Contributors:Howan Leung, Cannon X.L. Zhu, Danny T.M. Chan, Wai S. Poon, Lin Shi, Vincent C.T. Mok, Lawrence K.S. Wong
High-frequencyoscillations... An example of the implantation schedule (patient #1) demonstrating areas with conventional frequency ictal patterns, ictal high-frequencyoscillations, hyperexcitability, and radiological lesions.
... An example of the implantation schedule (patient #7) demonstrating areas with conventional frequency ictal patterns, ictal high-frequencyoscillations, hyperexcitability, and radiological lesions.
... High-frequencyoscillations (HFOs, 80–500Hz) from intracranial electroencephalography (EEG) may represent a biomarker of epileptogenicity for epilepsy. We explored the relationship between ictal HFOs and hyperexcitability with a view to improving surgical outcome.... Summary table for statistical analysis. HFO=high frequencyoscillations, CFIP=conventional frequency ictal patterns.
Contributors:Iori Ito, Maxim Bazhenov, Rose Chik-ying Ong, Baranidharan Raman, Mark Stopfer
Odor Concentration Determines Oscillation Coherence, Not Frequency
(A) EAG traces revealed total ORN output increased with odor concentration. Example from one antenna; horizontal bar: 4 s.
(B) Summary. EAG amplitude (first 1 s, see bracket in A) evoked by a range of odor concentrations. Mean ± SE; n = 8; two-way ANOVA: fodor_concentration = 16.84, p oscillations. Initial portions of the odor response are shown. Scale bar: 50 ms.
(D) The frequency of fast oscillation changed not at all or only slightly across a broad range of odor concentrations. All results are shown (dots); bar graph shows means, n = 9. Leftmost bars: basal oscillatory power in absence of odorant. Hexanol: two-way ANOVA: fhexanol_concentration = 6.16, p 0.25, ns.
... Odors Evoked LFP Oscillations in the Moth MB and AL
(A) Recording site for LFP: center of the calyx in the MB. MB, mushroom body; mnsc, medial neurosecretory cells; OL, optic lobe; AL, antennal lobe.
(B) LFP oscillations (black traces) with simultaneously recorded electroantennogram (EAG, green traces) evoked by different pulse durations of 1% benzyl alcohol, a plant volatile. Black horizontal bars: odor pulses. Color bars: time windows (500 ms) used to calculate the power spectra in (D).
(C) Brief odor pulses evoked fast oscillations; lengthy pulses evoked first fast, then slow oscillations. Normalized, average spectrograms from 18 trials obtained from six animals with three trials each (see Experimental Procedures). Black horizontal bars above each spectrogram: odor pulses.
(D) Power spectra of oscillatory LFP responses averaged from 22 moths and eight odors, total of 820 trials. Color brackets: 14 Hz-wide bands used to calculate the total oscillatory powers of fast (red, 30–44 Hz) and slow (blue, 10–24 Hz) oscillations in (E).
(E) Total oscillatory power of fast and slow LFP shifted significantly over lengthy odor pulses. Twenty trials tested for each odor were averaged before pooling, mean ± SE, n = 41; two-way ANOVA: fwindow(2) = 26.62, p oscillations); fwindow(2) = 9.09, p oscillations). Asterisks: significant differences (Tukey-Kramer multiple comparisons).
(F) LFP oscillations in the AL and MB were highly coherent. (Left) Example of odor-evoked LFP oscillations recorded simultaneously in the AL and MB; odorant: 1% cyclohexanone (4 s). Areas a and b are expanded in insets. Horizontal red (0.25–1 s) and blue (1–4 s) bars: times used for coherence analysis at right. (Right) Magnitude squared coherence between the AL and MB. Thin black line: coherence of the response shown. Thick black and dotted lines: average coherence and its one standard deviation range (five AL-MB combinations in four preparations, 20 trials each of two odorants), respectively.
... PN and LN Responses Were Strongly Phase Locked to the LFP
(A) Example simultaneous intracellular recordings from PN and LN, with LFP recorded in the MB. First 2 s after the odor onset shown; brackets: portions expanded beneath. Odorant: 1% benzyl alcohol.
(B) Subthreshold oscillations: five-trial average sliding window cross-correlograms show reliable LFP and subthreshold membrane potential oscillations for the PN (top) and LN (bottom) in (A). Spikes were clipped. Vertical bars: odor pulses.
(C) Spike-LFP phase relationships: Polar histograms show phase position, relative to LFP, of spikes recorded in PNs (n = 14) and LNs (n = 30) for fast and slow oscillations. Concentric circles: firing probability. Black arrows: mean direction.
(D) All recorded neurons were filled with dye and later morphologically identified. Example of PN and LN morphology. An Alexa Fluor-633 (red) filled PN and an Alexa Fluor-568 (yellow) filled LN are shown. Scale bar: 50 μm. AN: antennal nerve.
... In many species, sensory stimuli elicit the oscillatory synchronization of groups of neurons. What determines the properties of these oscillations? In the olfactory system of the moth, we found that odors elicited oscillatory synchronization through a neural mechanism like that described in locust and Drosophila. During responses to long odor pulses, oscillations suddenly slowed as net olfactory receptor neuron (ORN) output decreased; thus, stimulus intensity appeared to determine oscillationfrequency. However, changing the concentration of the odor had little effect upon oscillatory frequency. Our recordings in vivo and computational models based on these results suggested that the main effect of increasing odor concentration was to recruit additional, less well-tuned ORNs whose firing rates were tightly constrained by adaptation and saturation. Thus, in the periphery, concentration is encoded mainly by the size of the responsive ORN population, and oscillationfrequency is set by the adaptation and saturation of this response.... Spiking in KCs Is Sparse, Odor Specific, and Tightly Phase Locked to the LFP
(A) KCs showed odor-elicited subthreshold membrane potential fluctuations that were tightly correlated with LFP oscillations. Example: top, gray: LFP; bottom, black: simultaneous intracellular record of a KC. Bottom: details of fast and slow periods during oscillatory response. Odor: 4 s, 1% benzyl alcohol. Gray broken line: resting potential.
(B) Cross-correlations between LFP oscillations and KC subthreshold activity. Cross-correlation was calculated for times bracketed in (A). Black lines: correlation for the trial shown in (A); gray lines: 21 other trials from this cell. All eight KCs showing subthreshold oscillations revealed similarly shaped correlation functions, three with coefficients >0.3.
