Functional connectivity during real vs imagined visuomotor tasks: an EEG study

Size: px

Start display at page:

Download "Functional connectivity during real vs imagined visuomotor tasks: an EEG study"

Anna Perry
5 years ago
Views:

1 MOTOR SYSTEMS Functional connectivity during real vs imagined visuomotor tasks: an EEG study James M. Kilner, 1,CA Yves Paulignan and Driss Boussaoud Institut des Sciences Cognitives, 67 Boulevard Pinel, Bron, France 1 Present Address: Functional Imaging Laboratory,Wellcome Department of Imaging Neuroscience,12 Queen Square, London WC1N 3BG,UK CA Corresponding author: jkilner@ l.ion.ucl.ac.uk Received 8 December 2003; accepted 29 December 2003 DOI: /01.wnr It is proposed that real and imagined movements activate identical neural networks. Cortical oscillatory activity is proposed as a mechanism through which distributed neuronal networks may bind into coherent ensembles and coupling of oscillators is used as a tool to investigate modulations of cortical connectivity. The aim of the present study was to test the hypothesis that, although the same brain network is involved in both real and imagined movements, the functional connectivity within the network di ers. To do so, we measured interregional coupling, quanti ed using coherence between scalp EEG electrodes, during di erent periods of a prehension task during real and imagined movements. The results demonstrated a di erent pattern of coupling in the beta frequency range between electrodes overlying occipital and motor cortices during executed and imagined movements. These ndings are consistent with the hypothesis that the neural networks during real and imagined movements are not identical. NeuroReport 15:637^ 642 c 2004 Lippincott Williams & Wilkins. Key words: EEG; Imaging; Motor; Oscillations; Synchrony INTRODUCTION Differences between imagining and performing a motor act must be reflected in different neuronal dynamics, either the neuronal activity or the network connectivity. Neuroimaging studies of real and imagined movements have demonstrated that although some cortical areas are differentially activated during the two conditions [1], other cortical and sub-cortical areas, such as parietal and premotor cortex, the basal ganglia and the cerebellum, are similarly activated during both real and imagined movements [2 5]. However, these neuroimaging studies have exclusively tested hypotheses investigating task-dependent modulations in intra-areal neuronal activity. It is possible that any substantial differences between imagining and executing a movement may lie not in changes in neuronal activity within a given area but in modulations in the functional connectivity between activated cortical areas. Several studies have investigated task-dependent modulations of oscillatory activity within and between cortical areas known to be involved in visuomotor integration. Roelfsema et al. [6] recorded oscillatory activity in the visual, parietal and motor areas of the cat brain whilst the cat performed a visuomotor coordination task. They showed that oscillatory activity 4 20 Hz within each area became synchronised between areas during periods of visuomotor integration. Such zero-time-lag synchronisation was observed between parietal and motor electrodes and between parietal and visual electrodes. In humans, interareal coupling of cortical oscillators has also been demonstrated specifically during periods of visuomotor processing. Classen et al. [7] recorded whole-head scalp EEG while subjects performed either a visual force tracking task or tasks that required uniquely visual or motor processing. They demonstrated significant interelectrode coherence in the beta band between occipital and frontal electrodes. Furthermore, they showed that such coupling was only present when both visual and motor processing were required. They concluded that the coupling of oscillatory activity between visual and motor areas could underlie the process of visuomotor integration. The current study was undertaken to examine cortical oscillatory activity during prehension movements. We measured coupling of oscillatory activity within and between the sensorimotor, parietal and occipital cortex and found that intraregional neuronal activity is similar in both imagined and real movements. However, significant differences existed in the degree of coupling between these cortical regions. This suggests that any differences between real and imagined movement are realised not by the activity in a given area but by the functional interactions between areas. MATERIALS AND METHODS Subjects: Eight right-handed healthy volunteers, aged years (five male), took part in the current experiment. Each subject gave informed consent before the study, which had local ethical committee approval c Lippincott Williams & Wilkins Vol 15 No 4 22 March

