Figure 1.
Electrocorticographic (ECoG) activity in rolandic cortex during movement and rest (Subject 1).
(A) ECoG potential is measured from the brain surface. (B) ECoG electrodes in situ, embedded in silastic. (C) Electrode positions on the cortical surface. (D) Traces show simultaneous finger position color-coded as in inset (top) along with aspects of the potential timeseries. The raw ECoG voltage at bottom (black) is shown from the M1 site marked with green dot in A–C. “Motor rhythm” is 12–20 Hz bandpassed ECoG (gold trace). “Broadband spectral change” (pink) is the timeseries of an estimate of the coefficient in a power law in the power spectral density of the form
. (E) The power spectral density during movement (green) and rest (black) reveals a decrease in a peaked process at low frequencies (gray – 12–20 Hz), and a broadband increase across the rest of the frequency range during movement (60 Hz line noise and harmonics omitted). (F) The spatial distribution of sites showing a decrease in 12–20 Hz power associated with thumb movement. (color represents a signed r2 measurement, scaled to the maximum across the array: 0.38). In these figures the Central sulcus and Sylvian fissure are shown in yellow. (G) Broadband spectral changes associated with thumb movement are similarly shown (maximum: 0.64).
Figure 2.
Spectral changes in two adjacent precentral motor cortex sites (1 cm apart; Subject 2).
(A) Power spectral density (PSD) of the ECoG potential during epochs of thumb movement (blue) and rest (black). (B) The PSD re-constructed without the 2nd–4th principal spectral components (PSCs). (C) The PSD reconstructed with the 2nd–4th PSCs only. (D) Timecourse of broadband spectral change (pink), as in (B) averaged with respect to onset of the first cued thumb movement at time zero. (E) Average time-varying PSD. (F) Average time-varying envelope of 12–20 Hz filtered voltage (gold - “motor rhythm”). (G) Average finger flexion, timed to onset of thumb movement (thumb flexion – blue; index – green; others - gray). (H–J) Average 4–8 Hz amplitude (H), 12–20 Hz amplitude (I), and broadband change (J), during epochs of movement of thumb (dark blue, “T” at bottom), index (green, “I”), and little finger (light blue, “L”), as well as rest (dark gray). Right-most error bars in (J) are for rest epochs; the task epochs are zeroed with respect to mean of rest epochs. Arrow pointing to the tall bar denotes thumb specificity. (K–T) As in (A–J), but in an adjacent site 1 cm away, showing changes related to index finger movement.
Figure 3.
ECoG broadband power resolves somatotopic representation of fingers (Subject 3).
(A) Changes in broadband and 12–20 Hz beta power at different cortical sites for movement of thumb, index, and little finger. Colors denote a signed r2 measurement of increases and decreases in power with movement relative to rest (individually scaled with maximum to upper left of each plot). (B) Quantification of spatial overlap between changes associated with finger-movements. (1 is maximum possible overlap and 0 is no overlap; negative overlaps occur when increases overlap with decrease). Resampling significance, single cross denotes p<0.01, double denotes p<0.001, that the overlap happened by chance. (C) Traces of thumb (dark blue), index (green), and little finger (light blue) position, with corresponding timecourse of broadband spectral change (pink; approximated by the projection of the 1st PSC to the wavelet-obtained dynamic spectrum) for 3 cortical sites. (D) Quantification of overlap in finger movement activation for 8 other subjects (denoted “S#”).
Figure 4.
Relation between broadband power and phase of ECoG rhythms (subject 4).
(A) ECoG potential (in D) was measured from a pre-central motor cortex site (green dot). (B) Periodic flexion of index finger (green) and other fingers (gray). The entire movement and rest periods are all examined. (C) Fluctuations in broadband power (projection of the 1st PSC, here shown smoothed), extracted from ECoG potential. (D) The raw ECoG potential. (E) Example low frequency rhythm obtained by convolving ECoG with a 3 Hz wavelet (inset). The resulting time-series is color coded for instantaneous phase (relative to positivity peak of the potential). (F) Log values of the time-dependent broadband have a normal distribution. (G) The timeseries of the log of the broadband, color coded by the coincident phase of the low frequency 3 Hz rhythm. (H) The log-broadband signal is aligned with the phase of the low frequency rhythm (as color-coded). (I) The average of the log-broadband amplitude is obtained for phase bins. Error bars denote 3 times the standard error of the mean (3*SEM) for each phase bin. This can be appreciated in one dimension as the “3 Hz coupling row” in G. (J) The “Phase coupling palette” obtained by repeating the process detailed in E–I at each frequency from 1–50 Hz., showing modulation of broadband power with the full range of frequencies.
Figure 5.
The “coupling vector” and trial-by-trial characterization of modulation.
