Response inhibition alterations in migraine: An event-related potential study

Migraine is characterized by a hypersensitivity to environmental stimulation which climaxes during headache attacks but persists during attack-free period. Despite ongoing debates about the nature of the mechanisms giving rise to this abnormality, the presence of decient inhibitory cortical processes has been proposed to be one possible mechanism underlying its pathogenesis. Empirical evidence supporting this notion is mainly based on previous ndings showing functional cortical hyperexcitability in the sensory domain. Considering that a general inhibitory control process can play an important role across early to later stage of information processing, this may in turn indicate the important role other dimensions of inhibitory control can play in migraine disability. To this end, the present study was designed to examine the pathophysiological basis of inhibitiory control that takes place during suppression of prepotent responses. P3], CPz, Pz], [C4, CP4, P4] N2; CP3, P3], [FCz, Cz, CPz, Pz], [FC4, C4, CP4, P4] P3) [C3, Cz, C4], CPz, CP4], parietal [P3, Pz, P4] for N2;frontocentral [FC3,


Introduction
Migraine is a common episodic neurological disorder which is mainly characterized by recurrent attacks of headache, which has a great detrimental in uence on quality of life (1, 2). The annual prevalence of migraine estimated in the worldwide population were around 9% and migraine predominantly affects females (3)(4)(5)(6), resulting in an increasing nancial burden on global economies (7,8). Given such a situation, considerable empirical efforts have been devoted into understanding the causes of migraine because it would help with identifying risk factors, establishing correct diagnosis and developing a range of effective therapeutic interventions (9). Although the exact pathogenetic mechanisms for migraine have not been completely elucidated, migraine, at its core, is suggested to be a complex brain disorder (10,11).
To search for evidence of brain dysfunction in migraineurs, measuring neural activities with electrophysiological methods (i.e., Electroencephalogram (EEG)) has been proven to be effective in improving our understanding on the physiopathology of the migraine brain (12)(13)(14)(15).
The majority of electrophysiological studies on stimulus responsivity in visual, auditory, somatosensory and nociceptive domains strongly suggest that migraine is associated with a state of functional cortical hyperexcitability after sensory stimulation and a lack of the physiological habituation to repeated sensory stimulation especially during the intervals between attacks (12,16,17). Despite ongoing debates about the nature of the mechanisms giving rise to these abnormalities (10,18), the presence of de cient inhibitory cortical processes has been argued to be one possible mechanism underlying the pathogenesis of these conditions (18,19). Available evidence supporting this claim has come mainly from previous studies on the cortical response to external sensory stimuli in migraineurs (12,16,20,21). For example, reliable differences in cortical excitability in response to sensory stimulation between migraineurs and healthy controls have been found, usually re ected by increased amplitude of evoked responses (19,22,23) and decreased activity of GABA-mediated inhibition of the sensory cortex (24). Given that inhibitory control does not take place only at the sensory level and can play an important role across early to later stage of information processing (25,26), it is tempting to understand the possible role of other dimensions of inhibition control, such as inhibition of an initiated response, can play in migraine disability. However, to date, it is somewhat surprising that few researchers have turned their attetion in this direction.
To address this issue, the present study was designed to examine neural activity supporting response inhibition by using a stop-signal task (SST) in combination with event-related potentials (ERPs) in patients with migraine without aura (MwoA) during the interictal period. The stop-signal task is a typical paradigm for measuring inhibitory control of an ongoing motor response (27,28). In this experimental paradigm, participants are instructed to respond quickly and e ciently to a primary Go stimulus while inhibiting their go responses when a Go stimulus is occasionally and unexpectedly followed by a stopsignal. Considering that the promise of ERP-based biomarkers of cognitive dysfunction has been increasingly recognized in psychiatric (29,30) and neurological disorders (31) including migraine (12,14,15), the usage of this technique can thus greatly help in examining how the brain detects the stop signal and decides to stop a prepotent motor response in migraineurs. Previous ERP studies have consistently identi ed two typical ERP components related to response inhibition: the stop-N2 and stop-P3. The stop-N2 refers to a negative wave that occurs 200-300 ms following a stop signal, with its maximum amplitude over frontocentral, central and centroparietal scalp regions, while the stop-P3 is a large positive wave that peaks around 300-600 ms following a stop signal, with its maximum amplitude over central and centroparietal locations (32). In spite of some debates about the functional signi cance of these two components, they are considered to re ect different sub-processes that underline response inhibition.
Speci cally, the stop-N2 is proposed to primarily re ect con ict monitoring during early, non-motoric stages of inhibition, while the stop-P3 is mainly thought to re ect late-stage inhibition of the motor system itself and cognitive evaluation of motor inhibition (32)(33)(34)(35). Given that common time-domain ERP measures may not accurately re ect multiple processes underlying response inhibition, the use of timefrequency (TF) analysis can provide complementary information about underlying processes behind the stop N2-P3 complex in response inhibition. Indeed, event -related theta (4-8 Hz) and delta (1-4 Hz) oscillations have been argued to index two separable, but highly overlapping processes underlying the stop N2-P3 complex in response inhibition, although their functional signi cance has yet to be fully clari ed (32,36). Given that de cits in sensory inhibition and inhibitory attentional control have been reported in migraineurs (12,37), one could expect that patients with MwoA display a de cit in response inhibition as evidenced by prolonged reaction times to a stop signal and by increased amplitudes of both stop-N2 and stop-P3. Moreover, time-frequency decomposition has been employed to characterize separable but overlapping processes underlying the stop N2-P3 complex in response inhibition in patients with MwoA.

