Learning Processes and Brain Connectivity in A Cognitive-Motor Task in Neurodegeneration: Evidence from EEG Network Analysis

Abstract

Electroencephalographic (EEG) rhythms are linked to any kind of learning and cognitive performance including motor tasks. The brain is a complex network consisting of spatially distributed networks dedicated to different functions including cognitive domains where dynamic interactions of several brain areas play a pivotal role. Brain connectome could be a useful approach not only to mechanisms underlying brain cognitive functions, but also to those supporting different mental states. This goal was approached via a learning task providing the possibility to predict performance and learning along physiological and pathological brain aging. Eighty-six subjects (22 healthy, 47 amnesic mild cognitive impairment, 17 Alzheimer’s disease) were recruited reflecting the whole spectrum of normal and abnormal brain connectivity scenarios. EEG recordings were performed at rest, with closed eyes, both before and after the task (Sensory Motor Learning task consisting of a visual rotation paradigm). Brain network properties were described by Small World index (SW), representing a combination of segregation and integration properties. Correlation analyses showed that alpha 2 SW in pre-task significantly predict learning (r = –0.2592, p < 0.0342): lower alpha 2 SW (higher possibility to increase during task and better the learning of this task), higher the learning as measured by the number of reached targets. These results suggest that, by means of an innovative analysis applied to a low-cost and widely available techniques (SW applied to EEG), the functional connectome approach as well as conventional biomarkers would be effective methods for monitoring learning progress during training both in normal and abnormal conditions.

Keywords

Alpha band Alzheimer’s disease EEG eLORETA functional brain connectivity graph theory learning mild cognitive impairment precision medicine

INTRODUCTION

During motor learning, specific changes in the intensity of activation in a range of brain regions are observed from pre- to post- training epochs [1 –15]; moreover, motor learning is associated with changes in cortical excitability [16, 17]. Visuomotor skill training induces changes in white matter structure and subsequent structural connectivity [18], and increased intra-regional functional connectivity in the primary motor cortex has been reported while following-up training tasks [19]. Coynel and colleagues [2] reported training-related changes in functional connectivity in a finger tapping task, while Bassett [20] examined dynamic configuration of human brain networks using functional connectivity measurements of functional magnetic resonance imaging (fMRI) brain activity during motor learning.

A recent review [21] reported that training-induced improvements in motor performance are accompanied by increases in functional connectivity in the task-related brain networks across a wide spectrum of connectivity metrics. Furthermore, graph-theoretical network analysis was able to emphasize a contribution of specific brain regions to network function.

Another important aspect with regard to the training-related changes in functional network connectivity is the influence of feedback and perception during motor learning. Indeed, motor control is guided by sensory (such as visual) feedback to optimize motor performance and to learn new motor skills where the so called perceptual learning plays an important role particularly in the early learning stages, where somatosensory aims of movement are poorly defined [11 , 22–26].

Motor learning requires the progressive adjustment of motor commands and the development of a sensory plan, a trajectory of desired sensory values that regulates the generation of movement sequence. However, in situations where sensory and motor systems are active in tandem, the source of the neural and behavioral changes with learning is uncertain, with changes being attributable to motor learning driving sensory change, to perceptual learning driving movement, or the two in combination.

A recent analysis of the literature [27] suggests that perceptual learning may produce changes to motor areas of the brain that may be functionally independent from those observed in sensory systems. Recent studies indicate an effect of perceptual training on motor learning; brief periods of reinforced perceptual training have durable effects on the rate and extent of motor learning [28, 29]; somatosensory discrimination training increases primary motor cortex excitability and improves measures of motor learning [30]; passive movement of the arm increases the extent of motor learning [31].

Thus, the investigation of the effects of cognitive training on the dynamics of brain connectivity could be of great relevance to understand what is actually accomplished by the cognitive training.

Recent developments in neuroimaging techniques and related mathematical tools have extended our understanding of neural mechanisms underlying brain functions.

The changes in brain activations after cognitive training can be captured by a variety of neuroimaging techniques [28], including electroencephalography (EEG) [29].

Finally, it is worth mentioning that some biomarkers can be used to predict individuals’ intelligence or learners’ subsequent learning performance [30]. If optimal mental states and the related brain connectivity dynamics for ongoing training can be recognized and intercepted online, the effectiveness of the training process could be improved.

