Cognitive Network Neuroscience

John D. Medaglia1, Mary-Ellen Lynall2,3, and Danielle S. Bassett4

Abstract
■ Network science provides theoretical, computational, and work science to cognitive neuroscience. We describe the meth-
empirical tools that can be used to understand the structure odology of network science as applied to the particular case of
and function of the human brain in novel ways using simple neuroimaging data and review its uses in investigating a range
concepts and mathematical representations. Network neuro- of cognitive functions including sensory processing, language,
science is a rapidly growing field that is providing considerable emotion, attention, cognitive control, learning, and memory.
insight into human structural connectivity, functional connectiv- In conclusion, we discuss current frontiers and the specific chal-
ity while at rest, changes in functional networks over time lenges that must be overcome to integrate these complemen-
(dynamics), and how these properties differ in clinical popu- tary disciplines of network science and cognitive neuroscience.
lations. In addition, a number of studies have begun to quantify Increased communication between cognitive neuroscientists
network characteristics in a variety of cognitive processes and and network scientists could lead to significant discoveries
provide a context for understanding cognition from a network under an emerging scientific intersection known as cognitive
perspective. In this review, we outline the contributions of net- network neuroscience. ■

INTRODUCTION
processing mechanisms of the human brain. Moreover,
The conceptual frameworks that we use to understand the mathematical formalism is both generalizable (not
the brain and guide empirical and theoretical investi- being limited to applications to a single type of data or
gations have evolved slowly over several centuries. at a single spatial or temporal resolution) and flexible
Phrenology gave way to a focus on the interactions (enabling group comparisons, statistical inference, and
between brain areas or smaller computational units (con- model development).
nectionism) and the symbolic language of thought itself As with any new conceptual or mathematical frame-
(computationalism). During this evolution, cognitive work, it is critical to determine whether the novel ap-
psychologists reached out to mathematical frameworks proach is actually enlightening. Scientific enlightenment
developed in other disciplines—physics, mathematics, can take one of three forms: (i) the discovery of fun-
and engineering—to capture the brain’s function in for- damental principles that govern observed phenomena;
mal models. Artificial neural networks, for example, (ii) validated relationships with other known variables;
provided an early means of simulating information pro- and (iii) utility in uncovering novel processes, structures,
cessing paradigms inspired by biological neural systems. or phenomena that assist us in interpreting (but cannot
The landscape of potential frameworks and mathe- simply be explained by) prior empirical or principled
matical tools to examine complex dynamical systems like knowledge ( Woodward, 2011). In the first case (funda-
the human brain changed dramatically in the last few mental principles), it may be that there are governing
decades with the popularization and further develop- attributes of dynamical networks in general that apply
ment of network science (Newman, 2010). The use of to the special case of brains and the minds that depend
networks in neuroimaging has provided new means to upon them, a notion to which we will return in Current
investigate key questions in cognitive neuroscience. In Frontiers, below. In the second case (validation), confi-
this scheme, brain regions are treated as network nodes dence can be afforded by demonstrated network cor-
and the anatomical connections or putative functional relates of behavior (Reijmer, Leemans, Brundel, &
interactions between these regions are treated as net- Biessels, 2013), network alterations in psychiatric condi-
work edges (Figure 1). The network representation tions or neurological disorders (Fornito, Zalesky, Pantelis,
provides a parsimonious description of heterogeneous & Bullmore, 2012; Bassett & Bullmore, 2009; He, Chen,
interaction patterns thought to underlie the information Gong, & Evans, 2009), and network predictors of future
brain function or behavioral performance (Ekman, Derrfuss,
Tittgemeyer, & Fiebach, 2012; Heinzle, Wenzel, & Haynes,
1
Moss Rehabilitation Research Institute, 2University of Cam- 2012; Bassett, Wymbs, et al., 2011). In the third case (novel
bridge, 3University of Oxford, 4University of Pennsylvania utility), network-based approaches provide new information

© Massachusetts Institute of Technology Journal of Cognitive Neuroscience X:Y, pp. 1–21

doi:10.1162/jocn_a_00810
about brain function that cannot be derived from what we intrinsic functional connectivity (see Seeley, Menon,
already know about a person and their psychological, Schatzberg, Keller, & Glover, 2007 regarding the use of
clinical, or other status. In this case, the application of this term) of the brain (see Raichle, 2011); relationship
network science allows us to observe new phenomena, between the brain’s intrinsic functional networks, organi-
rather than explaining an already-observed phenomenon. zation during cognition, and underlying structure (Smith
A strong criterion for achieving enlightenment is et al., 2009); methodological approaches to examining
whether fundamental mechanisms have been identified. brain networks (Craddock et al., 2013); and relevance of
Mechanisms are “entities and activities organized in such network approaches to cognitive neuroscience (Sporns,
a way that they are responsible for the phenomenon” 2014). In this review, we demonstrate that neuroimaging
(Illari & Williamson, 2012). Mechanism discovery pro- studies to date have uncovered many new network phe-
ceeds gradually, and we propose that network science nomena in the human brain and their associations with
has the potential to uncover fundamental mechanisms cognitive processes. We first provide a comprehensive
in cognitive neuroscience. In principle, it is uniquely description of the conceptual and mathematical frame-
able to represent the brain in all its complexity. A major work of networks as applied within cognitive neuro-
advantage of network techniques is the explicit repre- science in the study of what has become known as the
sentation and assessment of both neural components human connectome (Sporns, 2011, 2012; Sporns, Tononi,
(neurons or brain regions) and their interactions with & Kötter, 2005; see also Kopell, Gritton, Whittington, &
one another (synapses or functional connections). Kramer, 2014). Then, we address key questions in cog-
The application of network techniques to neuroimag- nitive neuroscience using noninvasive neuroimaging mea-
ing data entails a coarse-grained perspective of lower- surements in humans. We conclude with a discussion of
level dynamical processes. Such techniques have begun exciting new frontiers and important theoretical and
to characterize brain network features relevant to cogni- methodological considerations.
tion that cannot be observed from the sole perspective of
functional localization. Numerous studies have applied
network methods to brain structural data and functional CONCEPTUAL AND
neuroimaging data during rest. Applications of network MATHEMATICAL FRAMEWORK
methods to understand cognition have been relatively
few. We suggest that cognitive network neuroscience is Conceptual Framework
in an early phase of enlightenment and that increased Many network systems comprise complex and diverse
communication between cognitive neuroscientists and interactions (Newman, 2010). Network representations
network scientists can lead to substantial discoveries. have the unique advantages of (i) enabling the quantita-
We review here some early successes in this new field tive analysis of these heterogeneous interactions within a
and outline the potential for cognitive network neuro- unified mathematical framework and (ii) enabling the
science to enrich our understanding of human cognition. examination of higher order multivariate patterns rather
This complements several other recent reviews on the than simply pairwise interactions. These advantages are

Figure 1. From nodes to networks. (A) Brain regions are organized into cytoarchitectonically distinct areas. (B) Each cytoarchitectural configuration
has structural properties with different implications for computational functions. (C) Cytoarchitectural regions can be represented as nodes in
a network. The nodes have functional associations, represented as edges, that extend beyond spatial boundaries evident in cytoarchitectural
organization. Subsystems can be described as network modules. Modules have varying intraconnectivity and intermodule connectivity in the
human brain. (D) An example topology of the modular organization of functional brain networks demonstrating the communication between
computational resources of different types.