Participants
In the present study, 22 patients with MwoA (age = 30.86 ± 5.69 years; 19 females) and 25 healthy controls (age = 30.24 ± 3.52 years; 18 females) were recruited. All participants had normal or corrected-tonormal vision and were right-handed. Both groups were matched on the basis of sex, age and years of education. For healthy controls, they reported no history of neurological or psychiatric disorders and no migraine in rst-degree relatives. For patients with MwoA, they underwent neurologic and physical evaluations by trained neurologists (Z.D. and S.Y.) as well as standard neuropsychological assessment by neuropsychologists (G.C.). The inclusion criteria for patients were used: 1) ful lling the diagnosed criteria for migraine without aura according to the International Classi cation of Headache Disorders, 3rd edition (ICHD-3); 2) at least 2 year's history of migraine and at least one migraine episode per month; 3) no migraine attacks 72 h before and after the experiment and 4) outside migraine attacks during the experiment (the interictal period). Moreover, the following exclusion criteria were used: 1) neurological diseases (i.e., epilepsy, neuromuscular disorders); 2) psychiatric symptoms (i.e., anxiety and depression); 3) mental retardation; 4) a current or past history of substance dependence, 5) receiving prophylactic antimigraine therapy and 6) having suicide ideation and/or previous suicide attempts; 7) the presence of periodic limb movement disorder (i.e., nocturnal hyperkinesias) and recurrent parasomnias (> 3 episodes per week). All female participants took no oral contraceptives for at least 1 week. Demographic and clinical characteristics are described in Table 1.