The aim of the present study was to use a learning task in order to understand the possibility to predict the improvement of motor performance along physiological and pathological brain aging (such as in neurodegenerative patients). In the present paper it is proposed that the brain connectome, namely the study of brain connectivity, can be a useful approach not only to mechanisms underlying brain cognitive functions, but also to those supporting different mental states. Furthermore, we have introduced the functional connectome approach, which is mainly based on the graph theory [31]. The graph theoretical network metrics can be biomarkers of cognitive states, as shown in previous studies on cognitive workload [32, 33] or mental fatigue [34, 35]. Several studies have shown that the brain functional and structural network connectivity can be altered through cognitive interventions, and the graph theoretical network metrics have shown the reorganization of topological architecture of the brain functional network over multiple temporal scales (i.e., minutes, days, weeks). These results suggest that the functional EEG connectome approach would represent an effective method for investigating and possibly boosting learning progress of learners due to training.

In this context, the introduction of neurodegenerative subjects allowed to collect various degrees of network deficit spectrum. Alzheimer’s disease (AD) is characterized by progressive loss of memory and other cognitive functions. Mild cognitive impairment (MCI) is a clinical and neuropsychological state of the elderly brain intermediate between normal cognition and dementia, being mainly characterized by objective evidence of memory impairment at the neuropsychological examination not yet encompassing the definition of dementia [36, 37].

SUBJECTS AND METHODS

Participants

86 subjects were recruited, 22 normal elderly (Ctrl, age 68.3±1.0 standard error mean, Mini-Mental State Examination (MMSE) 28.6±1.2, education 13.7±1.0), 47 aMCI (age 72.8±0.8, MMSE 25.9±0.33, education 10.5±0.6), 17 AD (age 70.1±1.5, MMSE 20.1±0.5, education 10.4±1.1). All subjects were right-handed on the basis of the Handedness Questionnaire. Individual informed consent was obtained and the study was approved by the local Ethical Committee. Experimental procedures were conformed to the Declaration of Helsinki and national guidelines.

Inclusion and exclusion criteria

All subjects took part to a battery of neuropsychological tests assessing attention, memory, executive functions, visuo-construction abilities, and language.

Each subject was also submitted to brain MRI and SPECT, MMSE, Clinical Dementia Rating, Geriatric Depression Scale, Hachinski Ischemic Score, and Instrumental Activities of Daily Living scale to confirm diagnosis and to exclude other causes of dementia.

AD was diagnosed according to the National Institute on Aging-Alzheimer’s Association workgroups [38] and DSM IV TR criteria. Moreover, in all of them, there was a significant reduction in the hippocampal volume, increased width of the temporal horn and of the choroidal fissure ranging between 2 and 4 at the Likert scale as well as an abnormal pattern blood flow and oxygen consumption on SPECT.

Amnesic MCI was diagnosed according to guidelines and clinical standards [37 , 39–41]. The exclusion criteria for aMCI were: 1) mild AD, as diagnosed by standard protocols including National Institute on Aging-Alzheimer’s Association workgroups [38]; 2) evidence (including MRI procedures) of concomitant dementia such as frontotemporal, vascular, and reversible dementias (including pseudo-depressive dementia), marked fluctuations in cognitive performance compatible with Lewy body dementia and/or features of mixed dementias; 3) evidence of concomitant extra-pyramidal symptoms; 4) clinical and indirect evidence of depression as revealed by the Geriatric Depression Scale; scores lower than 14 (no depression); 5) other psychiatric diseases, epilepsy, drug addiction, alcohol dependence, use of neuro/psychoactive drugs including acetylcholinesterase inhibitors; and 6) current or previous uncontrolled or complicated systemic diseases (including diabetes mellitus) or traumatic brain injuries.

Task procedure

The Sensory Motor Learning (SMoL) task consisted of a visual rotation paradigm: mobilize a cursor on the screen of a PC in order to reach circles (target) arranged at 0°, 45°, 90°, 135°, 180°, 225°, 270° with respect to the central rest position [42]. The cursor was moved by a “trackball”, whose main axes were 90° rotated counterclockwise respect to the reference axes of the screen. The SMoL task requires the development of the ability in reaching targets in the space. The challenge is therefore to learn, as quickly as possible, how to correctly move the cursor to reach the different targets (see Fig. 1).

Fig. 1

SMoL presentation. Computer screen displayed between trials (left) and the presentation of a target.

The subject was required to reach and maintain for a minimum of 0.5 s the cursor over the black target, following each target as soon as it becomes black (go signal). The target remains black until the cursor moves away from it. Movements were required to be as precise and fast as possible, trying not to overstep the target. As soon as the cursor touches the target, the center becomes black and it remains black until the cursor moves away from it. The person was required to maintain the cursor in the central circle until the presentation of a new target.

Movement periods of about 30 s alternate with fix periods of about the same duration in a block design fashion. The task consisted of a learning phase of 30 s duration, in which the circles blackened alternately and so random until the target was not centered from the cursor, followed by a resting phase of about 30 s, in which targets remained white. Each session consisted of 8 learning-rest sequences of 8 min of duration, for 4 repetitions of 32 min total duration time.

