https://doi.org/10.1038/nn.3803
- PubMed
- Google Scholar
1. Jacobs J
(2014) Hippocampal theta oscillations are slower in humans than in rodents: implications for models of spatial navigation and memory
Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences 369:20130304.
https://doi.org/10.1098/rstb.2013.0304
- PubMed
- Google Scholar
1. Jones SR
(2016) When brain rhythms are ’'t `` rhythmic'': implication for their mechanisms and meaning
Current Opinion in Neurobiology 40:72–80.
https://doi.org/10.1016/j.conb.2016.06.010
- PubMed
- Google Scholar
(2009) Gamma-Band synchronization in the macaque hippocampus and memory formation
The Journal of Neuroscience 29:12521–12531.
https://doi.org/10.1523/JNEUROSCI.0640-09.2009
- PubMed
- Google Scholar
(2013) Oscillatory activity in the monkey hippocampus during visual exploration and memory formation
PNAS 110:13144–13149.
https://doi.org/10.1073/pnas.1302351110
- PubMed
- Google Scholar
1. Katz CN
2. Patel K
3. Talakoub O
4. Groppe D
5. Hoffman K
6. Valiante TA
(2020) Differential generation of saccade, fixation, and image-onset event-related potentials in the human mesial temporal lobe
Cerebral Cortex 30:5502–5516.
https://doi.org/10.1093/cercor/bhaa132
- PubMed
- Google Scholar
1. Katz CN
2. Schjetnan AGP
3. Patel K
4. Barkley V
5. Hoffman KL
6. Kalia SK
7. Duncan KD
8. Valiante TA
(2022) A corollary discharge mediates saccade-related inhibition of single units in mnemonic structures of the human brain
Current Biology 32:3082–3094.
https://doi.org/10.1016/j.cub.2022.06.015
- PubMed
- Google Scholar
(2003) Brain-state- and cell-type-specific firing of hippocampal interneurons in vivo
Nature 421:844–848.
https://doi.org/10.1038/nature01374
- PubMed
- Google Scholar
1. Klausberger T
2. Somogyi P
(2008) Neuronal diversity and temporal dynamics: the unity of hippocampal circuit operations
Science 321:53–57.
https://doi.org/10.1126/science.1149381
- PubMed
- Google Scholar
1. Kleen JK
2. Chung JE
3. Sellers KK
4. Zhou J
5. Triplett M
6. Lee K
7. Tooker A
8. Haque R
9. Chang EF
(2021) Bidirectional propagation of low frequency oscillations over the human hippocampal surface
Nature Communications 12:2764.
https://doi.org/10.1038/s41467-021-22850-5
- PubMed
- Google Scholar
(2018) The bispectrum and its relationship to phase-amplitude coupling
NeuroImage 173:518–539.
https://doi.org/10.1016/j.neuroimage.2018.02.033
- PubMed
- Google Scholar
(1975) Two types of hippocampal rhythmical slow activity in both the rabbit and the rat: relations to behavior and effects of atropine, diethyl ether, urethane, and pentobarbital
Experimental Neurology 49:58–85.
https://doi.org/10.1016/0014-4886(75)90195-8
- PubMed
- Google Scholar
(2016) Reward expectancy strengthens CA1 theta and beta band synchronization and hippocampal-ventral striatal coupling
The Journal of Neuroscience 36:10598–10610.
https://doi.org/10.1523/JNEUROSCI.0682-16.2016
- PubMed
- Google Scholar
1. Lasztóczi B
2. Klausberger T
(2014) Layer-specific GABAergic control of distinct gamma oscillations in the CA1 hippocampus
Neuron 81:1126–1139.
https://doi.org/10.1016/j.neuron.2014.01.021
- PubMed
- Google Scholar
1. Leonard TK
2. Mikkila JM
3. Eskandar EN
4. Gerrard JL
5. Kaping D
6. Patel SR
7. Womelsdorf T
8. Hoffman KL
(2015) Sharp wave ripples during visual exploration in the primate hippocampus
The Journal of Neuroscience 35:14771–14782.
https://doi.org/10.1523/JNEUROSCI.0864-15.2015
- Google Scholar
1. Leonard TK
2. Hoffman KL
(2017) Sharp-wave ripples in primates are enhanced near remembered visual objects
Current Biology 27:257–262.
https://doi.org/10.1016/j.cub.2016.11.027
- PubMed
- Google Scholar
1. Lisman JE
2. Jensen O
(2013) The θ-γ neural code
Neuron 77:1002–1016.
https://doi.org/10.1016/j.neuron.2013.03.007
- PubMed
- Google Scholar
1. Liu AA
2. Henin S
3. Abbaspoor S
4. Bragin A
5. Buffalo EA
6. Farrell JS
7. Foster DJ
8. Frank LM
9. Gedankien T
10. Gotman J
11. Guidera JA
12. Hoffman KL
13. Jacobs J
14. Kahana MJ
15. Li L
16. Liao Z
17. Lin JJ
18. Losonczy A
19. Malach R
20. van der Meer MA
21. McClain K
22. McNaughton BL
23. Norman Y
24. Navas-Olive A
25. de la Prida LM
26. Rueckemann JW
27. Sakon JJ
28. Skelin I
29. Soltesz I
30. Staresina BP
31. Weiss SA
32. Wilson MA
33. Zaghloul KA
34. Zugaro M
35. Buzsáki G
(2022) A consensus statement on detection of hippocampal sharp wave ripples and differentiation from other fast oscillations
Nature Communications 13:6000.
https://doi.org/10.1038/s41467-022-33536-x
- PubMed
- Google Scholar
(2018) Parsing hippocampal theta oscillations by nested spectral components during spatial exploration and memory-guided behavior
Neuron 100:940–952.
https://doi.org/10.1016/j.neuron.2018.09.031
- PubMed
- Google Scholar
1. M Aghajan Z
2. Schuette P
3. Fields TA
4. Tran ME
5. Siddiqui SM
6. Hasulak NR
7. Tcheng TK
8. Eliashiv D
9. Mankin EA
10. Stern J
11. Fried I
12. Suthana N
(2017) Theta oscillations in the human medial temporal lobe during real-world ambulatory movement
Current Biology 27:3743–3751.
https://doi.org/10.1016/j.cub.2017.10.062
- PubMed
- Google Scholar
(2003) Cellular and network mechanisms underlying spontaneous sharp wave–ripple complexes in mouse hippocampal slices
The Journal of Physiology 550:873–887.
https://doi.org/10.1113/jphysiol.2003.044602
- Google Scholar
1. Mao D
2. Avila E
3. Caziot B
4. Laurens J
5. Dickman JD
6. Angelaki DE
(2021) Spatial modulation of hippocampal activity in freely moving macaques
Neuron 109:3521–3534.
https://doi.org/10.1016/j.neuron.2021.09.032
- PubMed
- Google Scholar
(2004) Measuring fundamental frequencies in local field potentials
Journal of Neuroscience Methods 138:97–105.
https://doi.org/10.1016/j.jneumeth.2004.03.014
- PubMed
- Google Scholar
(1975) Relationship between hippocampal theta activity and running speed in the rat
Journal of Comparative and Physiological Psychology 88:324–328.
https://doi.org/10.1037/h0076177
- PubMed
- Google Scholar
(2017) Hippocampal gamma-band synchrony and pupillary responses index memory during visual search
Hippocampus 27:425–434.
https://doi.org/10.1002/hipo.22702
- PubMed
- Google Scholar
1. Moroni F
2. Nobili L
3. Curcio G
4. De Carli F
5. Fratello F
6. Marzano C
7. De Gennaro L
8. Ferrillo F
9. Cossu M
10. Francione S
11. Lo Russo G
12. Bertini M
13. Ferrara M
(2007) Sleep in the human hippocampus: a stereo-EEG study
PLOS ONE 2:e867.
https://doi.org/10.1371/journal.pone.0000867
- PubMed
- Google Scholar
1. Nuñez A
2. Buño W
(2021) The theta rhythm of the hippocampus: from neuronal and circuit mechanisms to behavior
Frontiers in Cellular Neuroscience 15:649262.
https://doi.org/10.3389/fncel.2021.649262
- PubMed
- Google Scholar
1. O’Keefe J
2. Recce ML
(1993) Phase relationship between hippocampal place units and the EEG theta rhythm
Hippocampus 3:317–330.
https://doi.org/10.1002/hipo.450030307
- PubMed
- Google Scholar
1. Pesaran B
2. Vinck M
3. Einevoll GT
4. Sirota A
5. Fries P
6. Siegel M
7. Truccolo W
8. Schroeder CE
9. Srinivasan R
(2018) Investigating large-scale brain Dynamics using field potential recordings: Analysis and interpretation
Nature Neuroscience 21:903–919.
https://doi.org/10.1038/s41593-018-0171-8
- PubMed
- Google Scholar
(2021) Phase precession in the human hippocampus and entorhinal cortex
Cell 184:3242–3255.
https://doi.org/10.1016/j.cell.2021.04.017
- PubMed
- Google Scholar
(2015) Theta and beta oscillatory dynamics in the dentate gyrus reveal a shift in network processing state during cue encounters
Frontiers in Systems Neuroscience 9:96.
https://doi.org/10.3389/fnsys.2015.00096
- PubMed
- Google Scholar
1. Rex CS
2. Colgin LL
3. Jia Y
4. Casale M
5. Yanagihara TK
6. Debenedetti M
7. Gall CM
8. Kramar EA
9. Lynch G
(2009) Origins of an intrinsic hippocampal EEG pattern
PLOS ONE 4:e7761.
https://doi.org/10.1371/journal.pone.0007761
- PubMed
- Google Scholar
(2014) Timing of single-neuron and local field potential responses in the human medial temporal lobe
Current Biology 24:299–304.
https://doi.org/10.1016/j.cub.2013.12.004
- PubMed
- Google Scholar
1. Roux F
2. Parish G
3. Chelvarajah R
4. Rollings DT
5. Sawlani V
6. Hamer H
7. Gollwitzer S
8. Kreiselmeyer G
9. Ter Wal MJ
10. Kolibius L
11. Staresina BP
12. Wimber M
13. Self MW
14. Hanslmayr S
(2022) Oscillations support short latency co-firing of neurons during human episodic memory formation
eLife 11:e78109.
https://doi.org/10.7554/eLife.78109
- PubMed
- Google Scholar
1. Royer S
2. Sirota A
3. Patel J
4. Buzsáki G
(2010) Distinct representations and theta dynamics in dorsal and ventral hippocampus
The Journal of Neuroscience 30:1777–1787.
https://doi.org/10.1523/JNEUROSCI.4681-09.2010
- PubMed
- Google Scholar
(2010) Human memory strength is predicted by theta-frequency phase-locking of single neurons
Nature 464:903–907.
https://doi.org/10.1038/nature08860
- PubMed
- Google Scholar
1. Sainsbury RS
(1998) Hippocampal theta: a sensory-inhibition theory of function
Neuroscience and Biobehavioral Reviews 22:237–241.
https://doi.org/10.1016/s0149-7634(97)00011-0
- PubMed
- Google Scholar
Book
1. Saleem K
2. Logothetis N
(2012)
Atlas of the Rhesus Monkey Brain in Stereotaxic Coordinates