2 J. M. KILNER,Y. PAULIGNAN AND D. BOUSSAOUD Behavioural task: Subjects sat on a comfortable chair placed in front of a table in a dimly illuminated room. Placed on the table were a switch and a cylinder (diameter 30 mm, height 50 mm). The switch and the cylindrical object were orientated at the beginning of each recording session such that the subject could press the switch with their right thumb and then make a small movement, from left to right, with the same hand to grasp the object without lifting their right elbow from the table top. The distance between the switch and the object was B15 cm. Light emitting diodes (LED) that cued the movements were positioned behind the object such that the subject could see any colour change without moving their eyes (Fig. 1b). Subjects performed two task conditions: one (real) in which they were instructed to grasp the object between index finger and thumb and subsequently turn it clockwise through B901; and a second in which they were instructed to imagine making the same movement. In order to reduce eye movement artefacts subjects were instructed to fixate on the object throughout each trial. In both conditions subjects initiated every trial by closing the switch with their thumb, such that the thumb and index finger were in contact, as in a pinch. At a random interval of s later the LED directly behind the object was illuminated red and remained red for a further random interval of s before it changed colour to green (Fig. 1a). Subjects were instructed to either grasp the object, in the real condition, or imagine grasping the object, in the imagined task, as soon as the LED had changed colour from red to green. In the imagined condition subjects were instructed to release the switch when they had finished the imagined movement. This terminated the trial. In the real condition each trial was (a) Attention Preparation Movement Green ATT 400 ms (b) (c) LED Sensorimotor Partial Occipital Fig.1. The behavioural task and EEGrecordings. (a) Schematic of the tasks performed.time increases from left to right. Data was analysed during three windows of the task each 512 ms long.these were the ATT, and periods and these windows are shown with respect to the times of the colour changes. Based on previous studies three groups of electrodes overlying the sensorimotor, parietal and occipital cortex were chosen for further analysis. These are shown in (c)bygreycircles.(b) Schematic of the task setup. Subjects made movements from left to right with the right hand, releasing a switch, to grasp a small cylinder. 638 Vol 15 No 4 22 March 2004