(A) The selected precentral motor cortex electrode (green, Subject 3). (B) The phase-coupling palette for this site identifying range of low-β band frequencies (12–20 Hz). (C) Mean amplitude (+/−3*SEM) of broadband power as function of the Hilbert phase (φ) of the band-passed 12–20 Hz signal. (D) The complex “coupling vector” reflects the magnitude, Zmod, and phase, φc, of the peak modulation of broadband activity with the rhythm in the complex plane. (E) Epochs of finger movement and rest (upper traces) with simultaneous broadband power (pink) and β motor rhythm (gold). Green shows index finger position, and gray shows the other fingers. (F) Segments of the data traces corresponding to periods of rest (dark gray), thumb movement (green), and other finger movements (light gray). (G) From each behavioral epoch, a single trial coupling vector can be obtained. (H) Distribution of coupling vectors across all movement epochs. (I) Calculation of the mean coupling vector for the rest condition (). The distribution of coupling values for this behavior is obtained by projecting individual coupling vectors onto the mean complex coupling vector direction (dark pink line). The distribution of these projected values (upper right histogram) gives the error bar indicating 3*SEM for Zmod. The corresponding vectors from epochs in G are circled. (J) As in (I), for thumb movement epochs. (K) Distribution of rhythm amplitudes (envelope of motor rhythm amplitude in E) during epochs of movement and rest. T = Thumb movement epochs, I = Index finger, M = middle finger, L = little finger, R = rest. (L) Coupling vector amplitudes (projected as in H and I) for different behavioral types. (M) Z-score of log broadband values.
Figure 6.
Phase coupling at the lateral cortical surface during baseline fixation task (subject 1).
(A) Data were obtained from grid of electrodes on the lateral cortical surface. (B) The phase coupling palette for the green site in (A). Inset number (0.17) denotes maximum scaling. White lines identify 4, 8, 12, 20 Hz. (C) Phase coupling palettes for each cortical site on the grid region in (A), with white lines and scaling maxima noted as in (B). (D) Colors of electrode sites delineate cortical regions for average phase palettes in (G). (E) 4–8 Hz modulation in the grid using the band-pass & Hilbert transform. The strength of color and diameter denote the magnitude of coupling, and the color denotes the preferred phase of coupling. Note that coupling to the theta range is quite widespread. (F) Distribution of 12–20 Hz modulation, as in (E). Note that the modulation is strongest in dorsal peri-central areas. (G) Average phase palette by cortical regions delineated in D. The “Dorsal Rolandic” palette combines Dorsal Pre- and Post- central cortex.
Figure 7.
Pooled phase coupling at the lateral cortical surface during baseline fixation task, (subjects 1–4, 7–12).
(A) Average palette by region, across all subjects. (B) Coupling in the 4–8 Hz range. Each dot is an electrode. Black denotes statistically significant coupling. Gray is not significant. (C) Average coupling to 4–8 Hz range across all electrodes in each region (color code of bars indicate areas identified in A. Error bars are 95th percentiles obtained by repeated vector projection of only half of electrodes, randomly selected each time). (D and E) As in B and C, but for 12–20 Hz.
Figure 8.
Coupling motifs and movement-associated change in subject 4.
(A) For the electrode site denoted “A” on the middle cortical rendering. On the far left, the Broadband amplitude (mean +/− 3*SEM) during epochs of movement of the thumb, index, and little fingers, and rest (right-most error bars are for rest, all sub-distributions are zeroed with respect to mean of rest epochs), as in Figure 2J. The coupling palette is as in Figure 4, with the number denoting the maximum scaling for the palette. The top bars to the right of the palette are the amplitude of the 4–8 Hz filtered potential for different behaviors, and the top bars on the far right show the magnitude of coupling during the different behaviors. On the bottom right are filtered amplitudes and the coupling for the 12–20 Hz range. (B–J) As in A, but for the electrode sites denoted B–J on the middle cortical rendering. (K) The spatial distribution of shift in broadband magnitude during thumb movement compared with rest (a signed r2 measurement – maximum scaling denoted by number above the cortical rendering). (L) The 4–8 Hz shift in amplitude during thumb movement (also a signed r2 measurement). (M) The shift in coupling of broadband to 4–8 Hz phase during thumb movement (a signed r2 measurement comparing the distributions of projected epoch values, as in Figure 5). (N) Coupling of broadband to 4–8 Hz phase during rest epochs. As in Figure 6E, the strength of color as well as electrode diameter denote the magnitude of coupling, and the color itself denotes what the preferred phase of coupling is. In this case, the number above the cortical rendering denotes the maximum Zmod in the array. (O to Q) As in L–N, but for 12–20 Hz.
Figure 9.
Movement associated shift in rhythmic modulation of local cortical activity for different cortical areas.
