Abstract

Human voluntary response inhibition has frequently been investigated using go/no-go RT tasks. Recent studies have indicated that not only the traditional averaging waveforms of EEG activities (ERPs) but also the power and phase dynamics of single-trial EEG are important in studying the neural correlates of various human cognitive functions. Therefore, here, we aimed to undertake a detailed study of the time/frequency power and phase dynamics of single-trial EEG during go/no-go RT tasks, with focus particularly on the no-go-specific power and phase dynamics, which are presumed to involve the voluntary response inhibition processes. Thus, we demonstrated no-go-specific theta band EEG power increases and intertrial phase-locking in the midline-frontal areas, which are related to no-go-specific midline-frontal negative–positive ERP waveforms (no-go N2/no-go P3). In addition, we observed no-go-specific alpha band EEG intertrial phase-locking with an adjacent dephasing phenomenon, which is mainly associated with the early part of no-go N2. The estimated time point when the no-go-specific midline-frontal dephasing phenomenon occurred corresponded to the initial part of the voluntary response inhibition process (decision to withhold). Moreover, the no-go-specific phase dynamics in the midline-frontal areas just before and around the no-go N2 peak latency, unlike the power modulations, were affected by changes in the no-go stimulus probability, suggesting the dependence of only phase dynamics on no-go stimulus probability. From these results, we conclude that the complex power and phase dynamics of the theta and alpha band EEG in the midline-frontal areas are specific to no-go trials, being the underlying bases of the no-go-specific ERP waveforms, and suggest that the phase dynamics just before and around the no-go N2 peak latency may involve, at least, the initial part of the voluntary response inhibition process (decision to withhold).

INTRODUCTION

Human voluntary response inhibition, an important function for normal behavior in daily life, has frequently been investigated using go/no-go RT tasks. In one of the commonly used go/no-go tasks, the subject is asked to produce a task-specific motor response against a go stimulus as quickly as possible, but not against a no-go stimulus. However, in most trials of the go/no-go task, the subject prepares for the go response in advance before the presentation of the stimulus, and after detection and discrimination of the presented stimulus, he decides whether or not to withhold the response. Further, if the subject decides to withhold the go response for which he is prepared, the subject inhibits it; however, if the subject decides otherwise, he executes the response. Therefore, it is implicitly assumed that the cortical activities recorded in the no-go trials involve at least the voluntary response inhibition process. It is also assumed that the voluntary response execution process is included only in the go trials, and the stimulus detection and discrimination process is common to both go and no-go trials. Moreover, the initial part of the voluntary response inhibition process, which involves the process of making the “decision to withhold,” must precede the RT in the go trials.

The advance preparation for the go response is suggested by faster mean RTs in the go trials of a go/no-go task as compared to the mean RTs in a two-choice RT task because, in the latter, the subject can prepare for a selected response only after stimulus detection and discrimination (Miller & Low, 2001; Donders, 1969). In addition, the proportion of trials in which a subject prepares in advance for a go response is dependent on the probability of the no-go stimulus presentation, thereby accordingly changing the mean RT (Bekker, Kenemans, & Verbaten, 2004; Bruin & Wijers, 2002; Eimer, 1993). Thus, in most trials that have a combination of frequent go/rare no-go tasks, more frequent response preparation results in the shortening of the mean RTs in the go trials; in contrast, in most trials that have a combination of rare go/frequent no-go tasks, less frequent response preparation results in the lengthening of the mean RTs in the go trials. Therefore, the no-go stimulus probability affects the occurrence rate of the voluntary response inhibition process and also modulates the cortical activities recorded in the no-go trials.

Human neuroimaging studies have reported wide scattering of cortical/subcortical activities in no-go trials (Wager et al., 2005; Watanabe et al., 2002; Liddle, Kiehl, & Smith, 2001; Rubia et al., 2001; Garavan, Ross, & Stein, 1999). Although the functional correlates of each activated region in the no-go trials remain controversial, many researchers agree that right inferior frontal cortex relates to a common response inhibitory function (Aron, Robbins, & Roldrack, 2004; Rubia, Smith, Brammer, & Taylor, 2003; Garavan et al., 1999; Konishi et al., 1999). Further, some studies have suggested that medial frontal cortex, including anterior cingulate cortex and supplementary motor area, is engaged in the making and/or monitoring of the decision (Rushworth, Walton, Kennerley, & Bannerman, 2004; Liddle et al., 2001; Humberstone et al., 1997). However, the detailed time course of the cortical/subcortical activities in no-go trials cannot be revealed by neuroimaging studies because of the limitations of temporal resolution.

On the other hand, many studies that used ERPs, that is, averaged waveforms over a large number of single-trial EEG activities, have reported that after the early negative component that is dependent on the sensory modality in both go and no-go trials (N1), distinct negative–positive deflections over the midline-frontal area are observed mainly in the no-go trials (no-go N2 and no-go P3), whereas a positive component around the centro-parietal area is observed mainly in the go trials (go P3) (Bokura, Yamaguchi, & Kobayashi, 2001; Falkenstein, Hoormann, & Hohnsbein, 1999; Gemba & Sasaki, 1989; Pfefferbaum, Ford, Weller, & Kopell, 1985; Simson, Vaughan, & Ritter, 1977). Neural sources of no-go N2 and no-go P3 are localized mainly in the medial- and/or orbito-frontal cortex (Bekker, Kenemans, & Verbaten, 2005; Bokura et al., 2001). No-go N2 and no-go P3 have a similar appearance in nonmotor (Bruin & Wijers, 2002; Pfefferbaum et al., 1985) and motor deactivation (Yamanaka et al., 2002) go/no-go tasks. Similar no-go-specific midline-frontal negative–positive ERPs also appear in auditory (Falkenstein et al., 1999) and somatosensory (Nakata et al., 2004) go/no-go tasks. In addition, some previous ERP studies (Bekker et al., 2004; Bruin & Wijers, 2002; Eimer, 1993) have reported that no-go N2 and no-go P3 are enhanced on decreasing the probability of presentation of the no-go stimulus. These results suggest that no-go N2 and no-go P3 are independent of motor commands and sensory modalities, and are modulated by the no-go stimulus probability. Therefore, they are likely to be major candidates reflecting the human voluntary response inhibition process. However, the temporal relationship between the no-go N2/no-go P3 peak latencies and the voluntary response inhibition process still remains a critical problem. The peak latency of no-go P3 is generally observed after the mean go RTs (Falkenstein et al., 1999). Moreover, in the case of a go/no-go task with a simple stimulus and response, even the peak latency of no-go N2 often appears at almost the same time as or slightly after the mean go RTs (Filipović, Jahanshahi, & Rothwell, 2000; Gemba & Sasaki, 1989). This implies that the peak latencies of no-go N2 and no-go P3 are too late to reflect, at least, the initial process of voluntary response inhibition.

Recently, some researchers have emphasized that ERP waveforms represent a mean EEG response phase-locked to an event and do not entirely capture the event-related power and phase dynamics of single-trial EEG (Makeig, Debener, Onton, & Delorme, 2004; Pfurtscheller & Lopes da Silva, 1999). Therefore, the time/frequency power and phase dynamics of single-trial EEG during various psychophysical tasks have been investigated in greater detail (Fell et al., 2004; Rizzuto et al., 2003; Tallon-Baudry, Bertrand, Delpuech, & Pernier, 1996); for instance, increased intertrial phase consistency without power enhancement in the alpha frequency band has been observed in the single-trial EEG obtained around the time period of N1 from the parieto-occipital areas after visual stimuli (Hanslmayr et al., 2007; Gruber, Klimesch, Sauseng, & Doppelmayr, 2005; Klimesch et al., 2004; Makeig et al., 2002) and from the central areas after auditory stimuli (Fuentemilla, Marco-Pallarés, & Grau, 2006), suggesting at least a partial contribution of phase-resetting to N1. These results not only reveal the underlying single-trial EEG dynamics of averaged ERP waveforms but may also provide new insights into the functional correlates of single-trial EEG activity. From the go/no-go ERP waveforms, we can consider the existence of go/no-go-common, go-specific, and no-go-specific power and phase dynamics in single-trial EEG. However, the power and phase dynamics of single-trial EEG, in combination with traditional ERP analysis during go/no-go tasks, have not been studied in detail, except some previous research on only the EEG power dynamics during go/no-go trials (Alegre et al., 2004; Hummel, Andres, Altenmüller, Dichgans, & Gerloff, 2002; Leocani, Toro, Zhuang, Gerloff, & Hallett, 2001).

Therefore, in this investigation, we aim to study go/no-go-common, go-specific, and no-go-specific power and phase dynamics in single-trial EEG during visual go/no-go RT tasks, which can unravel the underlying basis of the ERP waveforms in go and no-go trials and may provide new insights of their functional correlates. In particular, we focus on the early component of no-go-specific power and phase dynamics, which is assumed to be related to the initial part of voluntary response inhibition. Furthermore, we search for no-go-specific power and/or phase dynamics of single-trial EEG that are dependent on the no-go stimulus probability in order to verify their functional correlates with the voluntary response inhibition process.

