resting state alpha band oscillations in migraineeprints.lincoln.ac.uk/30360/1/migraine paper...

48
1 Resting state alpha band oscillations in migraine Louise O’Hare* 1 , Federica Menchinelli 1 and Simon J. Durrant 1 1 University of Lincoln, Brayford Pool, Lincoln, LN6 7TS *corresponding author, [email protected]

Upload: others

Post on 20-Oct-2020

3 views

Category:

Documents


0 download

TRANSCRIPT

Resting state alpha band oscillations in migraine

33

Resting state alpha band oscillations in migraine

Louise O’Hare*1, Federica Menchinelli1 and Simon J. Durrant1

1University of Lincoln, Brayford Pool, Lincoln, LN6 7TS

*corresponding author, [email protected]

Abstract

Migraine groups show differences in motion perception compared to controls, when tested in between migraine attacks (interictally). This is thought to be due to an increased susceptibility to stimulus degradation (multiplicative internal noise). Fluctuations in alpha-band oscillations are thought to regulate visual perception, and so differences could provide a mechanism for the increased multiplicative noise seen in migraine. The aim of this paper was to characterise resting state alpha band oscillations (between 8 and 12Hz) in the visual areas of the brain in migraine and control groups. Alpha band activity in the resting state (with eyes closed) was recorded before and after a visual psychophysics task to estimate equivalent noise, specifically a contrast detection task. The lower alpha band (8-10Hz) resting-state alpha band power was increased in the migraine compared to the control group, which may provide a mechanism for increased multiplicative noise. In agreement with previous research, there were no differences found in the additive (baseline) internal noise, estimated using an equivalent noise task in the same observers. As fluctuations in alpha band oscillations control the timing of perceptual processing, increased lower alpha band (8-10Hz) power could explain the behavioural differences in migraine compared to control groups, particularly on tasks relying on temporal integration.

Key Words

lower alpha band; EEG; spectral analysis; contrast discrimination

Introduction

Migraine groups show differences in performance compared to control groups on visual tasks in between the migraine attacks themselves (see O'Hare and Hibbard, 2016, for a review). In particular, there are reports of poorer performance for migraine groups studied interictally compared to controls on orientation discrimination tasks (Tibber et al., 2006), certain motion-based tasks (also known as global motion, or motion coherence tasks) (e.g. Antal et al., 2005; McKendrick et al., 2006; Ditchfield et al., 2006; Webster et al., 2011; Braunitzer et al., 2012; Shepherd et al., 2012; Tibber et al., 2014), and also contrast sensitivity tasks (Shepherd et al., 2008). It has been suggested that differences in performance in migraine groups compared to controls is due to internal noise in the processing of visual stimuli (Wagner et al., 2010). Internal noise is random variability in the output of a system that originates within the system itself. This is different from external noise, which is typically a degradation of the incoming stimulus. Two possible types of internal noise are: a constant, baseline level of internal noise (additive, possibly due to background neural firing); and internal noise that scales with the signal degradation (multiplicative, increased uncertainty when the signal is unclear). Additive and multiplicative internal noise are both present in the widely used Perceptual Template Model (Lu and Dosher, 1998). Formally, additive noise processes are included in the model just before the decision stage, whereas multiplicative noise scales with the total input to the system (signal strength plus external noise). Additive noise processes add a fixed amount of noise, while multiplicative noise scales with the total input to the system (signal strength plus external noise). Including this multiplicative noise term improves the fit of the model to psychophysical data (Lu and Dosher, 1999). An example of multiplicative noise is a greater detriment to performance in a migraine group compared to control groups when external noise is introduced to the stimulus on a masking task (Wagner et al., 2010), showing a greater cost of stimulus uncertainty in the migraine group compared to the controls. Similarly, Tibber et al., (2014) used equivalent noise tasks to estimate the level of internal noise in migraine compared to control groups. Using tasks involving orientation discrimination, motion coherence, and size discrimination, Tibber et al., (2014) showed there to be no difference in baseline (additive) internal noise estimates for migraine and control groups. However, there was evidence of a greater reduction in performance in the migraine group when high levels of external noise were introduced to the stimulus, suggestive of multiplicative noise, a greater cost of signal degradation in migraine.

Adding external noise to a stimulus (i.e. degrading the incoming signal) has a particularly detrimental effect on migraine groups, but the mechanism for this is unclear. One possible reason for the increased effects of external noise in migraine groups is the rate of information processing in the brain. The timing of visual processing is thought to be controlled by alpha band (approximately 8-12Hz) oscillations (Jensen et al., 2014). In the absence of a visual stimulus, each individual has a resting state alpha oscillation with a peak at a particular frequency, which can be measured using EEG or MEG. There are two characteristics of alpha band activity; the peak frequency and the magnitude of the oscillation. Peak frequency is defined as the frequency at which the maximum amplitude occurs within the band. Alpha power is related to the variation in amplitude of the oscillation in a given frequency band.

Alpha band oscillations are thought to have a role in the visual system in controlling the timing of perceptual processes (e.g. Klimesch et al., 2007; Jensen et al., 2014; Klimesch et al., 2012). Alpha power over parietal-occipital cortex is related to visual attention (Foxe et al., 1998), and also to anticipation of a stimulus (Bastiaansen and Brunia, 2001; Rohenkohl and Nobre, 2011). Decreased alpha power immediately before stimulus presentation (prestimulus power) is associated with improved visual performance (Erenoglu et al., 2004; van Dijk et al., 2008; Hanslymayr et al., 2005; Hanslmayr et al., 2007). Lower prestimulus alpha power is also associated with higher evoked potentials (Becker et al., 2008), and easier induction of phosphenes via TMS on a trial-by-trial basis (Romei et al., 2008). The effects of prestimulus alpha power depends on the phase relationship: the time-course of the oscillation on each individual trial, and the exact timing of the stimulus relative to the peaks and troughs of this waveform. When exactly presentation of the stimulus occurs in the alpha band cycle (the phase relationship) is known to correlate with perception - incoming stimuli coinciding with troughs in the cycle are more likely to be perceived than those coinciding with the peaks (Becker et al., 2008), in keeping with the inhibitory role of the oscillations. Therefore, it has been suggested that the alpha band controls the flow of information, and provides a ‘window of excitability’ (Dugue et al., 2011).

