Spontaneous pre-stimulus fluctuations in the activity of right fronto-parietal areas influence inhibitory control performance.

Inhibitory control refers to the ability to suppress planned or ongoing cognitive or motor processes. Electrophysiological indices of inhibitory control failure have been found to manifest even before the presentation of the stimuli triggering the inhibition, suggesting that pre-stimulus brain-states modulate inhibition performance. However, previous electrophysiological investigations on the state-dependency of inhibitory control were based on averaged event-related potentials (ERPs), a method eliminating the variability in the ongoing brain activity not time-locked to the event of interest. These studies thus left unresolved whether spontaneous variations in the brain-state immediately preceding unpredictable inhibition-triggering stimuli also influence inhibitory control performance. To address this question, we applied single-trial EEG topographic analyses on the time interval immediately preceding NoGo stimuli in conditions where the responses to NoGo trials were correctly inhibited [correct rejection (CR)] vs. committed [false alarms (FAs)] during an auditory spatial Go/NoGo task. We found a specific configuration of the EEG voltage field manifesting more frequently before correctly inhibited responses to NoGo stimuli than before FAs. There was no evidence for an EEG topography occurring more frequently before FAs than before CR. The visualization of distributed electrical source estimations of the EEG topography preceding successful response inhibition suggested that it resulted from the activity of a right fronto-parietal brain network. Our results suggest that the fluctuations in the ongoing brain activity immediately preceding stimulus presentation contribute to the behavioral outcomes during an inhibitory control task. Our results further suggest that the state-dependency of sensory-cognitive processing might not only concern perceptual processes, but also high-order, top-down inhibitory control mechanisms.

Introduction

Inhibitory control, the ability to suppress planned or ongoing cognitive or motor processes is necessary to ensure flexible and adapted goal-directed behavior in ever-changing environments (Aron, 2007; Dillon and Pizzagalli, 2007). Converging functional neuroimaging, transcranial magnetic stimulation and lesion data indicate that inhibitory control relies on a cortico-subcortical network involving the right inferior frontal gyrus (rIFG), the pre-supplementary motor area (SMA) and the basal ganglia (Garavan et al., 1999; Aron et al., 2003; Chambers et al., 2009; Majid et al., 2012), and manifesting at latencies of 150–400 ms after the onset of the stimuli associated with the inhibition goals (Kaiser et al., 2006; Smith and Douglas, 2011).

Although inhibitory control performance mostly depends on how the stimuli triggering the inhibition are processed, mounting evidence indicates that the cognitive- and brain-states preceding the presentation of these stimuli also contribute to the success of the inhibition. In Eriksen flanker tasks, contrasts between event-related potentials (ERPs) time-locked to the motor responses to trial preceding error vs. accurate trials revealed a specific ERP component peaking 50–100 ms post-response onset over frontal electrode sites [The “error related positivity” (EPP) Ridderinkhof et al., 2003; Allain et al., 2004; Hajcak et al., 2005]. Britz and Michel (2010) extended these results by showing that dorsolateral prefrontal cortices were engaged differentially during the 100 ms preceding the stimulus onset in error vs. correct trials during a classical color stroop task. These collective results suggest that errors are foreshadowed by a disruption of prefrontal task monitoring systems before the actual need for inhibitory control (see also Eichele et al., 2010; Masaki et al., 2012; Steinhauser et al., 2012).

Additional support for the critical role of pre-stimulus brain states in inhibitory control performance comes from studies in which the occurrence of the inhibition stimuli was predictable. In such cases, proactive inhibitory mechanisms are engaged before the stimulus presentation and interact with stimulus-driven reactive inhibitory mechanisms to eventually enhance inhibitory control (Claffey et al., 2010; Jahfari et al., 2010; Aron, 2011; Cai et al., 2012; Duque et al., 2012; Majid et al., 2012). By manipulating the degree of predictability of inhibition trials, these studies showed that when response inhibition can be prepared, effector-selective proactive inhibition mechanisms, mediated by the dorsolateral prefrontal cortex, are engaged to support fronto-basal reactive inhibition mechanisms (Aron, 2011 for review). Criaud et al. (2012) further demonstrated that proactive mechanisms persist once established at the beginning of each trial, suggesting that proactive inhibition has not only a transient effect but also modifies the general response mode of the participants. Taken together, these findings suggest that inhibitory control performance does not solely depend on how participants manage the conflict induced by the stimulus but also on the state of the task-monitoring and control systems before the trial.