(C) Polar histograms show strong phase locking between spikes in KCs and the LFP oscillations. Histograms show spikes recorded from 20 KCs during fast and slow oscillations. Arrows: mean phase position.
(D) Example of KC morphology; posterior view of MB; KC filled with Alexa Fluor-633. Scale bar: 50 μm. Arrow: soma; CaM: medial calyx; CaL: lateral calyx.
... Odor-Evoked Oscillations in Model of Moth AL
(A) Full-scale, map-based model included randomly connected populations of 820 PNs and 360 LNs. Odor pulse input was simulated by external currents delivered to a subset of neurons.
(B) Amplitude of the input was set to resemble the EAG (bottom). LFP (top) and neuronal (middle) responses resembled those recorded in vivo. The input to the model was tuned to match results of our physiological recordings and corresponded to points “1” and “2” in the parameter space shown in (E).
(C) Raster plots show spikes in all PNs (top) and all LNs (bottom) evoked by one odor pulse (applied from 500 to 2500 ms).
(D) Interspike interval (ISI) distributions during fast and slow phases of LFP oscillations. Many PNs fired two spikes in a single oscillatory cycle (ISI frequency was typically limited to the LFP frequency.
(E) Frequency of LFP, PN, and LN oscillations as a function of input from ORNs to PNs and LNs. Sweeping the points between “1” and “2” in parameter space mimicked the changes in the ISI distribution (compare D and Figure S7) and the abrupt change in oscillatory frequency (compare B and the Figure 1C) we observed in vivo.
Contributors:Li Wang, Qingmei Kong, Ke Li, Yunai Su, Yawei Zeng, Qinge Zhang, Wenji Dai, Mingrui Xia, Gang Wang, Zhen Jin, Xin Yu, Tianmei Si
Brain regions showing significant group and frequency (slow-4 and slow-5) interaction effects on ALFF.
... The group and frequency (slow-4 and slow-5) interaction effects on ALFF. The regions showing significant group and frequency interaction effects on ALFF (hot colors): the left ventromedial prefrontal cortex (a), the left inferior frontal gyrus/precentral gyrus (b), and the bilateral posterior cingulate cortex/precuneus (c). The bar maps show the mean ALFF values for these regions.
... Low-frequencyoscillation... Frequency dependence... We conducted this fMRI study to examine whether the alterations in amplitudes of low-frequencyoscillation (LFO) of major depressive disorder (MDD) patients were frequency dependent.... Main effects of group and frequency on ALFF. (a) The group main effects on ALFF. Hot colors represent increased ALFF in the MDD group compared with HC, while the blue colors represent the opposite. (b) Frequency main effects on ALFF. Hot colors represent increased ALFF in the slow-5 as compared to slow-4 band, while the blue colors represent the opposite. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
Contributors:Eric Maris, Marieke van Vugt, Michael Kahana
In the analysis of the activation period wPLFs, 36 reliable PAC patterns were identified: two subjects had four reliable PAC patterns, eight had two, 12 had one, and four had none. In the analysis of the baseline period wPLFs, 17 reliable PAC patterns were identified: four subjects had two reliable PAC patterns, nine had one, and 13 had none. Reliability was defined in terms of the split-half correlation between two independent estimates of the spatial and the frequency spectra produced by the tensor decomposition. All identified PAC patterns had reliabilities much larger than what can be expected under the hypothesis of a random PAC (see Materials and methods). A representative selection of reliable PAC patterns is shown in Supplemental Fig. 3.... Our conclusion about the number of sources involved in PAC depends on our definition of a source. We defined a source in terms of two patterns: (1) a spatial map that specifies how strongly source activity affects the measurements at the sensor level, and (2) its frequency spectrum. We showed that, with this source definition, the array of PAC-measures (wPLFs) can be written as a tensor product of two spatial maps (one complex- and one real-valued) and two frequency spectra (both real-valued), which is exactly the structure that is extracted by our tensor decomposition (see Supplemental Methods). However, we cannot exclude PAC-generating source configurations that cannot be characterized in this way. In fact, we have argued that PAC-patterns may also be generated by a source configuration that would be considered a single source when viewed from the perspective of the mechanism that generates the physiological signal. This confronts us with the problem that sources can be defined both in terms of their formal characteristics (i.e., in terms of a spatial map and a frequency spectrum) and in terms of the neuronal network that generates the physiological signal. The difference between the two definitions is most clear if the physiological mechanism consist of multiple components, such as networks of inhibitory neurons that are connected to one or multiple classes of principal neurons, each with its own network topology. This whole multi-component network may be considered as a single source, but also as multiple sources, each one corresponding to one component. Importantly, if these components differ with respect to their spatial maps and frequency spectra, then they can be extracted by means of tensor decomposition. This shows there may be a need for a linking of the set of extracted source configurations on the basis of the neuronal interactions that may have produced them.... Package of figures showing the spatial maps and the frequency spectra of a representative set of reliable cross-frequency patterns extracted from the activation period wPLFs. The figures belonging to one cross-frequency patterns are each in one folder, of which the name refers to the patient. The files that have freqspectra_and_compass as a part of their name contain a figure of which the left panel show the frequency spectra of a coupling oscillation and the associated phase-coupled bursts, and the right panel shows the preferred phases of the PAC in a compass plots. The files that have _IA_ (instantaneous amplitude) as a part of their name contain figures of the magnitudes of the spatial maps for the phase-coupled bursts. Different figures show different subsets of the electrodes, each from on optimal viewpoint. Electrode subsets on the medial side of the brain are shown from two viewpoints, lateral and occipital. The files that have _AWIP_ (amplitude-weighted instantaneous phase) as a part of their name contain figures of the spatial maps of the coupling oscillation. Again, different figures show different subsets of the electrodes, each from on optimal viewpoint.
... Spatial maps of the phase-coupled bursts are smaller than the spatial maps of the associated coupling oscillation. Both panels show scatter plots of the extent of the spatial maps of the phase-coupled bursts (horizontal axis) against the extent of the corresponding spatial maps of the coupling oscillation (vertical axis). Panels a and b show the scatter plots for, respectively, the baseline and the activation period. Panels a and b show the scatter plots for, respectively, the baseline and the activation period. In both periods, for most of the PAC patterns, the extent of the spatial map of the phase-coupled bursts is substantially smaller than the extent of the spatial map of the associated coupling oscillation.