3 REALVS IMAGINED VISUOMOTORTASKS terminated 5 s after the LED turned green. In order to minimise the possibility of scalp muscle artefacts in the real condition, subjects were instructed to make a slow movement to the object without lifting their elbow from the table. Prior to recording, subjects performed practice trials to ensure that they understood the task and that they could make comfortable movements to the object. Subjects performed two blocks of each task condition with B35 trials in each. For each subject at least one block of the real condition was performed prior to the first block of the imagined condition to facilitate imagining the movement. Recordings: EEG was recorded with a 65 channels Geodesic Sensor Net through AC-coupled high input impedance amplifiers (200 MO, Net Amps, Electrical Geodesics Inc., Eugene, OR, USA). Data were recorded at a sampling rate of 500 Hz, band-passed filtered at Hz, using recording site Cz from the international system as reference. The impedance of each electrode was kept o 50 ko throughout the experiment (as in [8]). Any electrodes with an impedance 4 50 ko were rejected from further analysis. Vertical and horizontal eye movements, electro-oculograms (EOGs), were recorded using electrodes integral to the geodesics recording system. Event markers were recorded at the time the switch was pressed, the times the red light and green light were illuminated, and the time when the switch was released. Analysis: Off-line the data for each trial were divided into three task periods, each aligned to one of three task markers such that there was a window of 1 s before and 2 s after each event. The EEG recordings for each period were then converted by subtraction to average reference [8,9]. Trials contaminated by eye blinks or eye movements were rejected from further analysis. In addition, all trials in which artefacts were found on one or more electrodes were rejected. In total 21% of trials were rejected, resulting in an average of 40.6 trials per task condition. Spectral analyses: Power spectra were calculated for each electrode for three task periods, each 512 ms long as shown in Fig. 1a,b: these were a background attention period (ATT) ms before the red light; a preparation period () ms before the green light; and a movement period () ms after the green light. The power spectra were calculated using a 256 point FFT with a raised Hamming window and were linearly detrended before calculation. Based on the results of previous studies [6,7] three regions of interest (ROI) were defined, these are shown in grey circles in Fig. 1c: electrodes overlying the left sensorimotor electrodes, contralateral to the moving hand (5,9,17,22,18), electrodes overlying the left parietal cortex (28,29,32,3334) and electrodes overlying the occipital cortices (36,37, 39,40,44). Coherence spectra were calculated between each of the electrodes overlying the left sensorimotor cortex and all other electrodes for the three periods of the task for both the real and imagined conditions. Based on previous studies [6,7], data were subsequently analysed in two frequency bands: alpha ( Hz; corresponding to bins 5 7 of the power spectrum) and beta ( Hz; corresponding to bins of the power spectrum). The power at each electrode was converted into task-related power (TRPow) modulations as follows: Pfend n¼fstart PowTask n PowATT n PowATT n TRPow f ¼ 100 fend fstart where f is the frequency range defined by the bins between fstart and fend; PowTask is the power in either the or task periods and PowATT is the power in the ATT background period. A similar conversion was performed for the coherence values to produce a task-related coherence (TRCoh) measure. Prior to statistical analysis the values of TRPow and TRCoh were averaged across all electrodes in each ROI to produce a single value per task period per task condition per subject. Statistical analysis: All statistical analyses were performed using repeated measure ANOVAs. For both the TRPow and TRCoh modulations four sets of ANOVAs were performed. Initially, a four-factor repeated measure ANOVA was performed with the factors being; ROI (motor/parietal/ occipital), task period (/), task condition (imagined/real) and frequency (alpha/beta). This ANOVA was further interrogated to test for interregional modulations. All ANOVAs were corrected for non-sphericity where appropriate using a Greenhouse Gaiser correction. RESULTS Modulations in TRPow: Figure 2 shows the modulation in TRPow across task period, task condition, frequency and ROI. The four-way repeated measure ANOVA showed significant main effects of ROI (F(1.28,8.97) ¼ , p ¼ 0.008) and task period (F(1.00,7.00) ¼ , p ¼ 0.010) and a significant interaction between these two factors (F(1.70,11.90) ¼ 4.723, p ¼ 0.035). These results reflect the clear attenuation of TRPow in both the alpha and beta frequency bands during the compared to the conditions that were modulated across the ROI. Furthermore this pattern of attenuation was observed during both the real and imagined tasks. At the electrodes of interest overlying the left sensorimotor cortex, a three-way repeated measure ANOVA (with the factors being task period, task condition and frequency) only revealed a significant main effect of task period (F(1.00,7.00) ¼ , p ¼ 0.010). At the electrodes of interest overlying the left parietal cortex a similar three-way repeated measure ANOVA revealed significant main effects of task period (F(1.00,7.00) ¼ , p ¼ 0.011) and frequency (F(1.00,7.00) ¼ , p ¼ 0.013) and a significant interaction between frequency and task condition (F(1.00,7.00) ¼ 7.074, p ¼ 0.032). At the electrodes of interest overlying the occipital cortices the same threeway repeated measure ANOVA revealed only significant main effects of task period (F(1.00,7.00) ¼ 6.116, p ¼ 0.043). Modulations in TRCoh: Figure 3 shows graphically the modulation in TRCoh across task period, task condition, frequency and ROI. Whereas the modulations in TRPow were mostly attenuations of oscillatory activity the modulations in TRCoh were both attenuated and augmented. The four-way repeated measure ANOVA showed a significant main effect of task period (F(1.00,7.00) ¼ , p ¼0.009) and a significant interaction between the task period, task condition and frequency (F(1.00,7.00) ¼ , p¼0.013). At Vol15 No4 22March