METHODS

Experiment 1

Fifteen healthy men (mean age ± SD = 28.3 ± 3.4 years) underwent a visual go/no-go RT task. All the subjects provided their informed consent, and the local ethics committee approved the experimental procedures. Each subject was comfortably seated on a chair in a dimly lit, electrically shielded room. At approximately 50 cm in front of the subject's eyes, red and green light-emitting diodes (LEDs) for imperative signals were vertically arrayed 1.5 cm apart on a black panel. Each trial began with a warning signal (a beep) followed by an imperative signal for 500 msec after a variable delay of 1.8 to 2.2 sec. Intertrial intervals were randomized to be between 3.5 and 7.5 sec. The green or red LED was illuminated in a random order with almost equal probability [the probability of a no-go signal (mean ± SD) was 51.2 ± 2.6% for all 15 subjects]. The subjects undertook four experimental blocks, each consisting of 50 trials. The subjects were instructed to push a button immediately after a go signal (green LED) or not to push it after a no-go signal (red LED). In two blocks, the subjects had to respond with the right index finger and in the other two blocks, with the left index finger. The subjects were instructed to respond as fast as possible.

Experiment 2

Nine healthy men (mean age ± SD = 28.4 ± 3.7 years) underwent go/no-go RT tasks in the same experimental setting except for the probability of presentation of the go and no-go signals. All the subjects provided their informed consent, and the local ethics committee approved the experimental procedures. After the practice trials, the subjects undertook eight experimental blocks, each consisting of 50 trials. The probability of a no-go signal was approximately 30% (29.6 ± 2.9%) in four blocks and approximately 70% (71.1 ± 2.8%) in the other four blocks. Before each block, the subjects were informed of the probability of presentation of the go and no-go signals in the upcoming block. The order of the experimental blocks was counterbalanced and randomized across the subjects.

EEG Recordings

The EEG was recorded from 19 tin electrodes mounted on an elastic cap (Electro-Cap International, Eaton, OH) corresponding to the International 10–20 System of electrode placement and from two additional electrodes attached to the left and right ears. Data were recorded against a reference placed at AFz and later, off-line re-referenced to averaged earlobes. An EOG was recorded with a pair of electrodes placed above and beside the left eye. Surface electromyograms (EMGs) from the right and left first dorsal interosseous muscles were also recorded. Electrode impedance was maintained below 10 kΩ. Using an EEG recording system (Neurofax EEG-2100 or EEG-1100; Nihon Koden, Tokyo, Japan), the EEG, EOG, and EMG signals were amplified and filtered (band-pass settings: 0.5–100 Hz for EEG and EOG and 50–300 Hz for EMG) and were stored with the record of the LED signals for the purpose of off-line analyses (sampling rate = 500 Hz).

Task Performance and ERP Analysis

The RT in the go trials was defined as the time interval between the stimulus onset and the EMG onset, which is the first point in time when the EMG signal crosses the threshold levels (defined individually for each subject) in the time window of 100 to 400 msec after the stimulus onset. Trials including the missed responses in the go trials and false alarm responses in the no-go trials were defined as error responses. Trials including the error responses and/or EOG artifacts (more than ±100 μV) were excluded from the following analyses. As a result, 96.2 ± 4.3 (mean ± SD) go trials and 97.7 ± 4.3 no-go trials per subject were analyzed in Experiment 1. Further, 138.3 ± 6.1 go and 54.7 ± 9.0 no-go trials per subject in the 70%-go/30%-no-go tasks and 56.1 ± 6.5 go and 138.1 ± 4.5 no-go trials per subject in 30%-go/70%-no-go tasks were analyzed in Experiment 2.

For the ERP analysis, the EEG data of 19 channels were separately averaged for each subject and task condition over 700 msec with a 100-msec prestimulus baseline. Further, 19 grand mean go/no-go ERPs were obtained. The ERP amplitude and latency of the N1 and P3 peaks in the go trials and the N1, N2, and P3 peaks in the no-go trials, all of which generally appear in the go/no-go ERP waveforms, were measured for each subject as follows. At O2, go and no-go N1 comprised the largest negative peak between 100 and 200 msec; at Pz, go P3 was the largest positive peak between 250 and 350 msec; and at Fz, no-go N2 and no-go P3 were the largest negative and positive peaks between 180 and 280 msec, and between 300 and 400 msec, respectively. The recording sites used in the ERP peak measurements were the sites where the typical go/no-go ERP peaks are predominantly observed.

Event-related Power and Phase Dynamics across Trials

In order to investigate the power and phase dynamics in the single-trial EEG of 19 channels during the go/no-go tasks, we first extracted the artifact-free, single-trial EEG data over a time duration of 1100 msec before and 1100 msec after the stimulus onset for the correct go and no-go trials; this is long enough to convolve with the wavelet at the lowest frequency analyzed in this study. Subsequently, the signals were convolved with complex 4-cycle-long Morlet's wavelets. Their central frequencies were changed from 4 to 7 Hz (theta band) in 0.5-Hz steps, from 8 to 14 Hz (alpha band) in 1-Hz steps, and from 16 to 28 Hz (beta band) in 2-Hz steps. From the wavelet transformed signal, wk(t, f), of trial k at time t and frequency f, the instantaneous power pk(t, f) [pk(t, f) = RE(wk(t, f))2 + IM(wk(t, f))2] and instantaneous phase ϕk(t, f) [ϕk(t, f) = arctan{IM(wk(t, f))/RE(wk(t, f))}] were extracted (RE and IM symbolize the real and imaginary parts of a complex number, respectively). Using the instantaneous power pk(t, f), we obtained the event-related power (ERPow) triggered by the stimulus onset. The ERPow were obtained separately for the go and no-go trials as follows:
formula
Further, using the instantaneous phase ϕk(t, f), we calculated the phase-locking index (PLI) (Sauseng et al., 2007; Gruber et al., 2005; Klimesch et al., 2004), which is a measure of consistency across trials of the EEG phase angles for each time/frequency window and is analogous to the “phase-locking factor” (Mazaheri & Jensen, 2006; Tallon-Baudry et al., 1996) and the “intertrial coherence” (Fuentemilla et al., 2006; Delorme & Makeig, 2004; Makeig et al., 2002). The PLI was obtained separately for the go and no-go trials as follows:
formula
A PLI close to 1 reflects a high phase consistency across trials, whereas a PLI close to 0 reflects a high phase variability across trials.

In this analysis, we obtained the ERPow and PLI over the 600 msec before and 600 msec after the stimulus onset for each subject and task condition and defined the first 500 msec (from −600 to −100 msec) as the baseline epochs and the last 700 msec (from −100 to 600 msec) as the test epochs. As a statistical assessment procedure, first, the significant within-subject changes in the ERPow and PLI from the baseline epochs (p < .01, uncorrected for multiple comparisons) were assessed using a bootstrap distribution, extracted at random from the baseline epochs, and applied 200 times (Delorme & Makeig, 2004). Second, the ERPow and PLI at each time/frequency point were tested for significance across subjects by using binomial probability (p < .000001; Onton, Delorme, & Makeig, 2005; Makeig, Delorme, et al., 2004). Third, in order to assess intercondition differences of the ERPow and PLI time/frequency images in the test epochs, we first conducted a paired t test (in Experiment 1) or a repeated measures two-way ANOVA (in Experiment 2, within-subject factors of go/no-go and stimulus probability) for each time/frequency point (p < .05, uncorrected). Finally, to avoid the false positives from multiple comparisons, we applied a procedure for controlling any false discovery rate (FDR, q* < 0.05; Benjamini & Hochberg, 1995) to the time/frequency p value images of each channel. In all statistical results, the effects that were significant only in a few neighboring cells were excluded from interpretation.

Phase Dynamics over Time

In order to investigate the phase dynamics over time, we first obtained the phase-preservation indices (PPIs; Mazaheri & Jensen, 2006) using instantaneous phases ϕk(t, f) calculated as described above. The PPI quantifies the relationship between a phase angle at a certain reference time tref and that at a subsequent time tsub as follows:
formula
A PPI close to 1 implies that the phases are well preserved from the reference time, whereas a PPI close to 0 implies that the phases disperse randomly. Because the PPIs generally decrease over time from the reference point, we can assess the dephasing rate at the reference point from the decreasing-PPI curve. Therefore, we estimated the dephasing time using the least-square fit of the decreasing-PPI curve to the following sigmoid function:
formula
For this fit, 2-cycle-long baseline (PPI = 1) and 4-cycle-long raw PPI values were used. Here, parameter a denotes the final level attained by the PPI (PPI ≈ 0), parameter b relates to a slope of the sigmoid curve (maximum slope = −a * b/4), and parameter c denotes the halfway time point from the initial value (PPI = 1) to the final level attained, namely, a. Using these parameters, we defined 2 * (ctref) as the dephasing time at the reference point (see Figure 4A).

In this analysis, we obtained the dephasing times for all reference points from 300 msec before to 300 msec after the stimulus onset at 5, 10, and 20 Hz (corresponding to the theta, alpha, and beta frequency bands, respectively) for each subject and task condition. We were unable to define baseline periods of sufficient length uncontaminated by EEG data after the stimulus onset, which enable the calculation of the dephasing times by instantaneous phases ϕk(t, f); therefore, only the intercondition differences between the dephasing times were assessed statistically in the same manner as the statistical analysis of the ERPow and PLI time/frequency images.

