The present study attempted to replicate the effect of validity described in previous studies that used the Posner paradigm [34, 35]. The valid condition, that in which the subject's attention is focused on a region of the visual space, results in a faster response (33 ms) compared to the invalid condition, where the focus of attention has to be reoriented toward another region of the visual field.
Regarding the global differences among the groups, reaction time results indicate a poorer execution of the task in RRMS and BMS patients than in controls. Indeed, the BMS group exhibited fewer correct responses than the control group. Both results indicate a certain cognitive deterioration of the MS groups, noticeably more marked in the BMS group.
It is remarkable that the differences between the pathological and the control group are not caused by a speed- vs accuracy trade-off. In this sense, MS groups showed impaired performance on both accounts, i.e. slower reaction times in BMS and RRMS, and lower percentage of correct responses (for BMS patients only). This higher degree of cognitive impairment in the BMS group than the RRMS group has been previously observed .
The conclusion in our study is that the development of a subclinical cognitive disability not detected by neurological exploration could progress to greater cognitive deterioration. Therefore, it should be possible to assess cognitive status using behavioural techniques and cognitive paradigms.
Regarding the quantitative analysis of the EEG (QEEG), the main result is that, in general, RRMS patients exhibit a larger amplitude for the high bands of the spectrum (beta-2 and gamma) compared to the control and BMS groups in specific regions of the scalp (occipital bilateral and right frontal regions). The rest of the bands (delta, theta, alpha and beta-1) showed no significant difference among the various experimental groups.
This increment for the spectral high bands coincides with other studies where similar increments have been related to psychiatric diseases , or in this particular case, to multiple sclerosis . However, the absence of this increment in the background activity in our study suggests that it is more sensitive to calculate spectral variations during the execution of the cognitive task.
Using the same approach, but with the oddball paradigm as the cognitive task, a similar increment in the high bands was found . In this case, a spectral modulation was observed specifically in frontal regions, not in the posterior areas of the scalp. Therefore, it seems that during attentional tasks for diverse sensorial modalities (visual or auditory), it is possible to find different spectral modulations considering the topographical factor. However, a remarkable difference exists between the two experiments about the time at which the window of the PSD was calculated.
In the auditory study, FFT analysis was done after the arrival of the imperative stimulus, with the logical contribution of the evoked potentials related to the stimulus, and hence the possible influence of ERPs on the lower limit of the beta-band values. In the present experiment, calculation of the PSD was carried out prior to the arrival of the imperative stimulus. In this interval, the average in the time domain shows a well-known component called "contingent negative variation" (CNV). The spectral profile of this component is concentrated mainly in the delta range (0.5. 4 Hz) and is unlikely to cause the modulations observed in the beta and gamma ranges. This experiment confirms that the increment observed for high bands of EEG are not due to the contribution of ERPs and should be considered as an abnormal correlate during attentional process in MS patients.
Before we attempt to define the possible explanation for this modulation, it is necessary to discard some alternatives. One possible concern in our case is ocular contamination. However, if ocular artefacts have an effect on frontal electrodes, similar contribution would be expected in both derivations (F3 and F4), which is contrary to what is found (only F4 showed the statistical difference).
Further evidence is that the differences observed in occipital regions must be considered genuine because these are distant from the ocular source, and the intermediate regions (central and parietal) do not exhibit any increment in the high bands.
The increment of high bands could also be caused in different ways by medication. For example, an increment in the beta band has been detected after administration of psychoactive drugs . Other changes in the EEG could be associated to immunomodulatory substances, such as interferon-beta (IFN-beta), which has been referred by other authors . However, it should be remembered that all the recruited patients were free of any kind of medication during their participation in the experiment (at least one month window).
Another possible interpretation for the increment in the beta-2 and gamma bands could be from a higher level of anxiety, a concerned raised by some authors . Although, this phenomenon could appear in some subjects during the recording, it is unlikely that only RRMS patients experience this anxious. This argument could be applied to other variables, such as a different level of motivation or arousal.
Another concern is the possibility that harmonic components (mainly alpha contribution) could be responsible for differences in other spectral components (beta2 and gamma), as suggested by other authors [42, 43]. However, differences were not statistically significant among the various groups for the slowest bands in the spectral EEG (delta, theta or alpha).
