International Journal of Psychophysiology 79 (2011) 296–304
Contents lists available at ScienceDirect
International Journal of Psychophysiology j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / i j p s yc h o
An MEG investigation of the neural mechanisms subserving complex visuomotor coordination☆ Jon S. Kennedy a, Krish D. Singh b, Suresh D. Muthukumaraswamy b,⁎ a b
Brain & Behavioural Sciences Centre, Department of Psychology, The University of Birmingham, UK CUBRIC, School of Psychology, Cardiff University, Cardiff, CF10 3AT, UK
a r t i c l e
i n f o
Article history: Received 3 September 2010 Received in revised form 15 November 2010 Accepted 16 November 2010 Available online 23 November 2010 Keywords: Motor cortex Visuomotor control Gamma band response Magnetoencephalography Neural oscillations
a b s t r a c t Fifteen human participants performed a manual and ocular tracking task with a continuously and unpredictably moving visual target, while magnetoencephalography (MEG) signals were recorded. Threedimensional source reconstructions were generated from the MEG signals, using synthetic aperture magnetometry (SAM). The SAM images indicated main effects of alpha band (8–15 Hz) and beta band (15–30 Hz) source power decreases, for manual tracking in the sensorimotor and parietal cortices, and for ocular tracking in the parietal and occipital cortices. Additionally, the manual tracking task evoked a clear, contralateral motor cortex response in the form of high gamma band (60–90 Hz) source power increases. Time-frequency spectrograms revealed the induced gamma band power increases were sustained for the duration of each ten second trial demonstrating these oscillations are not simply transients associated with movement onset. The onset of the gamma band response was characterised by higher initial onset power and frequency but no correlations were observed between oscillatory power and successful tracking performance. © 2010 Elsevier B.V. All rights reserved.
1. Introduction Sensory, cognitive and motor paradigms elicit changes in the amplitude and frequency of neuronal oscillations in the electro- and magneto- encephalographic (EEG/MEG) signals (Jensen et al., 2007; Pfurtscheller et al., 1994; Swettenham et al., 2009). The BOLD response commonly used in functional Magnetic Resonance Imaging (fMRI) is spatially correlated with these changes, namely with decreases in power in the lower frequency bands of the EEG/MEG signals (K.D. Singh et al., 2003; Singh et al., 2002), and increases in power in the higher frequency bands (Brookes et al., 2005; Muthukumaraswamy and Singh, 2008, 2009). It has been suggested that these higher frequency, gamma band oscillations, loosely defined as between 30 and 100 Hz, reflect sensory integration of information; the binding of features (Fries et al., 2007; Gray and Singer, 1989). Whilst most commonly studied in vivo in the visual system, for example, with visual motion stimuli (Krishnan et al., 2005; Lutzenberger et al., 1995; Muller et al., 1997; Zaehle et al., 2009), gamma frequency oscillations have also been detected in the motor cortex using electrocorticograms (Crone et al., 1998; Miller et al., 2007, 2009; Pfurtscheller et al., 2003) and intracerebral depth electrode recordings (Szurhaj et al., 2005, 2006). Several groups have now demonstrated that non-invasive techniques (EEG/MEG) give very similar patterns of results to invasive recordings (Cheyne et al., 2008; Donner et al., 2009; Gaetz ☆ Work originated at: CUBRIC, School of Psychology, Cardiff University, Cardiff, UK. ⁎ Corresponding author. CUBRIC, School of Psychology, Cardiff University, Cardiff, UK. E-mail address:
[email protected] (S.D. Muthukumaraswamy). 0167-8760/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.ijpsycho.2010.11.003
et al., 2010; Huo et al., 2009; Tecchio et al., 2008; Waldert et al., 2008) and the validity of using non-invasive approaches to study motor cortex gamma oscillations has been confirmed by direct comparisons of invasive and surface EEG/MEG (Ball et al., 2008; Dalal et al., 2008; Darvas et al., 2009). However, the functional role of these induced gamma oscillations in motor cortex is not clear (Muthukumaraswamy, in press); it has been suggested that they reflect proprioceptive reafference following motor activity (Szurhaj et al., 2006), although they could play a more active role in motor control (Cheyne et al., 2008) or even motor planning and decision making (Donner et al., 2009). More generally, it has been suggested that fast (beta and gamma) sensorimotor oscillations may be a mechanism for sampling peripheral data to guide subsequent motor acts (Mackay, 1997), and it has been recently posited that gamma oscillations (Muthukumaraswamy, in press) may specifically serve this function. One might therefore predict from this that the degree of oscillatory power could reflect successful performance during complex visuomotor tasks. In the present study, we used a task based closely on Miall and Jenkinson's (2005) visuomotor continuous tracking task. It provided a complex motor task requiring visuomotor coordination and motor learning. Furthermore, the non-periodic nature of the target trajectory offered the potential to tap dissociable aspects of motor learning. Thus, we hoped to assess MEG oscillatory responses in correlation with each of two measures of tracking performance, one reflecting the lag of motor actions behind the visual target, and the other reflecting the imprecision of the motor actions after taking the lag into account. Furthermore, if motor cortex gamma power increases were correlated with these behavioural measures, such a result could shed light on its functional role in the motor
J.S. Kennedy et al. / International Journal of Psychophysiology 79 (2011) 296–304
297
system. We used a factorial design to allow us to separate out the main effects of ocular tracking and manual tracking from any interaction effects reflecting visuomotor coordination or shared prediction processes in the combined ocular/manual condition. Further, the expectation was that brain areas involved in shared aspects of eye and hand tracking (e.g. prediction of visual motion for motor control) should present as subadditive interactions between the two independent variables; and that areas related to the coordinated use of both oculomotor and manual motor modalities should be found as superadditive interactions.
Participants were exposed in random order to four ‘Pursue’ blocks and four ‘Fixate’ blocks of 24 trials each. In each block, the first 12 trials formed a ‘Track’ sub-block, and the remaining 12 trials formed an ‘Observe’ sub-block. Thus, each participant experienced each of four distinct combinations of ocular and manual activity, with a total of 48 trials in each condition. During each Observe sub-block, participants rested their hands, and observed a playback of their performance from the previous Track sub-block. The MEG experiment took approximately 48 min to complete.