However, the studies conducted so far on the state-dependency of inhibitory control were based on response- or stimulus-evoked ERPs and thus could not reveal whether spontaneous, not-time-locked fluctuations of ongoing brain activity preceding unpredictable inhibition-stimuli impact on inhibitory control performance. The averaging of the EEG signal in ERP studies indeed canceled out the activity not time-locked to the event of interest and thus dismissed this substantial fraction of the variability of the raw electrical brain activity (e.g., Arieli et al., 1996).

Based on the current evidence for the state-dependency of behavioral and brain responses during various types of perceptual tasks (e.g., Lehmann et al., 1994; Ress et al., 2000; Fox et al., 2006; Fox and Raichle, 2007; Britz et al., 2009, 2011), we hypothesize that spontaneous brain states immediately preceding the presentation of inhibition-triggering stimuli might also influence the success or failure to inhibit responses during inhibitory control tasks. To address this question, we used well-established methods of single-trial topographic analyses of EEG to determine whether specific voltage topographies present at the moment of the onset of unpredictable were associated with correct rejection (CR) vs. false alarms (FAs) to NoGo stimuli during a Go/NoGo task (Lehmann et al., 1994; Kondakor et al., 1995, 1997; Koenig et al., 1999; Muller et al., 2005; Britz et al., 2009; Brodbeck et al., 2012). The current investigation was based on a reanalysis of the data from Manuel et al. (2010) in which EEG was recorded in eleven healthy participants during an auditory spatial Go/NoGo task.

Material and methods

Participants

Eleven healthy volunteers participated in the study, all male and right-handed (Oldfield, 1971), aged 22–39 years (mean ± SD, 29.4 ± 1.6 years). Each participant provided written, informed consent to participate in the study. No participant had a history of neurological or psychiatric illness, and all reported normal hearing. All procedures were approved by the Ethics Committee of the Faculty of Biology and Medicine of the Vaudois University Hospital Center and University of Lausanne.

Stimuli

Auditory stimuli were 150-ms noise bursts (200–500 Hz bandpass filtered; 5 ms rise/fall), lateralized by means of a right- or left-ear leading interaural time difference of 770 μ s resulting in a perceived lateralization of ~80° from the central midline (Blauert, 1997). The sounds were presented via ER-4P Etymotic earphones.

Procedure and task

The current study is based on a reanalysis of the data from Manuel et al. (2010), in which the procedure and task are already detailed; we thus present only the main task parameters here.

Participants were seated in an electrically shielded and sound-attenuated booth in front of a 19-in screen. Stimulus delivery and response recording were controlled using E-prime 2.0.

The paradigm comprised an auditory spatial Go/NoGo task in which participants had to respond as quickly as possible via a button to left-lateralized sounds (Go stimuli, hereafter termed LG) and to withhold responses to right-lateralized sounds (NoGo stimuli, termed RNG). Each trial started with the presentation of a centrally presented gray cross on a black background for a randomly determined duration ranging from 1000 to 1900 ms. At the same time that the cross disappeared, the LG and RNG sounds were presented and response collection window was opened. In the Go conditions, a feedback on accuracy and response speed was provided immediately after the response. LG and RNG trials were presented with an equal probability of 0.5.

The experiment was divided into three sessions. Each session started with a calibration block of 16 randomly presented trials (eight LG and eight RNG), followed by two test blocks of 80 randomly presented trials each (40 LG and 40 RNG). The calibration blocks were used to individually adjust the task difficulty and to maintain time pressure across the whole experiment. During each calibration phase, the mean response time (RT) to LG trials was calculated online and used to determine the individual participant's RT threshold (RTt), which was set at 80% of the mean RT from the calibration block. During the test block, a Go response RT was considered as correct if it was below the 80% RTt of the immediately preceding calibration phase. Otherwise, a feedback screen indicating “too late!” was displayed immediately after the Go response (slow hit). On each trial, the mean percentage of correct trials, including fast hit and CR, was displayed. Participants were not informed about this thresholding procedure. Except for the global accuracy, no visual feedback was displayed after fast hits or FAs (see Vocat et al., 2008 for a similar procedure). The whole Go/NoGo training session included a total of 528 stimuli (160 stimuli in the test block + 16 stimuli in the calibration block × 3 sessions = 528) and lasted for a total of ~ 35 min. After the completion of each session, a rest period of 10 min was provided to participants.