2 Journal of Cognitive Neuroscience Volume X, Number Y

particularly useful in the study of the human brain, where traditional approaches, cognitive network neuroscience
different brain areas have different structural properties focuses on complex interactions between spatially dis-
(cytoarchitectural configuration, volume, shape, white crete brain regions, represented by graphs, and seeks
matter tracts) and dynamics and are known to play dis- to link these patterns of interaction to measured behav-
tinct roles in cognitive function. ioral variables. One key epistemological consequence of
Conceptually, brain networks are simplified represen- using network representations is that they can describe
tations of region–region relationships. Functional brain and uncover higher-level complexity that depends on
networks capture temporal relationships between activity the interacting elements of the networked system, which
in different brain regions (e.g., based on estimates of traditional approaches cannot provide. As neuroscience
functional or effective connectivity; Friston, 1994), ana- progresses, these disciplines will converge on a common
tomical brain networks capture white matter links be- scientific understanding of how cognition is represented
tween brain regions (e.g., using diffusion tractography; in the human brain.
Hagmann et al., 2008), and morphometric brain net-
works capture structural relationships between brain
Mathematical Framework
regions based on covariation between regional volume
(Bassett et al., 2008), cortical thickness (He, Chen, & Mathematically, a brain network can be defined as a
Evans, 2007), surface area (Sanabria-Diaz et al., 2010), graph G composed of N nodes (brain regions) and E
and curvature (Ronan et al., 2012) over subjects. Although edges (region–region relationships). In network science,
structural and functional networks can be produced on a the term graph refers to the join-the-dots pattern of con-
subject-by-subject basis, morphometric networks rely on nections (edges) between nodes, rather than to a visual
data from multiple subjects. For example, to ascertain a representation of data on axes. We examine the pattern
connection (or correlation) between gray matter thick- of edges linking nodes by quantifying the graph’s struc-
ness in brain region x and in brain region y requires ture using a variety of diagnostics, which each provide
multiple measurements of gray matter thickness in both complementary but not necessarily independent infor-
regions and hence multiple subjects. An analogous ap- mation (Valente, Coronges, Lakon, & Costenbader, 2008).
proach could be taken within subjects over time during In this review, we will describe a few of these diagnostics
brain development to maturity. These three prongs of to illustrate the types of structures that one can probe
investigation stem from three types of neuroimaging but we point the reader to Newman’s recent textbook
measurements: functional imaging (fMRI, EEG; and “Networks: An Introduction” (Newman, 2010) for a more
MEG), diffusion imaging (diffusion spectrum imaging, comprehensive list and associated descriptions and mathe-
diffusion tensor imaging), and structural imaging (struc- matical formulae and to Sporns’ book “Networks of the
tural MRI, sMRI). Brain” (Sporns, 2010) for intuitive descriptions of several
Network representations in neuroimaging data have a diagnostics in the context of neuroscience.
different meaning than traditional representations in the Network diagnostics can be used to probe the organi-
cognitive sciences, computational neuroscience, and zation of functional or structural connections in the brain
cognitive neuroscience. Cognitive science tends to re- across a spectrum of spatial scales from the neighbor-
duce cognitive systems into models of representations hood surrounding a single brain area (local) to the sum-
paired with processes. Measured variables in empirical mary statistics of the connectivity in the whole brain
cognitive studies are often behavioral indices. It typically (global). Diagnostics that describe the organization of
describes the symbol level architectures of cognitive connections in between these two scales are referred
processes irrespective of their physical instantiation. to as mesoscale diagnostics.
Measured behavioral variables are in turn used as evi- A common example of a local diagnostic is the cluster-
dence for or against the predictive capabilities of a cog- ing coefficient, which can be defined as triangles in which
nitive model, which is in turn modified to better predict a node participates, divided by the number of connected
behavioral data. In computational neuroscience, opera- triples in which a node participates (Figure 2A). In brain
tions performed over various levels of neural tissue orga- networks, it is thought that the clustering coefficient
nization are modeled, including candidate operations might indirectly measure the degree of local information
that support cognition. In cognitive neuroscience, brain integration (Sporns, 2010; Bullmore & Sporns, 2009).
structures composed of complex organizations of neu- A common example of a global diagnostic is the aver-
rons are assumed to support cognitive functions. It de- age shortest path length L. The shortest path between
scribes the neural localization of cognitive processes in node i and node j is the fewest number of connections
the brain. Neuroimaging variables are used to predict that must be traversed to get from node i to node j, and
behavioral indices to make inferences about the opera- the average shortest path is the mean of this value over
tions of the underlying neural substrate. Relationships all possible pairs of nodes in the graph (Figure 2B). In
between behavior and imaging variables are used to brain networks, it is thought that the average shortest
modify the understanding of how functions are repre- path length might indirectly measure the degree of global
sented in particular brain structures. In contrast to these information segregation (Sporns, 2010; Bullmore &

Medaglia, Lynall, and Bassett 3

 and global types. The degree centrality of a node, for
example, is given by the number of connections with that
node and therefore quantifies a local property of the net-
work. The betweenness centrality of a node, v, is a more
global diagnostic, given by the number of shortest paths
between any two nodes in the network (e.g., node i
and j) that must pass through node v.
Mesoscale organization can take various forms (Rombach,
Porter, Fowler, & Mucha, 2012; Fortunato, 2010; Porter,
Onnela, & Mucha, 2009). Two interesting types of mesoscale
organization are (i) core–periphery organization (Borgatti
& Everett, 1999), in which a set of nodes forms a densely
connected core whereas a second set of nodes forms a
sparsely connected periphery (Figure 2C, left), and (ii) mod-
ular organization (Newman & Girvan, 2004), in which sets
of nodes form densely connected modules (Figure 2C,
right). In brain networks, both types of organization appear
to exist at both the structural (Bassett et al., 2013; van den
Heuvel & Sporns, 2011) and functional levels (Bassett,
Brown, Deshpande, Carlson, & Grafton, 2011; Meunier,
Lambiotte, Fornito, Ersche, & Bullmore, 2009). Core–
periphery organization could play a role in conferring
robustness to the brain’s structural core (van den Heuvel
& Sporns, 2011) and in enabling a balance between stability
and adaptivity in brain dynamics (Bassett et al., 2013). Mod-
ular organization provides a natural substrate for the
combined integration and segregation of information
processing arguably required during healthy brain function
(Bullmore & Sporns, 2009).
Figure 2. Network diagnostics. (A) The clustering coefficient is a
Although many network properties—such as high clus-
diagnostic of local network structure. The left panel contains a network
with zero connected triangles and therefore no clustering, whereas tering, short path length, core–periphery structure, and
the right panel contains a network in which additional edges (green) modularity—have consistently been found to character-
have been added to close the connected triples (i.e., 3 nodes connected ize connectivity patterns extracted from many types of
by 2 edges) to form triangles (i.e., 3 nodes connected by 3 edges), noninvasive neuroimaging data, these properties all vary
thereby leading to higher clustering. (B) The average shortest path
among people. There is mounting evidence that one can
length is a diagnostic of global network structure. The left panel
contains a network with a relatively long average path length. For reliably identify individual differences in structural (Owen
example, to move from the purple node (top left) to the red node et al., 2013; Dennis et al., 2012; Bassett, Brown, et al.,
(bottom right) requires one to traverse at least 4 edges. The right panel 2011) and functional (Braun et al., 2012; Wang et al.,
contains a network in which addition edges have been added to form 2011; Telesford et al., 2010; Deuker et al., 2009) brain
triangles (green) or to link distant nodes (peach), thereby leading to
network organization, suggesting that individual variation
a shorter average path length in comparison. (C) Mesoscale network
structure can take many forms. The left panel contains a network with in such architectures can be linked to individual variation
a core of densely connected nodes (green circles; green edges) and a in cognitive performance.
periphery of sparsely connected nodes (brown circles; gray edges).
The right panel contains a network with four densely connected
modules (green circles; green edges) and a connector hub (brown APPLICATIONS IN NEUROIMAGING
circle; gray edges) that links these modules to one another.
In this section, we describe recent applications of net-
work-based methods to questions in cognitive neuro-
Sporns, 2009). A related concept—network efficiency science. For each area of enquiry, we summarize the
(Latora & Marchiori, 2001, 2003)—is also calculated based studies that have applied network methods to questions
on shortest paths through a network, but in this case, a in cognitive neuroscience. We then highlight evidence
network will have high efficiency if it has a short path that we feel particularly illustrates the epistemological
length and the network will have low efficiency if it has gains that network neuroscience has brought to that area.
a long path length. In the brain, this network efficiency
is often interpreted to underlie efficiency of information
A Taxonomy for Nodes and Edges
processing (Sporns, 2010; Achard & Bullmore, 2007).
Centrality measures, which quantify the relative influ- The definition of nodes in a network is a necessary step with
ence or rank of a node in a network, include both local consequences for network modeling and interpretation.