Elsevier.
- Google Scholar
(2020) Awake delta and theta-rhythmic hippocampal network modes during intermittent locomotor behaviors in the rat
Behavioral Neuroscience 134:529–546.
https://doi.org/10.1037/bne0000409
- PubMed
- Google Scholar
(2016) Movement enhances the nonlinearity of hippocampal theta
The Journal of Neuroscience 36:4218–4230.
https://doi.org/10.1523/JNEUROSCI.3564-15.2016
- PubMed
- Google Scholar
(2007) Eeg sharp waves and sparse ensemble unit activity in the macaque hippocampus
Journal of Neurophysiology 98:898–910.
https://doi.org/10.1152/jn.00401.2007
- PubMed
- Google Scholar
1. Stangl M
2. Topalovic U
3. Inman CS
4. Hiller S
5. Villaroman D
6. Aghajan ZM
7. Christov-Moore L
8. Hasulak NR
9. Rao VR
10. Halpern CH
11. Eliashiv D
12. Fried I
13. Suthana N
(2021) Boundary-anchored neural mechanisms of location-encoding for self and others
Nature 589:420–425.
https://doi.org/10.1038/s41586-020-03073-y
- PubMed
- Google Scholar
1. Stewart M
2. Fox SE
(1991) Hippocampal theta activity in monkeys
Brain Research 538:59–63.
https://doi.org/10.1016/0006-8993(91)90376-7
- PubMed
- Google Scholar
1. Tai SK
2. Ma J
3. Ossenkopp KP
4. Leung LS
(2012) Activation of immobility-related hippocampal theta by cholinergic septohippocampal neurons during vestibular stimulation
Hippocampus 22:914–925.
https://doi.org/10.1002/hipo.20955
- PubMed
- Google Scholar
1. Takeuchi S
2. Mima T
3. Murai R
4. Shimazu H
5. Isomura Y
6. Tsujimoto T
(2015) Gamma oscillations and their cross-frequency coupling in the primate hippocampus during sleep
Sleep 38:1085–1091.
https://doi.org/10.5665/sleep.4818
- PubMed
- Google Scholar
(2016) Closed-Loop interruption of hippocampal ripples through fornix stimulation in the non-human primate
Brain Stimulation 9:911–918.
https://doi.org/10.1016/j.brs.2016.07.010
- PubMed
- Google Scholar
Preprint
1. Talakoub O
2. Sayegh PF
3. Womelsdorf T
4. Zinke W
5. Fries P
6. Lewis CM
7. Hoffman KL
(2019) Hippocampal and Neocortical Oscillations Are Tuned to Behavioral State in Freely-Behaving Macaques
bioRxiv.
https://doi.org/10.1101/552877
- Google Scholar
1. Tamura R
2. Nishida H
3. Eifuku S
4. Fushiki H
5. Watanabe Y
6. Uchiyama K
7. Vyazovskiy V
(2013) Sleep-stage correlates of hippocampal electroencephalogram in primates
PLOS ONE 8:e82994.
https://doi.org/10.1371/journal.pone.0082994
- PubMed
- Google Scholar
1. Thammasan N
2. Miyakoshi M
(2020) Cross-Frequency power-power coupling analysis: a useful cross-frequency measure to classify ICA-decomposed EEG
Sensors 20:7040.
https://doi.org/10.3390/s20247040
- PubMed
- Google Scholar
1. Torrence C
2. Compo GP
(1998) A practical guide to wavelet analysis
Bulletin of the American Meteorological Society 79:61–78.
https://doi.org/10.1175/1520-0477(1998)079<0061:APGTWA>2.0.CO;2
- Google Scholar
1. Tort ABL
2. Kramer MA
3. Thorn C
4. Gibson DJ
5. Kubota Y
6. Graybiel AM
7. Kopell NJ
(2008) Dynamic cross-frequency couplings of local field potential oscillations in rat striatum and hippocampus during performance of a T-maze task
PNAS 105:20517–20522.
https://doi.org/10.1073/pnas.0810524105
- PubMed
- Google Scholar
(2010) Measuring phase-amplitude coupling between neuronal oscillations of different frequencies
Journal of Neurophysiology 104:1195–1210.
https://doi.org/10.1152/jn.00106.2010
- PubMed
- Google Scholar
1. Trimper JB
2. Galloway CR
3. Jones AC
4. Mandi K
5. Manns JR
(2017) Gamma oscillations in rat hippocampal subregions dentate gyrus, CA3, CA1, and subiculum underlie associative memory encoding
Cell Reports 21:2419–2432.
https://doi.org/10.1016/j.celrep.2017.10.123
- PubMed
- Google Scholar
1. Uchida S
2. Maehara T
3. Hirai N
4. Okubo Y
5. Shimizu H
(2001) Cortical oscillations in human medial temporal lobe during wakefulness and all-night sleep
Brain Research 891:7–19.
https://doi.org/10.1016/s0006-8993(00)03154-1
- PubMed
- Google Scholar
1. Ulanovsky N
2. Moss CF
(2007) Hippocampal cellular and network activity in freely moving echolocating bats
Nature Neuroscience 10:224–233.
https://doi.org/10.1038/nn1829
- PubMed
- Google Scholar
1. Vanderwolf CH
(1969) Hippocampal electrical activity and voluntary movement in the rat
Electroencephalography and Clinical Neurophysiology 26:407–418.
https://doi.org/10.1016/0013-4694(69)90092-3
- PubMed
- Google Scholar
(2010) The pairwise phase consistency: a bias-free measure of rhythmic neuronal synchronization
NeuroImage 51:112–122.
https://doi.org/10.1016/j.neuroimage.2010.01.073
- PubMed
- Google Scholar
Preprint
(2022) Aperiodic Processes Explaining Rhythms in Behavior: A Matter of False Detection or Definition?
PsyArXiv.
https://doi.org/10.31234/osf.io/wzvfh
- Google Scholar
1. Vivekananda U
2. Bush D
3. Bisby JA
4. Baxendale S
5. Rodionov R
6. Diehl B
7. Chowdhury FA
8. McEvoy AW
9. Miserocchi A
10. Walker MC
11. Burgess N
(2021) Theta power and theta-gamma coupling support long-term spatial memory retrieval
Hippocampus 31:213–220.
https://doi.org/10.1002/hipo.23284
- PubMed
- Google Scholar
1. Watrous AJ
2. Lee DJ
3. Izadi A
4. Gurkoff GG
5. Shahlaie K
6. Ekstrom AD
(2013) A comparative study of human and rat hippocampal low-frequency oscillations during spatial navigation
Hippocampus 23:656–661.
https://doi.org/10.1002/hipo.22124
- PubMed
- Google Scholar
1. Whishaw IQ
2. Vanderwolf CH
(1973) Hippocampal EEG and behavior: changes in amplitude and frequency of RSA (theta rhythm) associated with spontaneous and learned movement patterns in rats and cats
Behavioral Biology 8:461–484.
https://doi.org/10.1016/s0091-6773(73)80041-0
- PubMed
- Google Scholar
(2011) Grid cells without theta oscillations in the entorhinal cortex of bats
Nature 479:103–107.
https://doi.org/10.1038/nature10583
- PubMed
- Google Scholar
1. Ylinen A
2. Bragin A
3. Nadasdy Z
4. Jando G
5. Szabo I
6. Sik A
7. Buzsaki G
(1995) Sharp wave-associated high-frequency oscillation (200 Hz) in the intact hippocampus: network and intracellular mechanisms
The Journal of Neuroscience 15:30–46.
https://doi.org/10.1523/JNEUROSCI.15-01-00030.1995
- Google Scholar
(2020) Long-Term memory and hippocampal function support predictive gaze control during goal-directed search
Journal of Vision 20:10.
https://doi.org/10.1167/jov.20.5.10
- PubMed
- Google Scholar
1. Zhou Y
2. Sheremet A
3. Qin Y
4. Kennedy JP
5. DiCola NM
6. Burke SN
7. Maurer AP
(2019) Methodological considerations on the use of different spectral decomposition algorithms to study hippocampal rhythms
ENeuro 6:ENEURO.0142-19.2019.
https://doi.org/10.1523/ENEURO.0142-19.2019
- PubMed
- Google Scholar