EEG acquisition and preprocessing

Continuous EEG was acquired at 1024 Hz through a 128-channel Biosemi ActiveTwo system referenced to the common mode sense/driven right leg ground. All the EEG analyses were conducted with the Cartool software (Brunet et al., 2011). Before the single trial analyses, data at artifact electrodes from each participant were interpolated (Perrin et al., 1987). EEG epochs of the 50 ms preceding the stimulus onset were extracted from the raw EEG data, for each participant, for RNG CR and for the RNG FAs conditions. Because there was less FA than CR (see the behavioral results), we balanced the number of epochs included in each condition before the analyses to ensure that any potential differences between the two conditions did not follow from difference in statistical power. First, the same number of CR as FA epochs was randomly extracted for each participant separately. On the resulting epochs, a ±80 μV artifact rejection criterion was applied to exclude trials with eye blinks or other artifacts. When necessary, the number of epochs was then again balanced across conditions to eventually result in the inclusion of 25.4 ± 16.8 (mean ± SD) epochs in each condition for the single-trial analyses.

Behavioral data analyses

Behavioral data were analyzed to determine whether FA commission occurred randomly within the sequence of NoGo stimuli (see the Discussion section). We analyzed the pattern of FA occurrence by calculating the temporal auto-correlation function of the FA response type. If there was a relationship between FA occurrence at one trial with FA occurrence at previous or subsequent trials, as could for instance occur during periods of decrease in attention inducing series of FA, it should manifest as an increase in the autocorrelation coefficient (Britz et al., 2009, 2011; Bernasconi et al., 2011). We first performed a binary classification on the FAs and the Hits and Misses (respectively responded and missed Go trials), as well as CR of NoGo trials. To determine if the patterns of FA of the participants were different from a random distribution, we permuted 1000 times the sequence of each participant and then compared the autocorrelation coefficients of the sequence of the participant to the distribution of the autocorrelation coefficients of the randomized sequence of the participant. The autocorrelation coefficients of the FA occurrence were computed for each trial n with that in trial n + m for m = 1 − 20.

Single-trial topographic analyses

The time-locked averaging of the EEG signal across multiple repetitions of an event (e.g., the presentation of a stimulus or a behavioral response) to build ERPs cancels out all the fluctuations of brain activity not time-locked to the event of interest, because by definition, the phase of these fluctuations varies across trials. To circumvent this problem and to investigate whether non-stimulus-locked variations in the prestimulus activity impact on inhibitory control proficiency, we utilized previously published methods of single-trial EEG analysis (Lehmann et al., 1994; Kondakor et al., 1995, 1997; Koenig et al., 2002; Mohr et al., 2005; Britz et al., 2009; see for review Britz and Michel, 2010; Eichele et al., 2010; Steinhauser et al., 2012). This approach is based on evidence that evoked and induced ongoing EEG signal is not random, but rather organized in a succession of quasi-stable topographies of the electric potentials. Since the configuration of the electric potentials at the scalp reflects the sum of all active sources in the brain at each moment in time (Lehmann et al., 1987), the observation of stable topographies or “maps” suggest that during these time intervals, the functional state of the brain is stable (e.g., Michel et al., 2009). The short periods of functional stability have been referred to as “functional microstate” and typically comprise 80–120 long segments of stable configuration of the scalp-recorded voltage topography of the electric potential (Lehmann and Skrandies, 1980; Koenig et al., 2002). While EEG topography remains stable during a microstate, the strength of the electric potentials, as indexed, for example, by the Global Field Power (GFP), increases and then decreases. The GFP represents a single-number index of the strength of electric potentials; it is calculated as the spatial standard deviation of the electric potentials: the square root of the sum of all squared potentials divided by the number of electrodes (Lehmann and Skrandies, 1980; Murray et al., 2008). Compelling relationships have been found between spontaneously occurring microstates preceding the presentation of a stimulus and the behavioral and brain response to this stimulus (Lehmann et al., 1994; Kondakor et al., 1995, 1997; Mohr et al., 2005; Muller et al., 2005; Britz et al., 2009), suggesting that microstate-based single-trial EEG topographic analyses enable reliable investigations of the effects of variations in the raw ongoing EEG signal.

