Abstract
Physiological and neuroimaging studies suggest that human actions are characterized by time-varying engagement of functional distributed networks within the brain. In this study, we investigated whether specific prestimulus interhemispheric connectivity, as a measure of synchronized network between the two hemispheres, could lead to a better performance (as revealed by RT) in a simple visuomotor task. Eighteen healthy adults underwent EEG recording during a visual go/no-go task. In the go/no-go task, a central fixation stimulus was followed by a green (50% of probability) or red visual stimulus. Participants had to press the mouse button after the green stimuli (go trials). Interhemispheric coupling was evaluated by the spectral coherence among all the electrodes covering one hemisphere and matched with those on the other. The frequency bands of interest were delta (2–4 Hz), theta (4–8 Hz), alpha 1 (8–10.5 Hz), alpha 2 (10.5–13 Hz), beta 1 (13–20 Hz), beta 2 (20–30 Hz), and gamma (30–40 Hz). The task-related results showed that interhemispheric connectivity decreased in delta and increased in alpha band. Furthermore, we observed positive delta and negative alpha correlations with the RT; namely, the faster the RT, the lower delta and the higher alpha connection between the two hemispheres. These results suggested that the best performance is anticipated by the better functional coupling of cortical circuits involved during the processing of the sensorimotor information, occurring between the two hemispheres pending cognitive go/no-go task.
INTRODUCTION
Physiological and neuroimaging studies provide evidence that human actions are characterized by engagement of functional distributed networks within the brain. One of the unanswered questions is how the brain integrates processes that are lateralized in opposite hemispheres (Johansson et al., 2006). Studies on patients with callosal lesions (Gazzaniga, 2000) and on healthy volunteers (Hellige, 1990) suggested that closely related tasks can also rely on different mechanisms of interhemispheric integrating organization. Such network organization is particularly significant for the higher functions—including abstract reasoning, memory, and action planning—and requires a high degree of integration of information flow occurring from different and often remote brain sources (Uhlhaas & Singer, 2006). The whole process is orchestrated by attention, which combines and organizes also the individual afferent sensorial elements (visual, acoustic, olfactory and tactile) in a context-dependent hierarchical order (Uhlhaas & Singer, 2006; Fries, Reynolds, Rorie, & Desimone, 2001; Engel, König, Kreiter, & Singer, 1991). These networks dynamically connect adjacent and/or remote cortical neuronal assemblies via cortico-cortical “fragile and time-varying” connections (Ferreri, Vecchio, Ponzo, Pasqualetti, & Rossini, 2014; Siegel, Donner, & Engel, 2012; Kandel, 2008; Varela, Lachaux, Rodriguez, & Martinerie, 2001; Shepherd & Erulkar, 1997; for a review, see D'Amelio & Rossini, 2012). Within this framework, transient synchronization of neuronal firing has been proposed as one of the most efficacious mechanism for determining rapidly time-varying binding/unbinding phenomena, which can orchestrate the dynamic linkage of separate and widely distributed neuronal assemblies within a unique and functionally coherent frame as happens, for example, in sensorimotor areas coordination and integration (Ferreri et al., 2014; Roelfsema, Engel, König, & Singer, 1997; Singer et al., 1997; Engel et al., 1991; Mioche & Singer, 1989). Network excitability degree can be used as connectivity marker. In fact, short-term changes in excitability are largely indicative of changes in synaptic efficacy because they effectively modified network connectivity. Specific methods have been developed to identify transient modifications in functional coupling and/or excitability of distributed neuronal assemblies via EEG and magnetoencephalography recordings (Vecchio et al., 2010; Zeitler, Fries, & Gielen, 2006; Lachaux, Rodriguez, Martinerie, & Varela, 1999; Rodriguez et al., 1999; Andrew & Pfurtscheller, 1996; Bressler, Coppola, & Nakamura, 1993).