4 Journal of Cognitive Neuroscience Volume X, Number Y

Since 1909, Brodmann’s areas (BA) have served as a useful number, and configuration of edges represent patterns of
reference point in the neurosciences (Brodmann, 1909). information flow within the brain.
These histologically defined cytoarchitectonic brain areas
have been proven to support functionally distinct opera-
Anatomical Network Correlates of
tions. In principle, differences between BAs may imply
Cognitive Processes
different dynamical properties (at the level of cortical col-
umns and below) that support critical operations during Discussion of associations between brain structure and
cognition. In functional neuroimaging, many hundreds or cognition provide an important context for understand-
thousands of cortical columns may be contained in sam- ing brain function. Anatomical brain networks represent
pled regions used to define graphs. Despite this, main- the mediating architecture over which functional dynam-
taining distinctions between the cytoarchitecture sampled ics operate. Diffusion weighted imaging data provides
can provide a guiding taxonomy for qualitatively different quantitative measurements of white matter microstruc-
nodes (see Figure 1). It is fundamentally important to cogni- ture, whose integrity is crucial for healthy cognitive function
tion that the neural microstructures underlying regions (Roberts, Anderson, & Husain, 2013). Age-related indi-
sampled in neuroimaging studies vary in a highly organized vidual differences in cognitive performance—particularly
manner. Unfortunately, there is no modern atlas that val- perceptual speed and executive functioning—are accom-
idly represents all BAs in probabalistic space, although one panied by variations in white matter integrity across neural
is underdevelopment (Eickhoff et al., 2007). If a valid pro- systems that display an anterior–posterior gradient across
babalistic atlas becomes available, the considerable vari- the lifespan (see Madden, Bennett, & Song, 2009, for a
ance in the size and morphology of BAs across recent review). White matter fiber bundles in temporal lobe
individuals will remain problematic. In light of these con- projections (uncinate fasciculus, fornix, cingulum, inferior
siderations, we reference estimated BAs where noted in longitudinal fasciculus, and superior longitudinal cortex)
primary references and mention other spatial reference were associated with better executive function. Global
systems (e.g., gross morphological areas) when they network compromise was related to deficits in processing
are used in primary sources. The study and modeling of speed.
variation in cytoarchitectonics will ultimately prove to be Injury to white matter can immediately alter cogni-
critical in understanding how brain networks support tive function in multiple domains (Silver, McAllister, &
cognition. Arciniegas, 2009) and can interact with normal aging to
Subcortical structures do not have an analogous sys- drive later cognitive decline (Moretti et al., 2012). Changes
tem to Brodmann’s mapping but have observably differ- in which matter microstructure correlate with cognitive
ent microstructures and associated functions. Subcortical impairments in processing speed (Niogi et al., 2008), work-
structures are often more obviously distinguishable from ing memory (Palacios et al., 2012; Kinnunen et al., 2011;
one another and the cortex on standard anatomical Wu et al., 2010), and motor skills (Leunissen et al., 2013;
scans. Each subcortical structure tends to contain homo- Farbota et al., 2012). The efficiency of larger-scale network
geneous circuits with fewer layers relative to cortical structure correlates with switching scores on an executive
systems. These circuits tend to be composed of parallel, function task (Caeyenberghs et al., 2012, 2014). These
closed-loop projections (McHaffie, Stanford, Stein, studies have demonstrated that the structure of single
Coizet, & Redgrave, 2005), and link anatomically with brain areas or tracts are not the only, and perhaps not
multiple other subcortical and cortical regions. It is even the best, predictors of cognitive function. Instead,
important to note that atlases used in network analyses cognition is supported by a pattern of connections be-
have oversampled the cortex relative to subcortical tween distributed sets of brain areas.
structures despite the fact that subcortical structures
contain well over half of the neurons in the brain.
Functional Network Correlates of
Most subcortical neurons are in the cerebellum, which
Cognitive Processes
has 3.6 times the number of neurons as the neocortex
(Herculano-Houzel, 2010). Cognitive processes are associated with altered brain activity
Finally, the definition of edges in this context typically and—by extension—functional connectivity (Siebenhuhner,
refers to measures of structural or functional connections Weiss, Coppola, Weinberger, & Bassett, 2013; Yu et al.,
between nodes. Structural edges may include measures 2013; Bassett, Nelson, Mueller, Camchong, & Lim, 2012;
of fractional anisotropy or streamline counts along white Zalesky, Fornito, & Bullmore, 2012). By identifying changes
matter pathways between regions. Functional edges may in functional connectivity patterns induced by experi-
include any of a number of measures of relatedness be- mental tasks, we can begin to uncover the distributed
tween signals from standard signal processing approaches. network processes underlying mental function and be-
Important for cognitive network neuroscience is that the havioral performance. See Figure 3 for an overview of the
type of information communicated along edges differs as representation of brain networks during cognitive states.
a result of computations performed in local regions, which In the following paragraphs, we will review recent studies
are transmitted along white matter pathways. The weight, that have focused on characterizing whole-brain functional

Medaglia, Lynall, and Bassett 5

connectivity patterns in four main areas: (i) vision, audition for healthy general cognitive function (Buckner et al.,
and motion; (ii) memory; (iii) learning; (iv) emotion; 2009; Bassett, Meyer-Lindenberg, Achard, Duke, &
(v) language; and (vi) attention and cognitive control. Bullmore, 2006). However, the spatial layout of func-
These domains are by no means exhaustive of cognition tional wiring and other local properties of functional
but represent the constructs studied in the majority of connectivity can simultaneously be drastically altered,
network analyses to date. Although methods based on both representing local reconfigurations to meet task de-
seed-based correlation (Biswal, Yetkin, Haughton, & Hyde, mands (Bassett et al., 2006). For example, the intrinsic
1995) and component analyses (McKeown et al., 1998) have functional connectivity involves predominantly short
also been applied to task-based data (Michael, Calhoun, and by extension potentially efficient (Bullmore &
Andreasen, & Baum, 2008; Hampson et al., 2006; Calhoun Sporns, 2012) functional wiring, which is lengthened
et al., 2002), here we constrain ourselves to network- or to perform simple motor tasks such as finger tapping
graph-based approaches. (Bassett et al., 2006). In addition, the local connectivity
of individual nodes has been shown to vary across cog-
nitive states, though nodes known to be recruited to
From Intrinsic Functional Networks to Cognition manage general tasks demonstrate reliable node degree
Understanding the intrinsic organization of brain net- across task conditions (Cao et al., 2014). Thus, functional
works provides a context for the reconfigurations ob- systems in the brain are perhaps best understood as a
served during tasks. Intrinsic functional connectivity has stable organization that supports a number of relatively
been found to be more stable than the synchrony of minor state reconfigurations that enable cognitive func-
elicited activation in a number of tasks (Cao et al., tions. The role of some brain regions in the network
2014) suggesting, surprisingly, that there may be greater changes flexibly to address specific task demands,
constraints on how the brain must globally organize to be whereas others are reliably interactive with the rest of
at rest than on how it must allocate resources to achieve the network to manage global demands.
a specific task. Between-atlas comparisons have demon-
strated reliable global and local network properties in
Primary Sensorimotor Regions and
intrinsic connectivity and task-related conditions across
Their Relationship to Cognitive Hubs
network parcellation methodologies (Cao et al., 2014).
An interesting question is how intrinsic functional net- What roles do specific brain regions play within this stable
works subtly reconfigure to support cognitive functions. organization, and how might those roles change during
In comparisons between task and nontask states, func- cognitive functions? Perhaps the simplest set of areas in
tional network organization can appear strikingly similar which to answer this question are the primary sensory
at a global level (Cole, Bassett, Power, Braver, & Petersen, and motor regions, whose functions have been well
2014), suggesting that a stable core network is necessary delineated for over a century. Using network-based

Figure 3. Cognitive network

neuroscience (C = Cognitive
state). A schematic
representation of functional
brain networks during
cognition. Cognitive modules
are indicated by collections
of identically colored nodes
organized into network
modules. The organization of
brain networks varies across
cognitive states and time.
Some features of functional
network organization may
remain relatively stable as a
system “core,” and others may
vary substantially. Modules
may merge and separate.
Connections within and
between modules may change
in strength, configuration,
and number. Network
organization may change
over time as a function of
learning processes.