The adaptation of the movement to the new sensory motor coordinates were measured evaluating, for each session, the number of reached targets. Improvement due to training and learning was evaluated with the number of reached targets for each of the first 8 repetitions. We introduced the “Percentage Improvement” as the difference of the averaged last four and first four repetitions.

Data recordings and pre-processing

The EEG recordings (BrainAmp 64MRplus system, Brain Products GmbH, Munich, Germany) were performed at rest, with closed eyes, and no-task conditions (at least 5 min), before and after the cognitive task. EEG signals were recorded from 64 electrodes positioned according to the augmented International 10–10 system. Skin/electrode impedances were lowered below 5 KΩ.

Data were analyzed in the same way of previous studies published by the same work group [43–47 , 68], but in the following, the procedure are briefly reported for the readers’ convenience. Matlab software and scripts based on the EEGLAB toolbox were used. The EEG recordings (256 Hz sampling rate) were band-pass filtered from 0.2 to 47 Hz. Ocular, muscular, cardiac, and other types of artifacts were removed.

Functional connectivity analysis

EEG functional connectivity analysis has been performed using the eLORETA (exact low resolution electromagnetic tomography) software [43 –47]. The eLORETA algorithm is a linear inverse solution for EEG signals that has no localization error to point sources under ideal (noise-free) conditions [48]. Following the result of previous evidence on young subjects, we selected the nodes of our graph in accordance with fMRI studies that find the higher activity during this task in the brain areas reported in Fig. 2. The selection of these nodes was based on the possibility to evaluate the modulation of brain networks with respect to the ideal networks.

Fig. 2

Result of previous evidence on young subjects to select the nodes of graph in accordance with fMRI studies that find the higher activity during this task in the brain areas.

To obtain a topographic view of the involved network, brain connectivity was computed with eLORETA software these 18 regions considering a sphere of 19 mm of diameter.

Of note, to check the specificity of the selected network, the same procedure was also performed on a different network. In particular, the Sensory Motor network, including BA2, 4, 6, 13, 22, 24, 37, 40, 44, and 45 of the dominant hemisphere, was used.

Regions of interest (ROIs) are needed for the estimation of the electric neuronal activity that is used to analyze brain functional connectivity. Among the eLORETA current density time series of the selected ROIs, intracortical Lagged Linear Coherence [49], was computed between all possible pairs of the ROIs for each of the seven independent EEG frequency bands of 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 (30–45 Hz) for each subject. The values of lagged linear connectivity computing between all pairs of ROIs for each frequency band were used as weights of the networks built in the graph analysis.

Graph analysis

A network is a mathematical representation of a real-world complex system. It is defined by a collection of nodes (vertices) and links (edges) between pairs of nodes. Nodes usually represent brain regions, while links represent anatomical, functional, or effective connections [50], depending on the dataset. Functional connections correspond to magnitudes of temporal correlations in activity and may occur between pairs of anatomically unconnected regions. A weighted graph is a mathematical representation of a set of elements (vertices) that may be linked through connections of variable weights (edges).

In the present study, weighted and undirected networks were built, the vertices of the network being the estimated cortical sources in the BAs, and the edges being weighted by the Lagged Linear value within each pair of vertices. The software instrument used here for the graph analysis was the Brain Connectivity Toolbox (BCT, ), adapted with our own Matlab scripts.

Small World (SW) parameter was evaluated on brain networks, since it measures the balance between the local connectedness and the global integration of a network, representing the brain network organization. SW organization is intermediate between that of random networks, the short overall path length of which is associated with a low level of local clustering, and that of regular networks or lattices, the high-level of clustering of which is accompanied by a long path length [51]. The measure of network small-worldness was defined as the ratio of the normalized Clustering Coefficient, and the normalized Path Length.

Statistical evaluation

eLORETA statistical evaluation was made on a graph analysis pattern extracted with sLORETA/eLORETA from the brain network. The normality of the data was tested using the Kolmogorov-Smirnov test, and the hypothesis of Gaussianity could not be rejected. In order to confirm the working hypothesis, a statistical ANOVA design was addressed for the SW difference between pre- and post- task in the factor Group (Ctrl, aMCI, AD) for each of the bands of interest (delta, theta, alpha 1, alpha 2, beta 1, beta 2, gamma).

Pearson’s linear correlations were tested between and the percentage improvement and the significance parameters (Bonferroni corrected to obtain p < 0.05).

Results

Behavioral results

Behavioral results confirmed the learning effect as reflected by improvement of SMoL performances. Statistical analysis reported in top of Fig. 3 of the first 8 repetitions showed (F(14,469) = 4.29; p < 0.00001) that all subjects increased the number of reached target, but, as expected, Ctrl significantly more than aMCI and AD. Furthermore, it is possible to observe a ceiling effect on the learning that occurs after the first half of repetitions. For this reason, to measure the improvement of the performance, we introduced the “Difference Target”, namely the difference of averaged last four and the first four repetitions.