Thus far, not much attention has been dedicated to the eventual relationship of the ongoing EEG properties immediately preceding an imperative stimulus delivered to the participants and the cognitive performance of the participant in response to the triggering stimulus. For example, RT is an important behavioral measure of the overall efficacy of sensorimotor processing and is acknowledged that it varies significantly from trial to trial. Previous works on how stimulus-evoked cortical responses contribute to RT variability have helped to delineate the stages of neuronal information processing. Much less is known about how brain organization immediately preceding the stimulus onset (prestimulus) could affect RT.
A recent animal study (Zhang, Wang, Bressler, Chen, & Ding, 2008) showed the degree of linear correlation between RT and prestimulus EEG spectral power computed over a wide range of frequencies. In the pFC, prestimulus power in the beta band (14–30 Hz) was negatively correlated with RT. It has suggested a possible activity in the mediation of top–down control of visuomotor processing in the beta frequency range. In the sensorimotor cortex, prestimulus power in the beta range was positively correlated with RT. It is consistent with the hypothesis that oscillations in this range support the maintenance of steady-state motor output. In visual occipital and temporal lobe areas, prestimulus power in the alpha/low beta range (8–20 Hz) showed positive correlations with RT; this result possibly reflects a spatially specific disengagement of visual anticipatory attention (Zhang et al., 2008). The same effect was also observed for the frontoparietal gamma band (Gonzalez Andino, Michel, Thut, Landis, & Grave de Peralta, 2005). In this study, the first working hypothesis was to evaluate whether a specific prestimulus EEG power distribution and the topographic synchronization could anticipate a better performance (as revealed by RT) in a visuomotor task.
Some microelectrodes studies on animals also provided evidence that not only changes in power but also changes in neural synchrony can be predictive for behavioral performance (for a review, see Engel, Fries, & Singer, 2001). For example, Riehle and colleagues (Riehle, Grammont, Diesmann, & Grün, 2000; Riehle, Gruen, Diesmann, & Aertsen, 1997) showed that, in a delayed reaching task, synchrony of neural firing occurred particularly at those times when a monkey was expecting a GO signal to appear on the screen. When the GO signal appeared after prolonged periods of expectation, the number of significantly synchronized events increased over the delay period, indicating a relationship between the growing stimulus expectancy and the synchronization of network activity. Other animal studies (König, Engel, & Singer, 1995; Engel et al., 1991) on visual cortex showed that, when synchronization of neuronal activity occurred between the two hemispheres, it is almost at all times associated with oscillatory firing patterns. All these data supported the idea that oscillatory activity could contribute to the establishment of long-range synchrony (as between the two hemispheres) in a network of reciprocally coupled neurons too.
The second working hypothesis of this work was to test whether a specific prestimulus interhemispheric connectivity—as a measure of synchronized network between the two hemispheres—can anticipate a better performance (as revealed by RT) in a visuomotor task. The present experimental design was a simple go/no-go task. Despite its simplicity, the processing of the go/no-go task implies the activation of several circuits of the cerebral cortex (Vecchio et al., 2012). Prefrontal–parietal areas may be involved in the regulation of visual attention and executive functions (i.e., inhibition of impulsive behavior). Occipital areas may be engaged in the processing of colors and shape of the visual stimuli. Frontal premotor areas may subserve visuomotor transformations, whereas frontal primary and secondary motor areas may subserve the preparation, execution, and control of motor response. Because of this brain widespread activation, we evaluated an interhemispheric integration as a rough index of brain connectivity.
Summarizing, two working hypothesis were carried out. The first was that the behavioral performance, as revealed by RT, is due to changes in the spectral power of the two hemispheres. The second one was that this behavioral performance is more correlated with changes in functional brain networks.
METHODS
Participants
Eighteen healthy volunteers (11 men and 7 women, mean ± SEM of age being 29.63 ± 2.01 years, range: 24–36 years) were enrolled in this study. All participants were right-handed at Handedness Questionnaire (Salmaso & Longoni, 1985). None of them had ever suffered from neurological or psychiatric disorders. They were instructed to refrain from caffeine or alcohol and to maintain their regular cycle of sleeping–waking in the days before the experiment.