6 Journal of Cognitive Neuroscience Volume X, Number Y

techniques, these early-evolving regions have been shown sensorimotor cortex and density of connections with the
to display intrinsic functional connectivity patterns that insula and superior temporal gyrus predicted mobility
converge on cognitive “hubs.” across the lifespan (Mishkin & Ungerleider, 2014).
Functional connections from sensorimotor regions are Furthermore, a study of motor execution and motor
partially integrated in a multimodal network between imagination revealed that the SMA was a key node with
sensorimotor systems and cognitive hubs (Sepulcre, high betweenness centrality during motor execution and
2012, 2014). The multimodal network consists of por- the right premotor area had high betweenness centrality
tions of the superior parietal cortex, operculum parietal, during imagination (Xu et al., 2014). In a study of self-
anterior insula in the frontal operculum, dorsal ACC/SMA initiated finger movements, network connectivity strengths
(BA 6/24/32), dorsolateral pFC (BA 10/46), and the con- predicted striatal activity. The connectivity strength be-
fluence of BA 19/22/37/39 in the TPJ (Sepulcre, 2012, tween the dorsolateral pFC and striatum was negatively
2014). Following partial integration in the multimodal correlated with RTs, whereas the connectivity between
network, functional connections all converge in cognitive the ventrolateral pFC and striatum was positively correlated
hubs. Cognitive hubs are those with high centrality in the to RTs (Nagano-Saito, Martinu, & Monchi, 2014).
brain overall (Buckner et al., 2009) or between many Thus, network studies may extend our classical under-
brain networks ( Warren et al., 2014). Hubs have been standing of the organization of the motor system with a
identified within the frontoparietal and default mode focus on the interactive processes between the motor
systems and include the posterior cingulate, lateral tem- system and the rest of the brain. It is possible that motor
poral, lateral parietal, and medial/lateral prefrontal corti- regions serve as conditionally recruited processing hubs
ces (van den Heuvel & Sporns, 2013; Buckner et al., in conjunction with frontal control regions. Network
2009). Damage to cognitive hubs is implicated in many studies have supported this notion by dissociating be-
diseases (Crossley et al., 2014) and can result in general- tween intrinsic and extrinsic distributed processes medi-
ized and catastrophic failures in cognitive function (Warren ated by motor regions (Xu et al., 2014) and identifying
et al., 2014). opponent interactions involving frontal control regions
Overall, this initial work has formed an important basis (Nagano-Saito et al., 2014).
for understanding the relationships between networked With respect to vision and audition, network studies
subsystems in the brain. These discoveries suggest that have shown that the strengthening of key functional con-
an overarching organization appears to be that multiple nections underlies tasks that demand sensory integra-
sensory and motor processes are integrated within a set tion. The spatial distribution of network modules in
of mediating multimodal systems. In turn, multimodal auditory and visual cortex and of hub-like areas became
systems are supervised by a complex frontoparietal net- more constrained to traditional anatomical boundaries in
work and the default mode network. This is consistent a multisensory task and displaying less variability across
with a view of cognition in which the gross organization subjects (Moussa et al., 2011). These results were quali-
is a parallel distribution of extrinsic inputs to the brain, tatively corroborated by Ma, Calhoun, Eichele, Du, and
intermediate parallel associating mechanisms that poten- Adali (2012) who showed task-induced increases in the
tially subserve the “binding” of sensory input, and an clustering coefficient during an auditory oddball task in
overarching supervisory network. Within each layer of comparison to rest. In preparatory intervals just before
this hierarchy, distinct computational operations pre- the performance of a trial in a visual discrimination task,
sumably occur across varying cytoarchitectural mecha- functional networks extracted from fMRI data showed a
nisms. Thus, the major organization of human functional dynamic adjustment in core–periphery interactions, in
systems forms a robust heterarchy with a balance of infor- which task-relevant visual areas move toward the core
mation segregating and integrating functions (see Bressler of densely and mutually interconnected regions (Ekman
& Richter, 2015; Buckner & Krienen, 2013; Passingham, et al., 2012). Notably, this reconfiguration predicted suc-
Rowe, & Sakai, 2013). To better elucidate the network cessful task performance.
basis of cognition within this organization, we now turn Network approaches can therefore contextualize the
to network analyses in specific cognitive behavioral do- local functions of primary sensory areas within systems
mains that have identified associations with functional that support dynamic sensory integration and consoli-
network features. dation. They have discovered that tasks with a heavy
emphasis on sensory processing and integration appear
to depend upon tightly communicating cognitive hubs
Motion, Vision, and Audition
and sensorimotor regions (Ma et al., 2012; Moussa
Whereas traditional approaches examine the topological et al., 2011). Validation for a network view is suggested
organization and activity of the sensorimotor cortices in by the finding that increasing functional network integ-
association with behavior, network analyses have discov- ration results in better performance (Ekman et al.,
ered underlying functional interactions within and between 2012). Sensory tracking and discrimination may require
these regions related to variation in motor function. For robust sustained communication between primary sensory
example, the consistency of community structures in the and nearby regions.

Medaglia, Lynall, and Bassett 7

Memory Henson, Smith, Nathan, & Bullmore, 2011). Functional
networks extracted from BOLD data demonstrated that
Network approaches have begun to enlighten our under- increased connection pruning (Ginestet & Simmons,
standing of the distributed processes supporting effective 2011) and decreased clustering (He et al., 2012) are asso-
memory performance. One fMRI study involving a yes– ciated with swift and accurate performance. Decreased
no odor recognition task demonstrated that the hippo- memory encoding recognition and encoding in aging indi-
campus, caudate nucleus, and anterior cingulate gyrus viduals was associated with increased path length and
more frequently belonged to the same functional module decreased efficiency (Wang, Li, Metzak, He, & Woodward,
during hits than all other conditions. Network-wide mod- 2010).
ularity values were negatively correlated with memory in Taken together, network findings in working memory
the hit condition and positively related to bias scores in may reveal that it is a process that is arbitrated in the
hit and false alarm conditions (Meunier et al., 2014), frontoparietal system. Network studies have led to the
meaning that increasing segregation of the network re- discovery that working memory may fundamentally rely
sulted in inaccuracy. Thus, integration in this memory upon sustained and efficiently organized frontoparietal
subsystem may be critical to some discriminative deci- and visual interactions especially in the alpha and beta
sion processes. In particular, achievement of true positive frequency regimes (Palva, Monto, Kulashekhar, et al.,
memories may require a cooperation of coordinating, 2010; Palva, Monto, & Palva, 2010). Validation for a net-
error monitoring, and memory indexing processes that work perspective was found within this organization: in-
is optimized within sufficiently organized brain networks. creases in brain-wide small-world organization support
In addition, network analyses have brought insight into increased working memory capacity (Stevens et al.,
the brain dynamics underlying working memory. They 2012), and increased network efficiency results in better
have demonstrated the efficient organization of func- memory performance (Kitzbichler et al., 2011; Bassett
tional networks in specific frequency domains across et al., 2009).
the brain support working memory function. During
visual memory maintenance, a combined EEG/ MEG
Learning
study revealed that alpha and beta band networks were
more clustered and small-world like with smaller global Learning takes place over multiple timescales, from hours
efficiency than delta- and theta band networks. The alpha to days to months. Network findings have begun to clarify
and beta band networks had truncated power law degree brain-wide functional changes that support learning
distributions (Palva, Monto, & Palva, 2010). Sustained across these timescales. For example, in a comparison
phase synchrony during retention was found among between pretraining (Day 0) and posttraining (Day 5)
frontoparietal and visual areas in alpha, beta, and gamma sessions for a bimanual motor learning task, improved
frequency bands (Palva, Monto, Kulashekhar, & Palva, performance was related to increased clustering coeffi-
2010). Overall, this suggests that working memory pro- cients, network degree, connection strength, and shortened
cesses are predominantly supported by an efficiently path lengths across the brain in auditory and feedback
organized network predominantly in the alpha and beta conditions (Heitger et al., 2012).
frequency regimes. Several recent studies have begun to capture dynamic
Specific aspects of functional network configurations are patterns of functional connectivity at finer temporal scales,
associated with variance in working memory capacity and from temporal networks (Holme & Saramäki, 2012)
performance. Network modularity and small-worldness in extracted from contiguous 2–3 min windows of fMRI data
intrinsic functional connectivity following a working (Bassett, Yang, Wymbs, & Grafton, 2014; Bassett et al.,
memory test predicted capacity (Stevens, Tappon, Garg, 2013; Mantzaris et al., 2013; Bassett, Wymbs, et al., 2011).
& Fair, 2012). Individuals with higher memory accuracy Dynamic community detection techniques (Mucha,
in a 2-back working memory task tended to have more Richardson, Macon, Porter, & Onnela, 2010) can uncover
cost-efficient functional network architecture, as measured changes in clusters of brain regions linked by strong func-
by MEG in the beta frequency band (12–20 Hz; Bassett, tional connectivity (i.e., putative functional modules;
Meyer-Lindenberg, Weinberger, Coppola, & Bullmore, Bassett, Porter, et al., 2012). Individuals with greater
2009). Synchrony in frontoparietal regions increased with network flexibility (changes in module allegiance over
load, and synchrony in a hub within the intraparietal sulcus time) tended to learn a motor sequence more effectively
predicted individual visual working memory capacity (Bassett, Wymbs, et al., 2011; Figure 4). Using dynamic
(Palva, Monto, Kulashekhar, et al., 2010; Palva, Monto, centrality techniques for the same task, Mantzaris and
& Palva, 2010). With increasing cognitive load (0-back to colleagues (2013) demonstrated that broadcast and
2-back), networks became more globally efficient, less receive centralities, which together measure the flow of
clustered, and less modular, with more long-distance syn- functional connectivity changes over time, decrease strik-
chronization between brain regions. This configuration ingly over the course of the experiment. In an extended
was greater in faster performing individuals than in slower 6-week learning experiment, the presence of a stable
performing individuals for the beta frequency (Kitzbichler, temporal core of primary task-related areas and a flexible