Stop-signal task
We used the SST to measure participants' response inhibition, which is similar to that described in previous studies (27,38). This task included 80% Go trials and 20% Stop trials (Fig. 1). On Go trials, participants were presented with a xation cross on a black computer screen lasting for 600-800 ms, which is immediately followed by a visual stimulus (the letter 'X' or 'O') (a Go signal) lasting for 1000 ms. Participants were instructed to judge the shape of the visual stimulus as accurately and quickly as possible via a button press with the index ngers of the left and right hands, which was counterbalanced across participants. On the remaining Stop trials, the stop stimulus (a red square appearing above the location of the go stimulus) appeared after the Go stimulus after a variable delay of 0-250 ms in steps of 50 ms (stop signal delay; SSD), cuing participants to withhold their responses to the Go stimulus. Each trial was followed by a blank screen for a variable intertrial interval (1500-2000 ms). There were totally four experimental blocks of 400 trials. The experiment was preceded by a short practice block. Task presentation was controlled via E-prime 2.0.
EEG data recording and analysis EEG data recording procedure is similar to that described in our previous studies (39,40). Speci cally, EEG was recorded (SynAmps ampli er, NeuroScan) with a quick cap carrying 64 Ag/AgCl electrodes placed at standard locations covering the whole scalp (the extended international 10-20 system). The reference electrode was attached to the right mastoid (M2), and the ground electrode was placed on the forehead. The vertical electrooculogram (VEOG) was recorded with electrodes placed above and below the left eye. The horizontal electrooculogram (HEOG) was recorded with electrodes placed beside the two eyes. The impedance was kept below 5 kΩ. The electrophysiological data were continuously recorded with a bandwidth 0.05-100 Hz and sampled at a rate of 1000 Hz.
O ine time-domain EEG data analysis was conducted using EEGLAB (41) and ERPLAB (42). Data were rst re-referenced to linked mastoid (M1 and M2). An independent component analysis (ICA)-based artifact correction was achieved by using the ICA function of EEGLAB. Independent components with topographies representing saccades, blinks, and heart rate artifact were thus removed according to published guidelines (43). The resultant EEG data were then epoched from 200 ms pre-stimulus to 1000 ms post-stimulus and digitally low pass ltered by 30 Hz (24 dB/octave). The 200 ms pre-stimulus period was used for baseline correction. In order to remove movement artifacts, epochs were rejected when uctuations in potential values exceeded ± 75 µV at any channels except the EOG channel. The ERPs were averaged separately for successful Stop trials and correct Go trials in each group.
Our time-frequency analysis was performed using the Matlab FieldTrip toolbox (44) and the procedures were similar to that described in a recent study (45). The ltered EEG data between 0.5-30 Hz was segmented 500 ms pre-stimulus onset to 1000 ms post-stimulus onset separately for Go and Stop trials for each group. Total event-related spectral power was obtained by transforming each epoch into the frequency domain using a sequential and overlapping unique Hanning window of 250 ms in steps of 25 ms with the multitaper time-frequency transformation (MTMCONVOL from ft_freqanalysis Fieldtrip software) method. In addition, the convolution function includes a 'Zero' type padding in order to cope with edging effects. After the transformation, we obtained a time-frequency spectrum with 1 Hz and 250 ms resolution. At each frequency, the results employed a dB transform [dB power = 10*log10 (power/baseline)] and were baseline corrected by subtracting the average baseline period (from − 200 to − 0 ms) from each data point. The obtained power values were then averaged over EEG epochs for trial types (i.e., Go trials and successful Stop trials) in each participant. Then, data were grand-averaged across MwoA patients and across healthy controls for each trial type.