Experimental Procedures
The experimental design included the EEG montage and the recording of a go/no-go visuomotor task. During the go/no-go task, the participants were seated in a comfortable chair, placed in a dimly lit and sound-damped room. They kept their forearms resting with the right index finger between two buttons of a computer-connected mouse. The monitor of the computer was placed in front of them at a distance of about 90 cm. A trial of the go/no-go task included the following events: (i) central visual fixation target lasting 2 sec, (ii) a green or red visual stimulus lasting 0.5 sec, and (iii) motor reaction to the green (go) stimuli pressing the left mouse button with right (dominant) hand. In total, the participants were presented with 400 stimuli (green and red visual stimuli were pseudorandomized with 50% of probability) lasting about 25 min. Experimenter's instructions emphasized the request of an optimal behavioral performance.
EEG Recordings
EEG data were continuously recorded (bandpass = 0.1–100 Hz, sampling rate = 256 Hz; MICROMED) from 19 scalp electrodes (cap) equispatially positioned over the whole scalp according to the 10–20 system. The electrical reference was located between the AFz and Fz electrodes, and the ground electrode was located between the Pz and Oz electrodes. The electrode impedance was kept below 5 kΩ. Simultaneously, the recording of bipolar electrooculographic data (EOG; bandpass = 0.1–100 Hz, sampling rate = 256 Hz) monitored blinking and eye movements.
Preliminary Data Analysis
The EEG data were segmented into single trials of 6-sec duration, each trial lasting from −2 sec to +4 sec after the onset of the visual stimulus. Data epochs showing instrumental, ocular, and muscular artifacts were offline identified and automatically eliminated by a computerized procedure. The EEG data affected by ocular artifacts were corrected with an autoregressive method (Moretti et al., 2003). Finally, two expert electroencephalographers visually double-checked and confirmed the automatic selection and correction of the EEG single trials. A special attention was reserved to the selection of the artifact-free EEG epochs. Therefore, only the EEG single trials totally free from artifacts were considered for the subsequent analyses. It is worth noting that the visual-evoked potentials were not subtracted by the artifact-free EEG data because we are interested in the prestimulus periods.
The artifact-free EEG data were re-referenced to common average and classified in go and no-go trials. For the aim of this study, we focused on the go trials having EEG and behavioral correlates. On average across participants, the artifact-free EEG data were formed by 185 single trials (±0.6 SEM) for the go condition. The trials of the go condition were separated by the mean values in two groups: Fast RT and Slow RT accordingly to the RT of the behavior performance.
Spectral Analysis of the EEG Data
The digital FFT-based analysis (Welch technique, Hanning windowing function, no phase shift) was evaluated to calculate the frequency power spectrum. The time window of interest was the second just before the onset of each stimulus. The frequency bands of interest were delta (2–4 Hz), theta (4–8 Hz), alpha 1 (8–10 Hz), alpha 2 (10–13 Hz), beta 1 (13–20 Hz), beta 2 (20–30 Hz), and gamma (30–40 Hz).
Estimation of the Functional Connectivity: Between-electrode Coherence Analysis
This equation is the extension of the Pearson's correlation coefficient to complex number pairs. In this equation, f denotes the spectral estimate of two EEG signals x and y for a given frequency bin (λ). The numerator contains the cross-spectrum for x and y (fxy), and the denominator contains the respective autospectra for x (fxx) and y (fyy). For each frequency bin (λ), the coherence value (Cohxy) is obtained by squaring the magnitude of the complex correlation coefficient R. This procedure returns a real number between 0 (no coherence) and 1 (max coherence).
In line with previous works (Vecchio et al., 2007, 2010; Babiloni, Brancucci, et al., 2006; Babiloni, Vecchio, et al., 2006; Rossini et al., 2006), the spectral coherence between couples of electrodes was calculated by a home-made software developed under Matlab 6.5 (Mathworks, Inc., Natrick, MA; 1-Hz frequency resolution).
