Computational protocol: Mismatch negativity reflects asymmetric pre-attentive harmonic interval discrimination

Similar protocols

Protocol publication

[…] The data analysis was conducted off-line with Curry Neuroscan Software. All data were baseline corrected before being off-line filtered using a 1 Hz to 20 Hz Butterworth bandpass filter (24 dB/octave) and a 50 Hz notch filter. Bad blocks due to muscular activity were removed manually. The EOG channels were inspected automatically for artifacts, and if the sample amplitudes were not between -150 μV and 150 μV, all channels were corrected using principal component analysis. For those participants with few blinking artifacts, the correction range was changed to between -100 μV and 100 μV. The first PCA component was always identified as blinking and then removed. Depending on the length of the blinking, the interval was 100 ms or 200 ms before and 200 ms or 300 ms after the detected artifact was corrected.The preprocessed EEG data were segmented into epochs of 500 ms, with a 100 ms pre-stimulus beginning. Standard and deviant epochs were identified for both conditions and separately averaged. The two standard epochs recorded after a deviant and the last standard before a deviant were excluded. Epochs with a noise level larger than 1.6 times the average noise level were excluded.The remaining epochs were further analyzed with MATLAB 2015 software, including the eeglab 13_6_5b toolbox. The nose electrode was used as reference. The base line correction was made with a 100 ms pre-stimulus interval. Bad channels with high impedances were excluded. The group mean waveforms were calculated for every electrode out of the averaged individual standard and deviant epochs for each subject. Difference waveforms for the fifth condition were calculated by subtracting the standard epochs of the major third condition (i.e., the responses to the fifths as standards) from the deviant epochs of the fifth condition; the difference waveforms for the major third condition were calculated by subtracting the standard epochs of the fifth condition (i.e., the responses to the major thirds as standards) from the deviant epochs of the major third condition. The latency of the MMN response was measured at the minimum of the group mean waveform at electrode Fz in a time window from 0 to 350 ms after stimulus onset. For each subject, the individual MMN amplitude was calculated as the mean voltage in a 40 ms time interval centered on the MMN peak latency of the group average waveform. The noise floor for statistical comparisons was calculated as mean amplitudes in a time interval with the same length as the interval around the MMN peak from 70 ms to 30 ms before stimulus onset.The statistical analysis of the amplitudes was performed with SPSS20 software (IBM, Ehningen, Germany). The assumption of amplitude normality for both the MMN and the noise floor distributions was tested by the Kolmogorov-Smirnov test. The significance of the MMN amplitudes was tested with t-tests for paired samples, comparing the amplitude of the individual MMN in both conditions with the respective noise floor. The level of significance was reduced by the Bonferroni correction for multiple comparisons.Source reconstruction was performed with BESA Research 6.1 software. One pair of symmetric dipoles and one dipole with free orientation were assumed and fitted on a 4-shell ellipsoidal with the LORETA algorithm. Sources were reconstructed for significant MMN amplitudes and also to compare the group average N1 response of the major third condition. The MMN source reconstruction used the interval from 100 to 200 ms after stimulus onset. Based on the zero crossings of the Cz potential of the major third condition, the interval for reconstructing the N1 sources was selected as 95 to 135 ms after stimulus onset (in a 40 ms time window around the minimum of the group average).For the psychoacoustical part, the individual hit rates and false alarm rates were calculated for the two target intervals using Curry Software. The sensivity index d’ was calculated. […]

Pipeline specifications

Software tools EEGLAB, SPSS
Applications Miscellaneous, Clinical electrophysiology