Finally, it is necessary to discard inference due to muscle activity (we are grateful to one of the referees for this comment). The same argument for the anxiety or motivational level could be employed. Again, it is improbable that only the RRMS patients showed this activity. Two additional analyses (omitted for brevity) were conducted related to this issue. In the first case, to discard the threat from muscle artefact, we analysed the temporal electrodes (T5 and T6) and no increase in beta and gamma bands was detected on them. On the other hand, another analysis was performed to check if the increase of the spectral modulations was present all along the bands or specifically in the high bands (beta and gamma). A relative power analysis of all the spectra indicated that only beta and gamma bands showed an increase for multiple sclerosis patients and the relative power of slow bands was higher in control subjects.
In the individual analysis of QEEG scores, a relevant finding was the presence of abnormal high band activity in RRMS patients and the absence from the BMS and the control group. About 42% of RRMS patients showed abnormal beta-2 activity (over 2 SD), and 37% of these patients for the gamma band.
This result is truly outstanding; although the sensitivity of the technique is modest for the detection of those subjects with RRMS (42%), the probability that a control subject or a BMS patient would be a false positive can be considered null, at least after this preliminary study and with a discrete sample of patients.
Another relevant result that helps in interpreting the high EEG modulations is the absence of any correlation between QEEG scores and cognitive impairment assessed on behavioural grounds. This result agrees with other studies that failed to find a correlation between physiological parameters and cognitive deterioration [30, 44]. One possible cause is that the physiological index (QEEG of high bands) and the cognitive process are not related directly. Indeed, the increment of the high bands of the EEG has been associated with diverse psychiatric pathologies  where the cognitive deteriorations can be diverse.
If cognitive status is not related to these modulations, what could be the reason for the modulation of high beta bands of EEG in some RRMS subjects and not in the rest of the groups?
One proposal comes from studies that have suggested adaptive cortical functional changes in response to the progression of the disease [45, 46]. In these studies, an increment in the activity of brain areas normally devoted to the performance of a given task was found. But, there was also an additional recruitment of areas not activated in healthy people.
The lack of correlation between cognitive impairment and the QEEG scores in the present experiment suggests that the increment of activation (beta2- and gamma) in the cortex could not always be related to the performance of the cognitive task. Indeed, the BMS group showed a higher degree of cognitive impairment and the QEEG scores remained in the normal range. It is possible that there are several mechanisms activated during the MS course. One possibility that has been pointed out  could be an adaptive response in the brain to compensate for cognitive impairment, and on the other hand, a non-specific response in the brain (reflected by the increase of high bands of EEG) activating general compensatory mechanisms in response of the progression of the disease, but not strictly related to the cognitive impairment.
Studies have supported that notion that in the first phases of the disease, some processes of cortical reorganization appear  with redistribution of ionic channels  as lesions arise in certain tissues. These adaptive processes are thought to occur in the early lesions and remain subclinical . Our interpretation of the increment of the high bands of the EEG is that they are electrophysiological correlates of some of these processes occurring specifically in some of the RRMS patients.
But why do these modulations not happen with the rest of the patients with RRMS and in all BMS patients? First, this might be explained by some RRMS patients still being in early phases of the disease, and have not begun the reorganizational processes. Another possibility is that some of these have patients in fact are already drained of some form of "natural reserve", and therefore begin to develop a permanent disability  and no high band increment can be observed.
In the case of the BMS patients, perhaps a slow evolution of the disease is not activating the mechanisms of cortical reorganization and in consequence no modulations of EEG are shown. Some form of this "natural reserve" would be activated to compensate for cognitive impairment  but some more general cortical plasticity could not be started (failing to show the increase of high bands of the EEG).
Some challenges are opened after these results. First of all, it seems desirable to increase the sample in order to check if these results are consistently enough for MS population. Another important issue in the future would be to correlate these QEEG scores with MRI features (we are grateful to one of the referees about this comment). In the same sense, it would be interesting to check, from the early beginning of the disease, possible "high band profile" to understand the exact meaning of this correlate and the activation of plastic mechanisms. Particularly interesting would be a follow-up study with RRMS patients and the modulations in the high bands related to relapses or the conversion to the SPMS (secondary progressive multiple sclerosis) form.
This interpretation of the results, also suggests the possibility that the increment of high bands of the EEG is not in fact a specific marker for any pathology, but rather an indicator of a natural reaction in the brain before the appearance of lesions or dysfunctions that can be present in diverse pathologies. Regrettably, it is not possible to confirm this hypothesis definitively by the light of our results. More studies would be necessary to confirm this new point of view in the way that we interpret beta and gamma EEG modulations.