2. Materials and methods
2.2. Behavioural performance measures
2.1. Participants, stimuli and paradigm
There are delays inherent in human sensorimotor systems and the behaviour of real-world targets and obstacles may be non-periodic and less than fully predictable. These delays pose a problem for sensorimotor coordination tasks, in a number of ways. The initial effect of a sensorimotor delay, before any compensation, is simply for the motor system to lag behind the sensory target by that amount of uncompensated delay. There are various possible indirect effects of the delay. If an attempted movement appears not to have produced any change in sensed location, correcting the apparent error would entail attempting the movement again. The delayed movement and the delayed corrective movement in the same direction would combine to create or increase an overshoot of any turning point in the target's trajectory. Indeed, this pattern of increasing amplitude of overshoots with increased sensorimotor delays has been observed in tracking a one-dimensional sinusoid trajectory (Miall et al., 1985) and a semi-predictable two-dimensional trajectory (Foulkes and Miall, 2000). Thus, in addition to the direct impact of the delay to increase the motor system's average lag behind the target, the increase in the amplitude of overshoots would also increase the variability of performance around that lagged trajectory. Thus, for each of the two “Track” conditions we used two measures to capture two core behavioural effects of the delay: a measure of temporal inaccuracy, and a measure of spatial imprecision. Inaccuracy was taken as the lag, in milliseconds, of the cursor trajectory behind the target; that is, the lag which, when adjusted for, produced the smallest root mean square spatial error between the two trajectories. Imprecision was taken as the root mean square error, in pixels, between the target and the cursor trajectories, after adjusting for inaccuracy. For each participant and condition, from the per-trial behavioural measures, two summary measures were obtained: the mean error over all trials and the rate of change per-trial over the course of the condition. A negative rate of change would be an indication of learning across the experiment (either improved temporal accuracy or spatial precision).
Fifteen healthy right-handed participants including seven females, with normal or corrected-to-normal vision (age-range 23–38; mean 27.6), volunteered for the experiment with informed consent, and each was paid fifteen pounds. The Cardiff University School of Psychology Ethics Committee approved all procedures. Participants viewed a display consisting of a centred full-screen pale blue cross-hair, a green circle (the target) and a white square (the cursor) (see Fig. 1a). During each trial, the target followed a twodimensional trajectory defined, in each axis, as the sum of five inharmonic sinusoids with phases randomized independently for each trial, multiplied by a half cosine at the start and end of each trial to ensure that it started and finished at the centre of the display. This stimulus was derived closely from that used by Miall and Jenkinson (2005) to investigate behavioural adaptation to delayed visual feedback of motor actions. Each trial used a boxcar design, with 10 s of active stimulation, preceded by a 5 second baseline period, during which only the cross-hair was present on screen. Stimuli were presented using a Sanyo XP41 LCD back-projection system with the video-card driving the projection systems at 60 Hz at 1024 × 768 pixel resolution. The stimulus subtended 25 × 19° visual angle. A TTL pulse from the stimulation computer sent event-markers to MEG and eye-tracker acquisition computers at the start of each baseline and active period. A 2 × 2 repeated-measures design was used, with one factor of ocular activity (Pursue | Fixate) and an other factor of manual activity (Track | Observe) (see Fig. 1b). This created four conditions labelled; PT (Pursue and Track), PO (Pursue and Observe), FT (Fixate and Track) and FO (Fixate and Observe). Participants were instructed to fixate the central cross-hair during Fixate trials, and to pursue the target with their eyes during the Pursue trials. For Track trials, participants were asked to use an MEG compatible joystick (fORP, Cambride Research Systems) with their right hands to control the movement of a cursor (whose position on the screen depended linearly on the joystick angle) to track the target as closely as they could. Participants operated the joystick using a power grip hand position.
Fig. 1. a) Schematic of the stimulus display. A cross-hair divided the screen, and a circular target followed a software-generated trajectory, while a square cursor was controlled by the participant via an MEG compatible joystick which was held in a power grip position. b) The 2 × 2 within-subjects design of the experiment. In each block, participants either fixated the centre of the cross-hairs with their eyes (Fixate), or pursued the target around the screen (Pursue); and either allowed their hand to rest (Observe), or used it to control the joystick, and track the target with the cursor (Track). During Observe trials, cursor movements from the last Track block were replayed.
2.3. Eye-tracking acquisition and analysis For all participants we attempted to record eye-gaze position using a remote iViewX eye-tracking system (Sensomotoric Instruments). A 13 point calibration was used prior to MEG acquisition to calibrate gazeposition and the camera operated at 50 Hz frame rate. For all participants we were able to confirm compliance with experimental instructions (pursue or fixation) however, due to technical difficulties with remote eye-tracking high quality eye-tracking data that could be analysed for the entire length of the experiment was only obtained from four participants. For analysis of the continuous eye-tracker data it was first epoched into individual trials and missing values (for example, blinks) replaced using cubic spline interpolation. Data was then smoothed using a MATLAB's® Savitzky–Golay filter (3rd order, 51 frames). From these data the mean velocity of eye movement in each active condition was calculated. 2.4. MEG acquisition and analysis Whole-head MEG recordings were made using a 275-channel radial gradiometer system sampled at 600 Hz (0–150 Hz band-pass).