The length of the window of excitability relates to the frequency of the oscillation, and so frequency could provide a mechanism for the control of visual information processing by the alpha band oscillations. Resting alpha frequency has been shown to predict performance on a visual task (de Graaf et al., 2013). In the spatial cueing task, participants were asked to predict where the next stimulus in a sequence of flashes of light would appear. De Graaf et al., (2013) demonstrated that stimulation at approximately 10Hz (alpha band), and its first harmonic, showed a different pattern of results compared to stimulation at other stimulation frequencies. The stimulation at the alpha band frequency and the first harmonic resulted in cyclic patterns of performance. This cyclic pattern of performance was correlated with the individual resting alpha frequency. This supports the idea that there is a window of excitation from the resting-state alpha band oscillations that relates to performance on visual tasks. It is possible also that individual variation in alpha band oscillations at rest determine the rhythm of information processing. Romei et al., (2008) relate alpha band activity to noise: “This moment-by-moment variability in neuronal activity may be considered ‘‘physiological’’ noise”. Slower oscillations will result in reduced resolution and so reduced stimulus specificity (Wutz et al., 2016), thus increased internal noise. It is expected that reduced stimulus specificity would be multiplicative noise, rather than additive noise, a fixed level of noise. This is because stronger stimuli are able to be perceived earlier in the alpha cycle compared to weaker stimuli (Jensen et al., 2014), which might compound the problem of multiple stimuli in the same perceptual window. Also, the time window of multisensory integration (known as the “flash-beep” illusion) can also be predicted by the resting alpha frequency (Cecere et al., 2015) – those with slower alpha band oscillations also had longer time windows for multisensory integration.

As well as frequency, alpha band power relates to the window of excitability. There are two mechanisms for this, one is related to frequency: Greater lower alpha band (e.g. 7-10Hz) power compared to upper alpha band power (10-12Hz) indicates a greater probability of low frequency alpha waves compared to higher frequency alpha waves. A distinction is made in the literature between lower and upper alpha band power; the lower alpha band is defined as approximately 7-10Hz, and the upper alpha band as 10-12Hz (Pfurtscheller and Lopes da Silva, 1999). It has been suggested that event-related desynchronization in the lower alpha band might reflect attentional mechanisms that are regulated by thalamo-cortical networks (Brunia, 1999; Pfurtscheller and Lopes da Silva, 1999, lower alpha was defined by the authors as (defined by the authors as 6-10Hz and 7-10Hz respectively). It has been suggested that the lower alpha band is related to attentional processes (alertness) (Klimesch et al., 1996). Babiloni et al., (2006) suggest that the lower alpha band is associated with visual consciousness, demonstrating high power in the lower alpha band (6-10Hz) before a stimulus onset is associated with increased performance on a visually-based cueing task. They argue that high power in the lower alpha band before a stimulus is presented is associated with cortical excitation and widely distributed sensory attention. Babiloni et al., (2006) also show that whilst in the upper alpha band power is increased in the trials where the stimulus is not perceived compared to trials where a stimulus is perceived, the reverse is the case for the lower alpha band; there is stronger lower alpha band power when the stimulus is perceived compared to when not perceived. By contrast, the upper alpha band (approximately 10 -13.5Hz according to Klimesch et al., 2006) is associated with processing the semantics of the stimulus, and therefore has different effects depending on the band and depending on the task. Migraine is characterised by differences in low-level perceptual processing, and so the lower alpha band (8-10Hz) is of interest here. It has been suggested that the cortex is hyperexcitable in migraine, and so higher power in the lower alpha band is expected.

Alpha band power can also affect the window of excitability independently of frequency. Higher power at a given frequency will mean waveforms have greater amplitude, which in turn means that the time interval between the descending and ascending arms of the waveform at a certain amplitude threshold – the interval in which a stimulus can be perceived because the waveform amplitude is sufficiently low – will be longer (because the threshold point will occur relative higher up on a waveform with greater amplitude).There are therefore three possible contributing mechanisms for alpha band oscillations to influence perception: (1) frequency – longer time windows will allow more stimuli to be perceived simultaneously, resulting in a less specific, noisier response; (2) power related to frequency – more power in the lower compared to the upper alpha band would indicate a slower waveform on average and hence longer time windows between amplitude points of a given threshold; (3) power independent of frequency – the greater the power the steeper the slope of any individual waveform and more time spent beneath a given amplitude relative to the lower power waveform at the same frequency with the same amplitude threshold These are potential mechanisms for multiplicative internal noise because the strength of the stimulus affects the exact timing in the alpha band cycle a stimulus will be perceived (Jensen et al., 2014). Stronger stimuli are able to be perceived earlier in the alpha band cycle. Due to the interaction between stimulus strength and the relationship with the alpha band, it is expected that the effects of the alpha band will scale with stimulus strength, representing multiplicative, rather than baseline, internal noise mechanism.

There are many ways of adding noise to (and thereby degrading) a stimulus. One common method is to add pixel noise. Adding pixel noise has the potential confound of also systematically affecting the interactions between neurons, e.g. introducing additional lateral inhibition effects. It has been suggested that lateral inhibition may be different in migraine compared to control groups (Palmer et al., 2000). A task is needed to differentiate between the effects from lateral interactions between neurons and the effect of external (stimulus) noise. Baker and Meese (2012) developed an equivalent noise task to reduce this confound. The task they developed is based on contrast sensitivity, using a Gabor patch with contrast jitter. The aim of the observer is to detect the patch of greater contrast. The smallest amount of jitter that is detectable to the observer will give an estimate of baseline internal noise. As no additional pixels are added to the stimulus in order to introduce the noise, this technique does not systematically affect the lateral interactions between neurons, which is a potential confound for other methods of introducing noise to degrade a stimulus (e.g. masking, pixel noise). Therefore, the Baker and Meese (2012) behavioural task will be used to estimate additive internal noise, which is associated with background neuronal firing.

The equivalent noise task is designed to measure the additive, baseline levels of internal noise in the visual system of the individual. External (stimulus) noise is introduced in small increments until it can be perceived. The smallest perceptible external (stimulus) noise level that affects performance represents the amount needed to overcome the internal (neural) noise of the system (Dakin et al., 2005).