Decision letter

Martin Vinck

Reviewing Editor; Ernst Strüngmann Institute (ESI) for Neuroscience in Cooperation with Max Planck Society, Germany
Laura L Colgin

Senior Editor; University of Texas at Austin, United States
Antonio Fernandez-Ruiz

Reviewer; Cornell University, United States

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Decision letter after peer review:

[Editors’ note: the authors submitted for reconsideration following the decision after peer review. What follows is the decision letter after the first round of review.]

Thank you for submitting the paper "Theta- and γ-band oscillatory uncoupling in the macaque hippocampus" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and a Senior Editor. The following individual involved in the review of your submission has agreed to reveal their identity: Antonio Fernandez-Ruiz (Reviewer #2).

Comments to the Authors:

Although all three reviewers found the paper interesting and valuable, they all three agreed on major concerns with regard to several points: 1) One concern was that the data from stationary monkeys cannot be directly compared to moving rats and that the only freely moving data is an N=1 data point in the Supplement. 2) Other concerns were raised about the data analysis, e.g. the characterization, interpretation, and detection of theta oscillations. If you feel that you could however address point no. 1 and point no. 2, which would require adding new experimental data (or analyzing experimental data that you may have already collected that is not currently included in the manuscript) in case of point no. 1, we could consider a resubmission of a strongly improved manuscript. Since a paper of this kind would contain new data that has not yet been evaluated, it would be treated as a new submission with no guarantees of being sent out for full review or ultimate acceptance. If you decide to go this route, we would recommend to first draft a short (1-2 page) action letter, which would be shared with the reviewers, prior to submitting a new manuscript.