298
J.S. Kennedy et al. / International Journal of Psychophysiology 79 (2011) 296–304
An additional 29 reference channels were recorded for noise cancellation purposes and the primary sensors were analysed as synthetic third order gradiometers (Vrba and Robinson, 2001). 3 of the 275 channels were turned off due to excessive sensor noise. MRI images were coregistered to the MEG data by placing, during MEG data acquisition, fiduciary markers at fixed locations relative to the tragus and eye centres, which could be identified in the corresponding anatomic MRI images. The locations were confirmed using high-resolution digital photographs taken immediately prior to participants entering the MEG scanner. A multiple local-spheres forward model (Huang et al., 1999) was used for source localization, and this was derived by fitting spheres to the brain surface, extracted by FSL's Brain Extraction Tool (Smith, 2002). Offline data were band-passed filtered, using a fourth-order bidirectional infinite impulse response Butterworth filter, into alpha (8–15 Hz), beta (15–30 Hz), gamma (50–100 Hz) frequency bands. The synthetic aperture magnetometry (SAM) beamformer algorithm as implemented in the CTF software (Robinson and Vrba, 1999) was used to create differential images of source power (pseudo-T statistics) for 5 s of baseline (−5 to 0 s) compared to 5 s of task performance (2.5 to 7.5 s). Only 5 s of the active period was used to achieve balanced covariance estimation between stimulated and unstimulated states, and the centre part of task performance was used to avoid onset transients. Details of the calculation of SAM pseudo-T source image statistics are described elsewhere (Cheyne et al., 2003; Robinson and Vrba, 1999; Singh et al., 2002; Vrba and Robinson, 2001). An isotropic voxel resolution of 4 mm was used throughout the analysis. For group analysis, SAM images were then spatially normalised using FSL's FLIRT into MNI template space using an affine transform (Jenkinson and Smith, 2001). A non-parametric (permutation testing) 2 × 2 repeated-measures ANOVA, was conducted on the normalised images, within each frequency band, with 10 mm variance smoothing (Nichols and Holmes, 2002) performed using FSL's randomise. Clusterbased thresholding to correct for multiple comparisons (Hayasaka and Nichols, 2003) was then applied with an initial cluster forming threshold of t = 2.5 (K.D. Singh et al., 2003). A non-parametric Spearman correlational analysis was also conducted across participants, to assess correlations between the volumetric pseudo t statistic images and each of the behavioural measures (inaccuracy and imprecision and their respective slopes), for the two tracking conditions (cf Cornwell et al. (2008) for a similar approach). Virtual sensors were generated by using SAM beamformer coefficients obtained using the individual condition covariance matrices band-pass filtered between 0 and 100 Hz and returning time-series from peak locations of interest (Hillebrand et al., 2005), specifically the left motor cortex and right parietal cortex. Peak locations were single spatial points chosen from each individual's SAM image. To find the peak location we applied the reverse spatial normalisation transformation from the group-level statistic images to give a point in the individual's co-ordinate space. We then selected the peak location in the individual's image that was closest to this point in 3D space. Time-frequency analyses of these virtual electrodes were conducted using the Hilbert transform from 1 to 100 Hz in 0.5 Hz steps filtering with an 8 Hz wide band-pass, 3rd order Butterworth filter (Le Van Quyen et al., 2001) and represented as a percentage change from the average baseline (−4 to 0 s) value for each frequency band. Where single-trial estimates were obtained a subtracted baseline for each trial was used. 3. Results 3.1. Behavioural results Behavioural data were skewed and kurtotic, most notably in the temporal inaccuracy measures in both conditions. Thus, all analyses
were conducted using non-parametric statistics (Wilcoxon's signed ranks test), and averages are reported as medians, with interquartile ranges (IQR). See Fig. 2 for a summary of all behavioural data. Participants were more temporally inaccurate in their visuomanual tracking when fixating centrally (188 ms, IQR = 167 to 246 ms) than when pursuing the target with their eyes (169 ms, IQR = 137 to 194 ms), Z(14) = − 3.4, p = .001. They were also more spatially imprecise (after adjusting for temporal lag) when in the Fixate trials (59.9 pixel, IQR = 57.3 to 63.3 pixel) than in the Pursue trials (53.0 pixel, IQR = 48.3 to 56.9 pixel), Z(14) = − 3.4, p = .001. There was no significant difference in the rates of change of temporal inaccuracy (Z(14)= −1.1, p = .28) and spatial imprecision (Z(14) = −1.3, p = .19) between the Fixate and the Pursue conditions. In the Pursue condition, there was a non-significant negative slope of temporal inaccuracy (−0.22 ms/trial, IQR = −0.35 to 0.14 ms/trial, Z(14) = −1.4, p = .16) and a marginally non-significant negative slope of spatial imprecision (−0.17 pixel/trial, IQR = −0.32 to 0.03 pixel/trial, Z(14) = −1.7, p = .09). In the Fixate condition, there was a significantly negative slope of both temporal inaccuracy (−0.46 ms/trial, IQR = −0.81 to 0.20 ms/trial, Z(14) = − 2.0, p = .04) and spatial imprecision (−0.17 pixel/trial, IQR = −0.29 to −0.08 pixel/trial, Z(14) = −2.7, p = .01). Thus, participants improved their temporal inaccuracy, reducing their lag behind the target, and improved their spatial imprecision, reducing the spatial variability of their lagged trajectory, in both conditions, and significantly so in the Fixate condition. 3.2. Eye-tracking results Fig. 3 a) demonstrates an example target trajectory in a) from a single PT trial in the experiment with corresponding eye movement trace in b). The target pattern can be clearly seen in the eye-tracking data. Fig. 3 c) demonstrates a trial where a participant was instructed to Fixate while tracking (FT), with 3 d) giving the corresponding eye movement trace. Fig. 3 e) illustrates mean eye movement velocities in each of the four conditions for four participants. For all four participants there is clearly more eye movement present in the Pursue conditions. A fixed-effects analysis showed that there was also significantly more eye movement in the PO condition than in the PT condition (t = 6.51, p b .0001). 3.3. Effects of ocular and manual activity on neural oscillations For each of the three frequency bands, SAM pseudo t statistic images were submitted to a 2 × 2 repeated-measures ANOVA, with factors for ocular activity and manual activity. Thresholded t statistic images of main effects are presented in Fig. 4 with local maxima of the images summarised in Table 1. In general, main effects of manual activity were found in the bilateral sensorimotor and parietal cortices, as source power decreases in the alpha and beta frequency bands, and in contralateral sensorimotor cortex as source power increases in the gamma frequency band. Main effects of ocular activity were found as source power decreases with a more posterior distribution in bilateral parietal and occipital cortices and the cerebellum, in the alpha and beta frequency bands. No interaction effects were found between the two factors for any of the frequency bands. Virtual sensors were placed in the motor cortex, and timefrequency spectra were generated for each participant. Fig. 5 shows the mean spectra, across all participants. The morphology of the spectra was remarkably similar to that obtained in the visual cortex during visual tasks (Muthukumaraswamy and Singh, 2008). The motor cortex activity consisted of a sustained source power increase in a relatively narrow band (60–90 Hz) that was sustained for the duration of the task. Integrating data across the active period demonstrated that gamma source power increases existed in both tracking conditions (PT, t = 7.00; p b .000001; FT, t = 5.76, p b .0001; PT vs FT, t = 2.43, p b 0.03) but not in observation (no manual motor
J.S. Kennedy et al. / International Journal of Psychophysiology 79 (2011) 296–304
299
Fig. 2. Behavioural data from the experiment. a) The temporal inaccuracy for FT and PT conditions; b) The spatial imprecision for FT and PT conditions; c) The slope of temporal inaccuracy for FT and PT conditions; d) The slope of spatial imprecision for FT and PT conditions (negative slopes imply the participants' performance improved across the experiment). In each case, the data are presented as a box and whiskers plot. The box represents the median and the first and third quartiles, and the whiskers extend as far as the minimum and maximum values after outliers (values more than 1.5 interquartile ranges below the lower quartile or above the upper quartile) were excluded. Excluded outliers are displayed as circles or, when the values are more than 3 interquartile ranges outside the first and third quartiles, as stars.