This study will therefore investigate differences in lower alpha band (8-10Hz) oscillations in migraine. It is predicted that lower alpha band (8-10Hz) resting state oscillations will be increased in power in migraine as this is indicative of slower oscillations, and has been linked to increased cortical excitability (Babiloni et al., 2006).This will indicate a longer temporal integration window, and therefore less specific responses to incoming visual stimuli. As a result, this has the potential to provide a mechanism for the increased susceptibility to external stimulus noise in migraine compared to control groups that has been reported in previous behavioural tasks (e.g. Tibber et al., 2014). As a control measure, estimates of equivalent noise will be made using a basic visual task (contrast sensitivity), to ensure any differences cannot be explained in terms of baseline internal noise levels. Measurements of alpha band oscillations will be made before and after the visual task, as the visual task itself may have effects on the alpha band oscillations.

Method

Participants

42 young participants, both male and female, with corrected-to-normal vision participated in the study. Ethical approval was granted by the University of Lincoln Research Ethics Committee, and all experiments were in accordance with the Code of Ethics of the World Medical Association. Fully informed written consent was obtained. Participants were reimbursed for their time.

The migraine group was composed of thirteen individuals (85% female) who fulfilled the International Headache Society classification criteria for migraine (International Headache Society, IHS, 2013), or reported having diagnosis of migraine from a medical professional (7 of the participants were verified by a medical professional). Six participants recruited to the original sample, who reported migraine, were excluded as they did not fulfil the IHS (2013) criteria. Three participants of the original recruited group were excluded due to trauma to the head (car accident, stone thrown at head, unspecified trauma to the head). Three participants were excluded as their last attack was less than 2 days before the experimental testing session. The average time since the last migraine attack of those included in the sample was 15.58 days, SD = 16.13. The mean frequency of migraine attacks per month in the final sample was 2.75, SD = 2.65. 76.92% reported visual disturbances, 38.46% reported somatosensory disturbances during the attack. 23.08% reported disturbances of speech during the migraine attack. 53.85% of the final migraine group had an optical prescription. Table 1 shows migraine characteristics of those included in the study.

Obs

Monthly frequency of headaches

Visual disturbances during the attack

Speech disturbances during the attack

Medical professional diagnosis

Time since most recent attack

Gender

Age

3

> 1

Yes

No

Yes

5 days

U

U

14

1-3

Yes

Yes

Yes

7 days

F

19

16

1-3

No

No

No

2 days

M

18

18

> 1

No

No

Yes

28 days

F

19

27

3-10

Yes

Yes

Yes

2 days

F

19

28

3-10

No

Yes

No

7 days

F

19

31

1-3

Yes

Yes

No

56 days

F

19

34

1-3

Yes

No

No

U

F

18

37

> 1

Yes

Yes

No

21 days

F

20

39

> 1

Yes

No

Yes

28 days

F

18

40

3-10

No

No

No

3 days

M

19

41

1 to 3

Yes

No

Yes

7 days

F

44

44

1 to 3

Yes

No

Yes

21 days

F

20

Table 1: Migraine characteristics. U = unanswered.

The control group was composed of 17 individuals (82% female) who reported experiencing no migraine, and no headaches in general on a regular basis (less than one per month). Mean age of the control group was 19.24 (SD = 0.75) years, compared to a migraine group mean of 21 (SD = 7.27) years, with 2 males in the group. Results of an independent measures t-test showed there to be no statistically significant difference between the groups in terms of age (t(27) = 1.000, p = 0.3260, d = 0.368).

Apparatus

Experiments were conducted in a sound-attenuated, electrically-insulated, darkened room. Stimuli were presented on a 22 inch Illyama HM204DTA Vision Master Pro 514 Diamondtron U3-CRT monitor, calibrated with LS100 Minolta photometer. Mean luminance of the display used for the behavioural task was 39.24 cd/m2. Refresh rate was 60Hz, resolution was 1024 x 768 pixels. A Bits # signal processor (Cambridge Research Systems, Cambridge, UK) was used to convert the RGB signal into greyscale to allow for finer control of contrast in the behavioural experiment. Stimuli were created and presented using MATLAB R2014a and the Psychtoolbox v3 (Brainard, 1997; Pelli, 1997; Kleiner, 2007). Recordings were made using a 64-channel Biosemi Active-Two system, using Ag-AgCl electrodes. 64 electrodes were positioned using a 10/20 cap labelling system, with eight additional electrodes: two on the left and right mastoids, two infraorbital, two suborbital, and two on the outer canthi of the eyes. During recording signals were referenced to a CMS (common mode sense) electrode (see http://www.biosemi.com/faq/cms&drl.htm). Recording was sampled at 2048Hz, decimated to 256Hz offline. During recording signals were bandpass filtered between 0.16Hz and 100Hz to remove artefacts outside of the range of interest.

Procedure

Measurements of amplitude or power are both dependent on the quality of the signal, and are influenced by trivial factors such as skull thickness. Therefore, care must be taken to normalise this measure appropriately in order to compare between individuals. There is variation amongst researchers as to how this normalisation is achieved (e.g. Doppelmayr et al., 2002; Mantini et al., 2007; Ben-Simon et al., 2008). One widespread approach is to remove the appropriate baseline; however, resting-state alpha band oscillations are often measured by recording a 2 or 3 minute recording of the EEG (e.g. Ben-Simon et al., 2008). Splitting this into 10-second trials leaves sufficient data to perform a reliable spectral analysis but also has the advantage of allowing both the rejection of bad trials due to movements etc., and the removal of a baseline response. Therefore, resting state recordings were made using a trial-based design rather than a single continuous measurement. Each measurement of alpha was made using 20 x 10-second trials, initiated and ended with a flash of light, visible through the closed eyelids. The flash of light was minimum to maximum luminance of the monitor (151.1 cd/m2) for one visual frame only. The observer was instructed to close their eyes, initiate the trial using the spacebar, and maintain resting wakefulness until the next flash of light to indicate the end. The flash of light was bright enough to be easily seen through the closed eyelids. There were 20 x 10 second trials of this block design.

The visual task consisted of a 2-alternative-forced-choice contrast discrimination task, based on that of Baker and Meese, (2012). Participants viewed two gratings of 1 cycle/degree in a Gaussian-edged window, presented side-by-side. These were presented slightly peripheral to a small red fixation cross (0.7619° in width) which was present throughout. Both gratings were defined in contrast by a level that was jittered around a pedestal. The target consisted of one level of contrast (-10dB), plus the jittered contrast. The foil grating consisted of just the contrast jitter. There were 5 levels of contrast jitter, (-8 -10 -12 -14 -16dB). Stimuli reversed contrast polarity sinusoidally at a rate of 7Hz and 14Hz. The foil was another grating with a jittered level of contrast. The standard and the test grating were presented one after the other, in random order, so that approximately half the time the standard was first, the rest of the trials the test stimulus was presented first. Trials were presented in random order of contrast level and flicker rate, and this randomisation was different for each observer. The task of the observer was to indicate, using the arrow keys, whether the stimulus on the left or the right had the higher contrast. There were 100 repetitions of each of the contrast conditions for both flicker rates, resulting in a visual task lasting for approximately 45 minutes, depending on the speed of the participant.