The single-trial topographic analysis comprises the following processing steps, which mainly consist in reducing the raw EEG data into a limited number of common stable microstates within and then across participants (for similar procedures see also Koenig et al., 2002; Mohr et al., 2005; Britz et al., 2009, 2011).

The first step involved determining the topographies differentiating the FAs and CR conditions for each participant: (1) We extracted from each EEG epoch, for each participant and each condition the topography manifesting at the single time frame when the GFP was maximal (i.e., when the signal-to noise ratio of the microstate was the highest) within the 50 ms pre-stimulus onset. We selected this pre-stimulus time frame because it corresponds to the half of the duration of an average microstate and thus this procedure enables to extract topography at the time frame best representing the microstate present at the moment when the stimulus is presented (the total length of a microstate ranges between 80 to 120 ms; Britz et al., 2011); (2) For each participant separately, we applied a spatial k-means cluster analysis (e.g., Pascual-Marqui et al., 1995) on these topographies to identify the most dominant topographies among all topographies extracted in step (1). (3) The optimal number of clusters, or “template topographies,” was determined based on a modified Krzanowski-Lai criterion (Krzanowski and Lai, 1988); (4) A fitting procedure, in which each original topography of step (1) was relabeled with the template map of step (2) with which it best correlated (Pegna et al., 1997), enabled us to identify the two template topographies that best differentiated the FA and CR conditions: the template map with the highest frequency of occurrence in the FA and in the CR conditions were retained.

The second step comprised determining the topographies differentiating the FA and CR across participants: (5) We applied a second k-mean cluster analysis on the groups of FA and CR map identified during the first step for each participant and back-fitted the resulting template topographies on the original data using the same procedure as steps (1), (2), and (3). The statistical comparison between the frequency of occurrence of the maps in the FA and CR conditions revealed the topographies differentiating the two conditions at the group level.

Electrical source estimations

Electrical sources underlying the template map(s) of interest were estimated using a distributed linear inverse solution based on a local autoregressive average (LAURA) regularization approach (Grave De Peralta Menendez et al., 2001; Grave-De Peralta et al., 2004; see also Michel et al., 2004 for a comparison of inverse solution methods) implemented in the Cartool software (Brunet et al., 2011). LAURA selects the source configuration that mimics the biophysical behavior of electric potential (i.e., activity at one point depends on the activity at neighboring points). The solution space is based on a realistic head model and included 3005 solution points homogeneously distributed within the gray matter of the average brain of the Montreal Neurological Institute (courtesy of R. Grave de Peralta Menendez and S. Gonzalez Andino, University Hospital of Geneva, Geneva, Switzerland).

Results

Behavior

Reaction time to Go stimuli was 251 ± 21 ms (mean ± SEM). The percentage of FAs was 10.9 ± 0.95% (Manuel et al., 2010).

The autocorrelation coefficients of the sequence of FA of the participants were all between z = −0.37 and 0.64 SD from the mean of the coefficients from the randomized sequences, indicating no significant (−1.96 < z < 1.96; p > 0.05) differences between the distributions of FA in the random vs. actual sequences for lags 1–20 (Figure 1).

Figure 1

Analysis of the sequence of false alarms. For lags 1–20, the maximal (blue), mean (green), and minimal (red) autocorrelation coefficients are represented, for the sequences of the participants (plain lines) and for randomized sequences (dotted lines). The sequences of FA from the participant did not differ statistically from the random sequences (see the Results section), supporting that random fluctuations in the pre-stimulus period may account for FA occurrence.