task) conditions (PO, t = − 0.5; p = 0.6; FO, t = −1.15; p = 0.27). The grand mean gamma peak frequency was 68.6 Hz (range 63 to 74.5 Hz; std = 3.23 Hz). Gamma frequencies were not significantly different between conditions PT and FT but there was significantly more gamma power in the PT condition (t = 2.43, p b .03). Visual inspection of the grand-averaged tracking condition spectrograms indicated the existence of an initial burst of gamma activity that was slightly higher in frequency and power than the sustained frequency. To analyse this, peak frequency and power estimates were obtained between 60 and 90 Hz for the time epochs of 0.25 s to 0.75 s and 1 to 8 s respectively.
t tests showed significantly higher onset frequency for PT (t = 4.35, p b .001) and FT (t = 3.55, p b .005) as well as higher onset power for FT (t = 3.16, p b .007) and marginally higher onset power for PT (t = 1.98, p = .07). Beta power decreases were also present in all conditions in motor cortex virtual sensors (PT, t = −13.9 p b 1 × 10− 8; PO, t = −3.6, p b .003; FT, t = −13.3, p b 1 × 10− 9; FO, t = −5.53, p b 0.0001) but to a much greater extent in the tracking conditions (PT vs PO, t = − 6.60, p b .00001; FT vs FO, t = −7.95, p b 1 × 10− 6). Virtual sensors were also generated for locations in the right parietal cortex with sustained source power decreases seen in the
Fig. 3. Eye-movement data from the experiment. a) Example target trajectory from a single PT trial with the corresponding eye-movement position trace shown in b). c) Example target trajectory from a single FT trial with the corresponding eye movement position trace shown in d). e) Mean eye movement velocities for eye movement data from four participants for each of the conditions. Error bars represent the standard error of the mean; units are pixels/ms.
300
J.S. Kennedy et al. / International Journal of Psychophysiology 79 (2011) 296–304
Fig. 4. Group-level data showing the SAM source localisation results of the 2× 2 repeated measures permutation-based ANOVA for three frequency bands, alpha (8–15 Hz), beta (15–30 Hz) and gamma (50–100 Hz). Main effects of manual and ocular activity were seen but no interaction effects were present. Units are pseudo t scores and images are thresholded at pb .05 (corrected). Locations of local maxima in the images are presented in Table 1.
time-frequency analysis seen for the entire task period (Fig. 6) hence data were integrated for the full period into separate beta (15–30 Hz) and alpha (8–15 Hz) bands. Beta source power decreases were present in all conditions with greater decreases in the ocular movement conditions than fixation conditions (PO vs FO, t = −5.62, p b .00001; PT vs FT, t = 2.98, p b .01). Increased beta source power was also present in manual tracking conditions (PT vs PO, t = 5.92, p b 0.0001; FT vs FO t = 2.8, p b .02). Alpha power decreases were present in all conditions with the greatest decreases occurring in the ocular movement conditions (FO vs PO, t = 7.12, p b 1 × 10− 6; FT vs PT, t = 3.82, p b .002). For both Figs. 5 and 6 a small amount of power
increase can be seen in the early part of the baseline (− 5 to − 4 s), however, these were not included in the baseline period used for time-frequency estimation. These power increases probably reflect the tail end of the post movement beta rebound (Jurkiewicz et al., 2006; Pfurtscheller, 1992) from the previous trial. 3.4. Correlations with behavioural measures Two approaches were taken to probe for potential correlations between the behavioural and MEG data. In the first approach (Cornwell et al., 2008), for each of the measures of error and their
Table 1 Location of activation foci for significant (p b .05, corrected) main effects of manual and ocular activity for each frequency band. Corresponding images are presented in Fig. 4. No significant interaction effects were observed. The units of the peak values are pseudo-T scores and the XYZ co-ordinates are defined in Talairach space. Frequency
Alpha
Beta
Gamma
Peak
Manual activity X
Y
− 9.77 − 5.79 − 5.63 − 5.59 − 5.57 − 4.91 − 4.81 − 4.39 − 3.77
− 49 57 19 35 41 − 23 45 23 − 47
− 31 − 25 − 59 − 17 − 71 − 81 − 55 − 83 − 65
55 39 65 63 3 33 33 25 9
− 12.3 − 10.3 − 9.16 − 7.93 − 5.11 − 4.59 9.12
− 33 35 −1 49 39 41 − 27
− 33 − 23 − 31 −1 − 51 13 − 21
59 59 57 35 −1 − 15 53
Gyral location
Frequency
Peak
L postcentral gyrus R postcentral gyrus R superior parietal R precentral gyrus R middle occipital gyrus L precuneus R supramarginal R precuneus L middle temporal
Alpha
L postcentral gyrus R precentral gyrus L paracentral lobule R precentral gyrus R middle frontal gyrus R superior temporal gyrus L precentral gyrus
Beta
− 9.27 − 8.95 − 7.97 − 7.96 − 7.85 − 6.94 − 6.32 − 6.04 − 4.63 − 4.52 − 11 − 9.6 − 4.81 − 4.39
Z
Gamma
Ocular activity
Gyral location
X
Y
Z
21 7 41 45 − 37 15 − 35 − 55 − 53 55 33 23 37 − 55
83 − 85 − 69 − 65 − 83 − 65 − 65 − 59 − 31 − 17 − 71 − 77 − 45 −3
31 35 −1 − 13 −1 51 − 39 3 23 21 3 25 53 19
R cuneus R cuneus R inferior temporal gyrus R fusiform L middle occipital gyrus R precuneus L cerebellum L middle temporal gyrus L inferior parietal lobule R postcentral gyrus R middle occipital gyrus R precuneus R inferior parietal lobule L precentral gyrus
J.S. Kennedy et al. / International Journal of Psychophysiology 79 (2011) 296–304
301
Fig. 5. Grand-averaged time-frequency analysis of virtual sensors constructed from left (contralateral) primary motor cortex for each of the four conditions in the experiment. Units are % change from baseline. Pronounced gamma ERS and concurrent beta ERD is seen in both of the manual tracking conditions. Weaker beta ERD is seen in the passive observation conditions.
corresponding slopes, these summary statistics for each participant were correlated with the corresponding three dimensional SAM volumes in each condition for each of the three frequency bands (alpha, beta and gamma) using obtained. None of these correlations demonstrated significance at the p = .05 level (corrected). In the second approach, single-trial estimates of source power were obtained from time-frequency analysis of the virtual sensors analysed in Figs. 4 and 5 for alpha, beta and gamma frequency bands by integration (Muthukumaraswamy, in press). These analyses were performed firstly per participant (~48 trials, each of 10 s duration) and secondly in fixed-effects analyses after means were removed for
each participant (~720 trials, each of 10 s duration) and combined (~1440 trials). However, none of these analyses yielded systematic significant results. 4. Discussion In all tracking conditions, participants showed reductions in both temporal inaccuracy and spatial imprecision, but this reached significance only in the conditions in which participants fixated centrally. In the group-level MEG data, there were bilateral parietal, occipital and cerebellar source power decreases in the alpha and beta
Fig. 6. Grand-averaged time-frequency analysis of virtual sensors constructed from right parietal cortex for each of the four conditions in the experiment. Units are % change from baseline. Concurrent alpha and beta ERD is seen in all conditions but is most pronounced in the Pursue conditions.