After the equivalent noise task, the recording of the resting state EEG was completed again, as before.

Data analysis

Data were analysed using MATLAB R2015a and the EEGLAB toolbox, version 13_6_5b (Delorme and Makeig, 2004). Data from all 64 channels were re-referenced to the linked mastoids. The data from all 64 electrodes were divided into ten-second epochs. The mean activity in the 1000ms period before stimulus onset was subtracted from the rest of the epoch for each channel separately. Artefacts were removed using the EEGLAB automatic rejection procedure. This consists of multiple stages, first by detecting large fluctuations (1000μV) and excluding those epochs with channels containing large fluctuations. Then improbable data were rejected, which were defined as those with a standard deviation outside a certain value, initially SD = 5. A maximum of 5% epochs with data values outside this given standard deviation were rejected, and the standard deviation adjusted until the value where 5% epochs can be defined for rejection. Spectral analysis was conducted using the inbuilt EEGLAB function std_specplot, which is based on an FFT, with a default of 1s Hamming window length and 50% overlap. The spectral power estimated for all 64 channels were used for scalp activation maps. For the resting-state oscillations, the peak alpha frequency was defined as the frequency of the bin with the greatest power in the 8-12Hz range, pooled over the occipital electrodes (O1, O2, Oz and Iz). The mean power was the average power over the band.

Analysis was conducted using the statistical package "R" (R Core Team, 2013), using package “afex”, (Singman et al., 2017). Greenhouse Geisser corrections were used when the assumption of sphericity was violated. Effect sizes are given as Generalised Eta Squared (η2G). Post-hoc Tukey tests were conducted using the package "lsmeans" (Lenth, 2016).

Results

Resting state alpha

Figure 1 shows the scalp topography of the average power in the lower alpha band, specifically at 8Hz, for migraine and control groups, before and after the contrast discrimination task. As expected, the difference in power is localised over the occipital cortex, specifically over the occipital electrodes O1, O2, Oz, Iz. There appears to be greater power in the migraine compared to the control group, especially after the visual task. Scalp maps for higher (12Hz) and overall (10Hz) alpha band frequencies are shown in the supporting information.

Figure1: Scalp map at 8Hz. Top row shows the control group before (1) and after (2) the contrast discrimination task. Lower row shows the migraine group before (1) and after (2) the contrast discrimination task. Spectral power in 10xlog10(μV2/Hz).

Figure 2 shows the average power spectra for migraine and control groups, both for before, and after the visual task, averaged over electrodes O1, O2, Oz and Iz (electrodes included in analysis are plotted individually in the supplementary information). Figure 2 shows a bigger difference in the low alpha band (8-10Hz) compared to the high (10-12Hz), and overall alpha band (8-12Hz).

Figure 2: Average power spectra for migraine and control groups, before and after the visual task. Spectral power in 10xlog10(μV2/Hz) and averaged over electrodes O1, O2, Oz, Iz.

Figure 3 shows the average power within each group separately for lower and higher alpha bands, before and after the contrast discrimination task. There was a statistically significant effect of group on spectral power: the migraine group showed significantly higher power in lower alpha band (8-10Hz) compared to the control group (F(1,28) = 4.862, p = 0.0358, η2G = 0.118). This is not the case for the higher alpha power band (10-12Hz) (F(1,28) = 0.962, p = 0.335, η2G = 0.028). There is also a significant difference before compared to after a visual task in the lower alpha band (F(1,28) = 5.457, p = 0.026, η2G = 0.043), but no effect before compared to after in the upper band (F(1,28) = 2.966, p = 0.096, η2G = 0.017). There is no statistically significant interaction between task order (before or after) and group (migraine and control) for the low alpha band (8-12Hz) (F(1,28) = 0.978, p = 0.331, η2G = 0.008), or high alpha band (10-12Hz) (F(1,28) = 1.701, p = 0.203, η2G = 0.010). Taken together, these results show group differences and task effects, but specifically only in the lower alpha band.

Figure 3: Mean power for the lower (8-10Hz), and upper (10-12Hz) alpha band for the migraine and control group, before (left plot) and after (right plot) the visual task. These are averaged over electrodes O1, O2, Oz and Iz. Error bars are one standard error of the mean.

Figure 4 shows peak frequency before and after the visual task for the migraine and control groups. For the lower alpha (8-10Hz) band, there is no significant effect of migraine group on peak frequency (F(1,28) = 2.018, p = 0.167, η2G = 0.048), although there may be evidence for a small effect (Bakeman, 2005). There is no statistically significant effect of migraine group on peak frequency in the upper alpha band (F(1,28) = 0.093, p = 0.763, η2G = 0.003). There is no effect of before compared to after a visual task on peak alpha frequency for either the low (F(1,28) = 0.983, p = 0.330, η2G = 0.010) or high (F(1,28) = 1.387, p = 0.249, η2G = 0.006) alpha bands. There was no interaction effect in either the low (F(1,28) = 0.087, p = 0.770, η2G = 0.001) or the high (F(1,28) = 0.398, p = 0.533, η2G = 0.002) alpha band.

Figure 4: Peak frequency in the lower (8-10Hz) and upper (10-12Hz) alpha bands before (left) and after (right) a visual task for migraine and control groups. These are averaged over electrodes O1, O2, Oz and Iz. Error bars show one standard error or the mean.

Behavioural Results

The behavioural data were analysed by fitting a psychometric function using Psignifit (Fründ et al., 2011). The 75% correct threshold was obtained, as well as the slope of the psychometric function. An example psychometric function can be seen in the supplementary information. Two participants were excluded as their thresholds were more than three standard deviations from the mean. Figure 5 shows behavioural results, thresholds and slopes for migraine and control groups for the two levels of flicker (7 and 14Hz). There is no difference in equivalent noise estimates (75% threshold performance) between flicker speeds (F(1,26) = 0.053, p 0.821, η2G = 5.532x10-4). There is also no statistically significant difference in performance between migraine and control groups (F(1,26) = 0.0002, p = 0.990, η2G = 5.67x10-6). There is no interaction between speed and group (F(1,26) = 0.064, p = 0.802, η2G = 1.879x10-4).