Reviewer #1 (Recommendations for the authors):

This is a terse and easy-to-read manuscript, describing an overall lack of theta-dominance in the monkey hippocampus.

It is unfortunate that the most important data – active movements in the monkey – has N=1 and is relegated to the supplement. This weakens the impact of the paper. Furthermore, we don't even know if the electrodes were truly in the hippocampus (no MRIs or histology is shown), or perhaps this one monkey was just a bit unusual. I dislike making such important neuroscience claims based on a case report (this would not be acceptable in human studies, even in intracranial recordings). But I also acknowledge that this is sometimes done in monkey neuroscience.

The paper on its own is certainly thought-provoking, but the authors seem to be unfamiliar with the literature in humans, showing search-, memory-, and movement-related theta oscillations in the hippocampus. I'm not arguing that human hippocampal theta is the same as rodent theta (I don't think it is), but it's not really accurate to make claims about "primate" hippocampal theta based on these findings alone.

The y-axis limits in 2A-B are all very different. Indeed, the PPC magnitudes shown in panel A are basically at the noise level in panel B. That makes panels C-D difficult to interpret, because it's not clear whether the units are robustly (probably meaningfully) modulated by the LFP, or just statistically significantly so (that is, p<.05 but tiny effect size due to large sample size). This is not mentioned or discussed critically.

Perhaps get more data from monkey hippocampus and/or human hippocampus to substantiate these conclusions.

Reviewer #2 (Recommendations for the authors):

Abbaspoor and colleagues investigate in this work the interactions between different oscillatory patterns in the macaque hippocampus. This is an important topic that has been profusely investigated in other model species (mostly rodents), but the apparent differences with primates highlight the need for more comparative studies in the field. The present study has been conducted with the highest standard regarding both experimental and analytical methods. The results presented offer strong support for the main conclusion of the authors (the anticorrelation between theta and γ activities, contrary to what happens in rodents). My comments below are aimed at two main aspects of the study that can be further improved: a more comprehensive description of the data and a more extensive comparison with well-known rodent hippocampal physiology. Provided these corrections are incorporated, I believe the present study will add important evidence about macaque hippocampal oscillations and their comparison with those of rodents, thus being of potential interest for a large audience.

– Figure 1 presents a good description of the spectral properties of macaque hippocampal oscillations; however, spatial and temporal aspects of these activities can be better documented. In rodents, theta, γ and ripple oscillations have very stereotyped laminar profiles. Within the limitations of their recording methods, can the authors describe the laminar characteristics of macaque oscillations (e.g. is γ stronger in the apical CA1 dendrites, etc.)? In rodents, γ and ripples typically occur in a short burst while theta can be quasi-stationary for many seconds. What are the temporal dynamics of these oscillations in the macaque?

– As one of the main goals of this work is to compare theta-γ interactions in macaques versus rodents, it will be useful to reproduce the most common analysis used to illustrate this phenomenon in rodents: theta phase – γ amplitude cross-frequency coupling (Tort et al., 2007). This analysis will offer a complementary view to the power-power and bicoherence analysis already presented.

– In lines 78-80 the authors state 20-35Hz oscillations ('slow γ') have been shown in rodents to be the result of theta harmonics. This is a gross mischaracterization of the rodent literature. Surprisingly, the authors have chosen to focus on the only one outlier paper that makes this claim against the overwhelming body of work suggesting the opposite (existence of slow γ independently of theta harmonics). Just to cite some examples: Csicvari et al., Neuron, 2003; Colgin et al., Nature, 2009; Schomburg et al., Neuron, 2014; Fernandez-Ruiz et al., Neuron, 2017; Cabral et al., Neuron, 2014; Lopes dos Santos et al., Neuron, 2018; Lastozcni and Klausberger, Neuron, 2016; Belluscio et al., JNeurosci, 2012. This statement needs to be amended to provide a fair account of the rodent literature or, if that is the position of the authors, much further elaborated.

– In lines 122-125 a spike-SWR analysis is described but SWR has not been described or characterized before this point. Even if it is not the main point of this work, a basic characterization of SWRs needs to be provided. In Figure 1 there are a couple of events that look like ripples, if that is the case, that needs to be mentioned. How did the authors differentiate SWR from fast γ?

– Could the authors identify different γ sub-bands, as is the case in the rodents. If they couldn't, that is also a result worth mentioning, given the potential disparity with rodents.

– The methods pertaining to single unit isolation and classification are not described in the manuscript. How were individual units isolated and classified into cell types? Which cells were used for the spike-LFP analyses? Pyramidal cells and interneurons should be separated for these analyses.

– It needs to be acknowledged in the discussion that some of the reported differences between macaque and rodents could be due to the very different behavioral states in which oscillation has typically been measured in these cases (head-fixed visual search vs freely moving navigation).

– How were sampling periods selected for the analysis of Figure 3? Theta modulation of rodent ACGs will only be clearly present if periods of theta activity (or movement) were selected. How did the authors ensure that the rodent and macaque data are comparable? If no such preselection was done, it needs to be better explained in the results which exact data went into the analysis. A complementary analysis the authors could include in population CCGs, it may be possible that some rhythmic modulation appears in those that are missed in ACGs due to the sparse firing of pyramidal cells. Pyramidal cells and interneurons should not be mixed here.

Reviewer #3 (Recommendations for the authors):

The research on hippocampal oscillations has been heavily biased towards rodent models. Comparative studies between rodents and other species are lacking in this field, yet these are vital for the understanding of the mechanisms and functions of these rhythms. Therefore, this study by Abbaspoor, Hussin and Hoffman has to be appreciated for providing descriptive analyses of intracranial recordings from the macaque CA1 and relating them to the widely studied rodent CA1. More specifically, the authors focus on theta and related γ oscillations during rest and a visual search task. They found that CA1 LFPs present stronger 15-70Hz γ and reduced "theta" band power during the visual task compared to rest periods. Spike to phase coupling in the same region indicated that neurons coupled to three bands (3-10, 20-30Hz and 60-150Hz). Importantly, analysis of spike to field coherence indicated that in fact the "theta" power is dominated by irregular (non-oscillatory) activity nesting ripple-like events, i.e. most likely emerging from sharp wave events. The authors point out differences between these observations and what has been reported in the rodent CA1- notably, the lack of theta oscillations during "active" behavior.

Although I appreciate the objective and discussion of the work, the analyses need more refinement to truly support their claims. Further, there are a few apparent conceptual inconsistencies that should be addressed.

My first concern relates to the existence (or not) of true theta oscillations in the data.