... Illustration of a tensor decomposition of the four-dimensional array of weighted phase-locking factors (wPLFs). The two spatial maps and the two frequency spectra (see text) are each denoted by a different color (red, yellow, green, and blue) and a different index (i, i', f, f'). The same colors are used both for the boundaries of the panels and the symbols in the formula for the wPLF. In panels a and b, with red and index i, we show the complex-valued spatial map of the high-frequency bursts that are coupled to a common low-frequencyoscillation (the coupling oscillation). In panel a, we show the magnitude (absolute value) of this complex-valued spatial map (one colored circle per channel), which expresses the strength of the coupling. In panel b we show the phases of the coupling oscillation to which the high-frequency bursts are locked (one arrow per channel). In panel c, with yellow and index i', we show the spatial map of the coupling oscillation (one colored circle per channel). The more a coefficient bi’ deviates from zero, the more this channel is affected by the coupling oscillation. In panel d, with green and index f, we show the frequency spectrum of the phase-coupled bursts (shown on a logarithmic scale). In panel e, with blue and index f', we show the frequency spectrum of the coupling oscillation. The spectra shown in panels a, b, c, d and e are all in arbitrary units (a.u.). This is because the spectra are produced by a tensor decomposition which involves an arbitrary multiplicative scaling (see Supplemental Material). In panels f, h, and j, we show the magnitudes of the complex wPLFs for three selected channel pairs (see text). By means of arrows, we connect the channels in panels a and c for which these wPLFs were calculated, with the corresponding x-axes (showing the frequency of the coupling oscillation), respectively, y-axes (showing the frequency of the phase-coupled bursts), in panels f, h, and j. In panels g and i, we show the phases of the complex wPLFs that correspond to panels f and h, respectively.
... Spatially distributed coherent oscillations provide temporal windows of excitability that allow for interactions between distinct neuronal groups. It has been hypothesized that this mechanism for neuronal communication is realized by bursts of high-frequencyoscillations that are phase-coupled to a low-frequency spatially distributed coupling oscillation. This mechanism requires multiple physiologically different interacting sources, one generating the low-frequency coupling oscillation and the others generating phase-coupled high-frequencyoscillations. Using human intracranial EEG (iEEG) data, we provide evidence for multiple oscillatory patterns, as characterized on the basis of their spatial maps (topographies) and their frequency spectra. In fact, we show that the spatial maps for the coupling oscillations are much more widespread than the ones for the associated phase-coupled bursts. Second, in the majority of the patterns of phase-amplitude coupling (PAC), phase-coupled bursts of high-frequency activity are synchronized across brain areas. Third and last, working memory operations affect the PAC strength in a heterogeneous way: in some PAC patterns, working memory operations increase their strength, whereas in others they decrease it.... Simulated field potentials with bursts of gamma oscillations (60Hz) that are phase-coupled to the rising phase of a theta oscillation (5Hz).
... Spectral signature of the PAC patterns. Panels a and b show the results for, respectively, the baseline and the activation period. Both panels show a scatter plot of the central frequencies of the phase-coupled bursts (horizontal axis) against the central frequencies of the coupling oscillations (vertical axis). The central frequency of the phase-coupled bursts is always smaller than the one for the associated coupling oscillation. For all PAC patterns above the thick black line, the central frequency for the phase-coupled bursts is larger than the central frequency for the associated coupling oscillation. There is not a dominant frequency, neither for the phase-coupled bursts, nor for the associated coupling frequencies.
Contributors:Bassam V. Atallah, Massimo Scanziani
Excitation Balanced by Proportional Inhibition during Gamma Oscillations In Vivo
(A) Whole-cell recording of EPSCs in CA3 cell (red) and simultaneously recorded LFP (black, positivity is up) during gamma oscillations in anesthetized rat. IPSCs (cyan) and inverted LFP recorded from the same cell. Note correlated fluctuations in the amplitude of LFP and synaptic currents.
(B) Coherence between LFP and IPSCs (cyan) or EPSC (red); jack-knifed 95% confidence interval (thin lines); arrows mark peak coherence frequencies. Summary of peak coherence frequency (bottom) and peak coherence (right). Average shown as a vertical or horizontal bar (n = 7 cells).
(C) Oscillation triggered average (OTA) of EPSC (red), IPSC (cyan) and LFP. LFP was recorded simultaneously with EPSCs, IPSC (black and dotted traces, respectively). EPSC is inverted for illustration purposes. Overlaid POTH (green, data from Figure 1D, aligned to the LFP also in green) illustrates spike timing during oscillation cycle. Note that maximal spiking precedes peak of inhibition. (Bottom) Summary of EPSC-IPSC lag during an oscillation cycle; vertical bar is average.
(D) OTA of EPSCs (red), IPSCs (cyan) computed for four different bins of LFP oscillation amplitude (black; dotted and solid traces were recorded simultaneously with IPSC and EPSCs, respectively, same cell as A–C).
(E) Summary of correlation between average inhibitory (gI) and excitatory (gE) conductance in vivo; individual cells are each represented by a different color linear regression. Note, although excitation and inhibition are proportional, the inhibitory conductance is approximately five times larger (dotted line is at unity).
(F) OTA of IPSC (middle) computed for four different bins of LFP oscillation interevent interval (top). Vertical arrows illustrate IPSC amplitude and horizontal arrows the correlated changes in the time to the next oscillation event (IEI). (Bottom) IPSC amplitude during an oscillation event correlated with the time to the next oscillation in the LFP (IEI); blue dots correspond to the four OTA shown above.
... Gamma Oscillation Amplitude Predicts Latency to Next Oscillation Cycle
(A) (Top) Broadband (gray) and gamma-band filtered local field potential (LFP, 5–100 Hz) recorded in the stratum radiatum of area CA3 of an anesthetized rat. Raster plot marks the peak of each oscillation cycle. (Bottom, left) Autocorrelation of LFP and power spectral density of gamma-band LFP. (Bottom, right) Histograms of oscillation amplitude and interevent interval (IEI). (Inset) LFP recording illustrating the measurement of peak-to-peak amplitude and IEI (expansion of the recording marked by a horizontal bracket in the top panel). Positivity is up.
