Participants
Fifteen healthy volunteers (7 male, 1 left-handed) participated in the experiment. Two male participants had to be excluded from the analysis due to low signal-to-noise ratio. Mean age of the remaining thirteen participants was 22.92 years (range: 19 to 29 years). All participants reported normal hearing and normal or corrected-to-normal vision. None were taking any medication affecting the central nervous system. All participants received either course credit or payment for their participation. The experiment was undertaken with the understanding and written consent of each subject. The experimental protocol conformed to the Declaration of Helsinki and the ethics guidelines of the German Association of Psychology (ethics board of the Deutsche Gesellschaft für Psychologie, DGPs: http://www.dgps.de/dgps/aufgaben/ethikrl2004.pdf) and did thus not require any additional ethics approval.
Experimental conditions
Participants were asked to fixate on a grey cross constantly displayed on the center of a black screen. Small extensions of the fixation cross (from a visual angle of 0.69° to 0.74° with a distance to the monitor of 100 cm) were presented for 80 ms duration using a variable stimulus onset asynchrony (SOA) of 5–15 s. These extended fixation crosses were not predictable for the participants. Using a mixed experimental design self-initiated and externally-initiated sounds were presented in the same block (Figure 5). Participants were instructed to press a button with their left or right thumb (depending on handedness) with self-paced intervals of 5–8 s (mean: 6.5 s). In 50% of the trials button presses initiated a 50 ms sine tone of 1000 Hz (including 10-ms rise and 10-ms fall times) which was presented immediately after the button press through headphones (Sennheiser HD 25–1) (motor-auditory condition in the blocked design, MA). The intensity of the sounds was adjusted to a comfortable loudness by the participant with soft foam earplugs inserted to attenuate any other sounds. In the remaining 50% of the trials button presses were not followed by any sound (motor-only condition in the blocked design, M). For the participants it was not predictable whether the button press would initiate a sound or not. Additionally, externally-initiated sounds (with the same physical parameters as the self-initiated sounds) were presented randomly between button presses (auditory-only condition in the blocked design, A). Externally-initiated sounds were unpredictable in their occurrence. The SOA between two externally-initiated sounds ranged randomly between 5–8 s. All sounds were generated with MATLAB (http://www.mathworks.com). To avoid a possible overlap with preceding self-initiated sounds, externally-initiated sounds were always presented at least 1 s after the occurrence of a button press. When the SOA between a preceding externally-initiated sound and a button press (initiating a sound or not) was smaller than 1 s both trials were excluded, but the respective number of trials were added at the end of the block to avoid loss of data. In addition to the self-initiation task the allocation of attention was manipulated block-wise. Three attention conditions were included (Attention Sound, Attention Motor, Attention Visual). In the Attention Sound (AS) condition participants were instructed to count all sounds they could hear, including self-initiated and externally-initiated ones. In the Attention Motor (AM) condition participants counted all button presses they made. In the Attention Visual (AV) condition they were asked to count all extended fixation crosses they saw on the screen. Thus, less attention should be directed to the sounds when participants attend to the motor act or to the visual stimuli than when they attend to the sounds.
Experimental procedure
During EEG recordings, participants were seated in a sound-attenuated and electrically shielded chamber. Auditory stimulation was run via MATLAB using the Cogent2000 toolbox (http://www.vislab.ucl.ac.uk/cogent_2000.php). Participants were instructed to press the button once every 5–8 s (mean: 6.5 s). They were informed that a button press would be followed by a sound or silence. Participants were informed about the occurrence of the externally-initiated sounds. However, they were not provided with further information about them. To get used to the self-initiation task participants received several training blocks before the experiment. In these training blocks visual feedback of the button press SOA was given after each button press. In the main experiment visual feedback about the mean button press interval and the responses that were too slow or too fast were only shown at the end of each block. To avoid data loss, a block was repeated whenever participants pressed the button more than 5 times too slow or too fast within one block. In addition to the self-initiation task, participants had to count either all the sounds they could hear (AS), all the button presses they made (AM) or all the extended fixation crosses they saw (AV). Participants were always informed before the beginning of each block about the respective task. After each block they reported the number of counted events. To make sure participants attended to the particular events effectively the block was repeated whenever they miscounted more than +/− 2. Meta-blocks, including all three attention conditions, were repeated eight times. Thus, the EEG experiment consisted of twenty-four experimental blocks. In the meta-blocks the attention conditions (AS, AM, AV) were pseudo-randomized.
Each block consisted on average of twelve (range: ten to fourteen) self-initiated sounds (MA) and silent button presses (M), respectively. This variation was included to make the counting task less predictable for the participants. A comparable number of externally-initiated sounds (A) was presented depending on the mean SOA of the self-paced button presses. In total a mean of 96 trials were analysed for each event (MA, A, M) for each attention condition (AS, AM, AV), respectively.