The time window of interest was 1 sec just before the onset of each stimulus. The frequency bands of interest were delta (2–4 Hz), theta (4–8 Hz), alpha 1 (8–10 Hz), alpha 2 (10–13 Hz), beta 1 (13–20 Hz), beta 2 (20–30 Hz), and gamma (30–40 Hz). For the evaluation of the hemispheric synchronization of the EEG frequency bands of interest, the spectral coherence was evaluated between each of the electrodes in the left hemisphere and each of the electrodes in the right hemisphere. Namely, given the 19 electrodes, the spectral coherence was evaluated between eight electrodes (excluding Fz, Cz, Pz) for each hemisphere. Sixty-four combinations were evaluated and averaged to obtain the interhemispheric coupling.
Statistical Analysis
To test the first working hypothesis to attest that the behavioral performance, as revealed by RT, could be because of changes in spectral power of the two hemispheres, a statistical ANOVA design was used to compare the power spectrum (dependent variable) in the two RT conditions. The ANOVA design included the factors Hemisphere (left and right), Condition (Fast RT, Slow RT), and Band (delta, theta, alpha 1, alpha 2, beta 1, beta 2, gamma). Mauchley's test evaluated the sphericity assumption. Correction of the degrees of freedom was made by the Greenhouse–Geisser procedure. Duncan test was used for post hoc comparisons (p < .05).
Furthermore, to test the second working hypothesis that the behavioral performance, as revealed by RT, could be due to changes in functional brain networks, a statistical ANOVA design was used to compare the coherence values (dependent variable) in the two RT conditions. The ANOVA design included the factors Condition (Fast RT, Slow RT) and Band (delta, theta, alpha 1, alpha 2, beta 1, beta 2, gamma). Mauchley's test evaluated the sphericity assumption. Correction of the degrees of freedom was made by the Greenhouse–Geisser procedure. Duncan test was used for post hoc comparisons (p < .05).
Finally, a linear correlation analysis was evaluated between the RT and the spectral power, and the coherence values in those frequency bands significantly resulted in the previous analyses, considering each participant's condition as a different case. Specifically, the linear correlation was computed with Pearson test (Bonferroni corrected, p < .05).
RESULTS
Behavioral Results
The means of the RT to the go stimuli were 345 msec (±17 msec) for the Fast RT condition and 465 msec (±29 msec) for the Slow RT condition. To compare these behavioral results, we performed a Student's t test for the RT of the two RT conditions. As expected, there was significant p value (p < .00001) pointing to faster RT to the go stimuli (i.e., better behavioral performance) in the Fast than Slow condition.
Power Spectrum Analysis
For illustrative purpose, Figure 1 shows the power spectrum analysis on the two hemispheres in the bands of interest for the Fast and Slow RT conditions. The results showed that no statistical differences were found (p > .5) between Fast and Slow conditions, neither using the average between the two hemispheres. Remarkably, the only significant result was observed in the main effect Band, F(6, 102) = 12.78, p < .00001, showing highest values of power spectrum in delta and alpha bands independently by Condition and Hemisphere.
Power spectrum analysis on the two hemispheres and the bands of interest for the Fast and Slow RT conditions. The results showed that no statistical differences were found (p > .5) between Fast and Slow RT conditions, even after using the average between the two hemispheres.
Functional Connectivity as Revealed by Spectral Coherence between Hemispheres
Figure 2 shows the results relative to a statistically significant ANOVA interaction, F(6, 102) = 6.21, p < .0001, between the factors Condition (Fast RT, Slow RT) and Band (delta, theta, alpha 1, alpha 2, beta 1, beta 2, gamma). The post hoc testing revealed that the interhemispheric coupling was lower in delta (p < .001) and higher in both alpha 1 (p < .001) and alpha 2 (p < .0005) in Fast in respect to Slow RT condition. To be remarked that the main effect Condition does not result significant, whereas the opposite result was true for Band, F(6, 102) = 21.63, p < .00001, showing highest values of functional coupling in alpha bands independently by Fast RT or Slow RT.