(B) (Top) IEI correlated against amplitude of the previous cycle illustrated in histogram. Note the correlation between oscillation amplitude and IEI. (Bottom) Summary of correlations, n = 6 rats. Vertical bar is average.
(C) Broadband extracellular recording (top), gamma-band LFP (middle, 5–100 Hz band-pass), multiunit spiking (green, 0.2–2 kHz) from stratum pyramidale of area CA3. Negativity is up.
(D) Oscillation triggered average of LFP, peri-oscillation spike-time histogram (POTH), and local linear fit to POTH (green).
(E) (Left) Average LFP and POTH fit calculated separately for large (mean amplitude = 313 μV) and small (99 μV, dotted) oscillation cycles. Arrows illustrate the increased latency between spiking events after large-amplitude cycles. (Inset) Small POTH scaled to the peak of the large POTH. (Right) Summary of full-width at half-maximum (FWHM) of POTH for large (solid) and small (open) oscillation cycles (n = 6 rats). Averages are illustrated with horizontal bars. Note that spiking occurs in a narrow time window during each oscillation cycle independent of oscillation amplitude.
... Larger, Longer Hyperpolarization of Pyramidal Cells following Large-Amplitude Oscillation Cycles
(A) LFP and simultaneously recorded membrane potential (Vm; whole-cell current-clamp configuration: IC) during in vitro gamma oscillations (dotted line is mean Vm). Positivity is up.
(B) Oscillation cycles were binned according to LFP amplitude and the oscillation triggered average (OTA) of Vm computed for each bin (different colors): average time course of LFP in four bins of increasing amplitude (top) and corresponding (color coded) Vm averages (middle). Note that Vm undergoes larger and longer hyperpolarization during large-amplitude oscillation cycles. (Bottom, left) Cycle-by-cycle correlation between the peak hyperpolarization and LFP amplitude. Bins in upper panels are illustrated with solid dots of respective colors. (Bottom, right) Summary of correlation (n = 11 cells).
(C) (Top) Oscillation cycles were binned according to LFP interevent interval and the OTA of membrane potential computed for each bin (different cell than A and B). Arrows illustrate “recovery time,” i.e., time from onset of oscillation cycle till membrane potential recovers to mean Vm (horizontal dotted line). (Bottom) LFP interevent interval plotted as a function of Vm recovery time. Colored dots and black line correspond to the above cell, other cells shown in gray. Note, mean slope, m = 1.16; SD = 0.3, suggesting that changes in the time for recovery from hyperpolarization in individual cells can account for the entire range of oscillation intervals observed in the LFP.
... Excitation Instantaneously Balanced by Proportional Inhibition during Each Gamma Oscillation Cycle
(A) (Top) Broadband (gray) and gamma-band filtered (black) LFP recorded in the stratum radiatum of area CA3 in acute hippocampal slice. Raster plot marks the peak of each oscillation cycle. (Bottom, left) Autocorrelation of LFP and power spectral density of gamma-band LFP. (Bottom, right) Histograms of oscillation amplitude and interevent interval (IEI). (Inset) LFP recording illustrating the measurement of peak-to-peak amplitude and IEI (expansion of the recording marked by a horizontal bracket in the top panel). Positivity is up.
(B) (Top) IEI correlated against amplitude of the previous cycle. (Bottom) Summary of correlations, n = 6 slices. Vertical bar is the average. Note the correlation between oscillation amplitude and IEI.
(C) Dual patch-clamp recording from two neighboring CA3 pyramidal cells. Oscillations are monitored with an LFP electrode (black, positivity is up). EPSCs (red) and IPSCs (cyan) simultaneously recorded by holding two cells at the reversal potential for inhibition (−3 mV) and excitation (−87 mV), respectively. Note the correlated fluctuations in the amplitude of excitation and inhibition.
(D) (Left) Average time course of EPSC and IPSC (same cell as C) during an oscillation cycle recorded in the LFP, i.e., oscillation triggered average. EPSC is inverted for illustration purposes. LFPs recorded simultaneously with EPSCs and IPSCs are shown as black and gray traces, respectively. (Right) Summary of EPSC-IPSC lag during an oscillation cycle. Horizontal bar is the average.
(E) (Top) Cycle-by-cycle correlation between excitatory and inhibitory conductances recorded in the pair shown in (C). Summary of correlation between excitation and inhibition (bottom) and ratio of mean excitatory and inhibitory conductances (right) (n = 8 pairs). Vertical and horizontal bars illustrate respective averages.
... Correlated Amplitude and Frequency in Simple Model of CA3 Circuit
(A) Average excitatory (gE, red) and inhibitory (gI, cyan) synaptic conductance received by model pyramidal cells. LFP (black) is approximated as the sum of the two conductances.
(B) (Top) Autocorrelation and power spectrum of simulated LFP. (Bottom) Interevent interval correlated against amplitude of the previous cycle.
(C) The membrane potential (Vm) of an individual pyramidal cell in modeled circuit (spike truncated), gE (red) and gI (cyan); dotted line illustrates the average Vm.
(D) Oscillation cycles were binned according to gI amplitude and the oscillation triggered amplitude of Vm computed for each bin (different colors): average time course of gI in four bins of increasing amplitude (middle) and corresponding (color coded) Vm averages (top). The arrows illustrate that it takes longer for Vm to recover to the average potential (horizontal dotted line) after large-amplitude cycles. (Bottom) Cycle-by-cycle correlation between Vm hyperpolarization and the gI. Bins in upper panels are illustrated with solid dots of respective colors.
... Neurons recruited for local computations exhibit rhythmic activity at gamma frequencies. The amplitude and frequency of these oscillations are continuously modulated depending on stimulus and behavioral state. This modulation is believed to crucially control information flow across cortical areas. Here we report that in the rat hippocampus gamma oscillation amplitude and frequency vary rapidly, from one cycle to the next. Strikingly, the amplitude of one oscillation predicts the interval to the next. Using in vivo and in vitro whole-cell recordings, we identify the underlying mechanism. We show that cycle-by-cycle fluctuations in amplitude reflect changes in synaptic excitation spanning over an order of magnitude. Despite these rapid variations, synaptic excitation is immediately and proportionally counterbalanced by inhibition. These rapid adjustments in inhibition instantaneously modulate oscillationfrequency. So, by rapidly balancing excitation with inhibition, the hippocampal network is able to swiftly modulate gamma oscillations over a wide band of frequencies.