Fig. 3

Top: Learning effect as reflected by improvement of Sensory Motor Learning task performances. All subjects increased the number of reached target, but Ctrl significantly more than aMCI and AD. Furthermore, it is possible to observe a floor effect on the learning that occurs after the first half of repetitions. For this reason, to measure the improvement of the performance, we introduced the “Difference Target”, namely the difference of averaged last four and the first four repetitions. Bottom: The ANOVA of the Difference Target (F(2,62) = 6.91; p < 0.002) shows that Ctrl improved their performance significantly more than both aMCI (p < 0.04) and AD (p < 0.03).

The ANOVA of the Difference Target (F(2,62) = 6.91; p < 0.002) shows that Ctrl improved their performance significantly more than both aMCI (p < 0.04) and AD (p < 0.03) (Fig. 3, bottom).

Graph theory parameter analysis

The ANOVA for the evaluation of the SW showed no statistically significant interaction between Group (Ctrl, aMCI, AD), Band (delta, theta, alpha 1, alpha 2, beta 1, beta 2, gamma) and Time (pre and post) factors. Namely, very similar SW properties were found in pre- and post-learning in each of the three groups of subjects, but it was possible to observe a marked difference in alpha 2 band correlated with the severity of neurodegeneration. In particular, we introduced a SW alpha 2 difference index as the post- minus pre- values (F(2,83) = 3.0732; p < 0.05, Fig. 4), it was evident that Ctrl increase alpha 2 SW, while AD decrease and aMCI present only minimal differences in pre/post values. Planned post-hoc testing showed a significant difference between Ctrl and AD (p < 0.013614) and a trend for significant difference between Ctrl and aMCI (p = 0.08).

Fig. 4

SW alpha 2 difference index as the post- minus pre-values (F(2,83) = 3.0732; p < 0.05). It was evident that Ctrl increase alpha 2 SW, while AD decrease and aMCI present only minimal differences in pre/post values. Planned post-hoc testing showed a significant difference between Ctrl and AD (p < 0.013614) and a trend for significant difference between Ctrl and aMCI (p = 0.08).

Correlation between alpha 2 SW and learning (Difference Target)

Correlation analyses showed a negative correlation between alpha 2 SW in pre-period and behavioral learning (r = –0.2592, p < 0.0342, Fig. 5 top), namely lower alpha 2 SW, higher the learning in the first 8 repetitions.

Fig. 5

Top: Correlation analyses showed a negative correlation between alpha 2 SW in the pre-period and behavioral learning (r = –0.2592, p < 0.0342), namely lower alpha 2 SW, higher the learning in the first 8 repetitions. Middle: Furthermore, it is present a correlation (r = 0.2671, p < 0.0289) between SW and Target differences, namely the higher SW increase, the higher the improvement of reached targets. Bottom: Alpha 2 SW pre-period presented a strong negative correlation with SW difference (r = –0.5298, p < 0.00000008). This result means that lower SW pre-, the higher the possibility to increase and this increment correlates with the behavioral improvement, as observed in the number of reached targets.

Furthermore, a correlation (r = 0.2671, p < 0.0289, Fig. 5 middle) is present between SW and Target differences, namely the higher SW increases, the higher the improvement of reached targets.

Finally, alpha 2 SW pre-period presented a strong negative correlation with SW difference (r = –0.5298, p < 0.00000008, Fig. 5 bottom). This result means that the lower SW pre-, the higher the possibility to increase and this increment correlates with the behavioral improvement, as observed in the number of reached targets.

Control analyses

As a control analysis, to check the specificity of the selected network, the main analyses were also performed on a different network. In particular, when the Sensory Motor network was used, no correlation was observed.

We also performed a statistical ANOVA of the Difference Target (F(2, 60) = 6.2238, p = 0.00350) using Alpha 2 pre SW as covariate, very similar results were obtained showing again that Ctrl improved their performance significantly more than both aMCI (p < 0.003) and AD (p < 0.002).

Finally, the same correlations of the main results were also tested on Clustering coefficient and Characteristic Path Length. Results showed no correlation between Difference Target and Alpha 2 pre (p > 0.1), no correlation between Difference Target and Difference Alpha 2 (p > 0.2), negative correlations between Alpha 2 pre and Difference Alpha 2 (Clustering r = –0.27, p = 0.008; Path Length r = –0.26, p = 0.01).