302
J.S. Kennedy et al. / International Journal of Psychophysiology 79 (2011) 296–304
bands in the ocular pursuit conditions, compared with the ocular fixation conditions. The motor cortex showed effects of manual activity in the form of bilateral alpha and beta source power decreases in sensorimotor and parietal cortices and contralateral motor cortex gamma power increases. These effects were sustained for the duration of each trial. Neither subadditive nor superadditive interactions between ocular and manual activity were observed; nor were there correlations between the behavioural measures and the MEG data. While the group-level SAM source reconstructions from the current experiment show relatively diffuse changes in source power it is important to be careful in the interpretation of these images. Unlike fMRI where image voxels are largely spatially independent, adjacent voxels (and beyond) in beamformer source images exhibit high degrees of correlation. Moreover, as source power decreases the FWHM of the images actually become larger (Barnes and Hillebrand, 2003). That is, it is important not to interpret greater volumes of source power change as being indicative of more cortex being actively modulated. For group-level analyses, errors in, spatial normalisation, individual co-registration of MRI to MEG and spatial smoothing can also further contribute to the diffuseness of the images as well as inherent individual variability in (functional) anatomy and task strategy. That said, many of the areas found to be active in this experiment (Table 1) are consistent with the known functions of these areas, including sensorimotor and parietal cortices for manual tasks and bilateral parietal, and MT for the ocular task. Somewhat surprisingly, we did not find any modulation of areas corresponding to frontal eye fields in our source reconstructions for the ocular movement factor, as MEG has been demonstrated to be sensitive to FEF activity although generally in event-related contexts (Ahlfors et al., 1999; Herdman and Ryan, 2007). It has been demonstrated that some induced gamma oscillations recorded from the scalp using EEG may directly reflect eye-muscle artefacts from microsaccades (Yuval-Greenberg et al., 2008) or indeed neck and shoulder muscle artefacts (Whitham et al., 2008) and given the extensive eye-movements in this paradigm this is worth addressing. It must be taken into account that MEG is much less susceptible to these problems than EEG as MEG is a) a reference-free recording technique and b) contains much less volume conduction (field spread) than EEG recordings. Previous analyses of full sensor MEG topographic maps during movement induced gamma power increases showed no indication of eye-movement contamination (Muthukumaraswamy, in press) with very similar motor cortex timefrequency spectra to those we obtained here. It was also demonstrated in that paper that movement-related gamma power increases are not phase-locked to either EMG (not recorded here) or stimulus onset and as such we regard the power changes here as induced power changes. 4.1. Behavioural correlates of motor learning Participants' performance improved over the course of the experiment, with their lag behind the target reducing, and the variable error around the lagged trajectory reducing also. There are several types of compensation for errors in tracking non-periodic targets in the presence of sensorimotor delay: modification of various aspects of a forward model of the effector, or of various aspects of an inverse model controlling the effector. An inverse model may seek to null the temporal error between the target and the effector; in other words, the target's future trajectory may be predicted and a more anticipatory response made to catch up with it. If an inverse model were successfully adapted to the delay, the lag would be reduced. Any imprecision in the inverse model's prediction of the target behaviour (as there must be, if the target is not periodic or wholly predictable) would add to the variable error of the tracking. If a forward model compensated for the delay in sensory feedback, overshoot errors due to delayed feedback would be reduced, thus
decreasing variable error, as well as the temporal lag. This sort of forward model has been described by Smith (1959) in relation to engineering, and has been discussed by Miall et al. (1993) in relation to human performance. Variable error could also be reduced through more precise prediction of the target behaviour or more precise execution of movements. An alternative strategy, identified by Miall et al. is to reduce the gain and speed of feedback error corrective movements, which would tend to produce more sluggish but less variable tracking. Thus, the extent to which these various strategies are employed can be assessed by measuring changes in temporal inaccuracy and spatial imprecision in a tracking task. Given that no artificial sensorimotor delay is involved in the task, modification of the forward model should not arise. Reduction in temporal inaccuracy should reflect a modification of the input to an inverse model to increase the anticipation of the target, whereas an increase in temporal inaccuracy should reflect the reduced feedback gain strategy identified by Miall et al. (1993). Other things being equal, precision should be increased by the reduced correction strategy and decreased by increased anticipation. Increases in precision could reflect increased refinement of inverse models, perception or the execution of actions. Reduction in temporal inaccuracy in a task with only normal sensorimotor delays (neural visuomotor delays and the imperceptible delay due to the hardware) should, therefore, reflect changes in the gain of the temporal error (between the controlled cursor and the target) in an inverse model, since the forward model is presumably already tuned to normal sensorimotor delays. Thus, performance overall was consistent with increased temporal accuracy in prediction of the target and, additionally, increased precision of perception, prediction or execution (since increased gain on an inverse model alone should not predict increased precision). Performance was inconsistent with any reduction in the gain of the error in the inverse model over the course of the experiment, as this would have resulted in increasingly sluggish responses, hence a decrease in temporal accuracy. In the absence of any perceptual measures, our behavioural data cannot speak to the source of the increased precision in the sensorimotor loop. In sum, then, participants' behaviour indicated that, over the course of the experiment, they developed an increasingly anticipatory behavioural strategy, with increasingly (temporally) accurate and precise tracking of the target. 4.2. Neural correlates of motor learning Motor cortex activity accompanied manual tracking, contrasted with observation, and both occipital and cerebellar activity accompanied ocular tracking, contrasted with fixation. The main effects alone cannot identify which aspects of manual and ocular tracking elicited these neural signals, since tracking (relative to rest or fixation) has both motor and sensory consequences. Correlations with behavioural measures, if detected, could have complemented these results to clarify the motor learning correlates of the MEG data. In addition to any main effects of ocular and manual tracking, we had expected to find interactions between ocular and manual tracking; hence the factorial design. Brain areas involved in shared aspects of eye and hand tracking (for example, prediction of visual motion for motor control) should present as subadditive interactions between the two IVs; and that areas related to the coordinated use of both oculomotor and manual motor modalities should be found as superadditive interactions. However, no significant interactions were detected in the ANOVA. Aside from insufficient power to detect what may be relatively small effects, there are possible explanations for the lack of interactions. The lack of a subadditive interaction (for shared prediction of the target across ocular and manual tracking) could reflect a lack of prediction in the ocular tracking. Perhaps the error signal due to a lag between two visual objects (the target and the controlled cursor)