Consistent with previous applications of the current single-trial topographic analysis (Britz et al., 2009, 2011), the k-means cluster analysis applied during the first step of the analysis (see the Method section) identified on average 5.9 ± 1.6 maps for each participant as optimally explaining the data. These template maps accounted for 72.7 ± 7.3% of the global explained variance. The maps were then grouped for each conditions across participants and a second cluster analysis was applied. The best clustering during the second step explained 93% of the variance with 11 template maps. The resulting template maps were then back-fitted to the original individual subject data for each conditions to determine their frequency of occurrence. Two-tailed pairwise t-tests were conducted to compare the frequency of occurrence of each template map between the CR and FA conditions. A single topographic map was found to differ in its frequency of occurrence between the FA and CR conditions (Figure 2, violet square): This topography occurred significantly more often in the CR than FA condition [t(10) = 2.83; p = 0.018; d = 0.9; Figure 2]. To test for the probability of type 1 errors, we generated 1000 randomized permutations of the raw data maps and computed, for each of the permutation, the same t-tests as those we applied to the actual data. The results revealed a probability of 1.7% to have the same pattern as in the real data (i.e., one p-values < 0.019 among the 11 tests) in the 1000 permutations, i.e., when there was no structure in the data.

Figure 2

Topographies of the prestimulus microstates as revealed by the cluster analysis. The map that significantly differentiated the correct rejections is framed in violet. The red lines indicate the t-test p < 0.05 significance threshold and an asterisk is placed where significance has been reached.

The estimation of the electric sources at the origin of the template topography differentiating the CR from the FA condition showed that it resulted from the activation of a right hemispheric parieto-frontal network extending from the inferior parietal lobule to the frontal gyri (Figure 3). The largest hub of activation was centered around the supramarginal gyrus/primary auditory cortex. Importantly, this result is only a visualization of the sources underlying the EEG topography and not the result of a statistical contrast; it should thus be interpreted with caution.

Figure 3

LAURA distributed source estimations corresponding to the topography identified as best characterizing the pre-stimulus period before correct rejections identified a right hemispheric fronto-temporal prestimulus activation.

Discussion

The present study suggests that performance in a classical Go/NoGo task is influenced by the momentary state of the ongoing brain activity immediately preceding the onset of unpredictable NoGo stimuli. A specific EEG voltage topography manifested more frequently before successful response-inhibition to NoGo stimuli than before FAs. There was no evidence for topographies specifically preceding FA trials. Electrical source estimations localized the source of the topography preceding successful inhibition within a right fronto-temporal network.

Our results contribute to current knowledge on the influence of pre-stimulus brain state on inhibitory control by indicating that inhibition performance is modulated by the momentary brain state present when the stimuli are presented. Fluctuations in the ongoing brain activity have been shown to represent almost 90% of the variability in the EEG signal (Eichele et al., 2010). This source of variability was mostly dismissed in previous ERPs studies on the effect of pre-stimulus activity because time-locked averaging of the EEG signal were applied to extract ERPs (e.g., Hajcak et al., 2005; Masaki et al., 2012); in such procedures, the signal not time-locked to the event is cancelled out and usually considered as background physiological noise. In line with our findings, mounting evidence indicate that ongoing spontaneous EEG fluctuations are actually functionally relevant and account for a substantial fraction of the variability in the behavioral and brain responses to stimuli in various experimental contexts (Picton et al., 2000; O'Connell et al., 2009 for the role of pre-stimulus endogenous modulation in oscillatory activity in task performance; Britz and Michel, 2010 for discussion). However, the dependency of sensory-cognitive processes to single trial variability in brain activity was so far demonstrated in perceptual but not executive tasks. For instance, Mohr et al. (2005) showed that a specific topographic map with a left anterior-right posterior dipole orientation spontaneously manifesting before word presentation predicted an enhanced discrimination of emotional word when the word was presented in the left visual field but only for men during a bilateral lexical decision task. Lehmann et al. (1994) and Kondakor et al. (1995, 1997) further demonstrated that the brain responses to identical tones depended on the spontaneously occurring topography preceding their presentation. Corroborating and extending these findings, our results suggest that spontaneous fluctuations of ongoing brain activity not only modulate perceptual processes but also high-order, top-down executive mechanisms as those supporting inhibitory control in conflict tasks.