4 J. M. KILNER,Y. PAULIGNAN AND D. BOUSSAOUD Sensorimotor Parietal Occipital 20 REAL IMAG REAL IMAG REAL IMAG REAL IMAG REAL IMAG REAL IMAG 10 % TRPow BETA Fig. 2. Task-dependent modulations in TRPow for the alpha (black bars) and beta (white bars) ranges respectively. TRPow were meaned across electrodes of interest for motor electrodes, parietal electrodes and occipital electrodes and then across subjects. Error bars are s.e.m. calculated across subjects. ALPHA the electrodes of interest overlying the left sensorimotor cortex a three-way repeated measure ANOVA (with the factors being task period, task condition and frequency) revealed no significant main effects but did reveal a significant interaction between task condition and task period (F(1.00,7.00) ¼ , p ¼ 0.006). At the electrodes of interest overlying the left parietal cortex a similar three-way repeated measure ANOVA revealed a significant main effect of task period (F(1.00,7.00) ¼ , p ¼0.014) and a trend to significance for the interaction between frequency and task condition (F(1.00,7.00) ¼ 5.209, p ¼0.056). At the electrodes of interest overlying the occipital cortices the same threeway repeated measure ANOVA revealed only a significant interaction between the three factors (F(1.00,7.00) ¼ 6.267, p ¼ 0.041). DISCUSSION The current study was designed to investigate modulations in cortical oscillatory activity during natural prehension movements and to test the hypothesis that differences between real and imagined movements may lie at the level of neuronal connectivity, as measured by interelectrode coherence, and not at the level of regional neuronal activity, as measured by the amplitude of oscillatory activity. The results of the study support this hypothesis. In the analyses performed on both the TRPow and TRCoh measures the critical factor for assessing the effect of executed vs imagined movements was task condition. In all the analyses performed on TRPow only one interaction that was dependent upon the task condition factor was significant, the interaction between frequency and task condition at parietal electrodes. As can clearly be observed in Fig. 2, this lack of statistical dependency on the task condition factor reflects the fact that during the real and imagined conditions both the alpha and beta band oscillatory activities showed the same pattern of taskdependent modulations. In both conditions, there were no systematic modulations during motor preparation compared to background, however both alpha and beta band oscillatory activities were significantly attenuated during the periods of motor execution/imagination. This is reflected in the fact that the factor task period was significant in all anovas performed on the TRPow data sets. 640 Vol15 No4 22March2004

5 REALVS IMAGINED VISUOMOTORTASKS Sensorimotor Parietal Occipital 15 REAL IMAG REAL IMAG REAL IMAG REAL IMAG REAL IMAG REAL IMAG 10 5 % TRCoh ALPHA BETA 20 Fig. 3. Task-dependent modulations intrcoh for the alpha (black bars) and beta (white bars) ranges respectively.trcoh were meaned across electrodes of interest for motor electrodes, parietal electrodes and occipital electrodes and then across subjects. Error bars are s.e.m. calculated across subjects. The pattern of the attenuation observed here is consistent with modulations in oscillatory activity that have previously been reported for real movements [10 13] and imagined movements [11,14,15]. In contrast, the TRCoh measure was statistically dependent upon the task condition factor with all the analyses revealing a significant interaction, or a trend towards significance, involving the task condition factor. However, in comparison to the TRPow results those of the TRCoh show complex patterns of modulations with only coherence between motor and parietal areas showing a significant main effect, that of task period. Furthermore, this highlights a clear dissociation between the modulations in TRPow and TRCoh, i.e. the TRCoh does not simply reflect the modulations in TRPow. Previous studies investigating visuomotor integration using interelectode coherence as a measure reported an increase in beta coherence between occipital and motor electrodes during periods requiring coordinated visual and motor processing [7]. In the current study such an increase in beta occipital-motor coherence was observed during the period of the real task (Fig. 3), however during the period of the real task there was a decrease in coherence compared to background. Technical considerations: In the current study, interelectrode coupling was estimated using coherence analysis. Although this method is commonly used for assessing coupling between EEG electrodes [7,16 18] the measure can be modulated by factors other than modulations of cortical functional connectivity. In particular, the effects of a common reference and volume conduction can give rise to a false picture of cortical coupling [9]. To minimise the effect of the common reference the EEG was referenced to the average of all electrodes [8,9]. In all studies assessing interelectrode coupling there is a possibility that the results merely reflect artefactual coupling due to volume conduction. In the current study this would seem unlikely for two reasons. Firstly, there is dissociation between modulations in TRPow and TRCoh and secondly, there is a clear difference in the pattern of coherence between motor and parietal areas compared to the coherence between motor and occipital areas (Fig. 3). As the parietal electrodes lie midway between the occipital and sensorimotor electrodes it is therefore unlikely that the effects seen are simply due to volume conduction. CONCLUSION There is compelling evidence in the literature that many of the brain areas active when making a movement are similarly active when imagining the same movement. In Vol15 No4 22March