J.S. Kennedy et al. / International Journal of Psychophysiology 79 (2011) 296–304
better motivates prediction and anticipatory behaviour than does the error signal due to lagged eye movements tracking a single visual object. However, this explanation is inconsistent with Miall et al.'s (2001) behavioural data from a task similar to the one used in the present study, showing that eye and hand are synchronously coordinated in their predictive tracking of the target. An alternative explanation, which could also explain the lack of a superadditive interaction (for oculomanual coordination), would be that the neural loci for these processes are in areas such as the cerebellum (which would be consistent with Miall et al.'s, fMRI results) and that these processes may not be exhibited as changes in macroscopic oscillatory power as detected with MEG. Thus, a useful modification to the present study that could more powerfully detect the interactions would be to run it with fMRI in humans, or invasively in non-human animals. 4.3. Sustained motor cortex gamma In the present study, we detected induced gamma power changes in the motor cortex over sustained periods of time, with a continuous, closed-loop motor task. Previous detection of motor cortex gamma has been in brief bursts accompanying the onset of movements (Cheyne et al., 2008; Gaetz et al., 2010) and identical repetitively performed movements (Muthukumaraswamy, in press). The current data show that induced motor cortex gamma band responses can be sustained for long periods of time (10 s) similar to what has been reported in the visual system (Handel and Haarmeier, 2009; Hoogenboom et al., 2006; Koch et al., 2009; Muthukumaraswamy and Singh, 2008). It has previously been found (Muthukumaraswamy, in press) that for repetitive movement sequences that the first movement of the sequence is characterised by higher frequency and power. This previous finding is extended here to include non-repetitive movements and suggests a more general interpretation of the former, that is, when the motor cortex moves from a state of rest to activity that this is accompanied by initially higher frequency and power. This seems somewhat analogous to gamma oscillations in the visual cortex where an initially higher frequency and power burst is seen (Muthukumaraswamy et al., 2010). In our experience (Muthukumaraswamy, in press; Muthukumaraswamy et al., 2009, 2010; Muthukumaraswamy and Singh, 2008, 2009; Swettenham et al., 2009) gamma oscillations frequencies tend to be higher in frequency in primary motor cortex than visual cortex although this has not yet been demonstrated in the same set of individuals. Previous work in our laboratory has demonstrated with concurrent movement, EMG and MEG recordings (Muthukumaraswamy, in press) that gamma oscillation (and EMG) power is greater for larger amplitude movements. From this it was suggested that gamma oscillations may play a role in guiding ongoing motor behaviour. Indeed there is now extensive evidence which demonstrate correlations between gamma band activity and a number of behaviours including reaction time speed, response accuracy and memory performance (see Hermann et al. (2010) for a review). Here, we tested whether gamma power may reflect successful execution of a complex visuomotor task, but found no evidence for this, that is, we found no correlations between gamma oscillation power and any of our behavioural metrics. It may of course be that there was insufficient power in the current study to detect such an effect. In this experiment we were interested in the sustained aspects of brain activity that might support the visuomotor tracking task. Hence, the analysis here focussed on characterising sustained oscillations rather than onset transients (evoked fields) and for a similar reason in the design of this experiment we used relatively long (10 s active 5 s resting) but fewer trials to maximise tracking time. However, a significant amount of data contributed to each of the individual source localisations (~240 s (5 s × 48 trials)), which is more than enough for accurate covariance matrix estimation and source image reconstruction (Brookes et al., 2007). For power estimation in the virtual sensor timefrequency analyses this time was doubled to ~480 s for each condition
303
per individual. Alternatively, it may be that gamma oscillations reflect lower-level kinematic aspects of the task, for example, the amount the hand moved, or needs to move, in order to achieve the desired kinematic effect. This encoding could be relatively independent of successful tracking ability. In retrospect inclusion of a joystick only condition may have provided useful information on this, but nevertheless the current results add useful information in trying to elucidate what the functional role of gamma oscillations in motor control are. Due to the complex compound nature of manipulating the joystick in the current study we did not take measures of muscle activity or kinematics, unlike in previous work (Muthukumaraswamy, in press). In sum, participants adapted to a non-periodic visuomanual tracking task, showing reductions in temporal inaccuracy consistent with increased anticipatory prediction of the target. The manual tracking task produced motor cortex activity, and the ocular tracking task produced occipital and parietal cortex activity. The motor cortex activity included an induced component in the gamma frequency band, which was sustained for the duration of the trials and characterised by an initial burst that was higher in both frequency and power, however neither gamma or beta power in motor or parietal cortices correlated with successful task performance or improvement. Acknowledgements The work reported in this paper was funded by the Wales Institute for Cognitive Neuroscience and the School of Psychology, Cardiff University. References Ahlfors, S.P., Simpson, G.V., Dale, A.M., Belliveau, J.W., Liu, A.K., Korvenoja, A., Virtanen, J., Huotilainen, M., Tootell, R.B., Aronen, H.J., Ilmoniemi, R.J., 1999. Spatiotemporal activity of a cortical network for processing visual motion revealed by MEG and fMRI. Journal of Neurophysiology 82, 2545–2555. Ball, T., Demandt, E., Mutschler, I., Neitzel, E., Mehring, C., Vogt, K., Aertsen, A., SchulzeBonhage, A., 2008. Movement related activity in the high gamma range of the human EEG. Neuroimage 41, 302–310. Barnes, G.R., Hillebrand, A., 2003. Statistical flattening of MEG beamformer images. Human Brain Mapping 18, 1–12. Brookes, M.J., Gibson, A.M., Hall, S.D., Furlong, P.L., Barnes, G.R., Hillebrand, A., Singh, K.D., Holliday, I.E., Francis, S.T., Morris, P.G., 2005. GLM-beamformer method demonstrates stationary field, alpha ERD and gamma ERS co-localisation with fMRI BOLD response in visual cortex. Neuroimage 26, 302–308. Brookes, M.J., Vrba, J., Robinson, S.E., Stevenson, C.M., Peters, A.M., Barnes, G., Hillebrand, A., Norris, P.G., 2007. Optimising experimental design for meg beamformer imaging. Neuroimage 39 (4), 1788–1802. Cheyne, D., Gaetz, W., Garnero, L., Lachaux, J.