The sources of the scalp topography preceding more frequently CR than FAs were maximal within a right fronto-parietal network. The right rIFG has been repeatedly involved in inhibitory control by functional neuroimaging (Garavan et al., 1999; Rubia et al., 2003; Aron and Poldrack, 2006; Aron et al., 2007) and lesion studies (Decary and Richer, 1995; Aron et al., 2003; Rieger et al., 2003; Floden and Stuss, 2006; Picton et al., 2007). The rIFG is thought to trigger motor inhibition via its connections to the subthalamic nucleus (STN; Inase et al., 1999; Aron et al., 2007; Aron, 2011). Most of these studies, however, involved the rIFG in inhibitory control by contrasting the post-stimulus brain responses to stimuli associated with vs. without response inhibition. In the current study, we showed that random fluctuations in the activity of rIFG before the demand for inhibition also play a critical role in inhibitory control proficiency. In conditions when the onset of the NoGo stimuli cannot be predicted, if the NoGo stimuli are presented when the activity of the rIFG is high, the probability of a correct inhibition increases. Although speculative, an account for this effect could be that a pre-activation of the rIFG increases the speed of inhibition process because less time is needed to reach the elicitation threshold of the inhibitory command from frontal to subcortical structures.

Right inferior parietal areas have been previously involved in response inhibition (Garavan et al., 1999; Liddle et al., 2001; Menon et al., 2001; Rubia et al., 2003) or response conflict resolution (Braver et al., 2001; Van Veen et al., 2001). These studies suggest that inferior parietal cortices mediate attention to the task and might thus, in turn, modulate performance (see also Hampshire et al., 2010). Increase in the activity of the precentral gyrus in condition of successful inhibition has also been reported in stop-signal task (though more superior as in the current study; Li et al., 2006), and have been interpreted as “negative motor areas” (Ikeda et al., 2000; Yazawa et al., 2000), whose direct stimulation elicits response inhibition (Luders et al., 1995). Alternatively, since participants had to respond with their right hand, an activation of ipsilateral (right) motor areas might have inhibited contralateral, homotopic motor areas via interhemispheric inhibition mechanisms and in turn facilitated the rejection of NoGo stimuli (e.g., Ferbert et al., 1992). Hand and arm motor representations are, however, higher along the central sulcus as the area found in the current study. The precise role of the right temporal structures in inhibitory control is more difficult to delineate. These areas have been shown to interact with higher level prefrontal regions during inhibitory control tasks (Egner and Hirsch, 2005) and to be modulated in inhibition-related disorders (Tamm et al., 2004; Solanto et al., 2009), but their precise role remains unclear. An alternative account for the role of right temporal areas in FA commission would be their involvement in the processing of the auditory spatial features distinguishing Go from NoGo stimuli in the current task. Go and NoGo goals activation indeed first require discriminating whether the stimulus is presented on the left or right auditory hemifield. Right temporal areas have been repeatedly involved in the early stage of auditory spatial processing (Spierer et al., 2007, 2011) and their pre-activation could have facilitated the detection of NoGo stimuli and in turn response inhibition. In line with this hypothesis, Manuel et al. (2010) suggested an important role of early auditory analyses of the lateralization of the stimuli in inhibitory control proficiency during the current auditory spatial Go/NoGo task. Importantly, however, we would like to emphasize that our source estimations were only visualizations, rather than a statistical analysis, of the network likely underlying the topography preceding FA commission. These results should thus be interpreted with caution.

Although our pattern of result is highly consistent with current evidence for a crucial role of the rIGF in stimulus-induced inhibitory control processes, it contrasts with previous data on the brain regions whose pre-stimulus activity modulates inhibition proficiency. Electrophysiological studies on trials preceding error during an Eriksen flankers task identified a specific response-locked ERP component at 100 ms (EPP; Ridderinkhof et al., 2003; Allain et al., 2004; Hajcak et al., 2005). The EPP was specific to trials preceding an error and its amplitude was larger on frontal, central and parietal midline electrodes, compatible with a generator located within the anterior cingulate cortex (ACC). Using a stroop task, Britz and Michel (2010) showed a specific pattern of stimulus-locked fronto-parietal activity before erroneous vs. correct trials. The modulation of prefrontal activity before error commission found in the studies reviewed above was interpreted as reflecting a transient disruption of action monitoring and executive control systems. Our results complement these finding by demonstrating that in addition to the impact of maintaining a minimal level of action monitoring during the task found by response- or stimulus-locked ERP approaches, spontaneous ongoing activity within the networks mediating stimulus-induced response inhibition also influences inhibitory control performance.