Interhemispheric coupling (mean across participants) relative to a statistical ANOVA interaction between the factors Condition (Fast RT, Slow RT) and Band (delta, theta, alpha 1, alpha 2, beta 1, beta 2, gamma). Fast condition values (in blue rectangles) higher (p < .05) than Slow one (red rectangle at the opposite).
Interhemispheric coupling (mean across participants) relative to a statistical ANOVA interaction between the factors Condition (Fast RT, Slow RT) and Band (delta, theta, alpha 1, alpha 2, beta 1, beta 2, gamma). Fast condition values (in blue rectangles) higher (p < .05) than Slow one (red rectangle at the opposite).
Linear Correlation Analysis
Concerning the linear correlation analysis between the RT and the functional coupling between the two hemispheres, a positive correlation was observed in the delta band (r = 0.38, p < .03). Furthermore, there was a negative correlation in alpha 1 (r = −0.41, p < .02) and alpha 2 (r = −0.45, p < .01) bands. Namely, the faster the RT, the lower delta and the higher alpha connectivity between the two hemispheres, as reported in Figure 3.
Scatterplots showing the correlation between interhemispheric coupling and RT in the Slow and Fast conditions as a whole group. The r and p values relative to the Pearson correlation are reported within the diagram.
Control Analysis
To better explain the contribution of each macro cortical region, a statistical control analysis including only the frontal (F3-F4), central (C3-C4), and parietal (P3-P4) coupling was evaluated. The results of this analysis showed a statistical interaction, F(12, 204) = 1.87, p < .0402, among the factors Condition (Fast, Slow), Pair (frontal, central, parietal), and Band (delta, theta, alpha 1, alpha 2, beta 1, beta 2, gamma). The post hoc testing showed that, whereas the delta effect (higher values in the Slow RT condition) was widespread in the three considered pairs, the alpha effect (higher values in the Fast RT condition) was more evident in frontal and central rather than in parietal area.
DISCUSSION
Every behavioral, cognitive, or motor act requires a finely tuned balance between triggering and blocking processes to provide appropriate preparation, initiation, online control, and timely inhibition of this act. Inhibitory control is therefore an essential regulatory function. The go/no-go task demands high-level cognitive functions such as decision-making, response selection, and response inhibition. So, despite its simplicity, go/no-go task performance implies the activation of several brain networks. Because of the widespread brain activation, we evaluated an interhemispheric integration as a rough index of brain connectivity. Furthermore, the results of the present work demonstrated that the behavioral performance, as revealed by RT, is because of changes in functional brain networks rather than the changes in spectral power of the two hemispheres. Notably, a recent study (Zhang & Ding, 2010) evaluated the prestimulus Mu rhythm. They related cerebral rhythms before the stimuli with evoked potentials and found a nonlinear correlation with the two variables. This point could be taken into account in further studies including higher number of trials.
The patterns of coherence in ongoing activity can span several millimeters of cortical surface, indicating that specific long-range interactions are possible even in the absence of visual input. Moreover, these patterns are highly variable, and their fluctuations can largely account for the variability of the responses evoked by a subsequent visual stimulus (Arieli, Sterkin, Grinvald, & Aertsen, 1996). These data make it likely that ongoing activity contains structured information and therefore has an important role in cortical function (Tsodyks, Kenet, Grinvald, & Arieli, 1999). In the present work, we tested the hypothesis that a specific prestimulus interhemispheric connectivity in delta and alpha bands, as a measure of synchronized network among specialized networks in the two hemispheres, lead to a better performance (as revealed by RT) in a simple visuomotor task. The results showed a decreased delta and increased alpha band interhemispheric connection. Furthermore, positive delta and negative alpha correlations were observed with the RT, namely the faster the RT, the lower the delta and the higher the alpha connection among the two hemispheres before the stimulus.