-P., Ducorps, A., Schwartz, D., Varela, F.J., 2003. Neuromagnetic imaging of cortical oscillations accompanying tactile stimulation. Cognitive Brain Research 17, 599–611. Cheyne, D., Bells, S., Ferrari, P., Gaetz, W., Bostan, A.C., 2008. Self-paced movements induce high-frequency gamma oscillations in primary motor cortex. Neuroimage 42, 332–342. Cornwell, B.R., Johnson, L.L., Holroyd, T., Carver, F.W., Grillon, C., 2008. Human hippocampal and parahippocampal theta during goal-directed spatial navigation predicts performance on a virtual Morris water maze. Journal of Neuroscience 28, 5983–5990. Crone, N.E., Miglioretti, D.L., Gordon, B., Lesser, R.P., 1998. Functional mapping of human sensorimotor cortex with electrocorticographic spectral analysis. II. Eventrelated synchronization in the gamma band. Brain 121 (Pt 12), 2301–2315. Dalal, S.S., Guggisberg, A.G., Edwards, E., Sekihara, K., Findlay, A.M., Canolty, R.T., Berger, M.S., Knight, R.T., Barbaro, N.M., Kirsch, H.E., Nagarajana, S.S., 2008. Fivedimensional neuroimaging: localization of the time-frequency dynamics of cortical activity. Neuroimage 40, 1686–1700. Darvas, F., Scherer, R., Ojemann, J.G., Rao, R.P., Miller, K.J., Sorensen, L.B., 2009. High gamma mapping using EEG. Neuroimage 49, 930–938. Donner, T.H., Siegel, M., Fries, P., Engel, A.K., 2009. Buildup of choice-predictive activity in human motor cortex during perceptual decision making. Current Biology 19, 1581–1585. Foulkes, A.J., Miall, R.C., 2000. Adaptation to visual feedback delays in a human manual tracking task. Exp Brain Res 131, 101–110. Fries, P., Nikolic, D., Singer, W., 2007. The gamma cycle. Trends Neurosci 30, 309–316. Gaetz, W.C., MacDonald, M., Cheyne, D., Snead, O.C., 2010. Neuromagnetic imaging of movement-related cortical oscillations in children and adults: Age predicts postmovement beta rebound. Neuroimage 51, 792–807. Gray, C.M., Singer, W., 1989. Stimulus-specific neuronal oscillations in orientation columns of cat visual-cortex. Proceedings of the National Academy of Sciences of the United States of America 86, 1698–1702.
304
J.S. Kennedy et al. / International Journal of Psychophysiology 79 (2011) 296–304
Handel, B., Haarmeier, T., 2009. Cross-frequency coupling of brain oscillations indicates the success in visual motion discrimination. Neuroimage 45, 1040–1046. Hayasaka, S., Nichols, T.E., 2003. Validating cluster size inference: random field and permutation methods. Neuroimage 20, 2343–2356. Herdman, A.T., Ryan, J.D., 2007. Spatio-temporal brain dynamics underlying saccade execution, suppression, and error-related feedback. Journal of Cognitive Neuroscience 19, 420–432. Hermann, C.S., Frund, I., Lenz, D., 2010. Human gamma-band activity: a review on cognitive and behavioral correlates and network models. Neuroscience and Biobehavioral Reviews 34, 981–992. Hillebrand, A., Singh, K.D., Holliday, I., Furlong, P.L., Barnes, G.R., 2005. A new approach to neuroimaging with magnetoencephalography. Human Brain Mapping 25, 199–211. Hoogenboom, N., Schoffelen, J.M., Oostenveld, R., Parkes, L.M., Fries, P., 2006. Localizing human visual gamma-band activity in frequency, time and space. Neuroimage 29, 764–773. Huang, M.X., Mosher, J.C., Leahy, R.M., 1999. A sensor-weighted overlapping-sphere head model and exhaustive head model comparison for MEG. Physics in Medicine and Biology 44, 423–440. Huo, X., Xiang, J., Wang, Y., Kirtman, E.G., Kotecha, R., Fujiwara, H., Hemasilpin, N., Rose, D.F., Degrauw, T., 2009. Gamma oscillations in the primary motor cortex studied with MEG. Brain Development 32, 619–624. Jenkinson, M., Smith, S., 2001. A global optimisation method for robust affine registration of brain images. Medical Image Analysis 5, 143–156. Jensen, O., Kaiser, J., Lachaux, J.P., 2007. Human gamma-frequency oscillations associated with attention and memory. Trends in Neurosciences 30, 317–324. Jurkiewicz, M.T., Gaetz, W.C., Bostan, A.C., Cheyne, D., 2006. Post-movement beta rebound is generated in motor cortex: evidence from neuromagnetic recordings. Neuroimage 32, 1281–1289. Koch, S.P., Werner, P., Steinbrink, J., Fries, P., Obrig, H., 2009. Stimulus-induced and state-dependent sustained gamma activity is tightly coupled to the hemodynamic response in humans. Journal of Neuroscience 29, 13962–13970. Krishnan, G.P., Skosnik, P.D., Vohs, J.L., Busey, T.A., O'Donnell, B.F., 2005. Relationship between steady-state and induced gamma activity to motion. Neuroreport 16, 625–630. Le Van Quyen, M., Foucher, J., Lachaux, J.P., Rodriguez, E., Lutz, A., Martinerie, J., Varela, F.J., 2001. Comparison of Hilbert transform and wavelet methods for the analysis of neuronal synchrony. Journal of Neuroscience Methods 111, 83–98. Lutzenberger, W., Pulvermuller, F., Elbert, T., Birbaumer, N., 1995. Visual stimulation alters local 40-Hz responses in humans: an EEG-study. Neuroscience Letters 183, 39–42. Mackay, W.A., 1997. Synchronized neuronal oscillations and their role in motor processes. Trends in Cognitive Sciences 1, 176–183. Miall, R.C., Jenkinson, E.W., 2005. Functional imaging of changes in cerebellar activity related to learning during a novel eye-hand tracking task. Experimental Brain Research 166, 170–183. Miall, R.C., Weir, D.J., Stein, J.F., 1985. Visuomotor tracking with delayed visual feedback. Neuroscience 16, 511–520. Miall, R.C., Weir, D.J., Wolpert, D.M., Stein, J.F., 1993. Is the cerebellum a smith predictor? Journal of Motor Behavior 25, 203–216. Miall, R.C., Reckess, G.Z., Imamizu, H., 2001. The cerebellum coordinates eye and hand tracking movements. Nature Neuroscience 4, 638–644. Miller, K.J., Leuthardt, E.C., Schalk, G., Rao, R.P., Anderson, N.R., Moran, D.W., Miller, J.W., Ojemann, J.G., 2007. Spectral changes in cortical surface potentials during motor movement. Journal of Neuroscience 27, 2424–2432. Miller, K.J., Zanos, S., Fetz, E.E., den Nijs, M., Ojemann, J.G., 2009. Decoupling the cortical power spectrum reveals real-time representation of individual finger movements in humans. Journal of Neuroscience 29, 3132–3137. Muller, M.M., Junghofer, M., Elbert, T., Rochstroh, B., 1997. Visually induced gammaband responses to coherent and incoherent motion: a replication study. Neuroreport 8, 2575–2579. Muthukumaraswamy, S.D., in press. Functional properties of human primary motor cortex gamma oscillations. Journal of Neurophysiology 104, 2873–2885. Muthukumaraswamy, S.D., Singh, K.D., 2008. Spatiotemporal frequency tuning of BOLD and gamma band MEG responses compared in primary visual cortex. Neuroimage 40, 1552–1560.