Contributors:George M. Ibrahim, Simeon M. Wong, Ryan A. Anderson, Gabrielle Singh-Cadieux, Tomoyuki Akiyama, Ayako Ochi, Hiroshi Otsubo, Tohru Okanishi, Taufik A. Valiante, Elizabeth Donner, James T. Rutka, O. Carter Snead III, Sam M. Doesburg
Quantification of amplitude-to-phase cycle relationship. (A) Mean amplitudes of high frequencyoscillations, sorted by concurrent low-frequency phase into 60 bins of 0.105rad, for an example time window during a seizure; (B) an ideal cosine; and (C) a sine is modeled. (D) Phasor demonstrating the amplitude–phase cycle relationship. (E) The argument (angle) of the example phasor is a single contribution to be incremented onto a cumulative polar histogram spanning multiple subjects for one of ten given time periods in the seizure.
... Cross-frequency coupling... Modulation of high frequency amplitude by low-frequency phase. In the seizure-onset zone, significant modulation of high-frequency amplitude (40–300Hz) is observed, mainly by the phase of theta and alpha oscillations during the ictal period. In the interictal period, no specific CFC with slower oscillations is observed. There is also less cross-frequency coupling in the early propagation zone during seizures and no significant coupling is noted in the non-epileptogenic cortex. The z-axis demonstrates the modulation of amplitudes of different narrow-band frequencies (x-axis) by the phases of other narrow-band frequencies (y-axis). Lower and upper planes represent uncorrected and corrected statistical thresholds at p<0.05, respectively.
... High frequencyoscillations... Simulated data demonstrating expected polar histogram distribution. The high frequency amplitude is represented by the blue solid line, whereas the low-frequency phase is represented by the black dashed line. When the high frequency amplitude is maximal at the peak and trough of the low frequency phase, the polar histograms will indicate pi and 0, respectively.
... Individual seizure short-time Fourier transform spectrograms. The seizure onset zone contained predominantly low-frequency power, which was fairly heterogenous across the population. High frequency activity was also evident in all seizures as bursts of high power oscillatory activity.
... Pathological high frequencyoscillations (pHFOs) have been proposed to be robust markers of epileptic cortex. Oscillatory activity below this frequency range has been shown to be modulated by phase of lower frequencyoscillations. Here, we tested the hypothesis that dynamic cross-frequency interactions involving pHFOs are concentrated within the epileptogenic cortex. Intracranial electroencephalographic recordings from 17 children with medically-intractable epilepsy secondary to focal cortical dysplasia were obtained. A time-resolved analysis was performed to determine topographic concentrations and dynamic changes in cross-frequency amplitude-to-phase coupling (CFC). CFC between pHFOs and the phase of theta and alpha rhythms was found to be significantly elevated in the seizure-onset zone compared to non-epileptic regions (pfrequency oscillations at which pHFO amplitudes were maximal was inconsistent at seizure initiation, yet consistently at the trough of the low frequency rhythm at seizure termination. Amplitudes of pHFOs were most significantly modulated by the phase of alpha-band oscillations (p<0.01). These results suggest that increased CFC between pHFO amplitude and alpha phase may constitute a marker of epileptogenic brain areas and may be relevant for understanding seizure dynamics.... Topographic mapping of cross-frequency interactions in a representative subject. (A) Intraoperative image of grid demonstrating seizure onset and early propagation zones. (B) Fast-ripple amplitudes sorted by alpha phase for all grid electrodes. Cosine wave represents alpha phase from −π to π. Increased pHFO-to-low-frequency coupling occurs in the resected cortex (black borders). Values normalized by 95% confidence interval such values above 0 are significant at pfrequency modulation index for all electrodes, where the X-axis represents low frequency phase (1 to 40Hz; left to right), and the Y-axis denotes envelope amplitude (1 to 300Hz; top to bottom). Values exceeding Bonferroni correction threshold shown. Significant modulation of pHFO amplitudes by low-frequency phase is observed in the epileptogenic cortex.
... To characterize the ictal dynamics of relations between pHFO amplitude and low frequency phase, we measured the preferred slow oscillatory phase at which high amplitude pHFOs occurred at various times throughout the seizure. When the preferred phases from all bins from all subjects were plotted cumulatively on polar histograms, it was observed that pathological fast-ripple amplitudes preferentially occurred during the trough of alpha oscillations, whereas pathological ripple amplitudes preferentially occurred between 0rad and π/2rad of alpha and theta oscillatory cycles (Fig. 5; poscillations (p=0.14 and p=0.68, respectively). At seizure termination (i.e. the last bin), pHFO amplitudes occurred at the trough of the alpha oscillatory cycle (pathological ripple amplitude: p<0.01; pathological fast-ripple amplitude: p=0.03). Ripple amplitude maxima were also found at the peak of delta phase irrespective of the seizure progression (Supplementary Fig. S9). To ensure that differences in bin length did explain the measures of CFC, a reanalysis of the data with fixed length segments comprised of the first and last 2000ms of seizures, revealed the same pattern.
Contributors:Jiwoo Hong, Young Kwon Kim, Kwan Hyoung Kang, Joonwon Kim, Sang Joon Lee
Variations of resonance frequency according to drop viscosity. Dash denotes absence of resonance frequency.
... Temporal evolution of the base radius (colored symbol) of oscillating drops with different viscosities for different frequencies of (a) 32Hz and (b) 98Hz at η=0.25; (c) 28 and (d) 94Hz at η=0.39. The dashed lines denote subharmonic oscillation.
... Maximum amplitudes of oscillating drops with different viscosities along with AC frequency at different η; (a) and (b): η=0.25, (c) and (d) η=0.39.