8 Journal of Cognitive Neuroscience Volume X, Number Y

Figure 4. Brain network
dynamics during learning.
(A) The flexibility of brain
network dynamics—defined as
the frequency of a brain region
when it changes its allegiances
to network modules over
time—predicts individual
differences in learning: More
flexible individuals learn better
than less flexible individuals
(Bassett, Wymbs, et al., 2011).
Moreover, brain regions differ
in their flexibility. Regions
with greater flexibility form
a temporal network core,
whereas regions with less
flexibility form a temporal
network periphery. (B) The distribution of the temporal core and periphery nodes in the brain during learning. “Bulk” nodes are those that
do not significantly differ from a temporal network null model. (C) The relationship between region flexibility (f ), core–periphery separation (s),
and learning (parameterized by κ). Brain regions are represented using data points located at the polar coordinates (fs,fκ). Color indicates flexibility:
Blue nodes have lower flexibility, and brown nodes have higher flexibility. Poor learners (straighter spirals) tend to have a small separation
between core and periphery (short spirals), whereas good learners (curvier spirals) tend to have large separation between core and periphery (longer
spirals). The separation between core and periphery is a good predictor of individual differences in learning success. Adapted with permission
from Bassett et al. (2013).

temporal periphery of multimodal association areas was connectivity and hub properties of the left dorsolateral
identified (Bassett et al., 2013). Individuals with a greater pFC (Cisler et al., 2012). In depressed individuals, the
separation between their core and periphery learned centrality of the right dorsolateral pFC was negatively
better during the following 10 days of practice. correlated with the duration of disease. Centrality in
Thus, network neuroscience has discovered that learn- the right inferior frontal gyrus (pars triangularis) and
ing processes require sufficient network flexibility around efficiency in the right superior frontal gyrus were posi-
a stable core system during learning. In terms of vali- tively related to depression severity (Qin et al., 2014).
dation, learning was strongly related to brain flexibility Network analyses in emotion tasks have begun to
(Bassett et al., 2013; Bassett & Gazzaniga, 2011) and extend our understanding of the representation of dis-
the consolidation of network-wide effects following learn- tributed emotion regulatory processes in the brain. The
ing (Mantzaris et al., 2013). Whether these findings gen- dorsolateral pFCs (Koenigs & Grafman, 2009) and inferior
eralize to tasks that more heavily emphasize declarative frontal gyrus (Shamay-Tsoory, Aharon-Peretz, & Perry
memory remains to be seen. 2008) have been implicated as nodes in a distributed
emotion processing system with differentiable functional
roles. Thus far, network approaches have discovered net-
Emotion
work-wide processes in support of an emerging view that a
The involvement of distributed patterns of functional competing balance of function between the right and
connectivity is perhaps unsurprising in higher order cog- dorsolateral pFC has important implications for mood reg-
nitive functions that require cooperative activity from ulation (Qin et al., 2014; Cisler et al., 2012; Grimm et al.,
multiple brain systems. Perhaps as a result, “higher” cog- 2008). Validity was suggested by the finding that as com-
nitive domains have received considerable attention in ponents of the right frontal cortex increasingly dominate
network studies. Network studies of emotion are to date function across the network, negative mood results. In
rare but have also been informative. In response to emo- addition, environmentally induced reductions in broad net-
tion cues, Kinnison, Padmala, Choi, and Pessoa (2012) work communication may result in vulnerability to emo-
demonstrated increased global efficiency as well as de- tion dysregulation over the lifespan (Cisler et al., 2012).
creased modularity. In a network analysis of vulnerability
to reduced mood, resilient individuals demonstrated de-
Language
creased global connectivity and hub-like properties in the
right ventrolateral pFC and decreased local connectivity Whereas traditional views of language have placed an
in the dorsal ACC. Susceptible individuals demonstrated emphasis on the left hemisphere, network neuroscience
decreased local connectivity of the amygdala and ventro- has led to discovery in this domain as well. They have
lateral pFC and decreased hub properties of the amyg- implicated important interactions within supporting
dala and the dorsal ACC. Self-reported severity of early mechanisms of both the left hemisphere and the right
life stressors correlated negatively with global network hemisphere. A graph theoretical analysis of intrinsic

Medaglia, Lynall, and Bassett 9

functional connectivity data demonstrated significant Attention and Cognitive Control
connectivity between Broca’s area and right hemispheric
regions (Muller & Meyer, 2014). This suggests that lan- Cognitive control and attention presumably subserve
guage processing may be asymmetric but involves sys- numerous brain functions (Braver, 2012; Borgers &
tems in both hemispheres. Examining functional network Kopell, 2008; Corbetta & Shulman, 2002; Miller & Cohen,
configurations during language tasks and their rela- 2001; Miyake, Friedman, Emerson, Witzki, & Howerter,
tionships to language performance has led to additional 2000). Network approaches have begun to clarify the inter-
discoveries. A characteristic of network function under- actions between attention, cognitive control regions, and
lying language may be the extent to which specific nodes other brain systems in the service of cognitive functions.
mediate communication across the network and the Attention-demanding tasks activate a frontoparietal
extent to which hemispheres communicate. In the former network (see Parks & Madden, 2013, for a review), largely
case, a study of sentence comprehension revealed that subtended by structural wiring (Hermundstad et al., 2013).
engaging in an explicit task resulted in greater global Using nicotine replacement to modulate attention over
efficiency and increased betweenness centrality of the a prolonged task duration, Giessing, Thiel, Alexander-
left inferior frontal gyrus (Qin et al., 2013). In the latter Bloch, Patel, and Bullmore (2013) demonstrated that
case, in a split visual field experiment, Doron and col- modulations in this network correlated with intraindividual
leagues demonstrated that network-based measurements differences in cognitive function. Nicotine replacement in-
of interhemispheric coordination are generally weakest duced an increase in task performance accompanied by
when lexical stimuli are introduced to the language- functional network reconfiguration toward greater effi-
dominant (left) hemisphere and strongest when they are ciency, less clustering, and longer connection distance
introduced to the contralateral hemisphere (Doron, (Giessing et al., 2013). On the other hand, extended inter-
Bassett, & Gazzaniga, 2012). This novel observation of vals on-task induced a decrease in task performance
coordination highlights the underlying dynamic nature accompanied by a functional network reconfiguration
of brain communication during language processing and toward less efficiency, increased clustering, and shorter
specifically that interhemispheric mechanisms can tran- connection distance.
siently coordinate to subserve processing under challeng- Some distinctions between the network properties
ing conditions. underlying attention and cognitive control have been
An ongoing debate concerns the nature of language identified. The relatively global effects on network config-
with respect to other cognitive systems. In particular, uration during external modulation of attention contrast
one problem is whether language relies upon generalized against the region-centered signatures identified in cogni-
or exclusive and dedicated mechanisms. It has been pro- tive control studies (Cole, Laurent, & Stocco, 2013; Cole,
posed that language systems may be organized in func- Yarkoni, Repovs, Anticevic, & Braver, 2012; Dosenbach
tionally specialized cores with conditionally recruited et al., 2007). Dosenbach and colleagues (2007) demon-
regions on the periphery (see Figure 2C, left; Blank, strated dissociable network structure of two distinct sets
Kanwisher, & Fedorenko, 2014). In particular, cognitive of brain areas that appear to coordinate adaptive (cingu-
control and other regions that respond to multiple cog- lo-opercular network) and stable (frontoparietal network)
nitive demands may be peripheral mechanisms recruited task control on different timescales and using different
to support language functions depending on task de- mechanisms. Power and Petersen (2013) extended this
mands (Blank et al., 2014; Fedorenko, 2014). work and found that control-related regions tend to display
The application of network approaches has discovered start-cue, sustained, and error-related activity during cog-
that language processing involves interactions between nitive tasks. Employing a network analysis of a working
classically identified language systems, supporting right- memory task with high cognitive control demands, Cole
hemispheric systems, and cognitive control systems. In and colleagues (2012) identified the lateral pFC as the
particular, linguistic processing may involve dynamic re- single region whose functional connectivity within and out-
cruitment of homologous right hemispheric resources side of the frontoparietal system displayed a selective rela-
(Doron et al., 2012). The left inferior frontal gyrus may tionship with individual differences in fluid intelligence.
serve as a hub that mediates between conflicting repre- EEG recordings acquired during performance of a mathe-
sentations (Thompson-Schill et al., 1998), and this fea- matical task with high cognitive control demands found
ture may be represented in its fluctuating network that overall connectivity in the frontoparietal system was
betweenness centrality (Qin et al., 2013). These tech- differentially engaged. During transitions from subitizing
niques have begun to identify the conditions under (rapid counting of small numbers of objects) to retrieval,
which general control mechanisms are recruited to sup- increased local and global network efficiency were ob-
port relatively specialized language functions (Fedorenko served in the delta band. Difficult mathematics resulted
& Thompson-Schill, 2014). Examining associations be- in increased cliquishness in delta and theta bands (Klados
tween functional network properties and language per- et al., 2013).
formance may further validate network neuroscience Overall, these early network discoveries in attention
approaches in this domain. and cognitive control support and extend the view that

