Fig. 3.
Four surrogate signals consisting of 5 × 104 time points each were used to assess frequency variance as a proxy of phase resets. Shown are the profiles of the instantaneous frequency variance as a function of phase resets in the surrogate signal: (A) sinusoid, (B) pink (power law) noise with power law exponent α = 2, (C) signal exhibiting deterministic chaos generated using a one-dimensional logistic map (x0 = 0.3, r = 3.7), and (D) Gaussian white noise with a signal mean of zero and unity variance. Phase resets were added to surrogate signals at random time points using the amplitude inversion method. As expected, frequency variance is seen to increase almost monotonically with the number of phase resets for all surrogate signals except in the case of Gaussian white noise.