Pdf Level 3 Unit 8 P1

In this report, Kogenemaru et al examined whether neurofeedback about the amplitude of motor evoked potentials could lead participants to modulate the excitability of intracortical circuits within M1. They showed that the group receiving neurofeedback had a greater change in SICI than the control group. This apparent change in the excitability of intracortical circuits also resulted in shorter reaction times in a choice task. The study is interesting but I have a few issues with the paper in its current form (see below).

Major comments:
1) It is not clearly described in the manuscript how the ratios were calculated. Change rate = (post0/pre)-1. Are post0 and pre values of the MEPs obtained with paired-pulses or the ratios? If they are ratios,

then you need baseline measures with single-pulse MEPs for pre (pre paired-pulse MEP/pre single pulse MEP) and training (post0 paired-pulse MEP/post0 single pulse MEP). Alternatively, you might be using pre-single pulse MEPs to obtain ratios for post0 and post1. At the moment is not clear what is being presented in the figures. I apologise if I have missed any information here in this regard.
2) In addition to clarifying how baseline measures were used, a formal analysis of background EMG activity (e.g. RMS over 100 ms) preceding the test stimulus would increase my confidence in the main results. I understand that trials in which the 200 microvolts were exceeded were discarded but that does not mean that people could not modulate muscle activity below this level. At the end of the day, 200 microvolts is a relatively high level of activity for at rest conditions.
3) Still on that point, is there a justification for using 200microVolts as a criterion to exclude trials? I would be more comfortable with a proper statistical assessment to see whether or not people learned that they could try to reduce background activity to get a smaller circle. Background activity could be shown overtime (e.g. average of five trials) so that we can see that modulation of background activity is not the driving factor here. Why not discard trials in which rmsEMG activity exceeded a certain percentage (5%) of the peak rms activity during a certain window (100 ms) prior to the TMS stimulus.
4) More inhibition within M1 should result in negative rates of change.

Less inhibition should result in positive rates of change. Given the instruction is to make circles smaller, the authors asked them to inhibit more, is that correct? If you have more intracortical inhibition, why are RTs faster? For example, anodal tDCS shortens RT (Hummel et al. 2006, BMC Neuroscience) and decreases SICI (inducing larger MEP ratios, Kidgell et al. 2013, Neural Plasticity). Assuming participants can change intracortical excitability at will and this method is perhaps equivalent to using non-invasive stimulation (anodal tDCS), it would be expected that RT would become larger.
5) Participants are allegedly learning how to control intracortical activity within M1, thus some measure of performance during acquisition would be good (e.g. change in circle size over time).
6) It would be important to show the exact RT values for both groups before and after the intervention. In addition, the percentage of trials with incorrect responses should be analysed formally so that the RT data can be more meaningfully interpreted. For example, are there less errors for the intervention group in comparison to the control group? If you have more inhibition, I would expect less errors. Also, it was not clear in the results that improvement in RT was specific to the left-hand, but this mentioned in the discussion (line

270).

Minor comments:
7) The sample size justification is vague. Could you please provide more details so that others can reproduce your calculations? How could you base your sample size based on previous studies if neurofeedback learning studies used different measures and methods?

8) What does it mean “delay of feedback was within 40 ms after the paired-pulse occurred”? Line 170.

9) Were people aware of the timing of the TMS pulse? If so, could this affect the results?

10) Line 232: there was no significant difference found in the non-intentional state.

However, an inspection of Figure 3b suggests a qualitative similar pattern of results. I would suggest that the authors conduct a non-parametric permutational anova or one of the robust methods put forward by Wilcox (2012) to analyse the data. Because we are looking at ratios here, I wouldn’t be surprised if the non-significant results turned-out significant when you analyse the data with methods that are more robust to deal with extreme values and departures from normality. This could change the interpretation and discussion of the

results.

11) The labels in plot 3 and 4 need to have their orientation fixed.

12) Line 277: What are inner-thought situations? Please elaborate as it stands the sentence in not helpful.
13) Why did people get larger effects after 30 minutes? Typically learning experiments show a gradual decrement in performance (in this case ability to inhibit) over time when people don’t practice.

14) I find the suggestion that the adopted method is promising somewhat exaggerated. It would be easier to combined TMS and EEG to determine how modulations can be achieved and then simply use that knowledge with cheap EEG systems. TMS is not cheap or easy to use so I not sure how promising this would be in a clinical context.