The authors indicate in several parts of the manuscript (including the title) that they are analyzing "theta" activity. However, they have not presented convincing indications that they analyzed actual theta oscillations. In fact, spectral peaks do not necessarily demonstrate the presence of a true oscillation, as they write themselves. They find no evidence of genuine theta entrainment of spiking activity, arguing that the apparent theta modulation arises actually from sharp-wave-ripple (SWr) events. It is not clear to me if the authors are claiming the theta band in the LFPs arises solely from SWr in their data. If the authors are actually analyzing theta, they should present evidence this is the case. Although cells are not paced at the theta rhythm, if genuine theta oscillations are present in the LFPs it should be possible to show evidence of that (e.g. as Jutras et al. 2013, Fig1G). In fact, theta bouts/bursts have been reported in primate studies (e.g. Leonard et al. 2015). If the authors are convinced to have theta oscillations in their recordings through the BOSC algorithm, this should be better developed in the manuscript. Further, the authors could analyze changes in theta burst incidence if they want to claim theta is reduced during search, and, most importantly, show that this cannot be explained by a decrease of SWr incidence (which they do show to contaminate the "theta band") relative to rest. One of the main claims of the manuscript is the decoupling between 20-35Hz γ and theta, but this can only be addressed if the authors restrict their analysis to true theta oscillations rather than SWr events. Even by doing so, the authors should discuss the possibility that they do not see γ-to-theta modulation because they record from the pyramidal cell layer whereas γ oscillations have been shown to originate in deeper layers in rodents (e.g. Belluscio et al. 2012; Schomburg et 2014; Lasztóczi and Klausberger 2014). This modulation might be easier to record from the pyramidal cell layer in rodents as γ oscillations have stronger signals in these species due to their anatomical features (Buzsáki et al. 2003).

Alternatively, if the authors cannot show convincingly they are analyzing true theta oscillations, what they actually observe is a reduction (although very mild, as shown in Fig1D) of SWr incidence (reflected in a reduction of "theta" band power) during visual search compared to rest, which would be actually consistent with the rodent literature. What could be argued to be a contrast between species is that SWr rate is much higher in primates compared to rodents during "active behavior", but this has already been claimed by the same group (Leonard et al. 2015; Leonard et al. 2017). Moreover, primate visual search is taken here as analogous to rodent overt movement, which is arguably not a fair equivalence. In fact, SWr events have been shown to occur in rodents during decision making and planning when animals are relatively immobile but arguably attentive (Joo and Frank 2018). Here, the authors have a single freely moving animal (shown in the figure supplements), but lack of constant theta oscillations in similar conditions have been shown before (Ekstrom et al., 2005; Watrous et al., 2013; Talakoub et al., 2019). In fact, different correlates between theta and behavior in different species have been discussed for decades (e.g. Winson 1972). The authors should refine their discussion on the equivalence of "active behavior" between species and rewrite the paper according to the fact they are studying SWr events, and not genuine theta oscillations if that is the case. Also, regarding this point, they should clearly discuss what are the novel aspects of their current findings in relation to the literature.

My second point relates to the relevance of the decoupling between 20-35Hz γ and theta.

The authors highlight the observation that 20-35Hz γ oscillations are not coupled to theta. This has been shown before in primates during visual search (Leonard et al. 2015); and also has been shown in rodents. For example, 23-30 Hz (beta2) oscillations occur in the absence of theta (Berke et al. 2008; Figure 3) in freely moving mice, and γ can be recorded in slow-wave sleep in rats (Isomura et al. 2006 Fig7C). Thus, it is not established that γ oscillations in rodents can only occur coupled to theta. Therefore, the presence of this γ rhythm in the absence of theta reported in the manuscript does not strike me as a relevant contrast between primates and rodents. A contrast would be the absence of continuous theta in the primates, which was extensively reported before. I might have missed something, so I would suggest that the authors make clearer what the novel observations presented here, and how these represent an important contrast between species.

https://doi.org/10.7554/eLife.86548.sa1

Author response

[Editors’ note: the authors resubmitted a revised version of the paper for consideration. What follows is the authors’ response to the first round of review.]

Comments to the Authors:

Although all three reviewers found the paper interesting and valuable, they all three agreed on major concerns with regard to several points: 1) One concern was that the data from stationary monkeys cannot be directly compared to moving rats and that the only freely moving data is an N=1 data point in the Supplement. 2) Other concerns were raised about the data analysis, e.g. the characterization, interpretation, and detection of theta oscillations. If you feel that you could however address point no. 1 and point no. 2, which would require adding new experimental data (or analyzing experimental data that you may have already collected that is not currently included in the manuscript) in case of point no. 1, we could consider a resubmission of a strongly improved manuscript. Since a paper of this kind would contain new data that has not yet been evaluated, it would be treated as a new submission with no guarantees of being sent out for full review or ultimate acceptance. If you decide to go this route, we would recommend to first draft a short (1-2 page) action letter, which would be shared with the reviewers, prior to submitting a new manuscript.

We are pleased that the reviewers appreciated the value of the manuscript, and we took to heart the two main previous concerns raised. 1. As suggested, we have added new, additional data from another freely-moving monkey, demonstrating that the results hold even for the conditions considered more directly comparable to moving rats. 2. As requested, we have added more ‘raw’ data visualization and ran additional analyses, to clarify the (co)occurrence of oscillations, including theta oscillations.

Reviewer #1 (Recommendations for the authors):

This is a terse and easy-to-read manuscript, describing an overall lack of theta-dominance in the monkey hippocampus.

It is unfortunate that the most important data – active movements in the monkey – has N=1 and is relegated to the supplement. This weakens the impact of the paper. Furthermore, we don't even know if the electrodes were truly in the hippocampus (no MRIs or histology is shown), or perhaps this one monkey was just a bit unusual. I dislike making such important neuroscience claims based on a case report (this would not be acceptable in human studies, even in intracranial recordings). But I also acknowledge that this is sometimes done in monkey neuroscience.

We have added hippocampal CA1 recordings from a third animal (the second freely-behaving monkey) during waking movement and sleep, which showed similar patterns of results to the other monkeys: stronger, more prevalent 2-10Hz activity during sleep compared to stronger, more prevalent 20-80 Hz activity during alert freely-behaving states. Furthermore, theta and γ frequency bands did not show significant coupling. We also include the location of the implanted electrodes in our subjects which targets the CA1 field of the hippocampus (Figure 1 supplement 1) based on combinations of CT/MR coregistration, post-explant MR, and functional markers of the CA1 layer (Figure 2 supplement 1).

The paper on its own is certainly thought-provoking, but the authors seem to be unfamiliar with the literature in humans, showing search-, memory-, and movement-related theta oscillations in the hippocampus. I'm not arguing that human hippocampal theta is the same as rodent theta (I don't think it is), but it's not really accurate to make claims about "primate" hippocampal theta based on these findings alone.