DISCUSSION

On the basis of the reported results, it can be argued that brain connectome is a useful approach not only for understanding brain cognitive functions, but also for extracting biomarkers that could discriminate between different brain states. By using network metrics to represent a feature space for classification, detection of mental states of learners could be enhanced. An important element in examining brain function is the consideration that efficient brain activity depends on the collaboration and co-activation of many brain areas, for instance via transient synchronization and phase coherence of their oscillatory firing as reflected in the EEG rhythms, which operate as a coordinated network to achieve correct behavioral output [52]. The examination of brain activation from a network perspective is crucial in understanding the factors that drive several brain activities including our motor behavior. These brain connections could be organized in a network topology characterized by high degree of local clustering (segregation) and long-distance connections (integration). “Small-world” (SW) concept was introduced as a model of network organization allowing an optimal balance between local specialization and global integration [53]. This approach is promising for modeling brain functional architecture [54] and to correlate it with behavior (i.e., neuropsychological tests performance). It evaluates whether functional connectivity patterns between brain areas reproduce organization of more or less strongly bound networks based on the strength and duration of oscillatory firing synchronization between adjacent/remote neuronal assemblies [43 , 55–58]. Healthy brain organization reflects an optimal balance of functional integration and segregation; such scenario is termed small-world. SW characteristics reflect complex inhibitory and excitatory brain circuits consisting of functionally specialized regions that continuously and mutually cooperate to acquire, share and integrate information in a constant state of dynamic fluctuations also governed by a number of variables including attention, emotion, motivation, and arousal, and which finally could influence network performance.

In the present study, correlation analyses showed that alpha 2 SW in pre-period predicts behavioral learning (r = –0.2592, p < 0.0342), namely lower alpha 2 SW, higher the learning task. Concluding, the lower the level of SW alpha 2 in this specific network, the higher the possibility to increase its level during the task and better the learning of this task and consequently the number of reached targets.

Regarding the effects on brain rhythms, while low-frequency alpha rhythms (about 8–10 Hz) are supposed to reflect the regulation of global cortical arousal [59, 60], there is consensus that the high alpha rhythms (10.5–12 Hz) reflect the functional modes of thalamo-cortical and cortico-cortical loops that facilitate/inhibit the transmission and retrieval of sensorimotor information into the brain [60 –62] and reflect the “resonance” of specific neural systems for the elaboration of semantic information [63, 64].

Furthermore, many studies have looked at topological changes of brain networks with different modalities including structural and diffusion tensor imaging MRI, EEG/MEG, and fMRI recently reviewed by [65]. Therefore, AD brain topology can be represented by a progressive derangement of the brain organization in hub regions and long-range connections causing SW architecture alteration. In fact, due to the modulation of local and global connectivity parameters, the large-scale functional brain network organization in AD deviates from the optimal SW architecture toward a more “ordered” type in low frequency (as reflected by lower values of SW) and more “random” in the alpha band (as reflected by higher values of SW), producing a less efficient information exchange across brain areas in line with the disconnection hypothesis of AD [66]. In fact, the present findings are in line with previous studies [43 , 67] in which SW characteristics were found to be decreased in patients with AD and MCI in low frequency band. Moreover, significant differences between healthy elderly, and MCI and AD patients have been demonstrated by showing that physiological brain aging presents higher specialization of SW EEG rhythms characteristics being lower in alpha bands [55].

As a limitation of the chosen network, an ideal network was chosen to make the analyses on. Actually, brain networks evolve with individuals and continuously change not only with age but also with experience and learning processes. In fact, the best way could also be not to evaluate an average network based on elderly or AD subjects but to perform individual network on which graph theory application is made. This could be seen as a limitation of the study but it is also due to the difficulty for an AD subject to perform the task inside the MRI scan.

Concluding, the intrinsic characteristics of EEG rhythms contain relevant information on learning processes. Further EEG connectivity studies could be useful to test whether the functional connectome approach as well as conventional biomarkers would be effective methods for monitoring and eventually modulating learning processes during training.

Footnotes

ACKNOWLEDGMENTS

This work was partially supported by the Italian Ministry of Health for Institutional Research (Ricerca corrente) and for the project “NEUROMASTER: NEUROnavigated MAgnetic STimulation in patients with mild-moderate Alzheimer disease combined with Effective cognitive Rehabilitation” (GR-2013-02358430).

Authors’ disclosures available online (.

References

Anguera

, Reuter-Lorenz

, Willingham

, Seidler

(2011) Failure to engage spatial working memory contributes to age-related declines in visuomotor learning. J Cogn Neurosci 23, 11–25.

Coynel

, Marrelec

, Perlbarg

, Pelegrini-Issac

, Van de Moortele

, Ugurbil

, Doyon

, Benali

, Lehericy

(2010) Dynamics of motor-related functional integration during motor sequence learning. Neuroimage 49, 759–766.