Statistical analysis
For statistical analysis on the demographic data, a chi-square test was used to assess a between-group difference in sex ratio and independent sample t-tests were used to examine between-group differences in age, years of education, the rate of anxiety and depression as measured by the self-rating anxiety scale (SAS) and self-rating depresson scale (SDS), and body mass index (BMI). For statistical analysis on the behavioral data, accuracy on Go trials (Go ACC) and reaction times to Go stimuli (Go RT) and the Stop signal (stop signal reaction time: SSRT) were extracted for analyses. Independent sample t-tests were used to examine between-group differences in these behavioral data.
Regarding statistical analysis on electrophysiological data, our data were analyzed according to the topographical distribution of grand averaged ERP activity as well as the methods of previous ERP studies (3,32,34,46). The ERP statistical analysis involved two ERP indices of response inhibition: the stop-N2 and stop-P3. Mean amplitudes for the stop-N2 (time interval = 200-250 ms, at the C3, Cz, C4, CP3, CPz, CP4, P3, Pz, P4 electrodes) and the stop-P3 (time interval = 350-500 ms, at the FC3, FCz, FC4, C3, Cz, C4, CP3, CPz, CP4, P3, Pz, P4 electrodes) were calculated. In order to examine effects of migraine on these ERP components, we conducted a mixed analysis of variance (ANOVA), with group as a betweenparticipants factor (patients with MwoA versus healthy controls), and trial type (  (32), the same statistical analyses were conducted on eventrelated delta and theta power. Based on visual inspection of time-frequency plots and the methods of previous studies (45)(46)(47), the same area and laterality factors were included in such time-frequency analyses and mean power values in delta (1-4 Hz) and theta (4-8 Hz) frequency bands were extracted in the time windows used to extract mean amplitudes of the stop-N2 and stop-P3 components in order to disentangle the multiple processes underlying the stop N2-P3 complex in response inhibition.
All data were analyzed using IBM SPSS 19.0 (IBM Corp., Armonk, NY, USA). Statistical comparisons were made at p-values of p < 0.05, with the Greenhouse-Geisser correction when violations of sphericity occurred.

Participant demographic
There were no signi cant difference in age, sex ratio, education years, BMI, SAS, and SDS scores between MwoA and healthy controls (Table 1) The mixed ANOVA performed on the mean amplitude of the N2 revealed a signi cant main effect of trial type (F(1, 45) = 383.95, p < 0.001), with the N2 amplitude being larger on Stop trials than on Go trials (p < 0.001), athough there was not a signi cant main effect of group (F(1, 45) = 1.59, p = 0.21). Moreover, there was a signi cant interaction between group and trial type (F(1, 45) = 6.23, p < 0.05), which was due to larger amplitude of the N2 on Stop trials for patients with MwoA than for healthy controls (p < 0.05) (Fig. 2). There was also a signi cant main effect of area (F(2,90) = 8.00, p < 0.005), which was quanti ed by a signi cant interaction between area and trial type (F(2,90) = 7.53, p < 0.005). An analysis of simple effects revealed larger amplitude of the N2 in the central (p < 0.05) and centroparietal regions (p < 0.005) than in the parietal region on Stop trials, while larger amplitude of the N2 in the central region than in the centroparietal region (p < 0.001) on Go trials. Similarly, there was a signi cant main effect of laterality (F(2,90) = 53.87, p < 0.001), which was also quanti ed by a signi cant interaction with laterality and trial type (F(2,90) = 83.26, p < 0.001). An analysis of simple effects revealed larger amplitude of the N2 in the electrodes at the midline than that on the left (p < 0.001) and on the right (p < 0.001) on Stop trials and larger amplitude of the N2 in the electrodes on the right than that on the left on Stop trials (p < 0.001). No other signi cant effects were found.

P3 (350-500 ms)
The mixed ANOVA performed on the mean amplitude of the P3 revealed signi cant main effects of both group (F(1, 45) = 6.26, p < 0.05) and trial type (F(1, 45) = 330.19, p < 0.001), which was quanti ed by a signi cant interaction between group and trial type (F(1, 45) = 9.74, p < 0.005). An analysis of simple effects revealed larger amplitude of the P3 for patients with MwoA than for healthy controls on Stop trials (p < 0.005) (Fig. 2). In addition, there was a signi cant main effect of area (F(3,135) = 25.74, p < 0.001), which was quanti ed by a signi cant interaction between area and trial type (F(3,135) = 32.58, p < 0.001). An analysis of simple effects revealed larger amplitude of the P3 in the frontocentral (p < 0.001), central (p < 0.001) and centroparietal regions (p < 0.001) than in the parietal region on Stop trials, while larger amplitude of the P3 in the centroparietal region than other regions (all p < 0.005) on Go trials. Similarly, there was a signi cant main effect of laterality (F(2,90) = 71.18, p < 0.001), which was also quanti ed by a signi cant interaction with laterality and trial type (F(3,135) = 81.03, p < 0.001). An analysis of simple effects revealed larger amplitude of the P3 in the electrodes at the midline than that on the left and on the right (all p < 0.05) on Go trials, while larger amplitude of the P3 in the electrodes at the midline than that on the right and on the left (all p < 0.001) on Stop trials and larger amplitude of the P3 in the electrodes on the left than that on the right (p < 0.001) on Stop trials. No other signi cant effects were found.