... The effects of drop viscosity on oscillation dynamics of a sessile drop, such as resonance frequency and oscillation amplitude, in response to different AC voltages (80 and 100Vrms, corresponding to the electrowetting number η of 0.25 and 0.39, respectively) and frequencies (frequency range of 20–110Hz in which the first and second resonance frequencies exist) were investigated, based on both experiments and theoretical modeling. The results show that drop viscosity rarely affects resonance frequency, but strongly affects the oscillation amplitude and peak width of the resonance frequency. In addition, drop oscillation in the resonance mode is no longer observed, when the drop viscosity is over the critical value, which increases with applied AC voltage. A theoretical model predicts the oscillation dynamics of sessile drops within the error level of ±5%, except for the drop with high viscosity (60mPas). Moreover, the friction coefficient obtained by fitting the theoretical models is nearly proportional to the drop viscosity. Finally, an empirical relationship, in which the maximum amplitude is inversely proportional to the drop viscosity, is obtained.... Oscillation patterns of 5μL drops with different viscosities at η=0.25 for different frequencies of (a) 32Hz and (b) 98Hz. The patterns are obtained by superposing more than 20 images. The solid lines in the right column depict instantaneous drop deformation for theoretical P2 and P4 mode shape oscillation; short dashed lines represent initial quiescent shapes; arrows denote nodal points.
... Resonance frequency... Frequency response at harmonic and subharmonic frequencies.
... Drop oscillation
Contributors:Kenji Kirihara, Anthony J. Rissling, Neal R. Swerdlow, David L. Braff, Gregory A. Light
Amplitude and Intertrial Phase Coherence of Theta and Gamma Oscillations
... theta oscillations... The amplitude of stimulus-driven gamma oscillations is modulated by the phase of ongoing theta oscillations. This cross-frequency coupling indicates a hierarchical organization of cortical oscillatory dynamics in both healthy control subjects (black line) and schizophrenia patients (red line). The x axis indicates theta phase. The y axis indicates gamma amplitude.
... Heuristic model of phase-amplitude cross-frequency coupling. Gamma oscillations (red and blue lines) are largest in the excitatory versus inhibitory phase of ongoing theta oscillations (black line). Note that excitatory and inhibitory phase may vary according to tasks and neural sources.
... Schizophrenia patients have deficits across a broad range of important cognitive and clinical domains. Synchronization of oscillations in the gamma frequency range (∼40 Hz) is associated with many normal cognitive functions and underlies at least some of the deficits observed in schizophrenia patients. Recent studies have demonstrated that gamma oscillations are modulated by the phase of theta waves, and this cross-frequency coupling indicates that a complex and hierarchical organization governs neural oscillatory dynamics. The aims of the present study were to determine if schizophrenia patients have abnormalities in the amplitude, synchrony, and cross-frequency coupling of gamma and theta oscillations in response to gamma-frequency steady-state stimulation and if abnormal neural oscillatory dynamics are associated with cognitive deficits in schizophrenia.... cross-frequency coupling... gamma oscillations... Schizophrenia patients have normal theta-phase/gamma-amplitude cross-frequency coupling. The modulation index demonstrates the relative strength of cross-frequency coupling via comparison of observed (O) versus resampled or surrogate (S) electroencephalography data in healthy control subjects (black circle) and schizophrenia patients (red squares). The y axis indicates log transform of modulation index.
... neural oscillations... Schizophrenia patients (SZ) have increased theta amplitude and decreased gamma synchrony. The left column shows time-frequency maps from healthy control subjects (HC) and the middle column shows time-frequency maps from schizophrenia patients. The x axis indicates time in milliseconds and the y axis indicates frequency. Color indicates amplitude in the top row and intertrial phase coherence (ITC) in the bottom row. The right column shows difference between schizophrenia patients and healthy control subjects. Difference maps show only time-frequency points at p < .01.
The ictal EEG of a representative hyperthermia-induced seizure in a mutant rat in Group A at 7 weeks of age. The early (A), middle (B) and late (C) stages of the seizure. High-frequencyoscillations were typically observed in the clonic stage of seizure, which is represented by C. The ictal spike indicated by the arrow corresponds to the spike in the temporally expanded trace recorded at 7 weeks of age shown in Fig. 2A.
... Temporally expanded EEG traces during hyperthermia in wild-type rats. EEG traces are shown using the same three different low-cut filters (LCFs) and time–frequency spectra have the same width and frequency range (blue arrow temporally corresponding to black arrow) as in Fig. 2. (A) Data from a brief seizure recorded at 5 weeks of age (only left frontal and right occipital derivations shown because of poor contact of the other derivations); a spike (black arrow) includes an artifact-like irregular fast signal similar to those seen in association with jerks in typical seizures in mutant rats, but no clear HFOs are observed before the spike. The corresponding spectrum shows only artifact-like broad-band power increase with no high-frequency blobs. (B) Slow EEG activity including a sharp transient recorded at 7 weeks of age; no HFOs are found in the corresponding spectrum.
... High-frequencyoscillations... We examined high-frequencyoscillations (HFOs) in the ictal cortical EEGs of hyperthermia-induced seizures in a rat model of febrile seizures with an SCN1A mutation as a means of investigating the pathophysiological mechanisms underlying the generation of febrile seizures. We used 13 male homozygous Scn1a-N1417H mutant rats (F344/NSlc-Scn1aKyo811) and 10 wild-type control rats.... Age-related changes in high-frequencyoscillations (HFOs) in the ictal EEGs of the mutant rats. (A) Proportion of all spikes that were associated with HFOs. Each dot denotes the proportion in the corresponding rat at the indicated age. Data from each individual rat in Group A are connected with lines. (B) Changes in mean peak power. (C) Changes in mean peak frequency.