RESULTS

Experiment 1

In the task in Experiment 1, all subjects were almost free of errors, and the go RT was very fast; the error rates were 1.2 ± 1.9% and 3.4 ± 2.6% in the go and no-go trials, respectively, and the go RT was 228.6 ± 28.5 msec (mean ± SD). In the grand mean ERP waveforms (Figure 1), the occipital early negative peaks in both go and no-go trials (N1) were followed by a large centro-parietal positive deflection in the go trials (go P3), and midline-frontal negative–positive ERP deflections in the no-go trials (no-go N2 and no-go P3). The go/no-go ERP peak amplitudes and latencies (mean ± SD for all 15 subjects) are shown in Table 1.

Figure 1. 

Grand mean go/no-go ERP waveforms (left) and scalp topographies (right) of all 15 subjects. The waveforms at Fz, Cz, Pz, and O2 are shown as thick, colored lines, and the waveforms at the other sites are shown as thin, black lines. The scalp topographies at the N1 and go P3 peak latencies in the go trials and the N1, no-go N2, and no-go P3 peak latencies in the no-go trials are shown.

Figure 1. 

Grand mean go/no-go ERP waveforms (left) and scalp topographies (right) of all 15 subjects. The waveforms at Fz, Cz, Pz, and O2 are shown as thick, colored lines, and the waveforms at the other sites are shown as thin, black lines. The scalp topographies at the N1 and go P3 peak latencies in the go trials and the N1, no-go N2, and no-go P3 peak latencies in the no-go trials are shown.

Table 1. 

ERP Peak Amplitudes (Upper) and Latencies (Lower) during Go/No-Go Task in Experiments 1 and 2 (Mean ± SD)

Peak
Site

Experiment 1
Experiment 2
Go
No-go
70%-go
30%-No-go
30%-Go
70%-No-go
N1 O2 (μV) −7.6 ± 2.2 −7.2 ± 2.6 −8.4 ± 2.2 −8.1 ± 3.0 −8.4 ± 1.8 −7.4 ± 3.0 
  (msec) 153.6 ± 8.2 154.3 ± 10.1 147.8 ± 11.5 148.4 ± 6.2 150.2 ± 8.5 150.1 ± 8.7 
Go P3 Pz (μV) 13.2 ± 2.0  13.6 ± 3.9  12.3 ± 2.1  
  (msec) 286.3 ± 14.8  304.7 ± 17.6  289.1 ± 19.4  
No-go N2 Fz (μV)  −4.1 ± 3.6  −5.8 ± 5.0  −2.4 ± 3.5 
  (msec)  228.8 ± 16.1  232.2 ± 9.8  232.7 ± 9.8 
No-go P3 Fz (μV)  14.9 ± 4.2  13.8 ± 3.7  11.8 ± 3.5 
  (msec)  349.5 ± 25.0  358.0 ± 26.6  358.2 ± 25.4 
Peak
Site

Experiment 1
Experiment 2
Go
No-go
70%-go
30%-No-go
30%-Go
70%-No-go
N1 O2 (μV) −7.6 ± 2.2 −7.2 ± 2.6 −8.4 ± 2.2 −8.1 ± 3.0 −8.4 ± 1.8 −7.4 ± 3.0 
  (msec) 153.6 ± 8.2 154.3 ± 10.1 147.8 ± 11.5 148.4 ± 6.2 150.2 ± 8.5 150.1 ± 8.7 
Go P3 Pz (μV) 13.2 ± 2.0  13.6 ± 3.9  12.3 ± 2.1  
  (msec) 286.3 ± 14.8  304.7 ± 17.6  289.1 ± 19.4  
No-go N2 Fz (μV)  −4.1 ± 3.6  −5.8 ± 5.0  −2.4 ± 3.5 
  (msec)  228.8 ± 16.1  232.2 ± 9.8  232.7 ± 9.8 
No-go P3 Fz (μV)  14.9 ± 4.2  13.8 ± 3.7  11.8 ± 3.5 
  (msec)  349.5 ± 25.0  358.0 ± 26.6  358.2 ± 25.4 

The grand mean ERPow time/frequency images and scalp topographies for the go and no-go trials are shown in Figure 2; only the ERPow time/frequency images at Fz, Cz, Pz, and O2 (Figure 2A) and the ERPow scalp topographies at 5 and 10 Hz at some ERP peak latencies (Figure 2B) are shown. The first and third columns of the time/frequency images in Figure 2A and scalp topographies in Figure 2B show the grand mean ERPow values (not adjusted to the baseline). During the baseline epochs in both go and no-go trials, there were already large power values around 10 Hz (corresponding to the alpha band) distributed broadly over the scalp and centered in the parietal areas, whereas there were relatively small power values in the theta and beta bands. After stimulus onset, the ERPow values changed differently between the go and no-go trials. In order to highlight the ERPow changes from the baseline epochs, the second and fourth columns of the time/frequency images in Figure 2A show the grand mean baseline-adjusted ERPow values. The non-blue–green areas represent statistically significant ERPow changes from the baseline epochs across subjects as determined by binomial probability (p < .000001) based on the bootstrap significance calculations for each subject (p < .01).

Figure 2. 

(A) Grand mean time/frequency images of the event-related power (ERPow) during the go and no-go trials. The time/frequency images at Fz, Cz, Pz, and O2 are shown. The images in the first and third columns from the left represent the logarithmic ERPow values (not adjusted to the baseline). The left-hand color bars represent the mean power values in the baseline epochs. The images in the second and fourth columns from the left represent the baseline-adjusted logarithmic ERPow values in which the non-blue–green areas show statistically significant changes from the baseline epochs across subjects as determined by binomial probability (p < .000001) based on the bootstrap significance calculations for each subject (p < .01). The images in the right column represent significant ERPow differences between the go and no-go trials (p < .05; paired t test with FDR control; q* < 0.05, for multiple comparison correction). (B) Scalp topographies of grand mean logarithmic ERPow during the go and no-go trials. Scalp topographies of the mean baseline values and at some distinct ERP peak latencies, as described in the figure, are shown.

Figure 2. 

(A) Grand mean time/frequency images of the event-related power (ERPow) during the go and no-go trials. The time/frequency images at Fz, Cz, Pz, and O2 are shown. The images in the first and third columns from the left represent the logarithmic ERPow values (not adjusted to the baseline). The left-hand color bars represent the mean power values in the baseline epochs. The images in the second and fourth columns from the left represent the baseline-adjusted logarithmic ERPow values in which the non-blue–green areas show statistically significant changes from the baseline epochs across subjects as determined by binomial probability (p < .000001) based on the bootstrap significance calculations for each subject (p < .01). The images in the right column represent significant ERPow differences between the go and no-go trials (p < .05; paired t test with FDR control; q* < 0.05, for multiple comparison correction). (B) Scalp topographies of grand mean logarithmic ERPow during the go and no-go trials. Scalp topographies of the mean baseline values and at some distinct ERP peak latencies, as described in the figure, are shown.

In the theta band, the ERPow values increased in both go and no-go trials, but their distributions and time courses were different between the go and no-go trials. First, in both go and no-go trials, the ERPow values increased in the occipital areas around N1 peak latency. Subsequently, the ERPow values in the go trials peaked in the parietal areas around the go P3 peak latency, whereas those in the no-go trials peaked in the fronto-central areas around both no-go N2 and no-go P3 peak latencies. In the alpha band, the ERPow values changed in opposite directions between the go and no-go trials. The ERPow values in the no-go trials increased in the fronto-centro-parietal areas from around the no-go N2 peak latency, whereas those in the go trials transiently decreased in the centro-parietal areas from around the go P3 peak latency. In the beta band, there were some small clusters with significant ERPow changes in the go and no-go trials.

Consequently, significant ERPow differences between the go and no-go trials (p < .05; paired t test with FDR control; q* < 0.05) were confirmed as follows (right column of the time/frequency images in Figure 2A). As compared to the go trials, the theta band ERPow values in the no-go trials were significantly larger in the fronto-central areas from around stimulus onset to the end of the test epochs, but were significantly smaller in the parieto-occipital areas around the N1 peak latency. The alpha band ERPow values in the no-go trials were significantly larger than those in the go trials broadly over the fronto-centro-parietal areas from around the no-go N2 peak latency to the end of the test epochs. In the beta band, there were some small clusters with significant ERPow differences between the go and no-go trials.

The grand mean PLI time/frequency images and scalp topographies for the go and no-go trials are shown in Figure 3; only the PLI time/frequency images at Fz, Cz, Pz, and O2 (Figure 3A) and the PLI scalp topographies around 5 and 10 Hz at the some ERP peak latencies (Figure 3B) are shown. The non-blue–green areas of the images in the left and middle columns represent significant PLI changes from the baseline epochs across subjects as determined by binomial probability (p < .000001) based on the bootstrap significance calculations for each subject (p < .01).

Figure 3. 

