In the article “Only the Fastest Corticospinal Fibers Contribute to β Corticomuscular Coherence,” by J. Ibáñez, A. Del Vecchio, J. C. Rothwell, S. N. Baker, and D. Farina, which appeared on pages 4867–4879 of the June 2, 2021 issue, there was an error in the description of the original procedure. The authors note, “When initially testing ways to shift the phase of EEG by 90°, we tried both mathematical integration and differentiation. These gave similar results. The version of MATLAB code using differentiation was used in the article, but the text erroneously stated that integration had been used.
“Second, in the analysis code, we used the differential of the inverted EEG (negativity upward); however, this was not indicated or justified in the manuscript; this sign shift was required to make the peak with the largest amplitude in the cumulant positive. We have realized that this sign change is an important factor to explain the relation between surface recordings and local field potential activity in layer V of the motor cortex. We have accordingly now cited a reference to (Murthy and Fetz, 1996) that justifies this sign change.” The authors regret these errors and thank Peter Kirkwood for useful discussions leading to these corrections. These changes do not affect the conclusions of the article, and the online version has been updated to include the additional reference, the corrected Figure 6 and legend, and revised text as described below.
A, B, Average normalized cumulant densities obtained between the EEG and SMU activity (gray line) and between the phase-shifted EEG (EEGΦ) and SMU activity (black line). Panels A and B show data from the experiments with the TA and FDI, respectively. C, D, β Transmission delay estimates when based on EEG without phase modifications (left) versus when considering that spike trains in pyramidal tract neurons relate to EEGΦ (right). The horizontal solid line and gray area on each panel indicate the mean and SD of MEP latencies in previous studies using large sets of healthy individuals (Rossini et al., 1999). C, D, Data from the experiments with the TA and FDI, respectively.
On page 4874, in Results subsection PART IC, integrated EEG gives better delay estimates than raw EEG, first paragraph, the second and third sentences should instead appear as follows: “However, this is not correct, since pyramidal tract neuron spiking is related to the integral of nearby recordings of local field potential (Baker et al., 2003), and β oscillations recorded from deep cortical layers (close to pyramidal neurons) are π radians out of phase with more superficial recordings (Murthy and Fetz, 1996). This implies that EEG signals (recorded here with a polarity defined as negative upward) present a phase difference of –π/2 with pyramidal neurons.”
In the following paragraph, the second sentence should appear, “To correct for the phase shift between EEG and corticospinal neuron activity, we computed the first derivative (function 'diff' in MATLAB) of the inverted EEG (negativity upward); the cumulant determined with the phase-shifted EEG (EEGϕ) is illustrated with the black track in Figure 6A,B.”
In the following paragraph, the second sentence should appear: “By taking into account the phase shift between β activity in the EEG and in pyramidal neurons, the average cortical→SMU β transmission delay to the TA and FDI became 28.6 ± 3.1 and 21.2 ± 2.6 ms, respectively.”
In the following paragraph, the first sentence should instead read: “Compared with EEG→SMU delay estimates, cortical→SMU delay estimates including phase shift in EEG were much closer to previously reported measures of MEP latency for TA and FDI….”
On page 4877, in the Discussion subsection “Reliable estimates of cortical β transmission delay to muscles,” second paragraph, the last sentence should read, “We know that corticospinal cells discharge with a –π/2 phase shift relative to surface recordings of local field potential oscillations (Murthy and Fetz, 1996; Baker et al., 2003), this phase shift can be corrected for by inverting the EEG and computing its first derivative.”