Doyon

, Benali

(2005) Reorganization and plasticity in the adult brain during learning of motor skills. Curr Opin Neurobiol 15, 161–167.

Graydon

, Friston

, Thomas

, Brooks

, Menon

(2005) Learning-related fMRI activation associated with a rotational visuo-motor transformation. Brain Res Cogn Brain Res 22, 373–383.

Guillot

, Collet

, Nguyen

, Malouin

, Richards

, Doyon

(2008) Functional neuroanatomical networks associated with expertise in motor imagery. Neuroimage 41, 1471–1483.

Tamas

, Johansen-Berg

, Tomassini

, Bosnell

, Matthews

, Beckmann

(2008) Model-free characterization of brain functional networks for motor sequence learning using fMRI. Neuroimage 39, 1950–1958.

Miall

, Jenkinson

(2005) Functional imaging of changes in cerebellar activity related to learning during a novel eye-hand tracking task. Exp Brain Res 166, 170–183.

Muller

, Kleinhans

, Pierce

, Kemmotsu

, Courchesne

(2002) Functional MRI of motor sequence acquisition: Effects of learning stage and performance. Brain Res Cogn Brain Res 14, 277–293.

Orban

, Peigneux

, Lungu

, Albouy

, Breton

, Laberenne

, Benali

, Maquet

, Doyon

(2010) The multifaceted nature of the relationship between performance and brain activity in motor sequence learning. Neuroimage 49, 694–702.

10.

Vecchio

, Lacidogna

, Miraglia

, Bramanti

, Ferreri

, Rossini

(2014) Prestimulus interhemispheric coupling of brain rhythms predicts cognitive-motor performance in healthy humans. J Cogn Neurosci 26, 1883–1890.

11.

Puttemans

, Wenderoth

, Swinnen

(2005) Changes in brain activation during the acquisition of a multifrequency bimanual coordination task: From the cognitive stage to advanced levels of automaticity. J Neurosci 25, 4270–4278.

12.

Remy

, Wenderoth

, Lipkens

, Swinnen

(2008) Acquisition of a new bimanual coordination pattern modulates the cerebral activations elicited by an intrinsic pattern: An fMRI study. Cortex 44, 482–493.

13.

Ronsse

, Puttemans

, Coxon

, Goble

, Wagemans

, Wenderoth

, Swinnen

(2011) Motor learning with augmented feedback: Modality-dependent behavioral and neural consequences. Cereb Cortex 21, 1283–1294 .

14.

Toni

, Rowe

, Stephan

, Passingham

(2002) Changes of cortico-striatal effective connectivity during visuomotor learning. Cereb Cortex 12, 1040–1047.

15.

Vogt

, Buccino

, Wohlschlager

, Canessa

, Shah

, Zilles

, Eickhoff

, Freund

, Rizzolatti

, Fink

(2007) Prefrontal involvement in imitation learning of hand actions: Effects of practice and expertise. Neuroimage 37, 1371–1383.

16.

Muellbacher

, Ziemann

, Boroojerdi

, Cohen

, Hallett

(2001) Role of the human motor cortex in rapid motor learning. Exp Brain Res 136, 431–438.

17.

Perez

, Lungholt

, Nyborg

, Nielsen

(2004) Motor skill training induces changes in the excitability of the leg cortical area in healthy humans. Exp Brain Res 159, 197–205.

18.

Scholz

, Klein

, Behrens

, Johansen-Berg

(2009) Training induces changes in white-matter architecture. Nat Neurosci 12, 1370–1371.

19.

McNamara

, Tegenthoff

, Dinse

, Buchel

, Binkofski

, Ragert

(2007) Increased functional connectivity is crucial for learning novel muscle synergies. Neuroimage 35, 1211–1218.

20.

Bassett

, Wymbs

, Porter

, Mucha

, Carlson

, Grafton

(2011) Dynamic reconfiguration of human brain networks during learning. Proc Natl Acad Sci U S A 108, 7641–7646.

21.

Heitger

, Ronsse

, Dhollander

, Dupont

, Caeyenberghs

, Swinnen

(2012) Motor learning-induced changes in functional brain connectivity as revealed by means of graph-theoretical network analysis. Neuroimage 61, 633–650.

22.

Debaere

, Wenderoth

, Sunaert

, Van

, Swinnen

(2003) Internal vs external generation of movements: Differential neural pathways involved in bimanual coordination performed in the presence or absence of augmented visual feedback. Neuroimage 19, 764–776.

23.

Debaere

, Wenderoth

, Sunaert

, Van

, Swinnen

(2004) Cerebellar and premotor function in bimanual coordination: Parametric neural responses to spatiotemporal complexity and cycling frequency. Neuroimage 21, 1416–1427.

24.