Muthukumaraswamy, S.D., Singh, K.D., 2009. Functional decoupling of BOLD and gamma-band amplitudes in human primary visual cortex. Human Brain Mapping 30, 2000–2007. Muthukumaraswamy, S.D., Edden, R.A.E., Jones, D.K., Swettenham, J.B., Singh, K.D., 2009. Resting GABA concentration predicts peak gamma frequency and fMRI amplitude in response to visual stimulation in humans. Proceedings of the National Academy of Sciences of the United States of America 106, 8356–8361. Muthukumaraswamy, S.D., Singh, K.D., Swettenham, J.B., Jones, D.K., 2010. Visual gamma oscillations and evoked responses: variability, repeatability and structural mri correlates. Neuroimage 49, 3349–3357. Nichols, T.E., Holmes, A.P., 2002. Nonparametric permutation tests for functional neuroimaging: a primer with examples. Human Brain Mapping 15, 1–25. Pfurtscheller, G., 1992. Event-related synchronization (ERS): an electrophysiological correlate of cortical areas at rest. Electroencephalography and Clinical Neurophysiology 83, 62–69. Pfurtscheller, G., Neuper, C., Mohl, W., 1994. Event-related desynchronization (ERD) during visual processing. International Journal of Psychophysiology 16, 147–153. Pfurtscheller, G., Graimann, B., Huggins, J.E., Levine, S.P., Schuh, L.A., 2003. Spatiotemporal patterns of beta desynchronization and gamma synchronization in corticographic data during self-paced movement. Clinical Neurophysiology 114, 1226–1236. Robinson, S.E., Vrba, J., 1999. Functional neuroimaging by synthetic aperture manetometry (SAM). In: Yoshimoto, T., Kotani, M., Kuriki, S., Karibe, H., Nakasato, N. (Eds.), Recent Advances in Biomagnetism. Tohoku University Press, Sendai, pp. 302–305. Singh, K.D., Barnes, G.R., Hillebrand, A., Forde, E.M.E., Williams, A.L., 2002. Task-related changes in cortical synchronization are spatially coincident with the hemodynamic response. Neuroimage 16, 103–114. Singh, K.D., Barnes, G.R., Hillebrand, A., 2003a. Group imaging of task-related changes in cortical synchronisation using nonparametric permutation testing. Neuroimage 19, 1589–1601. Singh, M., Kim, S., Kim, T.-S., 2003b. Correlation between BOLD-fMRI and EEG signal changes in response to visual stimulus frequency in humans. Magnetic Resonance in Medicine 49, 108–114. Smith, O.J.M., 1959. A controller to overcome dead time. ISA 6, 28–33. Smith, S.M., 2002. Fast robust automated brain extraction. Human Brain Mapping 17, 143–155. Swettenham, J.B., Muthukumaraswamy, S.D., Singh, K.D., 2009. Spectral properties of induced and evoked gamma oscillations in human early visual cortex to moving and stationary stimuli. Journal of Neurophysiology 102, 1241–1253. Szurhaj, W., Bourriez, J.L., Kahane, P., Chauvel, P., Mauguiere, F., Derambure, P., 2005. Intracerebral study of gamma rhythm reactivity in the sensorimotor cortex. European Journal of Neuroscience 21, 1223–1235. Szurhaj, W., Labyt, E., Bourriez, J.L., Kahane, P., Chauvel, P., Mauguiere, F., Derambure, P., 2006. Relationship between intracerebral gamma oscillations and slow potentials in the human sensorimotor cortex. European Journal of Neuroscience 24, 947–954. Tecchio, F., Zappasodi, F., Porcaro, C., Barbati, G., Assenza, G., Salustri, C., Rossini, P.M., 2008. High-gamma band activity of primary hand cortical areas: a sensorimotor feedback efficiency index. Neuroimage 40, 256–264. Vrba, J., Robinson, S.E., 2001. Signal processing in magnetoencephalography. Methods 25, 249–271. Waldert, S., Preissl, H., Demandt, E., Braun, C., Birbaumer, N., Aertsen, A., Mehring, C., 2008. Hand movement direction decoded from MEG and EEG. Journal of Neuroscience 28, 1000–1008. Whitham, E.M., Lewis, T., Pope, K.J., Fitzgibbon, S.P., Clark, C.R., Loveless, S., DeLosAngeles, D., Wallace, A.K., Broberg, M., Willoughby, J.O., 2008. Thinking activates EMG in scalp electrical recordings. Clinical Neurophysiology 119, 1166–1175. Yuval-Greenberg, S., Tomer, O., Keren, A.S., Nelken, I., Deouelll, L.Y., 2008. Transient induced gamma-band response in EEG as a manifestation of miniature saccades. Neuron 58, 429–441. Zaehle, T., Frund, I., Schadow, J., Tharig, S., Schoenfeld, M.A., Herrmann, C.S., 2009. Interand intra-individual covariations of hemodynamic and oscillatory gamma responses in the human cortex. Frontiers in Human Neuroscience 3, 8.