Tiberio Sempronio [Tiberius Sempronius Gracchus] (principios del s.

The main outcome measures of the oddball task were the ERP components P3a and MMN. The P3a, a subcomponent of the P300, is a positive peak occurring between about 250 and 400 msec after stimulus onset, with maximal distribution over fronto- central electrodes. It has been suggested that the P3a is likely to reflect vigilance, sensory information processing and automatic orienting of attention (Yamaguchi & Knight, 1991). The MMN, an index of involuntary detection of changes in auditory information, is a negative deflection of the ERP signal that is maximal over fronto- central regions and it is usually detected between 100 and 250 msec after the onset of a deviant stimulus. The following procedures have been carried out to pre-process the raw EEG data collected during the auditory oddball task and extract the P3a and the MMN.

The pre-processing of EEG signal was performed with Brainstorm (Tadel et al., 2011). Firstly, the signal was band-pass filtered (0.05 - 30 Hz) and visual inspection of the filtered signal was carried out to manually exclude bad temporal segments of data from further analysis. Following this, power spectral density (PSD; Welch, 1967) was used to obtain an estimation of the power spectrum of the EEG signal for each electrode and over the entire recordings, so that flat or extremely noisy channels could be identified and rejected. Independent Component Analysis (ICA; Herault et al., 1985)

was then carried out on the continuous EEG to identify and remove artifacts associated with eye movements, blinks, muscular activity and any other temporary alterations of electrical activity not reflecting brain activity. After this step, the EEG signal at each electrode was re-referenced to the average of the signal at all remaining channels, before epochs locked to the stimulus onset were imported for standard and deviant tones, for each block of the task (social and non-social). The imported epochs were 800 msec long, and included a 100 msec pre-stimulus window, which was used as a baseline to normalise the signal on the 700 msec post-stimulus temporal window. Only epochs with electrical activity in the range ±100μV were further processed, to obtain four average ERP waveforms, reflecting the stimulus-locked synchronised brain activity for the standard and deviant tones, in the social and non-social blocks.

Similarly to previous studies, including the study from which our passive auditory oddball task was adapted (Whitehouse & Bishop, 2008), the P3a was calculated for both standard and deviant tones, while the MMN was calculated by subtracting the waveform to standard tones from the waveform to the deviant, for each block (social; non-social) (Näätänen et al., 2007). The P3a was determined as the maximal positive peak at the FCz electrode (fronto-central), in the single-subject ERP waveform between 250 and 400 msec after stimulus onset. Conversely, the MMN was identified as the most negative peak in the time window 100-250 msec in the single- subject difference waveform, at the same electrode (FCz). Latency and amplitude of the P3a and the MMN were extracted for further analysis.

Heart rate was recorded during the passive auditory oddball task, and different parameters were extracted. Raw heart rate data was utilised to extract time-domain and non-linear measures of HRV, i.e., the Cardiac Sympathetic Index (CSI), the Cardiac

Vagal Index (CVI) and the Root Mean Square of Successive Differences (RMSSD). Among the various time-domain measures which can be extracted from heart rate, the RMSSD is in fact one of the most reliable measure of parasympathetically mediated HRV (see Shaffer & Ginsberg, 2017, for an overview), and was calculated as following:

a) Raw heart rate signal collected from one of the free electrodes placed on participants’ wrists during the EEG session, was band-pass filtered (8-20 Hz) to reduce the baseline fluctuation of the cardiac signal and to minimise the impact of artifacts and high frequency noise (Fedotov, 2016).

b) Automatic detection of cardiac beats was carried out in Brainstorm (Tadel et al., 2011), followed by visual correction of potentially erroneous or missing peaks, before calculating the time differences (in msec) between each successive heartbeat, i.e., the inter-beat interval (IBI).

c) RMSSD was calculated, as following. First, the time differences between successive IBIs were squared and averaged, for the two blocks of the task and the 30-seconds-long resting blocks before the start of each block; then, the square root was calculated for each of these, to obtain the RMSSD (Shaffer & Ginsberg, 2017):

𝑅𝑀𝑆𝑆𝐷 = √ 1

𝑛 − 1 ∑ (I(k+1)− I(k)) 2 𝑘=𝑛−1

𝑘=1

with k = 1, 2, 3, …, (n - 1); n = number of IBIs within the period; I = IBI in milliseconds

Besides using RMSSD, we embraced the approach proposed by Toichi et al. (1997) to extract the Cardiac Sympathetic Index (CSI) and the Cardiac Vagal Index

(CVI), two indices of HRV which are likely to mirror activity of the sympathetic and parasympathetic branches of the ANS, respectively. To calculate CSI and CVI, a Poincaré plot is created by plotting every peak-to-peak interval (Ik+1) against the preceding interval (Ik), with k = (n – 1) and n = each of the cardiac beats extracted from the HR signal. This results in a two-dimensional graphical ellipsoid-shaped cloud of points, as represented in Figure 9. Two main parameters of this ellipsoid graph, i.e., SD1 and SD2, can be mathematically extracted from the distribution of R-R-intervals in a specific time window. Considering the line of identity as the 45° oriented line representing the identity Ik = Ik+1, SD1 is a measure of the dispersion of the points perpendicularly to the line of identity (i.e., the width of the ellipse), while SD2 represents the dispersion of points along the identity line (i.e., the length of the ellipse) (see Figure 9). More specifically, the mathematical calculations of SD1 and SD2 were carried out using the following equations:

SD1 = SD( 1 √2I(k)− 1 √2I(k+1)) SD2 = SD( 1 √2I(k)+ 1 √2I(k+1))

with k = 1, 2, 3, …, (n - 1); and n = number of cardiac beats within the period. SD = standard deviation of the sample

By multiplying SD1 and SD2 by four, it is possible to obtain an estimation of the transverse length (T) and the longitudinal length (L) of the ellipse, which are further used to calculate the CSI and the CVI (Toichi et al., 1997), as following:

CSI = 4 × 𝑆𝐷2 4 × 𝑆𝐷1 =

𝐿 𝑇

CVI = log₁₀(𝐿 × 𝑇)

Summarising, the ERP components P3a and MMN, besides the CSI, CVI and RMSSD, obtained from the analysis of HRV, were the outcome measures extracted from the passive auditory oddball task.

Figure 9. Example representation, based on collected HR data, of a Poincaré plot. Green line: identity line. Straight orange line: SD1; Dotted orange line: SD2.