Kovacs

, Shea

(2010) Amplitude differences, spatial assimilation, and integrated feedback in bimanual coordination. Exp Brain Res 202, 519–525.

25.

Kovacs

, Shea

(2011) The learning of 90 degrees continuous relative phase with and without Lissajous feedback: External and internally generated bimanual coordination. Acta Psychol (Amst) 136, 311–320 .

26.

Swinnen

, Wenderoth

(2004) Two hands, one brain: Cognitive neuroscience of bimanual skill. Trends Cogn Sci 8, 18–25.

27.

Vahdat

, Darainy

, Ostry

(2014) Structure of plasticity in human sensory and motor networks due to perceptual learning. J Neurosci 34, 2451–2463.

28.

Taya

, Sun

, Babiloni

, Thakor

, Bezerianos

(2015) Brain enhancement through cognitive training: A new insight from brain connectome. Front Syst Neurosci 9, 44.

29.

Maclin

, Mathewson

, Low

, Boot

, Kramer

, Fabiani

, Gratton

(2011) Learning to multitask: Effects of video game practice on electrophysiological indices of attention and resource allocation. Psychophysiology 48, 1173–1183.

30.

Langer

, Pedroni

, Gianotti

, Hanggi

, Knoch

, Jancke

(2012) Functional brain network efficiency predicts intelligence. Hum Brain Mapp 33, 1393–1406.

31.

Sporns

(2013) Structure and function of complex brain networks. Dialogues Clin Neurosci 15, 247–262.

32.

Ginestet

, Simmons

(2011) Statistical parametric network analysis of functional connectivity dynamics during a working memory task. Neuroimage 55, 688–704.

33.

Kitzbichler

, Henson

, Smith

, Nathan

, Bullmore

(2011) Cognitive effort drives workspace configuration of human brain functional networks. J Neurosci 31, 8259–8270.

34.

Breckel

, Thiel

, Bullmore

, Zalesky

, Patel

, Giessing

(2013) Long-term effects of attentional performance on functional brain network topology. PLoS One 8 e74125.

35.

Sun

, Lim

, Kwok

, Bezerianos

(2014) Functional cortical connectivity analysis of mental fatigue unmasks hemispheric asymmetry and changes in small-world networks. Brain Cogn 85, 220–230.

36.

Petersen

, Smith

, Ivnik

, Tangalos

, Schaid

, Thibodeau

, Kokmen

, Waring

, Kurland

(1995) Apolipoprotein E status as a predictor of the development of Alzheimer’s disease in memory-impaired individuals. JAMA 273, 1274–1278.

37.

Petersen

, Doody

, Kurz

, Mohs

, Morris

, Rabins

, Ritchie

, Rossor

, Thal

, Winblad

(2001) Current concepts in mild cognitive impairment. Arch Neurol 58, 1985–1992.

38.

McKhann

, Knopman

, Chertkow

, Hyman

, Jack

Jr. , Kawas

, Klunk

, Koroshetz

, Manly

, Mayeux

, Mohs

, Morris

, Rossor

, Scheltens

, Carrillo

, Thies

, Weintraub

, Phelps

(2011) The diagnosis of dementia due to Alzheimer’s disease: Recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement 7, 263–269.

39.

Petersen

, Smith

, Waring

, Ivnik

, Kokmen

, Tangelos

(1997) Aging, memory, and mild cognitive impairment. Int Psychogeriatr 9 Suppl 1, 65–69.

40.

Portet

, Ousset

, Visser

, Frisoni

, Nobili

, Scheltens

, Vellas

, Touchon

(2006) MCI Working Group of the European Consortium on Alzheimer’s Disease Mild cognitive impairment (MCI) in medical practice: A critical review of the concept and new diagnostic procedure. Report of the MCI Working Group of the European Consortium on Alzheimer’s Disease. J Neurol Neurosurg Psychiatry 77, 714–718.

41.

Tecchio

, Vecchio

, Ventriglia

, Porcaro

, Miraglia

, Siotto

, Rossini

, Rongioletti

, Squitti

(2016) Non-ceruloplasmin copper distinguishes a distinct subtype of Alzheimer’s disease: A study of EEG-derived brain activity. Curr Alzheimer Res 13, 1374–1384.

42.

Ghilardi

, Ghez

, Dhawan

, Moeller

, Mentis

, Nakamura

, Antonini

, Eidelberg

(2000) Patterns of regional brain activation associated with different forms of motor learning. Brain Res 871, 127–145.

43.

Vecchio

, Miraglia

, Marra

, Quaranta

, Vita

, Bramanti

, Rossini

(2014) Human brain networks in cognitive decline: A graph theoretical analysis of cortical connectivity from EEG data. J Alzheimers Dis 41, 113–127.