Figure 5: Left: 75% threshold contrast for migraine and control groups for Gabor stimuli flickering at 7 and 14Hz. Right: slope of the psychometric function for migraine and control groups for Gabor stimuli flickering at 7 and 14Hz.

There is no effect of flicker speed on the slope of the psychometric function (F(1,26) = 2.372, p = 0.136, η2G = 0.046). There is a trend towards steeper slopes in the migraine compared to the control group (F(1,26) = 4.036, p = 0.055, η2G = 0.067), and the effect size indicates the possibility of a medium effect. There is no interaction between speed and group (F(1,26) = 0.253, p = 0.620, η2G = 0.005).

Discussion

Results of the current study demonstrate differences in low alpha band (8-10Hz) power in those with migraine compared to controls. The alpha band is thought to control the processing of visual stimuli, and has been called the ‘window of excitability’ (Dugue et al., 2011). When two visual stimuli fall within the same alpha cycle, they may be perceived as the same stimulus, resulting in a less specific, i.e. “noisier” response. Reduced frequency in the alpha band will also be indicative of a longer time window in migraine compared to control groups. Some possible evidence for this was found in the current study, although this was only marginally significant. Increased lower alpha band power compared to upper alpha band power will mean a longer time window (on average) in those with migraine compared to controls, and so could result in less specific, “noisier” responses in the migraine group compared to the control group. Alpha band power can also effect the window of excitability independently of frequency, as higher alpha power will mean steeper-sloped waveforms (on average). This will result in a different full-width half-maximum and therefore longer window of excitability, depending on the threshold amplitude for the stimulus to be perceived. As there was evidence of slightly lower frequencies in the migraine compared to the control group, this seems to be less likely than the previous mechanism, however, this is also a possibility.

The frequency of the alpha band oscillations determine temporal resolution; lower alpha frequency leads to lower temporal resolution, and higher alpha frequencies are related to a finer resolution of visual stimuli (Samaha et al, 2015). The alpha band power also predicts the temporal resolution of the visual system; higher alpha power results in finer temporal resolution (Lange et al., 2013).

Temporal resolution is an important component of tasks such as motion processing and the integration of sensory information. Reduced performance on motion coherence tasks is one of the most robust findings in migraine performance when studied interictally (e.g. Antal et al., 2005; McKendrick et al., 2006; Ditchfield et al., 2006; Webster et al., 2011; Braunitzer et al., 2012; Shepherd et al., 2012; Tibber et al., 2014). Additionally, there is some preliminary evidence for greater susceptibility in migraine groups to multisensory illusions, specifically the sound-induced flash illusion (Yang et al., 2014). The significance of this is that the sound-induced flash illusion susceptibility is determined by the integration of visual and auditory signals. Integration over a longer time window will lead to a reduced ability to discriminate the separate signals and therefore increased susceptibility to the illusion. It has been shown that the window of integration in multisensory illusions is related to alpha power (Lange et al., 2013; Cecere et al., 2015). It has been suggested that there is increased integration of perceptual stimuli in migraine groups (Yang et al., 2014; Tibber et al., 2014). If migraine groups are characterised by higher power in the lower alpha band, then this might be a potential mechanism for the over-integration that could be the case in migraine. Over-integration could be the cause of the increased susceptibility to stimulus noise (multiplicative noise) seen in migraine compared to control groups.

Degraded (noisy) stimuli represent a weaker signal. Allard and Cavanagh, (2011) showed that pooling signals over time is a strategy in normal observers to increase signal to noise ratio of weak signals, when the noise is unrelated to the signal. However, if the noise is related to the signal, this will reduce the signal to noise ratio as the noise elements will be included in the estimate. An observer with a longer temporal integration window will be more susceptible to noise when it relates to the signal, which could account for the increase in susceptibility to stimulus noise (multiplicative noise) in the migraine group.

Internal noise has been suggested to be increased in migraine (e.g. Wagner et al., 2010; Webster et al., 2011), but there are two possible types of internal noise, additive (baseline) levels of internal noise, and multiplicative internal noise (increased susceptibility to the effects of signal degradation). It is theoretically important to determine which - additive internal noise is attributed to background neuronal firing (Dakin et al., 2005), whereas multiplicative internal noise must have another cause. In the current study, an equivalent noise task was used to control for the possibility of increased additive (baseline) levels of internal noise in the migraine group. The current study showed there to be no difference in behavioural threshold performance on a contrast-based equivalent noise task between the migraine and control group, supporting the findings of Tibber et al., (2014), who measured equivalent noise on orientation, size, and motion based tasks in migraine groups.

Previous work showed reduced alpha band (8-12Hz) power in migraine compared to control groups Neufeld et al., (1991), which is the opposite finding to the current study. Neufeld et al., (1991) tested 22 common and 22 classic (now called migraine-with-aura) and 20 control participants for routine EEG, including hyperventilation and photic stimulation. The authors reported no group differences in the response to photic stimulation. However, other authors have found the response to photic stimulation to be a relatively reliable indicator of migraine (e.g. Golla and Winter, 1959; Chorlton and Kane, 2000). Neufeld et al., (1991) conducted spectral analysis on the resting EEG of 13 common migraine, 10 classic migraine, 11 control participants, recording over the electrodes P3, O1, P4 and O2. There are some methodological differences that might account for the difference in results: Neufeld et al., (1991) considered the entire alpha band (8-11.5Hz) and did not investigate the lower alpha band separately. In the current study, there were no group differences found when measured over the whole alpha band (8-12Hz). However, for the lower alpha band only (8-10Hz) there were group differences between migraine and control groups, with migraine showing greater alpha power overall in this lower band. It might be argued that trivial differences in the signal e.g. skull thickness, skin conduction etc. might account for the differences between groups, however subtracting a baseline from the epoch allows for comparison between groups. Neufeld et al., (1991) did not report removing a baseline, whereas in the current study a baseline was removed before each epoch to allow for comparisons between observers. Additionally, migraine is characterised by differences in perception both immediately before (Lashley, 1941), during, in the case of photophobia (International Headache Society, 2013), and in between attacks (e.g. Marcus and Soso, 1989; Wagner et al., 2010; Shepherd et al., 2012; Tibber et al., 2014). There are differences in performance across the migraine cycle, and so it is important to note the time in the cycle. As neither the current study, nor that of Neufeld et al., (1991) had follow-up sessions, it might be the case that one group of subjects experienced a migraine attack shortly after the testing session, and so were in the pre-ictal stage. De Tommaso et al., (1998) a slower alpha band oscillation (peak around 8Hz) during the attack compared to interictal periods (around 9.5Hz peak) in those with migraine. Additionally, Bjork et al., (2009) showed that peak alpha frequency correlated negatively with disease duration, i.e. slower alpha band frequency in those who have experienced migraine for the most number of years. This highlights the need for longitudinal studies in migraine research.