(A) Grand mean time/frequency images of the phase-locking index (PLI) during the go and no-go trials. The time/frequency images at Fz, Cz, Pz, and O2 are shown. The images in the left and central columns represent raw PLI values in which the non-blue–green areas show statistically significant changes from the baseline epochs across subjects as determined by binomial probability (p < .000001) based on the bootstrap significance calculations for each subject (p < .01). The images in the right column represent significant PLI differences between the go and no-go trials (p < .05; paired t test with FDR control; q* < 0.05, for multiple comparison correction). (B) Scalp topographies of grand mean PLI during the go and no-go trials. Scalp topographies of the mean baseline values and at some distinct ERP peak latencies, as described in the figure, are shown. (C) Time/trial images of 5-Hz and 10-Hz EEG phase angles at Fz (upper) and O2 (lower) in all of the go and no-go trials for a typical subject (n = 93 for go and n = 100 for no-go). The vertical dotted lines in the upper right panel represent no-go N2 and no-go P3 peak latencies (206 and 330 msec, respectively), and those in the lower panels represent the N1 peak latency (150 msec) in this subject.

Figure 3. 

(A) Grand mean time/frequency images of the phase-locking index (PLI) during the go and no-go trials. The time/frequency images at Fz, Cz, Pz, and O2 are shown. The images in the left and central columns represent raw PLI values in which the non-blue–green areas show statistically significant changes from the baseline epochs across subjects as determined by binomial probability (p < .000001) based on the bootstrap significance calculations for each subject (p < .01). The images in the right column represent significant PLI differences between the go and no-go trials (p < .05; paired t test with FDR control; q* < 0.05, for multiple comparison correction). (B) Scalp topographies of grand mean PLI during the go and no-go trials. Scalp topographies of the mean baseline values and at some distinct ERP peak latencies, as described in the figure, are shown. (C) Time/trial images of 5-Hz and 10-Hz EEG phase angles at Fz (upper) and O2 (lower) in all of the go and no-go trials for a typical subject (n = 93 for go and n = 100 for no-go). The vertical dotted lines in the upper right panel represent no-go N2 and no-go P3 peak latencies (206 and 330 msec, respectively), and those in the lower panels represent the N1 peak latency (150 msec) in this subject.

In both go and no-go trials, significant theta-to-alpha band PLI increases appeared at the parieto-occipital areas around the N1 peak latency. Subsequently, only the theta band PLI values in the go trials were maintained until around the go P3 peak. On the other hand, the theta-to-alpha band PLI values in the no-go trials significantly increased in the fronto-central areas around the no-go N2 peak latency, and then, only the theta band PLI values were maintained until around the no-go P3 peak latency. Although the PLI values in the go trials had a similar significant modulation pattern to those in the no-go trials, they were generally small. In the beta band, there was no large cluster with significant PLI changes in either the go or the no-go trials.

Consequently, significant PLI differences between the go and no-go trials (p < .05; paired t test with FDR control; q* < 0.05) were confirmed as follows (right column of the time/frequency images in Figure 3A). As compared to the go trials, the theta band PLI values in the no-go trials were significantly larger in the fronto-central areas from around stimulus onset to the end of the test epochs, but were smaller in the parieto-occipital areas around the N1 peak latency. The alpha band PLI values in the no-go trials were significantly larger than those in the go trials mainly in the midline-frontal area around the no-go N2 peak latency. In the beta band, there was no large cluster with significant PLI differences between the go and no-go trials.

Furthermore, we aligned the phase angles of the wavelet-filtered go/no-go EEG signals at Fz and O2 in the theta (5 Hz) and alpha (10 Hz) frequency bands of all trials. An example for a typical subject is shown in Figure 3C. At O2, both the theta and alpha band phases were locked across trials and were aligned at approximately −π/2 radian (i.e., negative peak) around the N1 peak latency in both go and no-go trials. On the other hand, at Fz, both the theta and alpha band phases were locked across trials and were aligned at approximately −π/2 radian around the no-go N2 peak latency predominantly in the no-go trials.

The grand mean dephasing times of 5, 10, and 20 Hz in the go and no-go trials when the reference times were varied from −300 to 300 msec after the stimulus onset are shown in Figure 4; only the time courses of the dephasing times at Fz, Cz, Pz, and O2 (Figure 4B) and the scalp topographies of the 5- and 10-Hz dephasing times at some reference points in time (Figure 3B) are shown. When the reference time was −300 msec, in the broad areas in both go and no-go trials, the 10-Hz dephasing time was approximately 2 cycles or longer, whereas the 5- and 20-Hz dephasing times were approximately 1.5 cycles.

Figure 4. 

(A) Sigmoidal fitting of the phase preservation index (PPI) decreasing curve and definition of dephasing time. The dephasing time is twice the duration between the reference time (tref) and the time of the halfway-point (c). (B) Time courses of the grand mean dephasing times at 5, 10, and 20 Hz in the go and no-go trials when the reference times are varied from −300 to 300 msec after the stimulus onset. The gray bars on the x-axis indicate the significant differences between the go and no-go trials (p < .05; paired t test with FDR control; q* < 0.05, for multiple comparison correction). (C) Scalp topographies of grand mean dephasing times at 5 and 10 Hz during the go and no-go trials. Scalp topographies at some reference times, as described in the figure, are shown.

Figure 4. 

(A) Sigmoidal fitting of the phase preservation index (PPI) decreasing curve and definition of dephasing time. The dephasing time is twice the duration between the reference time (tref) and the time of the halfway-point (c). (B) Time courses of the grand mean dephasing times at 5, 10, and 20 Hz in the go and no-go trials when the reference times are varied from −300 to 300 msec after the stimulus onset. The gray bars on the x-axis indicate the significant differences between the go and no-go trials (p < .05; paired t test with FDR control; q* < 0.05, for multiple comparison correction). (C) Scalp topographies of grand mean dephasing times at 5 and 10 Hz during the go and no-go trials. Scalp topographies at some reference times, as described in the figure, are shown.

When the reference time was moved to the stimulus onset or later, the 5-Hz dephasing time increased in both go and no-go trials, but its distribution and time course differed between the go and no-go trials. The increase in the 5-Hz dephasing time in the go trials was centered in the parieto-occipital areas and peaked (>2 cycles) when the reference time was around 100 msec, whereas that in the no-go trials was centered in the fronto-central areas and peaked (>2 cycles) when the reference time was around 200 msec. Consequently, the 5-Hz dephasing time in the no-go trials was significantly longer at Fz when the reference time was between 114 and 282 msec, but was significantly smaller at O2 when the reference time was between 26 and 188 msec, as compared to the go trials (p < .05; paired t test with FDR control; q* < 0.05). The 20-Hz dephasing time showed no distinct modulations in any area in either the go or the no-go trials.

In both go and no-go trials, in contrast to the 5-Hz dephasing time, the 10-Hz dephasing time in the occipital areas decreased transiently to approximately 1.8 cycles when the reference time was around 0 msec. In addition, the 10-Hz dephasing time in the fronto-central areas decreased transiently to approximately 1.7 cycles when the reference time was around 100 msec, but only in the no-go trials. Consequently, significant differences between the go and no-go dephasing times were observed when the reference time was between 50 and 128 msec and between 82 and 112 msec after the stimulus onset at Fz and Cz, respectively (p < .05; paired t test with FDR control; q* < 0.05). Moreover, the 10-Hz dephasing time in the parieto-occipital areas decreased to approximately 1.8 cycles when the reference time was around 200 msec and after, but only in the go trials; moreover, there was no significant difference between the go and no-go dephasing times observed at Pz and O2, probably because of the large individual differences.

From the shortened dephasing times and their corresponding reference time periods, we estimated the time periods when the 10-Hz dephasing phenomena could occur as follows: reference time + (dephasing time)/2. As a result, the go/no-go-common 10-Hz dephasing phenomena in the occipital areas might occur around 90 msec after stimulus onset, and the no-go-specific 10-Hz dephasing phenomena in the midline-frontal areas might occur around 185 msec after stimulus onset. The former is compatible with those just before the N1 peak latency, and the latter is compatible with those just before the no-go N2 peak latency.

Experiment 2

In the task in Experiment 2, the mean RTs were significantly faster in the 70%-go trials [230.4 ± 28.0 msec (mean ± SD)]than in the 30%-go trials (259.6 ± 38.3 msec) [t(9) = 2.41, p < .05, paired t test], and commission error rates were significantly higher in the 30%-no-go trials (7.6 ± 8.1%) than in the 70%-no-go trials (2.1 ± 2.1%) [t(9) = 2.41, p < .05, paired t test]. In the grand mean ERP waveforms at Fz (Figure 5A), the no-go N2 amplitudes were significantly larger (more negative) in the 30%-no-go trials (−5.8 ± 5.0 μV) than in the 70%-no-go trials (−2.4 ± 3.5%) [t(9) = 4.62, p < .05, paired t test], whereas there was no significant difference in the other ERP peak amplitudes and latencies between the 30%-go and 70%-go trials. The go/no-go ERP peak amplitudes and latencies (mean ± SD) in Experiment 2 are also shown in Table 1.

Figure 5. 

