Fig 1.
TRF of the human visual cortex derived from a random broadband visual input train.
(A) The TFR filtered in the broad band (1–100 Hz; black line), the gamma band (40–100 Hz; blue line) and the alpha band (8–13 Hz; gray line). (B) Time-frequency representation of power of the TRF. Note that the TRF was computed for an occipital gradiometer that captures both the alpha band and gamma band TRF.
Fig 2.
TRF and the associated time-frequency representation of power for individual participants.
Note the robust response in 40–60 Hz gamma range. Gray lines depict the TFR at 1–100 Hz while the black lines show the response filtered at 40–100 Hz.
Fig 3.
Time-frequency representation of the power (relative change) induced by the gratings.
The dashed line indicates frequency of the induced gamma oscillations, and for comparison, the solid line indicates frequency of the gamma echo. Curves next to the time-frequency plots represent power averaged over the -0.5 to 3 s time interval.
Fig 4.
Topographies and source modelling of the gamma echo and induced gamma oscillations.
(A) The topography and the source modelling (LCMV beamformer) of the peak amplitude of the gamma echo (0.04–0.08 s) for the right flickering stimuli. (B) Same as (A) but for left flickering stimuli. (C) The topography and source modelling of the power (DICS beamformer) of the induced gamma oscillations within range 40–100 Hz.
Fig 5.
The PING model with constant input currents produces robust neuronal oscillations at around 48 Hz.
(A) Neuronal architecture; the simulated network consisted of interconnected E-cells (N = 400) and I-cells (N = 100). (B) Connectivity matrix between E-cells and I-cells. (C) Spike rastergram for E-cells (blue) and I-cells (orange) shows temporal synchronization among the cells in the presence of constant input current. (D) The average membrane potential of the E-cells exhibited prominent oscillations. (E) Power spectral density of the average membrane potential for the E-cells shows a clear peak at 48 Hz. Note that PSD was averaged over 20 trials.
Fig 6.
(A) The PSD with spectral peak at 48 Hz for input currents IE = 12.25 μA and II = 5.25 μA. (B, C) Power at the spectral peak (B) and corresponding frequency (C) of the network oscillations as a function of input current to E-cells and I-cells. Black, red, and blue circles indicate pairs of the currents producing oscillations at 48 Hz. (D) Spiking activity and corresponding average membrane potentials for three selected input currents that produce oscillations at 48 Hz.
Fig 7.
(A) Transients in spiking activity and average membrane potential after applying constant input currents to E-cells and I-cells. (B) The power (at the spectral peak) and corresponding frequency of the network as a function of the network size. The labels indicate the number of neurons in the network. Bars indicate standard deviation estimated in 20 trials.
Fig 8.
Broadband input current to E-cells produces damped oscillations in the TRF–the gamma echo.
(A) Broadband input current to E-cells (black line) and constant input current to I-cells (orange line). (B) Spike rastergram for E-cells and I-cells for broadband input current. (C) The average membrane potential of the E-cells in response to fluctuating input currents. (D) Power spectral density of average membrane potentials of the E-cells averaged over 20 trials. (E) TRF assessed for the average membrane potentials of the E-cells with respect to the broadband input current. Note the clear “gamma echo”. Gray and black lines depict respectively raw and the filtered TRF (40–100 Hz) averaged over 20 trials. (F) Time-frequency representation of power of the TRF.
Fig 9.
The amplitude of the broadband input current determines temporal characteristics of the gamma echo.
TRF (top panel) and its time-frequency representation (bottom panel). IBB denotes amplitude of the broadband input current.
Fig 10.
Oscillatory input current reveals an amplification of oscillatory dynamics at the ~48 Hz resonance frequency.
(A) Oscillatory input current to E-cells (black line) and constant current to I-cells (orange line). (B) Spike rastergram for E-cells and I-cells. (C) The average membrane potential of the E-cells. (D) Example of power spectral density for input current at 48 Hz (resonance frequency, black line) and at 78 Hz (non-resonance, orange line). (E) Power spectral density of the average membrane potential for oscillatory input over multiple frequencies, 1–100 Hz, and amplitude of 9 μA.
Fig 11.
Network output power at the spectral peak (A) and corresponding frequency (B) as a function of the oscillatory input. The amplitude and frequency of the input current varied within 0.5–10 μA and 5–100 Hz, respectively. Note that the spectral profile on A (right panel) for IE = 9 μA corresponds to the spectral profile shown on Fig 10E.
Fig 12.
(A) Grating stimuli were presented for 0.5 s and then the left and right gratings started contracting for 3 s either coherently (same direction) or incoherently (different directions). Yellows arrows on this figure (were not visible in the experiment) indicate coherent motion downwards. The cue (‘ = ‘) indicates coherent motion. (B) Luminance of the left and right grating stimuli was modulated by two independent broadband signals. The modulation (visual flicker) started at the time -0.5 s, together with the onset of grating stimuli.
Fig 13.
Model output frequency as a function of connectivity strength between I-cells.