10 Journal of Cognitive Neuroscience Volume X, Number Y

these processes are frontally mediated (Szczepanski & Network Dynamics and Cognitive Processes
Knight, 2014). Externally cued attention recruits efficient
The human brain is a dynamical system (Deco, Jirsa, &
global network activity, whereas dissociable cognitive con-
McIntosh, 2011), which has far-reaching implications for
trol demands rely upon dual networks that manage shifting
both basic neuroscience and clinical translation in, for
and sustained control processes (Dosenbach et al., 2007).
example, prosthetic design (Shenoy, Kaufman, Sahani,
Validation with performance was suggested by the finding
& Churchland, 2011), epileptic control (Ching, Brown,
that the left pFC engages in dynamic activity with the rest of
& Kramer, 2012), and the application of control theory
the brain to support general cognitive flexibility and fluid
to subcortical neural systems (Schiff, 2012). A critical
reasoning (Cole et al., 2012). Moreover, network ap-
frontier in the application of network-based methods to
proaches have led to the discovery of a spatiotemporal nest-
cognitive neuroscience lies in the development and
ing of processes: cognitive control may be mediated
application of dynamic network methods that character-
through distributed network processes via increased mod-
ize the evolution of putative communication patterns
ularization in dissociable frequency bands. In distinction to
over the multiple temporal scales in which cognitive
working memory, cognitive control may be executed across
function changes (Kopell et al., 2014). These are dynamic
brain networks in the delta and theta bands (Klados et al.,
centrality (Mantzaris et al., 2013) and dynamic commu-
2013). Limits to sustaining cognitive control may result
nity detection (Bassett, cognition-specific) in healthy
from network failure to maintain difficult-to-reach states
and diseased cohorts, and these dynamics can be used
(Giessing et al., 2013). Notably, tasks used in network
to predict fundamental human capacities. However, the
studies to date have not systematically decomposed certain
work in this area is limited, and the field is rich with
cognitive control contributions (see Miyake et al., 2000).
opportunities for both method development and appli-
How specific components of attention and cognitive con-
cations to further questions in cognitive neuroscience.
trol represented in particular nodes influence the rest of
the brain remains an area open for rigorous investigation.
Network-based Prediction of Cognitive Processes

CURRENT FRONTIERS Although network-based approaches show significant

promise to further basic understanding in systems and
Intersections with Cognitive Theory cognitive neuroscience, they are perhaps more pragmat-
So far, we have reviewed progress in an early phase of ically accompanied by novel possibilities for classification
cognitive network neuroscience. A prevailing challenge and prediction. The strength of individual putative func-
for network neuroscience is to provide the basis for tional connections between region pairs (Mokhtari
formal integration with established cognitive theories & Hossein-Zadeh, 2013; Shirer, Ryali, Rykhlevskaia,
(Sporns, 2014). That is: How do neural networks produce Menon, & Greicius, 2012; Chen et al., 2011; Richiardi,
cognitive states? Although we have described the tech- Eryilmaz, Schwartz, Vuilleumier, & Van De Ville, 2011;
niques and potential advantages of examining cognition Shen, Wang, Liu, & Hu, 2010) and more global graph
with network neuroscience, it is clear that studies to date properties themselves (Fekete et al., 2013; Bassett,
have only tangentially addressed classical issues in cogni- Nelson, et al., 2012; Heinzle et al., 2012; Supekar, Menon,
tive psychological theories. Additionally, only a small Rubin, Musen, & Greicius, 2008) form a new set of fea-
number of cognitive domains have received attention in tures that can be used to decode healthy brain states
cognitive network neuroscience. Importantly, as this area (Mokhtari & Hossein-Zadeh, 2013; Heinzle et al., 2012;
develops, several key issues will require exploration. In Shirer et al., 2012; Richiardi et al., 2011) or to classify dis-
particular, whether network neuroscience can account eased and nondiseased cohorts in an effort to develop
for semantics, conscious experience, and complex coor- neurodiagnostics (Fekete et al., 2013; Bassett, Nelson,
dination and competition between cognitive systems at et al., 2012; Richiardi et al., 2012; Chen et al., 2011; Shen
multiple levels of organization will be a stringent test of et al., 2010; Supekar et al., 2008) by exploiting robust
its explanatory power. Explicit focus on these issues may techniques from machine learning (Figure 5; see Richiardi,
encourage theoretical innovations and empirical verifica- Achard, Bunke, & Van De Ville, 2013, for a recent review).
tion. Progressive refinement of, integration with, and Complementary approaches such as mathematical
testing in light of the explanatory power of existing tradi- modeling (Raj, Kuceyeski, & Weiner, 2012) and statistical
tional cognitive architectures (Sporns, 2014; Langley, analyses (Bassett et al., 2013; Zhou, Gennatas, Kramer,
Laird, & Rogers, 2009; Anderson, 1995) will be necessary. Miller, & Seeley, 2012; Bassett, Wymbs, et al., 2011) of
Such investigations will address whether abstract cogni- functional and anatomical network structure have shown
tive architectures can accurately represent the mecha- promise in predicting individual brain maturity (Dosenbach
nisms of mind and whether we can determine the et al., 2010), disease progression (Raj et al., 2012; Zhou
physical basis of such architectures. To motivate this pro- et al., 2012), and potential receptivity for neurorehabi-
cess, we now highlight major frontiers that will define the litation efforts (Bassett et al., 2013; Bassett, Wymbs, et al.,
emergence of cognitive network neuroscience. 2011). The nontrivial heterogeneities in region–region

Medaglia, Lynall, and Bassett 11

Figure 5. Network-based prediction (M = graph theory metric, S = subject, V = vector). Two complementary approaches to network-based
prediction. (A) Trial level prediction. Node time series are sampled from brain regions, and their functional connectivity is estimated. (B) Functional
connectivity (e.g., correlation) matrices are created for different trials (e.g., accurate vs. inaccurate trials). The matrices can be reorganized into
vectors ( V) representing connectivity values from different trials. (C) Pattern classifiers can be used to associate observed connectivity patterns across
subjects to specific performance values. Finally, the connectivity classification can be used to predict whether vectors from a new subject are
associated with performance of different types. (D) Graph level prediction of cognitive states. An elaborated approach can be taken at the level of
network metrics during cognitive processes. A graph of nodes and edges can be constructed from brain data. Then, graph theoretical metrics for each
node (e.g., clustering coefficients, betweenness centrality, node degree) can be calculated. (E) A pattern classifier can be trained on the graph theory
metric patterns observed during different cognitive tasks. (F) Summaries of the absolute and relative importance of nodes, measures, and nodes
within measures can be used within the pattern classification scheme. (G) Finally, graph theoretical pattern features can be used to predict cognitive
states. Note that, in principle, pairwise functional connectivity such as that used to predict trial types (A–C) can be used for cognitive state prediction
(D–G) and vice versa.