(A) The spatial distribution of shift in broadband magnitude during index movement compared with rest (a signed r2 measurement – maximum scaling denoted by the number above the cortical rendering). The bars below denote the maximum signed-r2 increase/decrease in each region (Subject 5). (B) As in A, but for the shift analytic amplitude of the 12–20 Hz filtered voltage (β-rhythm). (C) The shift in the modulation of broadband spectral change with the phase of the 12–20 Hz phase – the signed-r2 between the index finger movement epoch and rest epoch projected distributions, as illustrated in Figure 5I&J. (D–F) As in A–C, but for Subject 6. (G) Maximum signed-r2 increase/decrease in broadband spectral change during index finger movement compared with rest, for each region separately for all subjects. Note that the effect is strong and specific to dorsal pre-central gyrus. The bar denotes the average, and each dot denotes a different subject (subjects 1–9). (H) As in G, but for the shift analytic amplitude of the 12–20 Hz filtered voltage (β-rhythm). (I) As in G, but for the shift in the modulation of broadband spectral change by the phase of the 12–20 Hz rhythm.
Figure 10.
The relationship between fMRI BOLD signal change and ECoG spectral change.
(A) The magnitude of the fMRI BOLD signal for epochs of thumb movement relative to rest is shown plotted on the brain surface of subject 13. The overlying electrodes show the corresponding shift in ECoG broadband power change (approximated with 65–135 Hz (100 Hz European line noise excluded), with the maximum r2 value noted above the cortical rendering. (B) As in A, except a comparison of fMRI BOLD shift and 12–20 Hz amplitude shift. (C) As in A, except a comparison of fMRI BOLD shift and shift in modulation of broadband by 12–20 Hz phase. (D) As in A, except a comparison of fMRI BOLD shift and absolute amount of modulation of broadband by 12–20 Hz phase specifically during rest – the strength of color and diameter denotes the magnitude of modulation, and the color itself denotes what the preferred phase of modulation is. The number above the cortical rendering denotes the maximum Zmod in the array. (E) Explicit quantification of spatial overlap from the plots in A–D. (F–K) As in A–E, but for subject 14.
Figure 11.
Pair-wise 12–20 Hz phase coherence with a peri-central index-finger specific electrode (subject 3).
The index-finger specific site is identified by magnitude of broadband shift, and denoted with a “ship wheel” symbol. (A) The pair-wise phase coherence between sites is calculated with the remainder of the array (during rest periods). The magnitude of the phase coherence (max = 0.32) is reflected in the strength of the color and the electrode diameter, whereas the relative phase-lag of the phase coherence is denoted by the color. Note that the spatial pattern of phase coherence clusters by gyral anatomy. (B) In order to more clearly isolate spatial changes in phase, the complex phase coherence at each site was projected onto the phase of the site of maximum absolute phase coherence. (C) The mean projected phase coherence from each region (shown color coded on inset cortical rendering) is quantified, with error bars denoting the standard error of the mean within each region.
Figure 12.
Pair-wise 12–20 Hz phase coherence with a peri-central index-finger specific electrode across subjects.
(A–C) As in Figure 11, for subjects 1, 5, and 9; maximum phase coherence noted in between cortical renderings. (D) Pooled data from subjects 1–9, showing that the 12–20 Hz pair-wise phase coherence is conserved within dorsal pericentral cortex, bounded by the pre-central sulcus anteriorly, and the post-central sulcus posteriorly. Note that the “anti-phase coherence” is most strongly due to introduced phase coherence in the common-average process (π out of phase) in the electrodes that do not otherwise have a large beta rhythm.
Figure 13.
Modes of neural activity with cortical beta rhythm states.
(A) Modulation of broadband amplitude by underlying rhythm can be thought of as population-averaged spike-field interaction. (B) “Released cortex” demonstrates a small amount of broadband power coupling to underlying rhythm phase, and the underlying spiking from pyramidal neurons is high in rate and only weakly coupled to the underlying rhythm phase. (C) “Suppressed cortex” demonstrates less broadband power but with higher modulation by the underlying rhythm, while underlying single unit spiking is low in rate but tightly coupled to the rhythm phase. (D) A simplified heuristic for how rhythms might influence cortical computation: During active computation, pyramidal neurons (PN) engage in asynchronous activity, where mutual excitation has a sophisticated spatio-temporal pattern. Averaged across the population, the ECoG signal shows broadband increase, with negligible beta. (E) During resting state, cortical neurons, via synchronized interneuron (IN) input, are entrained with the beta rhythm, which also involves extracortical circuits symbolized by the input from a synchronizing neuron in the thalamus (TN). The modulation of local activity with rhythms is revealed in the ECoG by significant broadband modulation with the phase of low frequency rhythms. (Note: D–E modified from [30], with permission).