(A) Grand mean ERP waveforms at Fz (left) and scalp topographies (right) during the go/no-go tasks with two stimulus probability conditions (go/no-go: 70%/30% and 30%/70%) for all nine subjects. The waveforms at Fz and the scalp topographies in the 30%- and 70%-no-go trials at the no-go N2 peak latencies are shown. (B) Grand mean time/frequency images and scalp topographies of the phase-locking index (PLI) during the go/no-go tasks with two stimulus probability conditions. The time/frequency images at Fz and the scalp topographies in 30%- and 70%-no-go trials at the no-go N2 peak latencies are shown. The images represent raw PLI values in which the non-blue–green areas show statistically significant changes from the baseline epochs across subjects as determined by binomial probability (p < .000001) based on the bootstrap significance calculations for each subject (p < .01). (C) Time/frequency images indicating the results of the statistical test (repeated measures two-way ANOVA; p < .05 with FDR control; q* < 0.05 for multiple comparison correction) in the PLI (upper) and ERPow (lower) during the go/no-go tasks with two stimulus probability conditions. The nongreen parts in the time/frequency images represent the significant main effect of go/no-go (red part) and, in addition, the significant main effect of stimulus probability (blue part) and the significant interaction between go/no-go and stimulus probability (purple part). Within the right-side boxes, the mean (±SD) of the PLI (upper) and ERPow (lower) between 150 and 250 msec after stimulus onset in the range of 4-to-7- and 8-to-14-Hz (within the black frame in the PLI and ERPow time/frequency maps) are presented. (D) Time courses and scalp topographies of the grand mean dephasing times at 5 and 10 Hz during the go/no-go tasks with two stimulus probability conditions when the reference times varied from −300 to 300 msec after stimulus onset. The time courses at Fz, and the scalp topographies at the time points as described in the figure, are shown.

Figure 5. 

(A) Grand mean ERP waveforms at Fz (left) and scalp topographies (right) during the go/no-go tasks with two stimulus probability conditions (go/no-go: 70%/30% and 30%/70%) for all nine subjects. The waveforms at Fz and the scalp topographies in the 30%- and 70%-no-go trials at the no-go N2 peak latencies are shown. (B) Grand mean time/frequency images and scalp topographies of the phase-locking index (PLI) during the go/no-go tasks with two stimulus probability conditions. The time/frequency images at Fz and the scalp topographies in 30%- and 70%-no-go trials at the no-go N2 peak latencies are shown. The images represent raw PLI values in which the non-blue–green areas show statistically significant changes from the baseline epochs across subjects as determined by binomial probability (p < .000001) based on the bootstrap significance calculations for each subject (p < .01). (C) Time/frequency images indicating the results of the statistical test (repeated measures two-way ANOVA; p < .05 with FDR control; q* < 0.05 for multiple comparison correction) in the PLI (upper) and ERPow (lower) during the go/no-go tasks with two stimulus probability conditions. The nongreen parts in the time/frequency images represent the significant main effect of go/no-go (red part) and, in addition, the significant main effect of stimulus probability (blue part) and the significant interaction between go/no-go and stimulus probability (purple part). Within the right-side boxes, the mean (±SD) of the PLI (upper) and ERPow (lower) between 150 and 250 msec after stimulus onset in the range of 4-to-7- and 8-to-14-Hz (within the black frame in the PLI and ERPow time/frequency maps) are presented. (D) Time courses and scalp topographies of the grand mean dephasing times at 5 and 10 Hz during the go/no-go tasks with two stimulus probability conditions when the reference times varied from −300 to 300 msec after stimulus onset. The time courses at Fz, and the scalp topographies at the time points as described in the figure, are shown.

The grand mean PLI time/frequency images and scalp topographies during the go/no-go tasks with two stimulus probability conditions (go/no-go: 70%/30% and 30%/70%) are shown in Figure 5B; only the PLI time/frequency images at Fz and only the PLI scalp topographies in the no-go trials around 5 and 10 Hz at the no-go N2 peak latency are shown. The PLI time/frequency images and scalp topographies reveal that the theta-to-alpha band no-go-dominant PLI increases in the midline-frontal areas around the no-go N2 peak latency were observed more clearly in the 30%-no-go trials than in the 70%-no-go trials.

The results of repeated measures two-way ANOVA of the PLI time/frequency images at Fz are shown in the left upper time/frequency images in Figure 5C. A significant main effect of go/no-go (red part, p < .05 with FDR control; q* < 0.05) was observed around the no-go N2 peak latency in the theta-to-alpha bands and around the no-go P3 peak latency in the theta band, which were almost the same as the upper panel of the right column in Figure 3A. In addition to the go/no-go main effect, significant main effects of stimulus probability (blue part, p < .05 with FDR control; q* < 0.05) and significant interaction between go/no-go and stimulus probability (purple part, p < .05 with FDR control; q* < 0.05) were observed only just before and around the no-go N2 peak latencies in the theta-to-alpha bands. The purple part of Figure 5C indicates that the effect of the stimulus probability is observed in either the go or the no-go trials; it is observed within the corresponding time/frequency window (150–250 msec/4–7 and 8–14 Hz) only in the no-go trials (right upper box panel in Figure 5C). Similar go- or no-go-specific PLI increases depending on the stimulus probability were not observed in any of the other areas or time/frequency windows.

The results of repeated measures two-way ANOVA conducted on the ERPow time/frequency images at Fz are shown in the left upper time/frequency images in Figure 5C. A significant main effect of go/no-go (red part, p < .05 with FDR control; q* < 0.05) was observed broadly in the theta and alpha bands from around the no-go N2 peak latency to the end of the test epoch. This was similar to the upper panel of the right column in Figure 2A. On the other hand, there was no significant main effect of the stimulus probability or significant interaction between go/no-go and stimulus probability, although the theta band ERPow in the 30%-no-go trials around the no-go N2 peak latency tended to be large (right lower box panel in Figure 5C), that is, the go/no-go differences in the theta-to-alpha band ERPow just before and around the no-go N2 peak latency were independent of the stimulus probability.

Further, increases in the 5-Hz dephasing time and transient decreases in the 10-Hz dephasing time in the midline-frontal areas were clearly observed in the 30%-no-go trials, but not in the 70%-no-go trials (Figure 5D). However, there was no significant main effect of go/no-go and stimulus probability and the interaction between them, probably because there were great differences in the time courses between the individuals. Therefore, we picked up and compared the maximum values of the 5-Hz dephasing time between 0 and 300 msec and the minimum values of the 10-Hz dephasing time between 0 and 200 msec in the 30%- and 70%-no-go trials for each subject. Consequently, the maximum 5-Hz dephasing time in the 30%-no-go trials (2.91 ± 0.56 cycles) was longer than that in the 70%-no-go trials (2.09 ± 0.24 cycles) [t(9) = 6.39, p < .05, paired t test], whereas their latencies were not significantly different (124.7 ± 44.7 msec and 157.3 ± 77.9 msec, respectively). Moreover, the minimum 10-Hz dephasing time in the 30%-no-go trials (1.56 ± 0.33 cycles) was shorter than that in the 70%-no-go trials (1.79 ± 0.27 cycles) [t(9) = 2.52, p < .05, paired t test], whereas their latencies were almost the same (104.7 ± 64.0 msec and 103.1 ± 60.7 msec, respectively).

DISCUSSION

In this study, we performed a detailed investigation of the power and phase dynamics in single-trial EEG and the traditional ERP waveforms during go/no-go RT tasks. The go/no-go ERP waveforms and behavioral results in Experiment 1 resemble those typically found in many previous go/no-go ERP studies (Bekker et al., 2004; Bruin & Wijers, 2002; Bokura et al., 2001; Falkenstein et al., 1999; Eimer, 1993; Gemba & Sasaki, 1989; Pfefferbaum et al., 1985; Simson et al., 1977). The differences in the go/no-go ERP waveforms and behavioral results between the two stimulus probability conditions in Experiment 2 are also in agreement with the results of previous studies (Bekker et al., 2004; Bruin & Wijers, 2002; Eimer, 1993). Therefore, the power and phase dynamics and their modulation by task manipulation as reported in this study are considered to be representative of single-trial EEG recorded from subjects performing go/no-go RT tasks.

Power and Phase Dynamics in Single-trial EEG during Go/No-go Task Trials

In the time/frequency images of single-trial EEG during the go and no-go trials in Experiment 1, we first confirmed that the alpha band power values in the baseline epochs were relatively large as compared to the theta and beta band power values. The result strongly suggests the existence of an alpha band oscillation before the stimulus onset. On the other hand, because the phase angles in the baseline epochs were randomly distributed across trials, the PLI values were small in all the analyzed frequency bands.

In the theta band, the time courses and scalp topographies of the ERPow and PLI after the stimulus onset were very similar. The parallel ERPow and PLI increases from the small baseline level probably reflect the existence of polarity- and latency-fixed deflections, the cycle length of which corresponds to the theta band (Sauseng et al., 2007). Therefore, it is suggested that the parallel theta band ERPow and PLI increases are the significant underlying basis of the go/no-go-common (occipital N1), go-specific (centro-parietal go P3), and no-go-specific (midline-frontal no-go N2 and no-go P3) ERP waveforms. On the other hand, in the beta band, the ERPow increases after the stimulus onset were relatively small, even though there were some small significant clusters in both go and no-go trials. There was no large cluster with significant PLI changes in the go and no-go trials. Therefore, we conclude that the beta band EEG is not substantially linked to the generation of the go/no-go ERP waveforms.

The alpha band PLI increases and the theta-to-alpha band phase alignments at the negative peak of phase angles were observed in the occipital areas around the N1 peak latencies in both go and no-go trials and in the midline-frontal areas around the no-go N2 peak latency predominantly in the no-go trials. These results indicate that the phase dynamics across trials related to the occipital N1 and the midline-frontal no-go N2 are similar, even though the former was go/no-go-common, whereas the latter was no-go-specific. Some previous studies have reported alpha band PLI increases related to occipital N1 (Hanslmayr et al., 2007; Gruber et al., 2005; Klimesch et al., 2004; Makeig et al., 2002) and have demonstrated theta-to-alpha band phase alignment related to occipital N1 (Gruber et al., 2005). In this study, we additionally report alpha band PLI increases and theta-to-alpha band phase alignment in relation to midline-frontal no-go-N2.