relationships require finely tuned models that are true to explicitly acknowledges the important aspects of this
underlying microscopic variables. Although most studies complexity is warranted for a comprehensive account of
have utilized data from only one imaging modality, multi- cognitive mechanisms (Turk-Browne, 2013; Bassett &
modal studies could arguably provide a more comprehen- Gazzaniga, 2011). As in traditional approaches to the brain
sive understanding of cognitive function and its alteration sciences, the analysis of dynamics on brain network archi-
in disease (Pandit et al., 2013; Sugranyes et al., 2012; tectures can occur on scales ranging from intracellular
Camchong, MacDonald, Bell, Mueller, & Lim, 2011; Steffens, mechanics to the macrolevel connectome. Understanding
Taylor, Denny, Bergman, & Wang, 2011; Pomarol-Clotet how cognition emerges within such a complex system
et al., 2010; Jeong, Wible, Hashimoto, & Kubicki, 2009). may necessarily involve the complementary use of simu-
lation and empirical techniques outside of neuroimaging
(e.g., intracranial recording, brain stimulation). As net-
Networks across Spatial and Temporal Scales
work neuroscience develops within levels, a focus should
Brain mechanisms are nonlinear and cross-multiple scales be placed on dynamics that emerge across different levels
of spatiotemporal and qualitative organization (Kopell of organization. For example, intersections between
et al., 2014). As a result, a default theoretical view that macroscale brain models such as the Virtual Brain (Ritter,

12 Journal of Cognitive Neuroscience Volume X, Number Y

Schirner, McIntosh, & Jirsa, 2013) and mesoscale models remain open in the extraction of anatomical, morpho-
such as for cortical columns (Chemla & Chavane, 2010) metric, and functional networks from neuroimaging data
and microscale models examining self-organized criti- (Bullmore & Bassett, 2011; Bassett & Bullmore, 2009;
cality (Plenz, Niebur, & Schuster, 2014) could form an Bullmore & Sporns, 2009; Bullmore et al., 2009).
important frontier. Without this strategy, some interlevel A division of the brain into regions is often termed a
dynamical properties of the brain relevant to cognitive parcellation, which can either be derived from known
processes may remain obscured. One additional challenge anatomical boundaries or from data-driven clustering of
is to consider the influences of rare events (Taleb, 2007) anatomical or state-induced functional connectivity. The
within and across levels of organization in nonlinear former a priori anatomical boundary parcellations have
brain systems. Thus, cognitive network neuroscience re- been defined across a range of spatial resolutions (Bassett,
searchers should seek and benefit from combined inno- Brown, et al., 2011), enable straightforward neurobiological
vative technologies and mathematical approaches in interpretation, and can be utilized similarly across each
network science. Across these efforts, the core challenge individual in a study, thereby simplifying group-based inter-
is to draw associations between multilevel brain dynamics pretations and group comparisons. The latter connectivity-
and cognitive states. It is possible that fundamental rules based parcellations have been derived for the whole
governing dynamical features of brain function can be cortex, individual lobes, and ROIs (see Zhang et al., 2014;
identified. These could be manipulated in experimental Power et al., 2011; Nelson et al., 2010, for examples).
designs to determine robust mappings from multilevel Connectivity-based parcellations provide a unique window
dynamics and cognition. into individual differences in brain structure and function
while somewhat complicating group comparisons. It is
important to note that all parcellation schemes can be
Networks Incorporating Subcortical Systems and
defined across a range of spatial resolutions, enabling the
the Cerebellum
exploration of multiresolution structure in the human
As mentioned previously, most approaches in cognitive brain (Bassett, Brown, et al., 2011; Bassett et al., 2010;
network neuroscience have undersampled or entirely Zalesky et al., 2010; Meunier et al., 2009). The choice of
omitted subcortical structures, especially the cerebellum. parcellation at any particular resolution affects the value
This is a crucial limitation of studies to date in light of of graph diagnostics but has not been found to impact
increasingly well-described contributions of subcortical on qualitative features of network organization (de Reus
regions to cognitive processes (Koziol & Budding, 2009; & van den Heuvel, 2013; Bassett, Brown, et al., 2011; Wang
Stoodley & Schmahmann, 2009) and dynamical signaling et al., 2009).
in brain systems (Buzsaki, 2011). Specifically, subcortical The definition of interactions between brain areas de-
systems should be given a fair chance at contributing to pends upon the type of network being constructed. Edges
estimated network structures. In conjunction with estab- in anatomical networks can represent white matter
lished atlases for cortical systems, investigators should streamline counts, mean fractional anisotropy, or indirect
consider including data for nodes sampled with atlases measurements of myelination (van den Heuvel, Mandl,
that validly represent subcortical systems. This can occur Stam, Kahn, & Hulshoff Pol, 2010; Hagmann et al.,
at a level of resolution such as substructures of the thala- 2008). Edges in functional networks can be defined based
mus (Behrens et al., 2003) or lobules of the cerebellum on statistical relationships between regional time series
(Diedrichsen, Balsters, Flavell, Cussans, & Ramnani, (Dawson, Cha, Lewis, Mendola, & Shmuel, 2013; Watanabe
2009). The software package FSL has useful documenta- et al., 2013; Gates & Molenaar, 2012; Smith et al., 2011; David,
tion on additional atlases that include subcortical regions. Cosmelli, & Friston, 2004), a common choice in fMRI
networks is the Pearson correlation (Zalesky et al., 2012),
although other measures of signal relatedness can be used,
METHODOLOGICAL AND
such as covariance, mutual information, and coherence.
INTERPRETATIONAL CONSIDERATIONS
The estimated strength of edges in both anatomical
It is necessary to identify and utilize optimal mathe- and functional networks is often complicated by noise
matical and computational techniques in each data modal- in the underlying measurements. Statistically less sig-
ity and cognitive domain of interest. In this section, we nificant edges are often removed from the subsequent
highlight three sets of key considerations of particular im- analysis to maximize power to detect true signals in con-
port in current research efforts: those pertaining to net- nectivity patterns. Binary networks (where edges are treated
work construction, statistical inference, and interpretation. as either present or absent) neglect potentially important
biological signatures present in their weighted network
counterparts (where edges maintain estimated strengths).
Network Construction
The complex interplay between mean edge weight, the
In the construction of any network, it is necessary to density of network connections, and the topology of the
identify system elements (network nodes) and inter- network itself (Langer, Pedroni, & Jäncke, 2013; Bassett,
actions (network edges; Butts, 2009). Such questions Nelson, et al., 2012; Ginestet, Nichols, Bullmore, &

Medaglia, Lynall, and Bassett 13

Simmons, 2011; van Wijk, Stam, & Daffertshofer, 2010) which appear both accurate and flexible in representing
has inspired a range of methodological developments to the topological characteristics of a brain network en-
accurately capture group differences in network organi- semble (Simpson et al., 2012; Simpson, Hayasaka, &
zations without resorting to arbitrary thresholds on the Laurienti, 2011). Representative organization decom-
edge weights themselves. These approaches include func- positions, such as the partition of network nodes into
tional data analysis (Bassett, Nelson, et al., 2012), iterative communities or modules, can be identified using the so-
cumulative and windowed thresholding (Bassett, Nelson, called consensus methods that identify reliable commu-
et al., 2012; Schwarz & McGonigle, 2011), and cost integ- nity structures. Consensus clustering methods can be
ration (Ginestet et al., 2011). utilized to uncover a modular decomposition that repre-
sents a group of subjects or time points (Lancichinetti &
Fortunato, 2012), and such methods can be made more
Statistics
statistically rigorous with the use of appropriate null models
Once networks have been constructed, a researcher must for comparison (Bassett, Porter, et al., 2012). Together,
determine whether its observed organization is simply these methods enable the identification of representative
consistent with a null model, that is, a control graph. and potentially more interpretable structure in sets of
The choice of a control graph depends on the goal. networks.
Often, we are not simply interested in whether a brain
network is different from a random network (one would
Generalized Utility in Understanding
hope so), but whether it is different from a network that
Cognitive Processes
shares some of the constraints of a brain but is not a
brain network. Network-based approaches outside of traditional neuro-
The development of null models for brain networks imaging data analysis could also have implications for
(Bassett, Porter, et al., 2012; Bassett & Gazzaniga, 2011) our understanding of cognitive function. For example,
and networks more generally (Rybarsch & Bornholdt, community structure is a network-based concept that is
2012; Milenkovic, Filippis, Lappe, & Przulj, 2009; Higham, intuitively comparable to the modular structures ob-
Rasajski, & Przulj, 2008; Kose, Budczies, Holschneider, & served in behavior, perception, and evolution and devel-
Fiehn, 2007; Thorne & Stumpf, 2007) is an area of ener- opment (Gerhart & Kirschner, 2007; Kirschner & Gerhart,
getic research. Null models fall into two general catego- 1998). One example is the chunking of motor movements
ries: models for static networks and models for temporal in humans, in which small sets of swift finger movements
networks. Common static null models include Erdos- are separated by pauses in a process similar to how we
Rényi networks with random connection probability remember three to five digits of a phone number sepa-
(Bollobás, 2001), Watts-Strogatz networks with small- rately from the next several digits. This temporal chunk-
world properties ( Watts & Strogatz, 1998), and fractal ing can be robustly identified (Bassett, Porter, et al., 2012)
networks with hierarchically modular structure (Sporns, in networks of behavior by applying dynamic community
2006). Dynamic null models include those that permute detection techniques to time-dependent similarities in
time, node identity, and connection patterns (Bassett intermovement durations and can be differentially linked
et al., 2013; Bassett, Porter, et al., 2012; Doron et al., to the recruitment of sensorimotor putamen and fronto-
2012; Bassett, Wymbs, et al., 2011), for example, using parietal cortex (Wymbs, Bassett, Mucha, Porter, & Grafton,
a rewiring rule that maintains the underlying degree dis- 2012). In a complementary application, the temporal com-
tribution (Maslov & Sneppen, 2002). Together, these null munity structure of task events was shown to give rise to
models are helpful for use in statistical inference; how- discrete neural representations in a manner similar to that
ever, the development of null models with more accurate thought to underpin the learning of semantic categories
neurobiological underpinnings will be critical. (Schapiro, Rogers, Cordova, Turk-Browne, & Botvinick,
Given the complexity of networks, a researcher is also 2013). Finally, the evolution of modular structure in synap-
likely to search for interpretative simplicity. One simplify- tic genes across a representative sampling of the animal
ing approach is to create a single network structure or kingdom can provide insight into the formation of complex
organizational decomposition from a larger set of data, cellular machines impacting on neural function (Conaco
whether that be, for example, a group of subjects in a et al., 2012). Together, these and related studies begin to
static network data set or a set of time points in a temporal demonstrate the utility of network-based methods to
network data set. Average network structures created by uncover meaningful structure in a wide range of data
taking the mean of individual networks within the group types with implications for cognitive neuroscience at large.
(Zuo et al., 2012; Song et al., 2009; Achard, Salvador,
Whitcher, Suckling, & Bullmore, 2006) often fail to ade-
Interpretational Caveats
quately reflect representative topological statistics of
the networks used to construct them (Simpson, Moussa, We close with a final word of caution in the interpretation
& Laurienti, 2012). A promising alternative is the use of of network diagnostics in neuroimaging data. Network
exponential random graph models (Newman, 2010), diagnostics can be intuitively interpreted in terms of