... High-frequencyoscillations (HFOs) observed in the ictal EEGs of hyperthermia-induced seizures in a mutant rat (the same rat as in Fig. 1). A. HFOs (underlined) with frequencies ranging from 200 to 400Hz were found to precede spikes in the temporally expanded EEG traces of all available cortical derivations at 5, 6 and 7 weeks of age. Each spike following HFOs was associated with a jerk that was indicated by artifacts in the cerebellar bipolar derivation (C2–C1). EEG traces are shown using three different low-cut filters (LCFs) at 1.5Hz (green), 80Hz (blue) and 250Hz (red) to demonstrate that HFOs are visible even in the traces filtered at 1.5Hz and that the detected HFOs were not artificial fast oscillations caused by the filtration. (B) Each time–frequency spectrum shows a high-frequency blob (orange arrow) corresponding to HFOs observed in the expanded EEG trace (A). The motion artifacts caused by convulsive jerks have a broad frequency range (blue arrow temporally corresponding to black arrow in A). The width of each spectral segment including HFOs was 250ms, and the frequency range was 100–500Hz. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)
Contributors:Alexandre Hyafil, Anne-Lise Giraud, Lorenzo Fontolan, Boris Gutkin
Phase–Amplitude Coupling (PAC) and Amplitude–Amplitude Coupling (AAC). (A) Dense-spiking continuous fast oscillation (FO). Plot shows FO amplitude as a function of slow oscillation (SO) phase for distinct FO frequencies (from 10Hz to 100Hz) for the dense-spiking continuous FO model of Figure 1E. Red, high FO power; blue, low FO power. Peak FO frequency varies throughout SO cycle (PFC) but the overall FO power remains constant. (B) Sparse-spiking continuous FO. PAC is present in a model identical to that of panel A with simply an increased level of noise, producing sparse-spiking FO oscillations. (Left) Raster plot showing that the number of FO spikes at each cycle varies as a function of SO phase. (Right) FO amplitude as a function of SO phase, showing strong PAC with maximal FO amplitude close to SO peak, on top of PFC. (C) Sparse-spiking continuous FO with adaptation. The model in (B) was modified by introducing an adaptation current, slowing down the dynamics of the FO oscillation in response to SO modulation. PAC plot shows that maximal FO power occurs at a later SO cycle compared with the no-adaptation model of (B). (D) Dual ING/PING alternation. (Leftmost and middle-left) The FO network alternates within a SO cycle between a period of pyramidal-interneuron-generated oscillations (PING mechanism) and another period of faster interneuron-generated oscillations (ING) (this is obtained by a SO modulation applied to the inhibitory instead of the excitatory population). (Middle-right) FO amplitude as a function of SO phase. PAC occurs for two distinct FO frequencies, with maximal FO power for these two frequencies in antiphase. Such a pattern with distinct frequencies (pointing to distinct oscillation-generation mechanisms) should not be confounded with the case of a frequency modulation within a single FO frequency band, as shown in (A). (Rightmost) Data from rat hippocampus recordings showing a similar alternation between slow gamma and fast gamma within a theta cycle (adapted from ). (D) Positive AAC for positive asymmetric SO. (Left panel) The cross-frequency coupling (CFC) network with sparse-spiking FO oscillations was simulated with a SO sinusoid which amplitude was modulated at a 1Hz rate, whereby an increase in SO amplitude is accompanied by a general increase in SO level (green curve, left panel). This leads to general increase of FO amplitude on top of an increase of PAC strength. (Middle panel) FO amplitude as a function of SO phase, separately for high SO amplitude cycles (red curve) and low SO amplitude cycles (black curve). PAC was present as for the non-modulated SO sinusoid (Figure 3B), but here PAC modulation increased with higher SO amplitude. (Right panel) FO amplitude as a function of SO amplitude, showing strong positive AAC. Figure S1 in the supplemental information online shows the absence of AAC for a symmetric SO. (F) Negative AAC for negative asymmetric SO. Same model as in (E) but here the increase in SO amplitude is accompanied by a general decrease in SO level. This in turn leads to general decrease of FO amplitude on top of an increase of PAC strength, hence a negative correlation between SO and FO amplitudes.... cross-frequency coupling... neural oscillations... Phase–Frequency Coupling (PFC) and Phase–Phase Coupling (PPC). (A) Continuous fast oscillation (FO). (Top) PPC in the continuous FO model of Figure 1E. The top panel shows the histogram of FO phases as a function of slow oscillation (SO) phase for a simulation of 30 s. Darker areas show higher phase concentration. The overall largely homogeneous pattern indicates that very limited PPC emerges, which is confirmed by computing the FO phase-locking factor (PLF) as a function of SO phase : low PLF indicates that SO phase had virtually no influence on FO phase. The green line shows the outline of the SO modulating signal. (Bottom) PFC in the same model. The black curve shows the frequency of FO bursts as a function of SO phase. (B) Intermittent oscillatory/quiescent FO. (Top) PPC in the intermittent oscillatory/quiescent FO model of Figure 1F. The top panel shows very strong FO phase concentration throughout the SO cycle, with nearly non-existent phase change during the quiescent period (i.e., frequency vanishes) and rapid phase dynamics in the oscillatory period. FO-PLF remains high during the whole SO cycle, with highest values during the quiescent period. (Bottom) PFC in the same model. FO frequency is modulated during the oscillatory period and vanishes during the quiescent period. (C) Intermittent oscillatory/asynchronous FO displaying PFC without PPC. (Top) PPC in the intermittent oscillatory/asynchronous FO model of Figure 1G. PPC hardly emerges in this network because no phase-resetting occurs during the asynchronous phase. (Bottom) PFC in the same model. FO frequency is modulated during the oscillatory period and vanishes during the asynchronous period. (D) Type II FO displaying PPC without PFC. (Top left) Raster plot of a network of a single class of interconnected excitatory neurons (adaptive quadratic integrate-and-fire) receiving rhythmic SO impulse. (Bottom left) Nearly symmetrical phase response curve (PRC) of FO neurons showing phase lead for later phases and phase lag for earlier phase, corresponding to a type II oscillator (Box 1). (Top right) Strong PPC in the network because type II oscillators phase-lock easily to external excitatory input. (Lower right) PFC is absent in the same network. Because of the symmetrical PRC, on average there is no acceleration or slowing down of the FO because phase leads and delays cancel out, and FO remains at a constant frequency throughout the SO cycle.