In contrast to the theta band, the PLI modulations in the alpha band were not in parallel with the ERPow modulations. The alpha band ERPow changes were oppositely directed between the go and no-go trials (decrease and increase, respectively) and were spread broadly over the scalp from around the no-go N2 peak latency and later, which is in agreement with the results of previous studies (Alegre et al., 2004; Hummel et al., 2002; Leocani et al., 2001). In case of long latency periods, these alpha band ERPow changes were hardly reflected in the ERP waveforms (see Figure 1) because they were not accompanied by PLI increases (ERP waveforms reflect the mean response of phase-locked activity, but do not reveal activity that is not phase-locked). One of the reasons for the dissociated modulation of the alpha band ERPow and PLI might be the existence of strong oscillations before the stimulus onset, that is, it remains possible that the underlying base of the PLI increases differ between the alpha and theta bands. Therefore, we further examined the time-dependent phase dynamics of single-trial EEG during the go and no-go trials.

In the dephasing time estimation, the 10-Hz (alpha band) dephasing time was relatively large as compared to the 5-Hz (theta band) dephasing time when the reference time was −300 msec. This suggests that the phase of existing strong alpha band oscillation before stimulus onset is preserved for a relatively long period, in agreement with the results of a previous study of MEG recordings (Mazaheri & Jensen, 2006). The difference between the 10- and 5-Hz dephasing times was similar to that between the baseline ERPow values, even though the dephasing times were computed using only the phase information. The increases in the 5-Hz dephasing time in the no-go trials appeared during the later time periods and in more anterior areas than in the go trials, which is also similar to the findings of the theta band ERPow values.

In contrast, the modulation patterns of the 10-Hz dephasing time are clearly different from those of the alpha band ERPow values. The 10-Hz dephasing time transiently decreased at Oz in both go and no-go trials when the reference time was around 0 msec and at Fz and Cz only in the no-go trials when the reference time was around 100 msec. If the decreases in the dephasing time were dependent solely on the decreases in the ERPow values, the ERPow values in the corresponding time periods would decrease. However, there was no significant alpha band ERPow decrease during those time periods and in the cortical areas. Therefore, the 10-Hz dephasing phenomena are estimated to occur, at least, in the occipital areas just before and around the N1 peak latency in both go and no-go trials and in the midline-frontal areas just before and around the no-go N2 peak latency only in the no-go trials. These results indicate that the phase dynamics over time in single-trial EEG underlying occipital N1 and midline-frontal no-go N2 are similar, even though the former is go/no-go-common and the latter is no-go-specific.

Considering the results of Experiment 1 together, the go/no-go-common, go-specific, and no-go-specific power and phase dynamics in single-trial EEG can be summarized as follows. In addition to the large baseline alpha band power, the go/no-go-common dynamics include transient theta band ERPow and PLI increases and transient alpha band PLI increases with adjacent dephasing phenomenon, both of which are related to occipital N1. The go-specific dynamics include transient theta band ERPow and PLI increases, which contribute to centro-parietal go P3, and transient alpha band ERPow decreases, which are hardly reflected in the ERP waveforms. The no-go-specific dynamics include theta band ERPow and PLI increases associated with midline-frontal no-go N2 and no-go P3, the alpha band PLI increases with adjacent dephasing phenomena that mainly contribute to midline-frontal no-go N2, and alpha band ERPow increases from around the no-go N2 peak latency, the later parts of which are hardly reflected in the ERP waveforms. These no-go-specific power and phase dynamics in single-trial EEG may involve the voluntary response inhibition process.

In addition, the estimated time period when the alpha band phase shifts, which probably reflects in the PLI increases and dephasing time decreases (around 185 msec after the stimulus onset), corresponds approximately to the time of the “decision to withhold” (after the go/no-go-common occipital N1 peak latency and before the mean go RTs). Therefore, particularly in the no-go-specific dynamics, the alpha band phase dynamics just before and around no-go N2 are presumably related to the initial part of the voluntary response inhibition process (decision to withhold).

Two Hypotheses for the Underlying Basis of ERP: Phase-resetting or Additive Response

Most of the previous studies investigating single-trial EEG power and phase dynamics have discussed two hypotheses for the mechanism of ERP generation: phase-resetting of the ongoing EEG oscillation and an additive latency- and polarity-fixed response to the ongoing EEG oscillation (see Sauseng et al., 2007 for a review). Further, it has been pointed out that intertrial phase-locking can be observed not only by phase-resetting but also by an additive response. As for the theta band PLI increases observed in this study, we suggest that they may mainly be attributed to an additive response with theta band oscillatory components because their baseline power is relatively small and they are accompanied by power increases. However, we do not rule out the contribution of theta band phase-resetting to ERP generation.

On the other hand, in case of alpha bands with a relatively large baseline power, which fulfill the prerequisites for phase-resetting (Hanslmayr et al., 2007; Mazaheri & Jensen, 2006; Shah et al., 2004), it is not easy to distinguish whether intertrial phase-locking reflects phase-resetting of the ongoing oscillation (Sauseng et al., 2007) because of the methodological uncertainty of phase estimation from single-trial EEG data with large ongoing oscillations and ERPow changes independent of intertrial phase-locking. Because the dephasing time is also influenced by uncertain phase estimation, shortening of the dephasing time is not hard evidence for phase-resetting; that is, from our results alone, it is difficult to conclude whether occipital N1 and midline-frontal no-go N2 are attributable to phase-resetting of the existing alpha band oscillations.

It is of note, however, that the dephasing time can reveal the differences in the phase dynamics between the alpha band intertrial phase-locking with relatively strong ongoing oscillations and the theta band intertrial phase-locking with relatively weak ongoing oscillations. Although further examination is necessary, time-dependent phase dynamics might be one of the important starting points for addressing this unsolved problem.

Modulations of Power and Phase Dynamics in Single-trial EEG Depending on No-go Stimulus Probability

In Experiment 2, we searched for the no-go-specific power and phase dynamics that undergo different modulations as a function of no-go stimulus probability. Before that, we compared the go/no-go ERP waveforms at Fz and confirmed the presence of the no-go-specific midline-frontal negative–positive ERP deflections even in the 30%-no-go trials (Figure 5A). Lavric, Pizzagalli, and Forstmeier (2004) also reported that on ERP comparison of the rare go and no-go trials recorded from the two groups of subjects, the no-go-specific midline-frontal negative–positive ERP deflections were observed to appear even in the rare no-go trials. In the statistical analysis of the PLI and ERPow at Fz, we observed the significant effects of go/no-go (Figure 5C, red part), the time/frequency maps of which were almost identical to those of the significant differences between the go and no-go trials in Experiment 1 (Figure 2A and Figure 3A). These results suggest the presence of no-go-specific ERP waveforms and the underlying power and phase dynamics, which are likely to be related to human voluntary response inhibition regardless of the stimulus probability.

In addition, the results of Experiment 2 suggested that enhancement of the theta-to-alpha PLI values in the midline-frontal areas around the no-go N2 peak latency was dependent on the stimulus probability only in the no-go trials. On the other hand, such a no-go-specific effect of stimulus probability was not observed in case of ERPow, that is, the modulation patterns of the PLI and ERPow against the stimulus probability were clearly different. Moreover, the no-go-specific lengthening of the 5-Hz dephasing time and the no-go-specific shortening of the 10-Hz dephasing time at Fz were also dependent on the stimulus probability. These results suggest that the theta-to-alpha band phase dynamics mainly contribute to the enhancement of midline-frontal no-go N2 depending on no-go stimulus probability.

It cannot be denied that the results reflect an effect of the rare stimulus, for example, a mismatch negativity, which is automatically elicited by a rare stimulus, around the latency of N2 (Rinne, Antila, & Winkler, 2001). However, the effect of the rare stimulus is thought to be similar in the 30%-no-go trials and 30%-go trials. Therefore, the 30%-no-go-specific modulation of the no-go N2 and underlying phase dynamics in single-trial EEG would be affected by the probability of the no-go-specific processes rather than by the probability of the visual stimulus.

To summarize the results of Experiment 2, there are two components in the no-go-specific ERP waveforms and the no-go-specific power and phase dynamics: components dependent on no-go stimulus probability and those independent of it. The former include the no-go N2 ERP waveform and the theta-to-alpha band phase dynamics just before and around the no-go N2 peak latency, which may be related to the probability-dependent processes in human voluntary response inhibition. The latter include the no-go P3 ERP waveform and the power and phase dynamics around the no-go P3 peak latency, which may be related to the size-variable processes in human voluntary response inhibition.