In line with this assumption, growing evidence indicate that as soon as the need for response inhibition can be predicted, proactive inhibitory control mechanisms are engaged to facilitate the inhibitory control of task-irrelevant responses (Claffey et al., 2010; Jahfari et al., 2010; see Aron, 2011 for review; Cai et al., 2012; Criaud et al., 2012). “Active breaking” mechanisms would be engaged before the stimulus onset to decrease the activity of motor area and in turn facilitate response inhibition (Cai et al., 2012; Duque et al., 2012). Anticipatory mechanisms have been shown to rely on a fronto-striatal network partly overlapping with the estimated sources of the topography prominently associated with successful inhibition in the current study. We cannot rule out that the frequency of occurrence of the topography preceding successful inhibition was influenced by proactive preparatory mechanisms in the current study. According to this hypothesis, the anticipation of NoGo stimuli would have increased the frequency of occurrence of the “facilitatory” topography or, in other word, pre-activated inhibitory control mechanisms. Speaking against this hypothesis, there was no strategic advantage of intentionally maintaining inhibitory control networks active in our task because such decrease in the response elicitation threshold would have resulted in a general decrease in response speed to Go stimuli. In our paradigm, emphasis was put on response speed: slow Hits (correctly responded Go stimuli but above the RTt determined in calibration blocks) were considered as error and reported explicitly as such by the visual feedback provided after each trial. Although there was a probability of 0.5 for NoGo stimuli occurrence, response prepotency was maintained very high by the recurrent feedback on responses speed. The threshold for considering a response to Go trials to slow was indeed individually adjusted to 80% of the mean response speed calculated during the calibration blocks intervening between experimental blocks (see the Method section). Hence, we interpret our results as indicating that when the fronto-temporal brain network was already (spontaneously) activated before a NoGo stimulus was presented, inhibition performance was improved. Compatible with this hypothesis, the analyses of the distribution of FA occurrence across the experimental session showed that participants committed FA following the same pattern as when FA were randomly distributed within the sequence. The adoption of any strategy by the participant would have most likely resulted in non-random sequences consisting of series of FA or CR because a specific task set could unlikely be maintained constantly during the whole experiment (Palva and Palva, 2012).

Our results cannot disentangle the relationship between proactive inhibition and the occurrence of specific microstates. To our knowledge, no study has so far investigated whether top-down, strategic implementation of task sets can impact on the frequency of occurrence-specific microstate and thus whether proactive mechanism could have influenced our pattern of result. Further studies focusing on longer epochs and manipulating the engagement of proactive inhibitory mechanisms are necessary to elucidate this question.

Of note, we found a specific EEG topography preceding successful trials, but no specific topography preceding errors. This pattern of result suggests that while the engagement of specific cognitive set may enhance motor responses inhibition in conditions of high response prepotency, many different deviations from this optimal brain-state could result in inhibition failure, in turn explaining that we did not identify a specific topographic map preceding errors. In this regard, a limitation of the current study is that by defining our FA and CR conditions only based on RT (and lack thereof), we possibly included in the CR condition some trials in which erroneous motor activation was actually engaged but inhibited before reaching activation threshold. Referred to as “covert” errors, these partial-error trials can only be detected with the recording of electromyography activations and have been suggested to depend on distinct mechanisms as overt errors (Burle et al., 2002; Allain et al., 2004; Boulinguez et al., 2008). In any case, the inclusion of covert error trials in the CR condition would have increased noise in the data and thus most likely increased the probability of type 2 but not type 1 errors. Second, because FAs were quite rare in our data, our single trial topographic analyses included only a limited number of trials. While the capacity to deal with small numbers of EEG epochs is an advantage of this approach and we reach reliable statistical results, how the number of epochs impacts the statistical outcome of our approach remains to be determined.

Collectively, our results point out that spontaneous fluctuations within inhibitory control networks at the moment when NoGo stimuli are presented influence inhibition performance. The question remains open, however, as to whether and how prestimulus modulations impact on the processing of the inhibition-related stimuli. It would be notably interesting to link directly the current results with previous literature on pre-target predictor of inhibition performance and to examine potential differences in the early sensory processing of the NoGo stimuli between FA and CR trials. Further studies comparing the pre-stimulus with the post-stimulus activity would be necessary to elucidate this question.

Conflict of interest statement

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.