Alpha power estimations in the current experiment were relatively stable before compared to after a visual task. However, there was a trend towards greater alpha power after performing a demanding visual task compared to before. During stimulation, multiple studies have found an increased response in migraine compared to control groups. Golla and Winter (1959) reported the H-response to 18Hz photic stimulation, which is highly characteristic of migraine. The H-response is a characteristic shape frequency response, which peaks at 10Hz, and the high response is maintained beyond the stimulation frequency. Control groups, by comparison, typically showed just a response at 10Hz, falling away around 14Hz. Some individuals also show a double peak, at both 10Hz, and the stimulation frequency, with a trough in between. This finding of the H-response commonly occurring in migraine, but not in controls, was confirmed by other authors with large sample sizes e.g. Smyth and Winter, (1964), with 202 patients, and also Chorlton and Kane, (2000) with 33 patients. There are many potential reasons for this, a slower temporal window in migraine would not be inconsistent – if the cells do not have sufficient time to return to baseline after responding, this could result in a sustained response to flickering stimuli.

There is also a literature showing a lack of habituation of increased visual evoked potentials (VEP) in migraine compared to control groups when measured over several blocks (for a review, see Coppola et al., 2009). Greater alpha power might be indicative of greater excitation levels, and so would not necessarily be inconsistent with the habituation literature. However, the current study compares before to after a visual task, and measures EEG activity at rest, and therefore it is not directly comparable to studies including several blocks of a visual task, and so this is speculative.

There is evidence that oscillations can be modulated, using either TMS (Thut et al.,2011) or repetitive visual stimulation (i.e. flicker) (Herrmann, 2001). This also implies that it may be possible to improve/worsen perception using artificial methods, such as neurofeedback, or visual stimulation. This would be of interest as therapy for migraine, however it appears that a simple visual task might not be a strong enough stimulus to cause changes in alpha band oscillations, and multiple sessions may be needed.

Conclusion

There are differences in low (8-10Hz) alpha band power between migraine and control groups: higher power for the migraine compared to the control group. In agreement with previous research (Tibber et al., 2014) there were no group differences in internal noise measured using an equivalent noise task. Migraine groups have previously been shown to be more susceptible to the effects of external noise added to a stimulus (e.g. Wagner et al., 2010; Wagner et al., 2013), therefore higher power in the low alpha band might be a potential mechanism for the over-integration of signals over time, which could have the effect of increased stimulus noise.

Acknowledgements

This research was supported by the University of Lincoln, College of Social Science College Research Fund.

Supporting information

Supporting information can be found at Figshare.com: 10.6084/m9.figshare.4680640.

References

Allard, R., and Cavanagh, P., (2011). Crowding in a detection task: external noise triggers change in processing strategy. Vision Research, 51(4): 408-416.

Angelakis, E., Lubar, J. F., Stathopoulou, S., and Kounios, J., (2004). Peak alpha frequency: an electroencephalographic measure of cognitive preparedness. Clinical Neurophysiology, 115: 887-897.

Antal, A., Temme, J., Nitsche, M. A., Varga, E. T., Lang, P., Paulus, W., (2005). Altered motion perception in migraineurs: Evidence for interictal cortical hyperexcitability. Cephalalgia, 25(10): 788-794.

Bakeman, R., (2005). Recommended effect size statistics for repeated measures designs. Behaviour Research Methods, 37: 379-384.

Baker, D. H., and Meese, T. S., (2012). Zero-dimensional noise: The best mask you never saw. Journal of Vision, 12(10): 20, 1-2.

Babiloni, C., Vecchio, F., Bultrini, A., Romani, G. L., Rossini, P. M., (2006). Pre- and poststimulus alpha rhythms are related to conscious visual perception: a high-resolution EEG study. Cerebral Cortex; 16: 1690-1700.

Bastiaansen, M. C. M., and Brunia, C. H. M., (2001). Anticipatory attention: an event-related desynchronization approach. International Journal of Psychophysiology, 43: 91-107.

Becker, R., Ritter, P., Villringer, A., (2008). Influence of ongoing alpha rhythm on the visual evoked potential. NeuroImage, 39(2): 707-716.

Ben-Simon, E., Podlipsky, I., Arieli, A., Zhandov, A., Hendler, T., (2008). Never resting brain: Simultaneous representation of two alpha-related processes in humans. PLoS ONE, 3(12): e3984.

Bjork, M., Hagen, K., Stovner, L., Sand, T., (2009). Interictal qualitative EEG in migraine: a blinded controlled study. J Headache Pain, 10, 331-9.

Bodenmann, S., Rusterholz, T., Dürr, R., Stoll, C., Bachmann, V., Geissler, E., Jaggi-Schwarz, K., Landolt, H. P., (2009). The functional Val185Met Polymorphism of COMT predicts interindividual differences in brain alpha oscillations in young men. The Journal of Neuroscience, 29(35): 10855-10862.

Brainard, D. H., (1997). The Psychophysics Toolbox. Spatial Vision, 10(4): 433-436

Braunitzer, G., Roksin, A., Kóbor, J., et al. (2012). Delayed development of visual motion processing in childhood migraine, Cephalalgia, 32, 492-496.

Brunia, C. H. M., (1999). Neural aspects of anticipatory behaviour. Acta Psychologia, 101: 213-242.

Cecere, R., Rees, G., Romei, V., (2015). Individual differences in alpha frequency drive crossmodal illusory perception. Current Biology, 25(2), 231-5.

Chorlton, P., and Kane, N., (2000). Investigation of the cerebral response to flicker stimulation in patients with headache. Clinical Electroencephalography, 31(2), 83-88.

Coppola, G., Pierelli, F., Schoenen, J., (2009). Habituation and migraine. Neurobiology of Learning and Memory, 92(2), 249-259.