We have endeavored to better articulate our claims. Over the last decade we have recorded in both humans and macaques as they perform similar tasks, giving us a relatively unique perspective on their commonalities and differences. This has required a stationary-subject preparation since the human participants were hospitalized for their medication-resistant seizures, and would have undergone a craniotomy and macroelectrode implantation with tethered cabling shortly before testing, such that mobile recordings were not feasible for them. (The limitations of this preparation, the reviewer may recognize, is in accordance with nearly all other reports of human MTL recordings.) In the present study we capitalize on the advantages of recording from healthy primates, with data obtained from 60 sessions in total across the 3 animals, each with experimentally-targeted and localized microelectrode recordings to one hippocampal subfield layer (Figure 2 supplement 1), across behavioral states including in movement and in stationary hippocampal-dependent search tasks. To better address how these results are situated relative to the human patient literature, we have separated our Discussion section from the results and extended it to discuss the observations in humans as well as other species (275-420). We point to similarities and differences across the main observations and offer potential explanations for some of the discrepancies. We agree with the reviewer -and now state explicitly – that further research will be needed to understand any discrepancies within the primate order. Overall, there were far more commonalities than differences when focusing on the most congruent preparations and methods.

The y-axis limits in 2A-B are all very different. Indeed, the PPC magnitudes shown in panel A are basically at the noise level in panel B. That makes panels C-D difficult to interpret, because it's not clear whether the units are robustly (probably meaningfully) modulated by the LFP, or just statistically significantly so (that is, p<.05 but tiny effect size due to large sample size). This is not mentioned or discussed critically.

Yes, we see various absolute magnitudes of PPC, in line with other PPC results in the literature. To assess whether the significant PPC values are meaningful, i.e. “whether the units are robustly modulated by the LFP”, we carried out the more conservative autocorrelogram fitting analysis. We now discuss the implications of differences across these methods, and offer caution in interpreting PPC phase locking as evidence of oscillatory coupling, when non-stationary signals dominate the LFP (see e.g. 300-308, 318-327, 400-405).

Perhaps get more data from monkey hippocampus and/or human hippocampus to substantiate these conclusions.

Reviewer #2 (Recommendations for the authors):

Abbaspoor and colleagues investigate in this work the interactions between different oscillatory patterns in the macaque hippocampus. This is an important topic that has been profusely investigated in other model species (mostly rodents), but the apparent differences with primates highlight the need for more comparative studies in the field. The present study has been conducted with the highest standard regarding both experimental and analytical methods. The results presented offer strong support for the main conclusion of the authors (the anticorrelation between theta and γ activities, contrary to what happens in rodents). My comments below are aimed at two main aspects of the study that can be further improved: a more comprehensive description of the data and a more extensive comparison with well-known rodent hippocampal physiology. Provided these corrections are incorporated, I believe the present study will add important evidence about macaque hippocampal oscillations and their comparison with those of rodents, thus being of potential interest for a large audience.

– Figure 1 presents a good description of the spectral properties of macaque hippocampal oscillations; however, spatial and temporal aspects of these activities can be better documented. In rodents, theta, γ and ripple oscillations have very stereotyped laminar profiles. Within the limitations of their recording methods, can the authors describe the laminar characteristics of macaque oscillations (e.g. is γ stronger in the apical CA1 dendrites, etc.)? In rodents, γ and ripples typically occur in a short burst while theta can be quasi-stationary for many seconds. What are the temporal dynamics of these oscillations in the macaque?

This is a fair point. We have added a figure showing the duration of the detected epochs of theta and γ band events as a function of behavioral state (Figure 1 supplement 4) and readers can hopefully get a sense of the prevalence and durations from the sample video (Video 1) scrolling through sleep and waking states, with detected bouts highlighted. We only used in-layer recordings here, positioned by microdrive, but agree that laminar profiles of these oscillations in different behavioral states is an important addition to clarify candidate homologies and mechanisms. We plan to address in future studies.

– As one of the main goals of this work is to compare theta-γ interactions in macaques versus rodents, it will be useful to reproduce the most common analysis used to illustrate this phenomenon in rodents: theta phase – γ amplitude cross-frequency coupling (Tort et al., 2007). This analysis will offer a complementary view to the power-power and bicoherence analysis already presented.

We follow this logic in principle, but in practice, we believe the bicoherence measure covers the effects of the Tort method while avoiding its specific danger of false positives. Bicoherence is generally a similar measure of PAC, but with less parameterization (Hyafil, Frontiers in Neuroscience, 2015). Further, bicoherence has a higher frequency resolution compared to other measures of PAC (Giehl et al. Neuroimage, 2021), allowing it to dissociate true from spurious cross-frequency couplings (Kramer et al. 2008).

– In lines 78-80 the authors state 20-35Hz oscillations ('slow γ') have been shown in rodents to be the result of theta harmonics. This is a gross mischaracterization of the rodent literature. Surprisingly, the authors have chosen to focus on the only one outlier paper that makes this claim against the overwhelming body of work suggesting the opposite (existence of slow γ independently of theta harmonics). Just to cite some examples: Csicvari et al., Neuron, 2003; Colgin et al., Nature, 2009; Schomburg et al., Neuron, 2014; Fernandez-Ruiz et al., Neuron, 2017; Cabral et al., Neuron, 2014; Lopes dos Santos et al., Neuron, 2018; Lastozcni and Klausberger, Neuron, 2016; Belluscio et al., JNeurosci, 2012. This statement needs to be amended to provide a fair account of the rodent literature or, if that is the position of the authors, much further elaborated.

We acknowledge that the statement was focused on a narrow critique of the slow γ oscillation that has been rebutted in the literature. We removed the statement from the paper and included many of the mentioned articles in the Discussion section.

– In lines 122-125 a spike-SWR analysis is described but SWR has not been described or characterized before this point. Even if it is not the main point of this work, a basic characterization of SWRs needs to be provided. In Figure 1 there are a couple of events that look like ripples, if that is the case, that needs to be mentioned. How did the authors differentiate SWR from fast γ?

We have added a description of ripple detection in the methods (490-506), and show examples from all 3 animals in Figure 2 supplement 1. We used similar methods to those described in Leonard et al., 2015. There, we used a clustering algorithm to show the difference in SWR detection versus fast γ and HFOs. Ideally, one needs to incorporate dendritic layer recordings for segregation; however, as described in the recent consensus statement on SWRs (Liu et al., 2022), that is not always used. As we begin to incorporate laminar recordings we hope to be able to elucidate the possibility of phase-amplitude coupling that is not due to SWRs, outside the pyramidal layer.

– Could the authors identify different γ sub-bands, as is the case in the rodents. If they couldn't, that is also a result worth mentioning, given the potential disparity with rodents.

We agree that identifying sub bands is an important addition to our understanding of the circuit function of CA1. While we note the division between low and high γ (e.g. Figures 1 and 2), we hesitate to make any strong claims about sub-bands, because those are typically measured from LFP outside the pyramidal layer, where we are recording. As indicated above, our future studies will be better suited to describe this.

– The methods pertaining to single unit isolation and classification are not described in the manuscript. How were individual units isolated and classified into cell types? Which cells were used for the spike-LFP analyses? Pyramidal cells and interneurons should be separated for these analyses.