6 J. M. KILNER,Y. PAULIGNAN AND D. BOUSSAOUD line with this view, in the current study we observed that there were no significant differences in oscillatory power in either the alpha or the beta range between imagined and executed movements. In contrast, there were clear statistical dependences upon the nature of the task, whether it was imagined or executed, on the modulation of interelectrode coherence. As the measure of interelectrode coherence has been postulated to reflect the underlying cortical functional connectivity [7,16 18] the observed difference at the level of the second-order spectral analysis suggests that either different networks are activated during imagined and real movements or that the same network is differentially activated under the two task conditions. REFERENCES 1. Roland PE, Larsen B, Lassen NA and Skinhoj E. Supplementary motor area and other cortical areas in organization of voluntary movements in man. J Neurophysiol 43, (1980). 2. Decety J, Perani D, Jeannerod M, Bettinardi V, Tadary B, Woods R et al. Mapping motor representations with positron emission tomography. Nature 371, (1994). 3. Stephan KM, Fink GR, Passingham RE, Silbersweig D, Ceballos-Baumann AO, Frith CD and Frackowiak RS. Functional anatomy of the mental representation of upper extremity movements in healthy subjects. J Neurophysiol 73, (1995). 4. Grafton ST, Arbib MA, Fadiga L and Rizzolatti G. Localization of grasp representations in humans by positron emission tomography. 2. Observation compared with imagination. Exp Brain Res 112, (1996). 5. Gerardin E, Sirigu A, Lehericy S, Poline JB, Gaymard B, Marsault C et al. Partially overlapping neural networks for real and imagined hand movements. Cerebr Cortex 10, (2000). 6. Roelfsema PR, Engel AK, Konig P and Singer W. Visuomotor integration is associated with zero time-lag synchronization among cortical areas. Nature 385, (1997). 7. Classen J, Gerloff C, Honda M and Hallett M. Integrative visuomotor behavior is associated with interregionally coherent oscillations in the human brain. J Neurophysiol 79, (1998). 8. Keil A, Muller MM, Ray WJ, Gruber T and Elbert T. Human gamma band activity and perception of a gestalt. J Neurosci 19, (1999). 9. Nunez PL, Wingeier BM and Silberstein RB. Spatial-temporal structures of human alpha rhythms: theory, microcurrent sources, multiscale measurements, and global binding of local networks. Hum Brain Mapp 13, (2001). 10. Gastaut H. Etude electrocorticographique de la reactivite des rythmes rolandiques. Rev Neurol (Paris) 87, (1952). 11. Pfurtscheller G and Neuper C. Motor imagery activates primary sensorimotor area in humans. Neurosci Lett 239, (1997). 12. Feige B, Aertsen A and Kristeva-Feige R. Dynamic synchronisation between multiple cortical motor areas and muscle activity in phasic movements. J Neurophysiol 84, (2000). 13. Kilner JM, Baker SN, Salenius S, Hari R and Lemon RN. Human cortical muscle coherence is directly related to specific motor parameters. J Neurosci 20, (2000). 14. Neuper C and Pfurtscheller G. Event-related dynamics of cortical rhythms: frequency-specific features and functional correlates. Int J Psychophysiol 43, (2001). 15. Schnitzler A, Salenius S, Salmelin R, Jousmaki V and Hari R. Involvement of primary motor cortex in motor imagery: a neuromagnetic study. Neuroimage 6, (1997). 16. Andrew C and Pfurtscheller G. Lack of bilateral coherence of postmovement central beta oscillations in the human electroencephalogram. Neurosci Lett 273, (1999). 17. Andres FG, Mima T, Schulman AE, Dichgans J, Hallett M and Gerloff C. Functional coupling of human cortical sensorimotor areas during bimanual skill acquisition. Brain 122, (1999). 18. Mima T, Matsuoka T and Hallett M. Functional coupling of human right and left cortical motor areas demonstrated with partial coherence analysis. Neurosci Lett 287, (2000). Acknowledgements: J.M.K. was supported by the WellcomeTrust, UK and Y.P. and D.B. were supported by the CNRS, France. 642 Vol15 No4 22March2004

Coupling of Oscillatory Activity Between Muscles Is Strikingly Reduced in a Deafferented Subject Compared With Normal Controls

J Neurophysiol 92: 790 796, 2004. First published April 7, 2004; 10.1152/jn.01247.2003. Coupling of Oscillatory Activity Between Muscles Is Strikingly Reduced in a Deafferented Subject Compared With Normal