200-250 ms
The mixed ANOVA performed on theta power (4 ~ 8 Hz) showed a signi cant main effect of trial type (F(1, 45) = 209.03, p < 0.001), with theta activity being larger on Stop trials compared to Go trials (p < 0.001), although there was not a signi cant main effect of group (F(1, 45) = 0.05, p = 0.83) and there was not a signi cant interaction between group and trial type (F(1, 45) = 1.09, p = 0.30). In addition, there was a signi cant main effect of area (F(2,90) = 16.71, p < 0.001), showing increased theta activity in the theta band in the central (p < 0.005) and centroparietal regions (p < 0.001) relative to the parietal region.
Similarly, there was a signi cant main effect of laterality (F(2,90) = 52.93, p < 0.001), which was quanti ed by a signi cant interaction between laterality and trial type (F(2,90) = 33.40, p < 0.001). An analysis of simple effects revealed increased theta activity in the electrodes at the midline compared with that in the electrodes on the left (p < 0.001) and on the right (p < 0.001) on Go trials, while increased theta activity in the electrodes at the midline relative to that in the electrodes on the left (p < 0.001) and on the right (p < 0.001) on Stop trials and increased theta activity in the electrodes on the left compared to that in the electrodes on the right (p < 0.001) on Stop trials. Finally, no other signi cant effects were found.
The mixed ANOVA performed on delta power (1 ~ 4 Hz) showed a signi cant main effect of trial type (F(1, 45) = 97.00, p < 0.001), showing increased spectral power in the theta frequency band on Stop trials compared to Go trials (p < 0.001). However, there was not a signi cant main effect of group (F(1, 45) = 0.67, p = 0.42) and there was not a signi cant interaction between group and trial type (F(1, 45) = 2.64, p = 0.11) either. In addition, there was a signi cant main effect of area (F(2,90) = 7.92, p < 0.005), which was quanti ed by signi cant interaction between area and trial type (F(2,90) = 16.28, p < 0.001). An analysis of simple effects revealed increased delta activity in the electrodes in the centroparietal region compared to that in the electrodes in the central (p < 0.001) and parietal regions (p < 0.05) on Go trials, while increased delta activity in the electrodes in the central (p < 0.05) and centroparietal regions (p < 0.005) compared to that in the electrodes in the parietal region on Stop trials (p < 0.001). Meanwhile, there was a signi cant main effect of laterality (F(2,90) = 38.35, p < 0.001), which was quanti ed by a signi cant interaction between laterality and trial type (F(2,90) = 41.69, p < 0.001). An analysis of simple effects revealed increased delta activity in the electrodes at the midline compared with that in the electrodes on the left (p < 0.01) and on the right (p < 0.001) on Go trials, while increased delta activity in the electrodes at the midline relative to that in the electrodes on the left (p < 0.001) and on the right (p < 0.005) on Stop trials and increased delta activity in the electrodes on the right relative to that in the electrodes on the left (p < However, there was a signi cant interaction between group and trial type (F(1, 45) = 9.42, p < 0.005), which was due to an increase in delta activity on Stop trials for patients with MwoA compared to healthy controls (p < 0.05) (Fig. 3). There was also a signi cant main effect of area (F(3,135) = 10.92, p < 0.001), which was quanti ed by a signi cant interaction between area and trial type (F(3,135) = 23.86, p < 0.001).
An analysis of simple effects revealed larger delta activity in the centroparietal region than in the central region on Go trials (p < 0.05), while stronger delta activity in the frontocentral (p < 0.005), central (p < 0.001) and centroparietal regions (p < 0.001) than in the parietal region on Stop trials. Similarly, there was a signi cant main effect of laterality (F(2,90) = 35.16, p < 0.001), which can also be quanti ed by a signi cant interaction with laterality and trial type (F(2,90) = 19.64, p < 0.001). The analysis of simple effects showed that delta activity reached its maximum in the electrodes at the midline compared with that on the left (p < 0.001) and on the right (p < 0.001) on Go trials, while on Stop trials, delta activity was stronger in the electrodes at the midline than that in the electrodes on the left (p < 0.001) and on the right (p < 0.001) and delta activity was larger in the electrodes on the right than that in the electrodes on the left (p < 0.05). Finally, No other signi cant effects were found. Moreover, there was not a signi cant interaction between group and trial type (F(1, 45) = 3.25, p = 0.08). However, there was also a signi cant main effect of area (F(3,135) = 32.24, p < 0.001), which was quanti ed by a signi cant interaction between area and trial type (F(3,135) = 37.21, p < 0.001). An analysis of simple effects revealed stronger theta activity in the centroparietal region than in the parietal region on Go trials (p < 0.01), while larger theta activity in the frontocentral, central and centroparietal regions than in the parietal region on Stop trials (all, p < 0.001). Similarly, there was a signi cant main effect of laterality (F(2,90) = 47.18, p < 0.001). which can also be quanti ed by a signi cant interaction with laterality and trial type (F(3,135) = 21.65, p < 0.001). The analysis of simple effects showed theta activity reached its maximum in the electrodes at the midline compared with that in the electrodes on the left (p < 0.001) and on the right (p < 0.001) on Go trials. In contrast, on Stop trials, theta activity was stronger in the electrodes at the midline than that in the electrodes on the left (p < 0.001) and on the right (p < 0.001) and theta activity was stronger in the electrodes on the right than that in the electrodes on the left (p < 0.001). Finally, No other signi cant effects were found.