We have included in the method section the criteria for spike sorting (476-480), which was also described in Leonard et al., 2015 and Hussin et al., 2020. For this study, we didn’t perform functional cell-type classification (as we did in Hussin et al., 2020), to be as inclusive as possible in our pool of cells that might show modulation. We couldn’t identify any strong example of theta rhythmicity in the single examples of autocorrelograms in our cell populations, therefore functional classification of the cell types would not change the results. Furthermore, we performed the same analysis on the rat dataset for a fair comparison and visualization when cells are not segregated by type.

– It needs to be acknowledged in the discussion that some of the reported differences between macaque and rodents could be due to the very different behavioral states in which oscillation has typically been measured in these cases (head-fixed visual search vs freely moving navigation).

Yes! Freely-moving navigation could easily make a difference to data obtained in head-fixed animals. To address this, we included the hippocampal LFP recordings from 2 freely-behaving monkeys during movement. As described in the text (98-110) in both subjects, the pattern of results remained qualitatively the same as in the stationary animal. For example, there was no correlation between theta and γ in these sessions (Figure 1 supplement 3). The power and prevalence of theta activity was much larger during sleep (Video 1, Figure 1 supplement 4), consistent with an even purer measure of the offline state in the overnight sleep recordings. Furthermore, we measured velocity in the video to demonstrate the relative weakness of theta during movement. See also reference to free movement in the discussion (272-274, 385-390).

– How were sampling periods selected for the analysis of Figure 3? Theta modulation of rodent ACGs will only be clearly present if periods of theta activity (or movement) were selected. How did the authors ensure that the rodent and macaque data are comparable? If no such preselection was done, it needs to be better explained in the results which exact data went into the analysis. A complementary analysis the authors could include in population CCGs, it may be possible that some rhythmic modulation appears in those that are missed in ACGs due to the sparse firing of pyramidal cells. Pyramidal cells and interneurons should not be mixed here.

The sampling period was the whole session. To ensure that our methods weren’t unfairly biasing us against seeing periodic spiking, we took the same unit populations (not segregated) from the homologous area of the rat, across the full session. Whereas ACGs as a population still showed strong theta periodicity overall in the rat, we found no evidence of theta periodicity in the monkey (Figure 3C), consistent with other reports mentioned in the text (e.g. Courelis et al., 2019). We discuss this in the context of other non-rodent studies of spiking oscillations (312 – 327).

Reviewer #3 (Recommendations for the authors):

The research on hippocampal oscillations has been heavily biased towards rodent models. Comparative studies between rodents and other species are lacking in this field, yet these are vital for the understanding of the mechanisms and functions of these rhythms. Therefore, this study by Abbaspoor, Hussin and Hoffman has to be appreciated for providing descriptive analyses of intracranial recordings from the macaque CA1 and relating them to the widely studied rodent CA1. More specifically, the authors focus on theta and related γ oscillations during rest and a visual search task. They found that CA1 LFPs present stronger 15-70Hz γ and reduced "theta" band power during the visual task compared to rest periods. Spike to phase coupling in the same region indicated that neurons coupled to three bands (3-10, 20-30Hz and 60-150Hz). Importantly, analysis of spike to field coherence indicated that in fact the "theta" power is dominated by irregular (non-oscillatory) activity nesting ripple-like events, i.e. most likely emerging from sharp wave events. The authors point out differences between these observations and what has been reported in the rodent CA1- notably, the lack of theta oscillations during "active" behavior.

Although I appreciate the objective and discussion of the work, the analyses need more refinement to truly support their claims. Further, there are a few apparent conceptual inconsistencies that should be addressed.

My first concern relates to the existence (or not) of true theta oscillations in the data.

The authors indicate in several parts of the manuscript (including the title) that they are analyzing "theta" activity. However, they have not presented convincing indications that they analyzed actual theta oscillations. In fact, spectral peaks do not necessarily demonstrate the presence of a true oscillation, as they write themselves. They find no evidence of genuine theta entrainment of spiking activity, arguing that the apparent theta modulation arises actually from sharp-wave-ripple (SWr) events. It is not clear to me if the authors are claiming the theta band in the LFPs arises solely from SWr in their data. If the authors are actually analyzing theta, they should present evidence this is the case.

This reviewer’s perspective helped us clarify our message -- see the additional Video, Figure 1 supplement 4 and Figure 2 supplement 1, and 308-311. We clearly see epochs containing theta oscillations, which are prevalent in rest and sleep, and not merely contamination by SWRs. Indeed, our results suggest that the acceptance of non-stationary, non-oscillatory activity in primates (e.g. during arousal) may have given a false impression about what oscillations are happening during which behavioral states. We will discuss further below.

Although cells are not paced at the theta rhythm, if genuine theta oscillations are present in the LFPs it should be possible to show evidence of that (e.g. as Jutras et al. 2013, Fig1G). In fact, theta bouts/bursts have been reported in primate studies (e.g. Leonard et al. 2015). If the authors are convinced to have theta oscillations in their recordings through the BOSC algorithm, this should be better developed in the manuscript. Further, the authors could analyze changes in theta burst incidence if they want to claim theta is reduced during search, and, most importantly, show that this cannot be explained by a decrease of SWr incidence (which they do show to contaminate the "theta band") relative to rest. One of the main claims of the manuscript is the decoupling between 20-35Hz γ and theta, but this can only be addressed if the authors restrict their analysis to true theta oscillations rather than SWr events. Even by doing so, the authors should discuss the possibility that they do not see γ-to-theta modulation because they record from the pyramidal cell layer whereas γ oscillations have been shown to originate in deeper layers in rodents (e.g. Belluscio et al. 2012; Schomburg et 2014; Lasztóczi and Klausberger 2014). This modulation might be easier to record from the pyramidal cell layer in rodents as γ oscillations have stronger signals in these species due to their anatomical features (Buzsáki et al. 2003).

We note that slow and fast γ oscillations are detectible in the layer (as indeed, we do see them!). Studies of laminar segregation find differences in relative magnitude and current sources, but the signal is measurable in the layer and indeed, phase amplitude coupling in rats and mice is evident in single LFP traces from the CA1 layer. See Author response image 1, from the cited article above, Belluscio et al., 2012, where the raw signal is taken from the pyramidal layer and the CSD plots show S/M/F γ in the filtered

traces at the Pyr layer (depth = 0). Consider the implication, if we are detecting slow and fast γ bouts in the layer, but that the PAC would happen with some invisible, distal γ not detected with our pyr-layer recordings. That would mean that there was another γ to be explained: whatever it was we were seeing in the layer. We will discuss more below, but appreciate the importance of the comment.