Further Functional Considerations of the Power and Phase Dynamics in Single-trial EEG during the Go/No-go Task Trials

With regard to alpha band power dynamics, many previous studies have reported results similar to the results of this study in that alpha band ERPow values over the broad motor-related areas decreased before and during the execution of the movement (Stančák, Feige, Lücking, & Kristeva-Feige, 2000; Babiloni et al., 1999; Leocani, Toro, Manganotti, Zhuang, & Hallett, 1997; Pfurtscheller & Berghold, 1989). Therefore, the alpha band ERPow decreases in our go trials might also reflect cortical activation related to the motor go response. On the other hand, Klimesch, Sauseng, and Hanslmayr (2007) recently proposed that the increase in the alpha band ERPow values distributed broadly over the scalp in different cognitive tasks reflects a state of top–down inhibitory control processes, which may be related to the alpha-band ERPow increases in our no-go trials. Our results further suggest that the alpha-band ERPow increases are, at least, related to the no-go-specific process regardless of the stimulus probability.

Because parallel modulations of the theta band ERPow and PLI values would substantially be related to the P3 ERP waveforms, the theta band power and phase dynamics might reflect functions that have already been suggested in previous go/no-go ERP studies. Polich (2007) reviewed many P3 ERP studies and stated that no-go P3 is a variant of an anterior-centered P3 subcomponent, namely, “P3a,” which is generally considered to originate from a stimulus-driven frontal attention mechanism. Falkenstein et al. (1999) reported that no-go P3 is not very sensitive to task manipulation and differences in performance. Our results of the no-go-specific midline-frontal theta band power and phase dynamics around the no-go P3 peak latency also demonstrate that they are not dependent on the stimulus probability.

In order to understand the functional correlates of the complex spatio-temporal patterns of the power and phase dynamics, determining whether all of them originate from an isolated cortical area remains a problem. Debener, Makeig, Delorme, and Engel (2005) applied independent component analysis to EEG data recorded during an auditory novelty oddball task and reported that P3 is decomposed into two components with anterior- and posterior-centered distributions. Further, Onton et al. (2005) applied independent component analysis to the EEG data recorded during a Sternberg memory task and reported that the midline-frontal theta band components are distinct from the centro-parietal alpha band components. In our preliminary results, the midline-frontal centered no-go N2/no-go P3 component and the parietal centered go P3 component were successfully separated, whereas the theta and alpha band components were not distinctly separated, probably because the number of recording electrodes in our study was relatively small. Further investigation using EEG data with a high spatial resolution would allow us to understand the functional correlates of single-trial EEG dynamics during go/no-go tasks.

Conclusion

In summary, we investigated the power and phase dynamics of single-trial EEG and the traditional ERP waveforms during go/no-go RT tasks and found the following no-go-specific power and phase dynamics: theta band EEG power increases and intertrial phase-locking, which are associated with the whole midline-frontal no-go N2 and no-go P3 waveforms, and alpha band EEG intertrial phase-locking with an adjacent dephasing phenomenon, which mainly contributes to the early part of the midline-frontal no-go N2 waveform. These no-go-specific power and phase dynamics presumably involve the voluntary response inhibition process. Moreover, we have confirmed that the no-go-specific theta-to-alpha band phase dynamics in the midline-frontal areas just before and around the no-go N2 peak latency are affected by no-go stimulus probability, whereas the power dynamics are not affected, suggesting that the phase dynamics reflect the probability-dependent processes in human voluntary response inhibition.

Acknowledgments

This work was supported by the Grant-in-Aid for Young Scientists (B) of the Ministry of Education, Culture, Sports, Science and Technology, Japan, and the Casio Science Promotion Foundation (K. Y.). We thank A. Delorme & S. Makeig for their Matlab scripts in the open source EEGLAB Toolbox (Delorme & Makeig, 2004). We also thank K. Kitajo of the RIKEN Brain Science Institute and the anonymous reviewers for their helpful comments on this manuscript.

Reprint requests should be sent to Kentaro Yamanaka, Department of Health Design, Showa Women's University, 1-7 Taishido, Setagaya-ku, Tokyo, 154-8533 Japan, or via e-mail: kentaro@swu.ac.jp.