Discussion
To the best of our knowledge, the present study is the rst to investigate the physiopathological basis of response inhibition during the interictal period of patients with MwoA during a stop signal task. Overall, consistent with the theory that cortical hyperexcitability is a key signature of migraine, the results in the present study revealed a decrease in suppressing prepotent responses in migraineurs which can be attributable to cortical hyperexcitability, implying the existence of dysfunctional inhibitory control at later stage of information processing.
At the behavioral level, patients with MwoA relative to healthy controls showed signi cantly slower RTs (SSRT) on Stop trials. However, accuracy on Go trials did not highlight any signi cant difference between patients with MwoA and healthy controls. Given that the stop signal reaction time (SSRT) provides an effective means of quantifying the latency of the inhibition mechanism, the slower SSRT implied a decrease or disruption in response inhibition in migraineurs. Such poorer performance during a stop signal task is in line with previous studies reporting that migraineurs during the interictal period relative to healthy controls showed signi cantly slower RTs on incongruent trials during the Stroop interference test without a between-group difference in commited errors (37,48). Given that the Stroop interference test measured cognitive interference or inhibition (26), the behavioral nding in the present study indicates that the impairment in inhibitory control in migraineurs can occur not only during the early and intermediate stage of information processing, but also during the late stage of information processing, such as response execution.
At the neural level, consistent with the notion that cortical hyperexcitability is a key signature of migraine, we rst found that patients with MwoA relative to healthy controls were characterized by a pronounced increase in the amplitude of the N2 in response to the stop signal in the central and parietocentral regions, but a nonsigni cant between-group diference in the amplitude of the N2 in response to Go stimuli. In spite of some controversy about its functional signi cance, the stop-N2 is proposed to be related to premotor cognitive processes such as con ict monitoring that is involved in the initial detection of con ict between response execution and inhibition (32,33). As a consequence, the increased stop-N2 amplitude seems to re ect a disruption of the degree of con ict monitoring that controls the concurrent and competing co-activation of response representations during early, non-motoric stages of response inhibition in patients with MwoA. Despite such ndings on the amplitude of the stop-N2, there was not a difference in the topographical distribution of this ERP index of response inhibition between patients with MwoA and healthy controls. The stop-N2 reached its maximum in the central and centroparietal regions across these 2 groups. Considering source-localization studies have found that the neural generator of the stop-N2 is located around the anterior cingulate cortex (ACC) (32,49,50), this seems to suggest that the presence of de cient inhibitory cortical processes can be localized beyond the sensory cortical system in the migraine brain. Although our observation of the increased amplitude of the stop-N2 provides compelling evidence in support of a disruption of premotor cognitive processes such as con ict monitoring during early, non-motoric stages of response inhibition, this can not answer the question of whether the other subprocess underying response inhibition is altered. Our further analysis on the amplitude of the P3 additinoaly help with addressing this issue. Regarding the P3, we found that patients with MwoA relative to healthy controls