Fig. 7.

Removing the latency confound.

Simulations were performed where the trials were divided into two conditions. In condition 1, the mean latency between ERS and ERD was 0.3 s and in condition 2 it was 0.6 s. In both conditions the latencies were drawn from a normal distribution with standard deviation of 0.125 s.

A. Average amplitude waveforms for the two conditions, time-locked to the ERS. Note that the apparent amplitude of the ERS is reduced in condition 1 due to greater overlap with the ERD—although in reality the amplitude of ERS is equal in both conditions.

B. Results of two-sample t-test between time–frequency images for the ERS of condition 1 and condition 2 performed across 10 repetitions of the simulation. Time–frequency images were computed using GLM (top row) or averaging (bottom row).

C. Average amplitude waveforms for the simulation where the amplitude of the ERS in condition 1 was adjusted to equate peak average ERS amplitudes for the two conditions.

D. Results of two-sample t-test between time–frequency images for the ERS of condition 1 and condition 2 performed across 10 repetitions of the simulation with adjusted ERS amplitude.