... (A,B) Spiking frequency of type I and type II neurons as a function of driving-current magnitude. (A) Cells with type I membrane excitability can spike at any arbitrary frequency once passed beyond the bifurcation point. (B) When a type II neuron goes through the bifurcation point it starts firing from a finite frequency. (C) Phase response curve of a type I neuronal oscillator: any small excitatory input induces a phase advance within the spiking cycle of the oscillator. (D) Phase response curve of a type II neuron: an excitatory pulse either advances or delays the next spike depending on the particular phase at which the input arrives within the spiking cycle. (E) When solicited with an external sinusoidal input (SO), individual neurons from Aplysia display Arnold tongue maps (i.e., different forms of m:n coupling depending on the input frequency and amplitude). Experimental results are in fair agreement with model simulations (adapted from ). (F,G) Phase–amplitude coupling modeled through a set of Wilson–Cowan equations to represent the activity of a specific population of interneurons in rat hippocampus (adapted from ). Experimenters recorded the response of excitatory and inhibitory neurons to sinusoidal inputs induced using optogenetics tools. (F) Schematic illustration of the excitatory–inhibitory hippocampal network and spontaneous FO visible in E and I population activity in absence of external noise. (G) Phase portrait of the FO model showing the stable limit cycle of the system (blue), the nullclines (i.e., the points where either the excitatory or the inhibitory activity would be constant; black: dashed, excitatory; solid, inhibitory), and the isochrones (the set of initial points from which the dynamics would evolve towards oscillations having the same phase; red). Phase and amplitude modulations of FO can be inferred from the phase portrait, given a specific set of initial conditions.
... Functions of Cross-Frequency Coupling (CFC). (A) Multi-item/sequence representation. (a) Multi-item representations through CFC is envisaged for working memory (Top, in hippocampus and prefrontal cortex) [54,116], spatial memory for previously and subsequently visited items (in hippocampus) [60,61], and visual attention to a series of items in visual field (Bottom, in visual cortex) . (b) Putative CFC architecture for multi-item representation. Intertwined CFC network with a common population of pyramidal neurons connecting to two populations of distinct types of inhibitory neurons, each responsible for the generation of one of the two oscillations (slow, SO; and fast, FO). Some models also include direct connections between the two inhibitory populations [6,20]. (c) Multi-item representation relies on intermittent/quiescent CFC regime: the nested FO oscillation only occurs during a defined phase range, with one item being represented at each FO burst. AAC (through asymmetrical SO) allows modulating the phase range of FO and thus the number of represented items. (B) Long-distance communication. (a) Putative ... Architectures for Cross-Frequency Neural Coupling. (A–D) Cross-frequency coupling (CFC) architectures. (A) Intertwined oscillators. Neural oscillations in distinct frequency bands are generated by partially overlapping neural populations. In the depicted example, excitatory neurons (represented by the triangle) participates in the generation of both a slow oscillation (SO) and a fast oscillation (FO), whereas separate inhibitory populations (represented by green and orange circles) are involved in the generation of respectively SO and FO. CFC arises though the dynamics of the neural population common to both oscillations. Intermingled CFC has been proposed to explain the emergence theta/gamma coupled oscillations in hippocampus [6,30,100], where a common excitatory population is coupled to fast-spiking (FS) cells (generating gamma oscillations) and oriens-lacunosum moleculare (O-LM) cells (generating theta oscillations). (B) Bidirectional coupling. Segregated populations are implicated in the generation of the SO and FO (in the depicted example, a population of excitatory and inhibitory neurons for both oscillations), and coupling is mediated by reciprocal coupling between SO neurons and FO neurons. This architecture was used in the first computational model of coupled theta–gamma oscillations in hippocampus, which featured two coupled inhibitory subpopulations with distinct GABA decay time . Moreover, precise spiking and local field potential (LFP) dynamics of in vitro cortical slices was explained by a sophisticated model whereby a beta1 rhythm concatenates two bidirectionally coupled oscillations, a gamma rhythm generated in superficial layers and a beta2 rhythm generated in deep layers [102,103]. (C) Unidirectional coupling. Distinct populations are implicated in the generation of SO and FO, and coupling arises though one population projecting onto the other population (here SO to FO). In a recent model of speech perception in auditory cortex, coupled theta–gamma oscillations were modeled by two separate excitatory–inhibitory subpopulations responsible for the generation of both rhythms , with the theta module projecting onto the gamma module. (D) Sensory entrainment. Sensory entrainment of neural oscillators is a special case of unidirectional coupling where a neural oscillator is modulated by slow modulations in sensory stimulus (e.g., visual rhythmic movements or amplitude-modulated sounds). This model has been tested in the context of visual processing: a PING module generating broad gamma rhythm responds to visual activity experimentally recorded in monkey thalamus that carries strong slow modulations . (E–G) Temporal dynamics of FO. (E) Continuous FO. (Left) Schematic phase diagram of FO dynamical state: continuous FO occurs when SO modulation shapes a trajectory within the region of existence of FO. (Right) The continuous model of CFC from Fontolan et al.  (see their Figure 5) was transformed into a mathematically equivalent model of quadratic-and-fire neurons and simulated (with membrane white noise of variance 0.01). SO is modeled as a simple modulatory signal while FO is composed of a population of pyramidal and inhibitory cells generating gamma oscillations through the PING mechanism. (Top right) SO modulating signal. (Bottom right) raster plot of FO spikes: pyramidal neurons (n=50, light orange); inhibitory neurons (n=50, dark orange). FO spiking occurs throughout the SO cycle. (F) Oscillation/quiescent-state intermittent FO. (Left) This type of intermittent FO occurs when SO modulation shapes a trajectory between the oscillatory and quiescent regions of FO. (Right) The intermittent model of CFC from  (see their Figure 7) was transformed into network of quadratic-and-fire neurons and simulated. FO alternates between a period of gamma oscillations when excitation from SO is large enough, and a quiescent state when SO signal is lower. (G) Oscillation/asynchronous-state intermittent FO. (Left) This type of intermittent FO occurs when SO modulation shapes a trajectory between the oscillatory and asynchronous regimes of FO. (Right) We modified the parameters of the gamma network of  (constant current, synaptic conductance, and membrane noise) such that the network alternates within SO cycle between a period of strong oscillation and a period of asynchronous spiking activity.
... Neural oscillations are ubiquitously observed in the mammalian brain, but it has proven difficult to tie oscillatory patterns to specific cognitive operations. Notably, the coupling between neural oscillations at different timescales has recently received much attention, both from experimentalists and theoreticians. We review the mechanisms underlying various forms of this cross-frequency coupling. We show that different types of neural oscillators and cross-frequency interactions yield distinct signatures in neural dynamics. Finally, we associate these mechanisms with several putative functions of cross-frequency coupling, including neural representations of multiple environmental items, communication over distant areas, internal clocking of neural processes, and modulation of neural processing based on temporal predictions.