**4. Discussion**

We found a linear relation, with a positive slope between the average power of source activity in the visual cortex and brain noise. The results show that the subjects with more powerful visual cortex activity demonstrate more substantial brain noise. This relationship can be explained as follows. The higher the power of the reconstructed sources, the more neurons are involved in realizing cognitive activity. In a larger network of neurons, the number of synapses would also be higher, and both the synapses and the neurons would feed the phase-destabilizing noise into the system [30].

The two independent methods essentially lead to the calculation of the difference in spectral activity inside the subject's brain, corresponding to the second harmonic of the stimulus frequency when the subject is observing a flickering image, as opposed to when the subject is gazing at a stationary stimulus. Averaging them over the respective regions of interest led to very similar trends between average event-related coherence or frequencyfiltered signal power and brain noise using either software program (Figures 1 and 3). One can see in Figures 2 and 4 that the subject with higher brain noise ("sub06") has a more extensive and intensely activated neuronal network, coherent with the stimulus, as distinct from the subject with lower brain noise ("sub02").

As we have already mentioned above, we set out to adapt the prescribed analysis pipelines of both FieldTrip and Brainstorm to our study. The two software programs gave congruent results following their independent analysis strategies. However, it should not be a surprise that if we try mixing the two analysis pipelines midway, the results will likely deteriorate. Figure 5 shows the result of such mindless mixing of the two methods. Even though the order of subjects' frequency-filtered signal powers remained conserved from Figure 1, the linear relation was lost.

Since we calculate brain noise from the phase fluctuation time series and the corresponding probability distribution, which in turn depends upon the signal-to-noise ratio (SNR) of the source waveforms in the visual cortex to be properly calculated, it can turn into a circular problem where, for very high brain noises, the SNR would be too low to correctly determine the phase fluctuations, which would make the calculation of brain noise impossible. This was the case with subjects "sub8" and "sub11". For these subjects, we did not see any frequency tags in the power spectrum during the flickering cube presentation (signal) and also in the power spectrum for the stationary cube presentation (noise). Thus, they had to be removed from the study. The subjects who showed frequency tags in the power spectrum also had clear bandpass-filtered waveforms in the 13–14 Hz frequency band used to calculate phase difference fluctuations.

We have to emphasize that all codes of our analysis and MEG data used for this study were made publicly available during the review period. The developed methods, along with the prescribed codes on the software documentations adapted to a generic MEG study starting with only a FIFF file, will be accessible to newcomers in the field.