Dakin, S. C., Mareschal, I., Bex, P. J., (2005). Local and global limitations on direct integration assessed using equivalent noise analysis. Vision Research, 45(24), 3027-49.

De Graaf, T. A., Gross, J., Paterson, G., Rusch, T., Sack, A. T., and Thut, G., (2013). Alpha-band rhythms in visual task performance: Phase-locking by rhythmic sensory stimulation. PLoS ONE, 8(3): e60035.

De Tommaso, M., Sciruicchio, V., Guido, M., Sasanelli, G., Specchio, L. M., Puca, F. M., (1998). EEG spectral analysis in migraine without aura attacks, Cephalalgia, 18, 324-8.

Delorme, A., and Makeig, S., (2004). EEGLAB: An open-source toolbox for the analysis of single-trial EEG dynamics. Journal of Neuroscience Methods, 134: 9-21.

Ditchfield, J. A., McKendrick, A. M., and Badcock, D. R., (2006). Processing of global motion and form in migraineurs. Vision Research, 46(1-2): 141-148.

Doppelmayr, M., Klimesch, W., Stadler, W., Pöllhuber, D., and Heine, C., (2002). EEG alpha power and intelligence. Intelligence, 30: 289-302.

Dugue, L., Marque, P., and vanRullen, R., (2011). The phase of ongoing oscillations mediates the causal relation between brain excitation and visual perception. The Journal of Neuroscience, 31(33): 11889-11893.

Ergenoglu, T., Demiralp, T., Bayraktaroglu, Z., Ergen, M., Beydagi, H., and Uresin, Y., (2004). Alpha rhythm of the EEG modulates visual detection performance in humans. Cognitive Brain Research, 20: 376-383.

Foxe, J. J., Simpson, G. V., Ahlfors, S. P., (1998). Parieto-occipital ~10Hz activity reflects anticipatory state of visual attention mechanisms. NeuroReport, 9: 3929-3933.

Fründ, I., Haenel ,N. V., and Wichmann, F. A., (2011). Inference for psychometric functions in the presence of non-stationary behaviour. Journal of Vision, 11(6): DOI: 10.1167/11.6.16

Golla, F. L., and Winter, A. L., (1959). Analysis of cerebral responses to flicker in patients complaining of episodic headache. Electroencephalogr. Clin. Neurophysiol., 11, 539-549.

Grandy, T. H., Werkle-Bergner, M., Chicherio, C., Schmiedek, F., Lövdén, M., and Lindenberger, U., (2013b) Individual alpha peak frequency is related to latent factors of general cognitive abilities. NeuroImage, 79: 10-18.

Hanslmayr, S., Klimesch, W., Sauseng, P., Gruber, W., Doppelmayr, M., Freunberger, R., and Pecherstorfer, T., (2005). Visual discrimination performance is related to decreased alpha amplitude but increased phase locking. Neuroscience Letters, 375: 64-68.

Hanslmayr, S., Aslan, A., Staudigl, T., Klimesch, W., Herrmann, C. S., and Bäuml, K. H., (2007). Prestimulus oscillations predict visual perception performance between and within subjects. NeuroImage, 37: 1465-1473.

Herrmann, C. S., (2001). Human EEG responses to 1-100Hz flicker: resonance phenomena in visual cortex and their potential correlation to cognitive phenomena. Experimental Brain Research, 137: 346-353.

International Headache Society: Headache classification committee of the International Headache Society (IHS). (2013). Classification and diagnostic criteria for headache disorders, cranial neuralgias and facial pain. Cephalalgia, 33(9): 629-808.

Jensen, O., Gips, B., Bergmann. T. O., and Bonneford, M., (2014). Temporal coding organized by coupled alpha and gamma oscillations prioritize visual processing. Trends in Neurosciences, 37(7), 357-369.

Kleiner, M., Brainard, D., and Pelli, D., (2007). What’s new in Psychtoolbox-3? Perception, 365(ECVP Supplement): s14.

Klimesch, W., Schimke, H., Doppelmayr, M., Ripper, B., Schwaiger, J., Pfurtscheller, G., (1996). Event-related desynchronization (ERD) and the Dm effect: Does alpha desynchronization during encoding predict later recall performance? International Journal of Psychophysiology, 24: 47-60.

Klimesch, W., Doppelmayr, M., and Hanslmayr, S., (2006). Upper alpha ERD and absolute power: their meaning for memory performance. Prog Brain Res, 159: 151-65.

Klimesch, W., Sauseng, P., and Hanslmayr, S., (2007). EEG alpha oscillations: The inhibition-timing hypothesis. Brain Research Reviews, 53: 63-88.

Klimesch, W., (2012). Alpha-band oscillations, attention and controlled access to stored information. Trends in Cognitive Sciences, 16(12): 606-617.

Lange, J., Oostenveld, R., Fries, P., (2013). Reduced occipital alpha power indexes enhanced excitability rather than improved visual perception. Journal of Neuroscience, 33(7): 3212-3220.

Lashley, K. S., (1941). Patterns of cerebral integration indicated by the scotomas of migraine. Archives of Neurology and Psychiatry, 46(2): 331-339.

Lu, Z. L., and Dosher, B., (1998). External noise distinguishes mechanisms of attention. Vision Research, 38, 1183-1198.

Lu, Z. L., and Dosher, B., (1999). Characterising human perceptual inefficiencies with equivalent internal noise. J. Opt. Soc. Am. A., 16(3), 764-778.

Lenth, R. V., (2016). Least-squares means: The R package lsmeans. Journal of Statistical Software, 69(1), 1-33.

McKendrick, A. M., and Backcock, D. R., (2004). Motion processing deficits in migraine. Cephalalgia, 24: 363-372.

McKendrick, A. M., Badcock, D. R., and Gurgone, M., (2006). Vernier acuity is normal in migraine, whereas global form and motion perception are not. Investigative Ophthalmology and Visual Science, 47(7): 3213-9.

Mantini, D., Perrucci, M. G., Del Gratta, C., Romani, G. L., and Corbetta, M., (2007). Electrophysiological signatures of resting state networks in the human brain. PNAS, 104(32): 13170-13175.

Marcus, D. A., and Soso, M. J., (1989). Migraine and stripe-induced visual discomfort. Archives of Neurology, 46(10):1129-1132.