were characterized by an increase in the amplitude of the stop-P3 in the frontocentral, central and centroparietal regions. Contrary to the stop-N2, this ERP component has been proposed to re ect late-stage inhibition of the motor system itself and cognitive evaluation of motor inhibition (32). As such, our observation of the increased amplitude of the stop-P3 in patients with MwoA indicates that patients with MwoA can also mainifest a de cit in response inhbition during the late stage of response inhibition (when inhibiting a motor response). Given that source-localization studies have also shown that the stop-P3 is mainly associated with neural activity of the primary motor cortex (M1) and the supplementary motor areas (SMA) (32), this therefore may re ect dysfunctional activity in these cortical regions in migraineurs. Taken together, the time-domain ERP results during a stop signal task are suggestive of neurocognitive de cits while suppressing a prepotent motor response as evidenced by the increased ERP indices of response inhbitionn (stop N2-P3 complex) in patients with MwoA relative to healthy controls.
Although these two time-domain ERP indices of response inhibition greatly contribute to our understanding of how the brain detects the stop signal and decides to stop a prepotent motor response in patients with MwoA, these ERP measures can not accurately re ect the overlapping processes that need to be disaggregated in order to re ect separable mechanisms underlying response inhibition. To address this issue, the time-frequency analysis was employed in the present study to characterize event-related theta (4-8 Hz) and delta (1-4 Hz) oscillations that index two separable, but highly overlapping processes underlying the stop N2-P3 complex in response inhibition in patients with MwoA. With regard to theta and delta oscillations, the time-frequency analysis found signi cantly higher theta and delta activities in response to the stop signal than in response to Go stimuli in the time windows used to extract mean amplitudes of the stop-N2 and stop-P3 components, which is consistent with ndings from previous studies (32,34,45,46). Although both theta and delta activities in response to either the stop signal or Go stimuli in the time window used to extract mean amplitude of the stop-N2 component did not highlight any signi cant difference between patients with MwoA and healthy controls, patients with MwoA were associated with increased delta activity (not theta activity) relative to healthy controls in response to the stop signal rather than in response to Go stimuli in the time window used to extract mean amplitude of stop-P3 component. Despite ongoing debates on the functional signi cance of these two oscillatory activities (32), the relationship of TF phase dynamics to time-domain ERP measures may substantially explain the time-domain experimental effects observed in the present study. Given that previous studies have found that delta activity relative to theta activity contributed more to the stop-P3 component (32,36), this may suggests that the increased time-domain stop-P3 observed in patients with MwoA relative to healthy controls is mainly driven by the delta activity in the same time window. In contrast to deltarelated stop-P3 component, we failed to nd an interaction effect between group and trial type on both theta and delta activities in the time window used to extract mean amplitude of the stop-N2. Given that theta and delta oscillations have been suggested to contribute uniquely in the opposite direction at the N2, one possible explanation may lie in the fact that their combination may lead to the non-signi cant interaction effect between group and trial type in the frequency domain in the time time used to extract the mean amplitude of the stop-N2.