The present experiment demonstrated that a reduction of delta and an increase of alpha bands' interhemispheric coupling before a motor–cognitive task could predict the task performance. The present results extend those of a previous magnetoencephalographic study (Yamagishi, Callan, Anderson, & Kawato, 2008), in which changes in neural intertrial coherence (2–14 Hz) before a target but after a cue stimulus were found predictive of behavioral performance.
At least two main issues require a careful discussion. The first issue regards the delta EEG band. It is well known that in the awake brain, alpha rhythms dominate in the posterior areas and delta rhythms are low in amplitude, thus reflecting a condition of likely alpha-delta “reciprocal inhibition” (Rossini et al., 2006); on the other hand it is well known that anatomical or functional disconnection from related cortical areas generates spontaneous slow oscillations in the delta range in virtually all recorded neurons (Gloor, Ball, & Schaul, 1977). On the basis of this theoretical framework, it can be speculated that the widely decreased coherence in delta band, immediately preceding a task performance, could produce a better interhemispheric integration, leading to better motor–cognitive performances. Increase of coherence in the delta range might therefore represent—at least in part—a functional unbinding when a highly concentration or active condition is required (see also Ferreri et al., 2014; Mountcastle, Lynch, Georgopoulos, Sakata, & Acuna, 1975).
The second issue is why connectivity results were observed at alpha rhythms (about 8–12 Hz). Low-frequency alpha rhythms (about 8–10 Hz) are supposed to reflect the regulation of global cortical arousal (Klimesch, 1999; Pfurtscheller & Lopes da Silva, 1999), whereas there is consensus that the high-frequency alpha rhythms reflect the functional modes of thalamo-cortical and cortico-cortical loops that facilitate/inhibit the transmission and retrieval of sensorimotor information into the brain during the performance of cognitive–motor tasks (Brunia, 1999; Klimesch, 1999; Pfurtscheller & Lopes da Silva, 1999; Steriade & Llinás, 1988). It can therefore hypothesize that the best performance could be obtained when there is the best interaction with the oscillatory activity of cortical circuits involved in the processing and in the transformation of sensorimotor information during the just following cognitive go/no-go task among the two hemispheres.
As a methodological remark, common reference signal and volume conduction potential might contaminate the data. Bipolar derivation and current source density (CSD) are ways to deal with this (Bollimunta, Chen, Schroeder, & Ding, 2009). The results of several works (Bollimunta, Chen, Schroeder, & Ding, 2008; Ding, Chen, & Bressler, 2006; Kaminski, Ding, Truccolo, & Bressler, 2001) suggest also that functional coupling analysis (in particular Granger causality), especially when combined with traditional techniques like CSD analysis, can improve the ability to understand the organization of cortical networks. Here the number of electrodes used may not be sufficient for attempting bipolar or CSD-based analysis, and further works should be performed with a higher number of electrodes to address this important point to understand better the dynamical organization of synchronous oscillatory cortical networks.
Conclusions
Evidence from this study provides direct confirmation for a stochastic linking of cortical areas, as revealed by oscillatory synchronization of the two hemispheres in selected EEG rhythms, in determining behavior performance in a cognitive–motor task. New developments could help in understanding the nature of the coordinated and gated brain dynamics underlying behavior performance (Ferreri et al., 2014). However, it is possible that the mechanisms subtending response variability could help to understand how brain optimizes the motor control, possibly providing a new probe to noninvasively test for brain functions and cortico-cortical connectivity in healthy and in neurological diseases.
Acknowledgments
Dr. Francesca Miraglia participated to this study in the framework of her PhD program at the Doctoral School in Neuroscience, Department of Neuroscience, Catholic University of Rome, Italy. The article is partially funded by the Italian Ministry of Instruction, University and Research MIUR (“Functional connectivity and neuroplasticity in physiological and pathological aging” PRIN project).
Reprint requests should be sent to Dr. Fabrizio Vecchio, Brain Connectivity Laboratory, IRCCS San Raffaele Pisana, Via Val Cannuta, 247, 00166 Rome, Italy, or via e-mail: fabrizio.vecchio@uniroma1.it, fabrizio.vecchio@sanraffaele.it.