Moretti, D. V., Bablioni, C., Binetti, G., Cassetta, E., Dal Forno, G., Ferric, F., Ferri, R., Lanuzza, B., Miniussi, C., Nobili, F., Rodriquez. G., Salinari, S., Rossini, P. M., (2004). Individual analysis of EEG frequency and band power in mild Alzheimer’s disease. Clinical Nuerophysiology, 115(2): 299-308.

Moretti, D. V., Miniussi, C., Frisoni, G. B., Geroldi, C., Zanetti, O., Binetti, G., and Rossini, P. M., (2007). Hippocampal atrophy and EEG markers in subjects with mild cognitive impairment. Clinical Neurophysiology, 118(12): 2716-2729.

Neufeld, M. Y., Treves, T. A., Korcyn, A. D., (1991). EEG and topographic frequency analysis in common and classic migraine. Headache: The Journal of Headache and Face Pain. 31(4): 232-236.

O’Hare, L., and Hibbard, P. B., (2016). Visual processing in migraine. Cephalalgia, 36(11): 1057-1076.

Palmer, J. E., Chronicle, E. P., Rolan. P., et al., (2000). Cortical hyperexcitability is cortical under-inhibition: Evidence from a novel functional test of migraine patients. Cephalalgia, 20: 525-532.

Pelli, D. G. (1997). The Videotoolbox software for visual psychophysics: Transforming numbers into movies. Spatial Vision, 10(4): 437-442.

Pfurtscheller, G., and Lopes da Silva, F. H., (1999). Event-related EEG/MEG synchronization and desynchronization: basic principles. Clinical Neurophysiology, 110: 1842-1857.

Posthuma, D., Neale, M. C., Boomsma, D. I., and de Guess, E. J. C., (2001). Are smarter brains running faster? Heritability of alpha peak frequency, IQ, and their interrelation. Behavior Genetics, 31(6): 567-579.

R Core Team (2013). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria, ISBN 3-900051-07-0, URL http://www.R-project.org/.

Rohenkohl, G., and Nobre, A. C., (2011). Alpha oscillations related to anticipatory attention follow temporal expectations. The Journal of Neuroscience, 31(40): 14076-14084.

Romei, V., Brodbeck, V., Michel, C., Amedi, A., Pascual-Leone, A., and Thut, G., (2008). Spontaneous fluctuations in posterior alpha-band EEG activity reflect variability in excitability of human visual areas. Cerebral Cortex, 18: 2010-2018.

Samaha, J., and Postle, B. R., (2015). The speed of alpha-band oscillations predicts the temporal resolution of visual perception. Current Biology, 25: 2985-2990

Shepherd, A. J., Beaumont, H. M., and Hine, T. J., (2012). Motion processing deficits in migraine are related to contrast sensitivity. Cephalalgia, 32(7): 554-570. doi: 10.1177/0333102412445222

Singmann., H., Bolker, B., Westfall, J., Aust, F., (2017). Afex: Analysis of factorial experiments. R package version 0.18-0, https://CRAN.R-project.org/package=afex.

Smit, D. J. A., Posthuma, D. I., Boomsma, D. I., and de Geuss, E. J. C. (2005). Heritability of background EEG across the power spectrum. Psychophysiology, 42: 691-697.

Smit, C. M., Wright, M. J., Hansell, N. K., Geffen, G. M., and Martin, N. G., (2006). Genetic variation of individual alpha frequency (IAF) and alpha power in a large adolescent twin sample. International Journal of Psychophysiology, 61: 235-243.

Strasburger, H., (2001). Converting between measures of slope of the psychometric function. Perceptual Psychophysics, 63(8), 1348-55.

Smyth, V. O. G. S., and Winter, A. L., (1964). The EEG in migraine. Electroencephalogr. Clin. Neurophysiol., 16, 194-202.

Thut, G., Veniero, D., Romei, V., Miniussi, C., Schyns, P., and Gross, J., (2011). Rhythmic TMS Causes Local Entrainment of Natural Oscillatory Signatures. Brain Topography, 22(4): 219-232.

Tibber, M. S., Guedes, A., and Shepherd, A. J., (2006). Orientation discrimination and contrast detection thresholds in migraine for cardinal and oblique angles. Investigative Ophthalmology and Visual Science, 47: 5599-5604.

Tibber, M. S., Kelly, M., Jansari, A., Dakin, S. C., and Shepherd, A. J., (2014). An inability to exclude visual noise in migraine. Investigative Ophthalmology and Visual Science, 55(4): 2539-46.

Van Dijk, H., Schoffelen, J. M., Oostenveld, R., and Jensen, O., (2008). Prestimulus oscillatory activity in the alpha band predicts visual discrimination ability. The Journal of Neuroscience, 28(8): 1816-1823.

Van Pelt, S., Boomsma, D. L., and Fries, P., (2012). Magnetoencephalography in twins reveals a strong genetic determinaion of peak frequency of visually induced gamma-band synchronisation. The Journal of Neuroscience, 32(10): 3388-3392.

Vogt, F., Klimesch, W., and Doppelmayr, M., (1998). High-frequency components in the alpha band and memory performance. Journal of Clinical Neurophysiology, 15(2): 167-72.

Webster, K. E., Dickinson, J, E., Battista, J., McKendrick, A. M., and Badcock, D. R., (2011). Increased internal noise cannot account for motion coherence processing deficits in migraine. Cephalalgia, 31(11): 1199-1210.

Wagner, D., Manahilov, V., Loffler, G., Gordon, G. E., and Dutton, G. N., (2010). Visual noise selectively degrades vision in migraine. Investigative Ophthalmology and Vision Science, 51: 2294-2299.

Wagner, D., Manahilov, V., Gordon, G. E., and Loffler, G., (2013). Global shape processing deficits are amplified by temporal masking in migraine. Investigative Ophthalmology and Vision Science, 54: 1160-1168.

Wang, J., Barstein, J., Ethridge, L. E., Mosconi, M. W., Takarae, Y., and Sweeney, J. A., (2013). Resting state EEG abnormalities in autism spectrum disorders. Journal of Neurodevelopmental Disorders, 5(24): 1-14.

Wutz, A., Muschter, E., van Koningsburggen, M. G., Weisz, N., Melcher, D., (2016). Temporal integration windows in neural processing and perception aligned to saccadic eye movements. Current Biology, 26(13), 1659-1668.

Yang, W., Chu, B., Yang, J., Yu, Y., Wu, J., and Yu, S., (2014). Elevated audiovisual temporal interaction in patients with migraine without aura. The Journal of Headache and Pain, 15, 44.