REFERENCES

REFERENCES
Alegre
,
M.
,
Gurtubay
,
I. G.
,
Labarga
,
A.
,
Iriarte
,
J.
,
Valencia
,
M.
, &
Artieda
,
J.
(
2004
).
Frontal and central oscillatory changes related to different aspects of the motor process: A study in go/no-go paradigms.
Experimental Brain Research
,
159
,
14
22
.
Aron
,
A. R.
,
Robbins
,
T. W.
, &
Roldrack
,
R. A.
(
2004
).
Inhibition and the right inferior frontal cortex.
Trends in Cognitive Sciences
,
8
,
170
177
.
Babiloni
,
C.
,
Carducci
,
F.
,
Cincotti
,
F.
,
Rossini
,
P. M.
,
Neuper
,
C.
,
Pfurtscheller
,
G.
,
et al
(
1999
).
Human movement-related potentials vs desynchronization of EEG alpha rhythm: A high-resolution EEG study.
Neuroimage
,
10
,
658
665
.
Bekker
,
E. M.
,
Kenemans
,
J. L.
, &
Verbaten
,
M. N.
(
2004
).
Electrophysiological correlates of attention, inhibition, sensitivity and bias in a continuous performance task.
Clinical Neurophysiology
,
115
,
2001
2013
.
Bekker
,
E. M.
,
Kenemans
,
J. L.
, &
Verbaten
,
M. N.
(
2005
).
Source analysis of the N2 in a cued Go/NoGo task.
Cognitive Brain Research
,
22
,
221
231
.
Benjamini
,
Y.
, &
Hochberg
,
Y.
(
1995
).
Controlling the false discovery rate: A practical and powerful approach to multiple testing.
Journal of the Royal Statistical Society: Series B, Methodological
,
57
,
289
300
.
Bokura
,
H.
,
Yamaguchi
,
S.
, &
Kobayashi
,
S.
(
2001
).
Electrophysiological correlates for response inhibition in a Go/NoGo task.
Clinical Neurophysiology
,
112
,
2224
2232
.
Bruin
,
K. J.
, &
Wijers
,
A. A.
(
2002
).
Inhibition, response mode, and stimulus probability: A comparative event-related potential study.
Clinical Neurophysiology
,
113
,
1172
1182
.
Debener
,
S.
,
Makeig
,
S.
,
Delorme
,
A.
, &
Engel
,
A. K.
(
2005
).
What is novel in the novelty oddball paradigm? Functional significance of the novelty P3 event-related potential as revealed by independent component analysis.
Cognitive Brain Research
,
22
,
309
321
.
Delorme
,
A.
, &
Makeig
,
S.
(
2004
).
EEGLAB: An open source toolbox for analysis of single-trial EEG dynamics including independent component analysis.
Journal of Neuroscience Methods
,
134
,
9
21
.
Donders
,
F. C.
(
1969
).
On the speed of mental processes.
Acta Psychologica
,
30
,
412
431
.
Eimer
,
M.
(
1993
).
Effects of attention and stimulus probability on ERPs in a Go/Nogo task.
Biological Psychology
,
35
,
123
138
.
Falkenstein
,
M.
,
Hoormann
,
J.
, &
Hohnsbein
,
J.
(
1999
).
ERP components in Go/Nogo tasks and their relation to inhibition.
Acta Psychologica
,
101
,
267
291
.
Fell
,
J.
,
Dietl
,
T.
,
Grunwald
,
T.
,
Kurthen
,
M.
,
Klaver
,
P.
,
Trautner
,
P.
,
et al
(
2004
).
Neural bases of cognitive ERPs: More than phase reset.
Journal of Cognitive Neuroscience
,
16
,
1595
1604
.
Filipović
,
S. R.
,
Jahanshahi
,
M.
, &
Rothwell
,
J. C.
(
2000
).
Cortical potentials related to the nogo decision.
Experimental Brain Research
,
132
,
411
415
.
Fuentemilla
,
L.
,
Marco-Pallarés
,
J.
, &
Grau
,
C.
(
2006
).
Modulation of spectral power and of phase resetting of EEG contributes differently to the generation of auditory event-related potentials.
Neuroimage
,
30
,
909
916
.
Garavan
,
H.
,
Ross
,
T. J.
, &
Stein
,
E. A.
(
1999
).
Right hemispheric dominance of inhibitory control: An event related functional MRI study.
Proceedings of the National Academy of Sciences, U.S.A.
,
96
,
8301
8306
.
Gemba
,
H.
, &
Sasaki
,
K.
(
1989
).
Potential related to no-go reaction of go/no-go hand movement task with color discrimination in human.
Neuroscience Letters
,
101
,
263
268
.
Gruber
,
W. R.
,
Klimesch
,
W.
,
Sauseng
,
P.
, &
Doppelmayr
,
M.
(
2005
).
Alpha phase synchronization predicts P1 and N1 latency and amplitude size.
Cerebral Cortex
,
15
,
371
377
.
Hanslmayr
,
S.
,
Klimesch
,
W.
,
Sauseng
,
P.
,
Gruber
,
W.
,
Doppelmayr
,
M.
,
Freunberger
,
R.
,
et al
(
2007
).
Alpha phase reset contributes to the generation of ERPs.
Cerebral Cortex
,
17
,
1
8
.
Humberstone
,
M.
,
Sawle
,
G. V.
,
Clare
,
S.
,
Hykins
,
J.
,
Coxon
,
R.
,
Bowtell
,
R.
,
et al
(
1997
).
Functional magnetic resonance imaging of single motor events reveals human presupplementary motor area.
Annals of Neurology
,
42
,
632
637
.
Hummel
,
F.
,
Andres
,
F.
,
Altenmüller
,
E.
,
Dichgans
,
J.
, &
Gerloff
,
C.
(
2002
).
Inhibitory control of acquired motor programmes in the human brain.
Brain
,
125
,
404
420
.
Klimesch
,
W.
,
Sauseng
,
P.
, &
Hanslmayr
,
S.
(
2007
).
EEG alpha oscillations: The inhibition-timing hypothesis.
Brain Research Reviews
,
53
,
63
88
.
Klimesch
,
W.
,
Schack
,
B.
,
Schabus
,
M.
,
Doppelmayr
,
M.
,
Gruber
,
W.
, &
Sauseng
,
P.
(
2004
).
Phase-locked alpha and theta oscillations generate the P1–N1 complex and are related to memory performance.
Cognitive Brain Research
,
19
,
302
316
.
Konishi
,
S.
,
Nakajima
,
K.
,
Uchida
,
I.
,
Kikyo
,
H.
,
Kameyama
,
M.
, &
Miyashita
,
Y.
(
1999
).
Common inhibitory mechanism in human inferior prefrontal cortex revealed by event related functional MRI.
Brain
,
122
,
981
991
.
Lavric
,
A.
,
Pizzagalli
,
D. A.
, &
Forstmeier
,
S.
(
2004
).
When “go” and “nogo” are equally frequent: ERP components and cortical tomography.
European Journal of Neuroscience
,
20
,
2483
2488
.
Leocani
,
L.
,
Toro
,
C.
,
Manganotti
,
P.
,
Zhuang
,
P.
, &
Hallett
,
M.
(
1997
).
Event-related coherence and event-related desynchronization/synchronization in the 10 Hz and 20 Hz EEG during self-paced movements.
Electroencephalography and Clinical Neurophysiology
,
104
,
199
206
.
Leocani
,
L.
,
Toro
,
C.
,
Zhuang
,
P.
,
Gerloff
,
C.
, &
Hallett
,
M.
(
2001
).
Event-related desynchronization in reaction time paradigms: A comparison with event-related potentials and corticospinal excitability.
Clinical Neurophysiology
,
112
,
923
930
.
Liddle
,
P. F.
,
Kiehl
,
K. A.
, &
Smith
,
A. M.
(
2001
).
Event-related fMRI study of response inhibition.
Human Brain Mapping
,
12
,
100
109
.
Makeig
,
S.
,
Debener
,
S.
,
Onton
,
J.
, &
Delorme
,
A.
(
2004
).
Mining event-related brain dynamics.
Trends in Cognitive Sciences
,
8
,
204
210
.
Makeig
,
S.
,
Delorme
,
A.
,
Westerfield
,
M.
,
Jung
,
T.-P.
,
Townsend
,
J.
,
Courchesne
,
E.
,
et al
(
2004
).
Electroencephalographic brain dynamics following manually responded visual targets.
PLoS Biology
,
2
,
747
762
.
Makeig
,
S.
,
Westerfield
,
M.
,
Jung
,
T.-P.
,
Enghoff
,
S.
,
Townsend
,
J.
,
Courchesne
,
E.
,
et al
(
2002
).
Dynamic brain sources of visual evoked responses.
Science
,
295
,
690
694
.
Mazaheri
,
A.
, &
Jensen
,
O.
(
2006
).
Posterior α activity is not phase-reset by visual stimuli.
Proceedings of the National Academy of Sciences, U.S.A.
,
103
,
2948
2952
.
Miller
,
J. O.
, &
Low
,
K.
(
2001
).
Motor processes in simple, Go/No-Go, and choice reaction time tasks: A psychophysiological analysis.
Journal of Experimental Psychology: Human Perception and Performance
,
27
,
266
289
.
Nakata
,
H.
,
Inui
,
K.
,
Nishihira
,
Y.
,
Hatta
,
A.
,
Sakamoto
,
M.
,
Kida
,
T.
,
et al
(
2004
).
Effects of a go/nogo task on event-related potentials following somatosensory stimulation.
Clinical Neurophysiology
,
115
,
361
368
.
Onton
,
J.
,
Delorme
,
A.
, &
Makeig
,
S.
(
2005
).
Frontal midline EEG dynamics during working memory.
Neuroimage
,
27
,
341
356
.
Pfefferbaum
,
A.
,
Ford
,
J. M.
,
Weller
,
B. J.
, &
Kopell
,
B. S.
(
1985
).
ERPs to response production and inhibition.
Electroencephalography and Clinical Neurophysiology
,
60
,
423
434
.
Pfurtscheller
,
G.
, &
Berghold
,
A.
(
1989
).
Patterns of cortical activation during planning of voluntary movement.
Electroencephalography and Clinical Neurophysiology
,
72
,
250
258
.
Pfurtscheller
,
G.
, &
Lopes da Silva
,
F. H.
(
1999
).
Event-related EEG/MEG synchronization and desynchronization: Basic principles.
Clinical Neurophysiology
,
110
,
1842
1857
.
Polich
,
J.
(
2007
).
Updating P300: An integrative theory of P3a and P3b.
Clinical Neurophysiology
,
118
,
2128
2148
.
Rinne
,
T.
,
Antila
,
S.
, &
Winkler
,
I.
(
2001
).
Mismatch negativity is unaffected by top–down predictive information.
NeuroReport
,
12
,
2209
2213
.
Rizzuto
,
D. S.
,
Madsen
,
J. R.
,
Bromfield
,
E. B.
,
Schulze-Bonhage
,
A.
,
Seelig
,
D.
,
Aschenbrenner-Scheibe
,
R.
,
et al
(
2003
).
Reset of human neocortical oscillations during a working memory task.
Proceedings of the National Academy of Sciences, U.S.A.
,
100
,
7931
7936
.
Rubia
,
K.
,
Russell
,
T.
,
Overmeyer
,
S.
,
Brammer
,
M. J.
,
Bullmore
,
E. T.
,
Sharma
,
T.
,
et al
(
2001
).
Mapping motor inhibition: Conjunctive brain activations across different versions of Go/No-Go and stop tasks.
Neuroimage
,
13
,
250
261
.
Rubia
,
K.
,
Smith
,
A. B.
,
Brammer
,
M. J.
, &
Taylor
,
E.
(
2003
).
Right inferior frontal cortex mediates response inhibition while mesial prefrontal cortex is responsible for error detection.
Neuroimage
,
20
,
351
358
.
Rushworth
,
M. S. F.
,
Walton
,
M. E.
,
Kennerley
,
S. W.
, &
Bannerman
,
D. M.
(
2004
).
Action sets and decisions in the medial frontal cortex.
Trends in Cognitive Sciences
,
8
,
410
417
.
Sauseng
,
P.
,
Klimesch
,
W.
,
Gruber
,
W. R.
,
Hanslmayr
,
S.
,
Freunberger
,
R.
, &
Doppelmayr
,
M.
(
2007
).
Are event-related potential components generated by phase resetting of brain oscillations? A critical discussion.
Neuroscience
,
146
,
1435
1444
.
Shah
,
A. S.
,
Bressler
,
S. L.
,
Knuth
,
K. H.
,
Ding
,
M.
,
Mehta
,
A. D.
,
Ulbert
,
I.
,
et al
(
2004
).
Neural dynamics and the fundamental mechanisms of event-related brain potentials.
Cerebral Cortex
,
14
,
476
483
.
Simson
,
R.
,
Vaughan
,
H. G.
, Jr., &
Ritter
,
W.
(
1977
).
The scalp topography of potentials in auditory and visual go/nogo tasks.
Electroencephalography and Clinical Neurophysiology
,
43
,
864
875
.
Stančák
,
A.
, Jr.,
Feige
,
B.
,
Lücking
,
C. H.
, &
Kristeva-Feige
,
R.
(
2000
).
Oscillatory cortical activity and movement-related potentials in proximal and distal movements.
Clinical Neurophysiology
,
111
,
636
650
.
Tallon-Baudry
,
C.
,
Bertrand
,
O.
,
Delpuech
,
C.
, &
Pernier
,
J.
(
1996
).
Stimulus specificity of phase-locked and non-phase-locked 40 Hz visual responses in human.
Journal of Neuroscience
,
16
,
4240
4249
.
Wager
,
T. D.
,
Sylvester
,
C.-Y. C.
,
Lacey
,
S. C.
,
Nee
,
D. E.
,
Franklin
,
M.
, &
Jonides
,
J.
(
2005
).
Common and unique components of response inhibition revealed by fMRI.
Neuroimage
,
27
,
323
340
.
Watanabe
,
J.
,
Sugiura
,
M.
,
Sato
,
K.
,
Maeda
,
Y.
,
Matsue
,
Y.
,
Fukuda
,
H.
,
et al
(
2002
).
The human prefrontal and parietal association cortices are involved in NO-GO performances: An event-related fMRI study.
Neuroimage
,
17
,
1207
1216
.
Yamanaka
,
K.
,
Kimura
,
T.
,
Miyazaki
,
M.
,
Kawashima
,
N.
,
Nozaki
,
D.
,
Nakazawa
,
K.
,
et al
(
2002
).
Human cortical activities during Go/NoGo tasks with opposite motor control paradigms.
Experimental Brain Research
,
142
,
301
307
.