Author response image 1

Download asset Open asset

Alternatively, if the authors cannot show convincingly they are analyzing true theta oscillations, what they actually observe is a reduction (although very mild, as shown in Fig1D) of SWr incidence (reflected in a reduction of "theta" band power) during visual search compared to rest, which would be actually consistent with the rodent literature. What could be argued to be a contrast between species is that SWr rate is much higher in primates compared to rodents during "active behavior", but this has already been claimed by the same group (Leonard et al. 2015; Leonard et al. 2017). Moreover, primate visual search is taken here as analogous to rodent overt movement, which is arguably not a fair equivalence. In fact, SWr events have been shown to occur in rodents during decision making and planning when animals are relatively immobile but arguably attentive (Joo and Frank 2018). Here, the authors have a single freely moving animal (shown in the figure supplements), but lack of constant theta oscillations in similar conditions have been shown before (Ekstrom et al., 2005; Watrous et al., 2013; Talakoub et al., 2019). In fact, different correlates between theta and behavior in different species have been discussed for decades (e.g. Winson 1972). The authors should refine their discussion on the equivalence of "active behavior" between species and rewrite the paper according to the fact they are studying SWr events, and not genuine theta oscillations if that is the case. Also, regarding this point, they should clearly discuss what are the novel aspects of their current findings in relation to the literature.

We have made substantial changes. This includes additions to the text, which has a new Discussion section, with the collection of additional data, and by performing additional analyses. The main claims of the paper are: (1) The power and prevalence of theta-band activity is greater during inactive behavioral states (sleep and rest in dark) compared to active behavioral states such as search, and this includes true theta oscillations. We included a Video (Video 1) which shows a recording example of hippocampal LFP during sleep and rest and the decomposition of the signal, and bout detection, demonstrating oscillation in rest not attributable to only SWRs (see also the duration difference in Figure 1 supplement 4). We included BOSC results for head-restrained and freelymoving animals in Figure1 supplement 1 and Figure 1 supplement 2. Thus, we do observe theta oscillations in the hippocampus of NHP but the behavioral correlates are different compared to rodents, and we show how some common analysis methods are prone to contamination. (2) Theta and γ activity are not correlated in NHP hippocampal CA1. We used power-power comodulation and bicoherence analysis to demonstrate this point. Both of these analyses measure the interaction between cross-frequency activity with high resolution, and none points to strong coupling between these two oscillatory modes. If anything, we see negative correlations between these two spectral bands in accordance with the divergence in their prevalent behavioral states. Our final finding we have attempted to clarify: (3) Sharp wave ripples in NHP hippocampus show one concrete example of how common analysis methods may produce inflated measures of low frequency “oscillations”. SWRs can have a strong post-ripple deflection in the LFP, producing spectral energy at <10Hz, which overlaps with the theta oscillations. Further, because SWRs were observed and reported during active states in NHP previously, they can contribute to the power of theta activity, coupling between theta and higher-frequencies, and spike-LFP coupling in the theta range. This does not account for the entirety of our findings, but rather, one contributor. We suggest that future research should keep this type of contribution in mind before making evaluations of theta activity in the primate hippocampus.

My second point relates to the relevance of the decoupling between 20-35Hz γ and theta.

The authors highlight the observation that 20-35Hz γ oscillations are not coupled to theta. This has been shown before in primates during visual search (Leonard et al. 2015); and also has been shown in rodents. For example, 23-30 Hz (beta2) oscillations occur in the absence of theta (Berke et al. 2008; Figure 3) in freely moving mice, and γ can be recorded in slow-wave sleep in rats (Isomura et al. 2006 Fig7C). Thus, it is not established that γ oscillations in rodents can only occur coupled to theta. Therefore, the presence of this γ rhythm in the absence of theta reported in the manuscript does not strike me as a relevant contrast between primates and rodents. A contrast would be the absence of continuous theta in the primates, which was extensively reported before. I might have missed something, so I would suggest that the authors make clearer what the novel observations presented here, and how these represent an important contrast between species.

We recognize that there are special cases of γ-theta decoupling in the literature; however, the distinction between existence and prevalence is paramount. It is difficult to overstate the implications of describing an unusual or niche appearance of an oscillation, versus a dominant, reliable feature of a behavioral state that our mechanistic models must account for. Currently, the dual oscillatory regime of theta-γ versus SWR is still being applied across species to describe online/offline states, respectively. If this paper’s results hold true, this conceptualization is wrong in primates, at least in the common cases.

https://doi.org/10.7554/eLife.86548.sa2

Article and author information

Author details

Saman Abbaspoor

Department of Psychology, Vanderbilt Vision Research Center, Vanderbilt Brain Institute, Vanderbilt University, Nashville, United States

Contribution
Data curation, Formal analysis, Investigation, Visualization, Methodology, Writing - original draft, Writing - review and editing

Competing interests
No competing interests declared

"This ORCID iD identifies the author of this article:" 0000-0002-3550-0617
Ahmed T Hussin

Department of Biology, Center for Vision Research, York University, Toronto, Canada

Contribution
Data curation, Investigation, Methodology

Competing interests
No competing interests declared
Kari L Hoffman
1. Department of Psychology, Vanderbilt Vision Research Center, Vanderbilt Brain Institute, Vanderbilt University, Nashville, United States
2. Department of Biology, Center for Vision Research, York University, Toronto, Canada
Contribution
Conceptualization, Data curation, Supervision, Funding acquisition, Investigation, Visualization, Writing - original draft

For correspondence
kari.hoffman@vanderbilt.edu

Competing interests
No competing interests declared

"This ORCID iD identifies the author of this article:" 0000-0003-0560-8157

Funding

National Institutes of Neurological Disorders and Stroke (R01NS127128)

Saman Abbaspoor
Kari L Hoffman

Whitehall Foundation

Kari L Hoffman

Alzheimer's Society of Canada Doctoral Award

Ahmed T Hussin

National Science and Engineering Research Council (Discovery Grant)

Ahmed T Hussin
Kari L Hoffman

NSERC CREATE Vision Science and Applications

Ahmed T Hussin
Kari L Hoffman

Brain Canada Multi-Investigator Research Initiative

Ahmed T Hussin
Kari L Hoffman

The Krembil Foundation

Ahmed T Hussin
Kari L Hoffman

National Eye Institute (P30 EY008126)

Kari L Hoffman
Saman Abbaspoor

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

The authors wish to thank A Maurer for thoughtful comments on the manuscript and W Zinke and I Hayes for technical support for the wireless recordings.

Ethics

This study was performed in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. All of the procedures were in accordance with a protocol approved by the local governing authorities. In the US this was the institutional animal care and use committee (IACUC # M1700152), and in Canada, this was the Canadian Council on Animal Care, local Animal Care Committee at York University (#2014-9).

Senior Editor

Laura L Colgin, University of Texas at Austin, United States

Reviewing Editor

Martin Vinck, Ernst Strüngmann Institute (ESI) for Neuroscience in Cooperation with Max Planck Society, Germany

Reviewer

Antonio Fernandez-Ruiz, Cornell University, United States

Publication history

Preprint posted: January 1, 2022 (view preprint)
Received: January 31, 2023
Accepted: March 13, 2023
Accepted Manuscript published: May 4, 2023 (version 1)
Version of Record published: May 24, 2023 (version 2)

Copyright

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.