44.

Vecchio

, Miraglia

, Bramanti

, Rossini

(2014) Human brain networks in physiological aging: A graph theoretical analysis of cortical connectivity from EEG data. J Alzheimers Dis 41, 1239–1249.

45.

Miraglia

, Vecchio

, Bramanti

, Rossini

(2016) EEG characteristics in “eyes-open” versus “eyes-closed” conditions: Small-world network architecture in healthy aging and age-related brain degeneration. Clin Neurophysiol 127, 1261–1268.

46.

Miraglia

, Vecchio

, Rossini

(2017) Searching for signs of aging and dementia in EEG through network analysis. Behav Brain Res 317, 292–300.

47.

Miraglia

, Vecchio

, Bramanti

, Rossini

(2015) Small-worldness characteristics and its gender relation in specific hemispheric networks. Neuroscience 310, 1–11.

48.

Pascual-Marqui

(2002) Standardized low-resolution brain electromagnetic tomography (sLORETA): Technical details. Methods Find Exp Clin Pharmacol 24 Suppl D, 5–12.

49.

Pascual-Marqui

(2007) Instantaneous and lagged measurements of linear and nonlinear dependence between groups of multivariate time series: Frequency decomposition.. arXiv preprint arXiv:0711 1455

50.

Friston

(1994) Functional and effective connectivity in neuroimaging: A synthesis. Hum Brain Mapp 2, 56–78.

51.

Rubinov

, Sporns

(2010) Complex network measures of brain connectivity: Uses and interpretations. Neuroimage 52, 1059–1069.

52.

Womelsdorf

, Schoffelen

, Oostenveld

, Singer

, Desimone

, Engel

, Fries

(2007) Modulation of neuronal interactions through neuronal synchronization. Science 316, 1609–1612.

53.

Watts

, Strogatz

(1998) Collective dynamics of ‘small-world’ networks. Nature 393, 440–442.

54.

Bassett

, Bullmore

(2006) Small-world brain networks. Neuroscientist 12, 512–523.

55.

Vecchio

, Miraglia

, Curcio

, Altavilla

, Scrascia

, Giambattistelli

, Quattrocchi

, Bramanti

, Vernieri

, Rossini

(2015) Cortical brain connectivity evaluated by graph theory in dementia: A correlation study between functional and structural data. J Alzheimers Dis 45, 745–756.

56.

Stam

, Jones

, Nolte

, Breakspear

, Scheltens

(2007) Small-world networks and functional connectivity in Alzheimer’s disease. Cereb Cortex 17, 92–99.

57.

de Haan

, Pijnenburg

, Strijers

, van der Made

, van der Flier

, Scheltens

, Stam

(2009) Functional neural network analysis in frontotemporal dementia and Alzheimer’s disease using EEG and graph theory. BMC Neurosci 10, 101.

58.

Caliandro

, Vecchio

, Miraglia

, Reale

, Della

, La

, Lacidogna

, Iacovelli

, Padua

, Bramanti

, Rossini

(2017) Small-world characteristics of cortical connectivity changes in acute stroke. Neurorehabil Neural Repair 31, 81–94.

59.

Klimesch

(1999) EEG alpha and theta oscillations reflect cognitive and memory performance: A review and analysis. Brain Res Brain Res Rev 29, 169–195.

60.

Pfurtscheller

, Lopes da Silva

(1999) Event-related EEG/MEG synchronization and desynchronization: Basic principles. Clin Neurophysiol 110, 1842–1857 .

61.

Brunia

(1999) Neural aspects of anticipatory behavior. Acta Psychol (Amst) 101, 213–242 .

62.

Steriade

, Llinas

(1988) The functional states of the thalamus and the associated neuronal interplay. Physiol Rev 68, 649–742.

63.

Klimesch

(1996) Memory processes, brain oscillations and EEG synchronization. Int J Psychophysiol 24, 61–100.

64.

Klimesch

(1997) EEG-alpha rhythms and memory processes. Int J Psychophysiol 26, 319–340.

65.

Xie

, He

(2011) Mapping the Alzheimer’s brain with connectomics. Front Psychiatry 2, 77.

66.

D’Amelio

, Rossini

(2012) Brain excitability and connectivity of neuronal assemblies in Alzheimer’s disease: From animal models to human findings. Prog Neurobiol 99, 42–60 .

67.

de Haan

, van der Flier

, Koene

, Smits

, Scheltens

, Stam

(2012) Disrupted modular brain dynamics reflect cognitive dysfunction in Alzheimer’s disease. Neuroimage 59, 3085–3093.

68.

Miraglia

, Vecchio

, Rossini

(2018) Brain electroencephalographic segregation as a biomarker of learning. Neural Netw 106, 168–174.