Data recording and analysis
EEG activity was recorded continuously with Ag/AgCl electrodes from 60 standard locations (Fp1, Fp2, AF3, AFz, AF4, F7, F5, F3, F1, Fz, F2, F4, F8, FT7, FC3, FC1, FCz, FC2, FC4, FC6, FT8, T7, C5, C3, C1, Cz, C2, C4, C6, T8, TP7, CP5, CP3, CP1, CPz, CP2, CP4, CP6, TP8, P7, P5, P3, P1, Pz, P2, P4, P6, P8, PO9, PO7, PO3, POz, PO4, PO8, PO10, O1, Oz, O2) according to the international 10–20 electrode system [41] including the left and right mastoid (M1, M2). An additional electrode was placed at the tip of the nose (serving as offline reference). EOG was measured using the setup described by [42] with one electrode at nasion and two electrodes at the outer canthi. EEG signals were sampled at 500 Hz.
Automatic eye movement correction was applied on the data according to the procedure described in [42], preceded by a 1 to 100 Hz offline band-pass filter. After EOG artifact correction, data were filtered with a 1–25 Hz band-pass filter (kaiser-window, ripple: 0.017, length: 5653 points). For each trial, an epoch of 600 ms duration including a 200 ms pre-stimulus baseline was extracted from the continuous EEG record. Epochs with amplitude changes exceeding 75 μV on any channel were rejected from further analysis. ERPs were averaged time-locked to stimulus onset separately for each event type, attention condition and participant. Button press errors (inter-press interval < 5000 ms or > 8000 ms) were removed from the EEG analysis.
To correct for motor activity present in responses to self-initiated sounds, the ERPs elicited by button presses followed by no sound were subtracted from the ERPs elicited to the self-initiated sounds. This motor-response-corrected ERP was then compared with the ERP of the externally-initiated sounds. In all figures and analysis, ERPs elicited by the self-initiated sounds were corrected this way. This approach has become an appropriate procedure in previous research (presenting MA and M conditions in separate blocks) to measure auditory processing activity in the presence of motor-related activity. However, presenting MA and M conditions introduces a possible confound, namely that it cannot be completely ruled out that non-motor responses, e.g. responses related to temporal expectations of the sound, might also be eliminated subtracting the ERPs elicited by button presses followed by no sound from the ERPs elicited to the self-initiated sounds. However, as the N1-suppression effect observed in the present study was virtually identical to the one reported in previous studies using no mixed design suggests that the suppression effects are not an artefact of the subtraction method of the mixed design.
Because of the multiple components with separate and potentially overlapping latencies underlying the N1 [25] we investigated three separate intervals in the N1 latency range which fit to the peaks N1a, N1b and N1c that have been described in the literature before [25, 26, 40, 43]. Intervals for the N1a and N1c peaks were defined to encompass the first and second peak of the N1 at temporal electrodes. The interval for the N1b peak was defined to encompass the broader N1 peak at central and frontal electrodes. Thus, ERP effects were investigated around the grand-average peaks in the latency range of 85–150 ms (N1b time window), 60–100 ms (N1a time window) and 115–150 ms (N1c time window) after stimulus onset (see Figure 1). ERP amplitudes were calculated from the individual averages as the mean amplitude within these specified analysis time windows. A repeated measurement analysis of variance (ANOVA) with the factors Attention (AS, AM, AV), Production (self-initiated vs. externally-initiated), Laterality (far left: F7, T7, P7; left: F3, C3, P3; midline: Fz, Cz, Pz; right: F4, C4, P4; far right: F8, T8, P8) and Anterior-Posterior (frontal: F7, F3, Fz, F4, F8; central: T7, C3, Cz, C4, T8; parietal: P7, P3, Pz, P4, P8) was computed for each N1 time window, on the mean amplitudes of the electrodes F7, T7, P7, F3, C3, P3, Fz, Cz, Pz, F4, C4, P4, F8, T8, P8. Moreover, in order to identify the sensory specific N1 component generated in auditory cortex, a further repeated measurement ANOVA with the factors Attention × Production was calculated for the mastoid signals in the latency range of 70-110 ms, since the generator for this component has a tangential orientation and results in N1 responses which are negative over frontocentral locations but are also recorded with inverted polarity on the mastoids.
For studying the scalp topographies in the interesting latency ranges, ERP voltage distributions were transformed into scalp current density (SCD) distributions, computing the second spatial derivative of the interpolated potential distribution [44, 45]. The maximum degree of the Legendre polynomials was chosen to be 50, and the order of splines (m) was set to 4. A smoothing parameter lambda of 10−4 was applied. For behavioural data a one-way repeated ANOVA with the factor Attention was computed to compare inter-press time intervals, total number of button presses and timing errors for the self-initiation task between the attention conditions (AS, AM, AV). Furthermore, the counting rates of the attention task for all attention conditions were compared. The counting rates represent the total number of correctly counted events in relation to the total number of actual events of each attention condition. Greenhouse-Geisser correction was applied where appropriate. Additional pairwise comparisons (p-value alpha-adjusted using the Bonferroni correction) were conducted when appropriate to clarify the origin of significant effects. Only interactions that are relevant for the addressed question are reported.