Limitations
In spite of such promising ndings, the present study still has some limitations. First, the present study included a relative small sample size, thus possibly tempering the strength of our conclusions. Going forward it is important that future research involves a larger sample size. Second, the present study only included patients suffering from migraine without aura. Given that migraine is a heterogeneous disease and the difference between migraine subtypes has been increasingly highligtened (51-53), this may affect "generalizability" of the results in migraineurs and therefore foster future research exploring the physiopathology of response inhbition in other subforms of this disease. Third, although the stop signal task is a frequent measurement of response inhibition, there are other types of task that can be used to mesaure inhbitory control such as the Go/no-go task. Moreover, accumulating evidence has revealed differences in these measures, raising the question of whether they tap into equivalent cognitive mechanisms underlying inhbitory control capacities (54). As a consequence, future research is required to examine other measures of response inhibition to better characterize the physiopathological basis of inhbition control at later stages of information processing such as response execution. Despite these limitations, we believe that our ndings are still robust and may foster further research on neuroanatomical characteristics and pathological signatures underlying inhibitory control function in migraineurs.

Conclusions
In conclusion, the present study was designed to characterize the physiopathological basis of response inhibition using event-related potentials (ERPs) in patients with MwoA during the interictal period. Our main nding is that patients with MwoA were associated with prolonged RTs to the stop signal relative to healthy controls, indicating a decrease in response inhibition in migraineurs. At the brain level, the amplitude of the stop-N2, a component of the ERPs related to con ict monitoring that is involved in the initial detection of con ict between response execution and inhibition, was signi cantly increased in patients with MwoA relative to healthy controls. Meanwhile, the amplitude of the stop-P3, a component of the ERPs re ecting late-stage inhibition of the motor system itself and cognitive evaluation of motor inhibition, was also signi cantly increased in patients with MwoA relative to healthy controls. These ndings further indicate a dysfunction in ERP indices of sub-processes underlying response inhibiton in patients with MwoA, which can be attributable to cortical hyperexcitability. Moreover, time-frequency decompositions have revealed increased delta activity in the time window used to extract to the mean amplitude of the stop-P3 in patients with MwoA relative to healthy controls. Taken together, the present study offers novel insights into how the brain detects the stop signal and decides to stop a prepotent motor response in migraineurs and provides both behavioral and neural evidence showing the existence of dysfunctional inhibitory control at later stage of information processing in migraineurs. These ndings not only aid in a growing body of literature showing the presence of a de cit inhbition control in migraineurs, but also may have important implications for further research charactering inhibitory control alterations in migraineurs.

Abbreviations
MwoA: migraine without aura; SST:Stop signal task; ERPs:Event-related potentials; SSRT:Stop signal reaction time Declarations Ethics approval and consent to participate This study was approved by the Ethics Committee of the Chinese PLA General Hospital and informed written consent was obtained for all participants.

Consent for publication
Written informed consent for publication was obtained.

Availability of data and materials
The datasets used and analyzed during the present study are available from the corresponding authors on reasonable request.

Competing interests
The authors declare that they have no competing interests.  Figure 1 Stop signal paradigm. This task included 80% Go trials and 20% Stop trials. On Go trials, participants were presented with a xation cross on a black computer screen lasting for 600-800 ms, which is immediately followed by a Go stimulus (the letter 'X' or 'O') (a Go signal) lasting for 1000 ms. Participants were instructed to judge the shape of the Go stimulus as accurately and quickly as possible via a button press with the index ngers of the left and right hands. On the remaining Stop trials (20%), the stop stimulus (a red square appearing above the location of the go stimulus) appeared after the Go stimulus after a variable delay of 0-250 ms in a step of 50 ms (the stop signal delay; SSD), cuing participants to withhold their responses to the Go stimulus. A variable intertrial interval was 1500-2000 ms.