14 Journal of Cognitive Neuroscience Volume X, Number Y

information processing: High clustering can suggest that nitive neuroscience could lead to a fundamental shift in
information is processed in local domains, whereas short our understanding of the mind. We encourage increased
path length can suggest that information is being trans- communication between researchers in each of these
mitted over longer distances within the network. How- fields to capitalize on early progress and promote dis-
ever, the biological meaning of these interpretations covery at this new scientific intersection.
requires a conceptual leap from topological to biological
terms, which are semantically equivalent but not neces- Reprint requests should be sent to Danielle S. Bassett, Depart-
sarily conceptually interchangeable. Biological efficiency, ment of Bioengineering, University of Pennsylvania, Philadelphia,
for example, has evolutionary implications, which may PA 19104, or via e-mail: dsb@seas.upenn.edu.
not apply to topological or cognitive efficiency (Poldrack,
2014). More generally, such interpretations of network UNCITED REFERENCES
diagnostics require empirical validation demonstrating
the relationship between quantitative estimates of infor- Allen et al., 2014
mation processing or biological efficiency and network Calhoun, Miller, Pearlson, & Adali, 2014
characteristics. Until then, the cautious interpretation of Kramer et al., 2010
network diagnostics need not hamper the utility of these Kramer et al., 2011
approaches in prediction and classification in the study of Kramer, Kolaczyk, & Kirsch, 2008
system level dynamics underlying cognitive function. Sasson, Doniger, Pasternak, Tarrasch, & Assaf, 2013
In addition, the dynamic nature of brain network orga- Wen et al., 2011
nization begs for knowledge of underlying neurophysio- Ystad et al., 2011
logical mechanisms. In a few initial forays into this area,
links between the strength of functional connectivity and REFERENCES
properties of regional activity time series such as power
Achard, S., & Bullmore, E. (2007). Efficiency and cost of
and entropy have been identified across brain areas
economical brain functional networks. PLoS Computational
(Bassett, Nelson, et al., 2012), within brain areas (Zalesky Biology, 3, e17.
et al., 2012) and between subjects (Siebenhuhner et al., Achard, S., Salvador, R., Whitcher, B., Suckling, J., & Bullmore,
2013; Yu et al., 2013). In a similar vein, the magnitude of E. (2006). A resilient, low-frequency, small-world human
CBF has also been linked to the strength and topology of brain functional network with highly connected association
cortical hubs. Journal of Neuroscience, 26, 63–72.
functional connectivity in individual regions within a
Allen, E. A., Damaraju, E., Plis, S. M., Erhardt, E. B., Eichele, T.,
wider network (Liang, Zou, He, & Yang, 2013). Perhaps & Calhoun, V. D. (2014). Tracking whole-brain connectivity
even more interestingly, recent work has linked regional dynamics in the resting state. Cerebral Cortex, Epub ahead
patterns of aerobic glycolysis to intrinsic functional con- of print, 663–676.
nectivity (Vaishnavi et al., 2010). Neuroplasticity (Arnsten, Anderson, J. (1995). The architecture of cognition (2nd ed.).
Psychology Press.
Paspalas, Gamo, Yang, & Wang, 2010) and genetic con-
Arnsten, A. F., Paspalas, C. D., Gamo, N. J., Yang, Y., & Wang, M.
tributors to neuromorphometry, synapse formation, and (2010). Dynamic network connectivity: A new form of
neuromodulatory systems likely all impact on observed neuroplasticity. Trends in Cognitive Science, 14, 365–375.
network organization. Indeed, evidence supports the Bassett, D. S., Brown, J. A., Deshpande, V., Carlson, J. M., &
claim that network organization is both heritable (Fornito Grafton, S. T. (2011). Conserved and variable architecture of
human white matter connectivity. Neuroimage, 54,
et al., submitted) and modulated by individual alleles
1262–1279.
(Brown et al., 2011; Esslinger et al., 2009). Together, these Bassett, D. S., & Bullmore, E. T. (2009). Human brain networks
studies circumscribe an open space of fascinating ques- in health and disease. Current Opinion in Neurology, 22,
tions yet to be answered about the neurophysiological 340–347.
and genetic drivers of network organization. Bassett, D. S., Bullmore, E. T., Verchinski, B. A., Mattay, V. S.,
Weinberger, D. R., & Meyer-Lindenberg, A. (2008). Hierarchical
organization of human cortical networks in health and
CONCLUSION schizophrenia. Journal of Neuroscience, 28, 9239–9248.
Bassett, D. S., & Gazzaniga, M. S. (2011). Understanding
In this review, we have described how network-based complexity in the human brain. Trends in Cognitive Sciences,
approaches provide a powerful means to reveal neuro- 15, 200–209.
Bassett, D. S., Greenfield, D. L., Meyer-Lindenberg, A.,
physiological correlates of behavior and new organiza-
Weinberger, D. R., Moore, S., & Bullmore, E. (2010). Efficient
tional principles governing cognitive function. Cognitive physical embedding of topologically complex information
network neuroscience is in an early stage of development processing networks in brains and computer circuits. PLoS
and is only beginning to demonstrate its utility. A number Computational Biology, 6, e1000748.
of discoveries and validations have already occurred, and Bassett, D. S., Meyer-Lindenberg, A., Achard, S., Duke, T., &
Bullmore, E. (2006). Adaptive reconfiguration of fractal
governing rules linking dynamics in brain networks to
small-world human brain functional networks. Proceedings of
cognition may be clarified in the future. Under the banner the National Academy of Sciences, 103, 19518–19523.
of this emerging field, the integration of dynamic systems Bassett, D. S., Meyer-Lindenberg, A., Weinberger, D. R.,
analysis, mathematical network theory, and classical cog- Coppola, R., & Bullmore, E. (2009). Cognitive fitness of

Medaglia, Lynall, and Bassett 15

 cost-efficient brain functional networks. Proceedings of the in brain imaging data and other biological systems.
National Academy of Sciences, 106, 11747–11752. Neuroimage, 47, 1125–1134.
Bassett, D. S., Nelson, B. G., Mueller, B. A., Camchong, J., & Bullmore, E., & Sporns, O. (2009). Complex brain networks:
Lim, K. O. (2012). Altered resting state complexity in Graph theoretical analysis of structural and functional
schizophrenia. Neuroimage, 59, 2196–2207. systems. Nature Review Neuroscience, 10, 186–198.
Bassett, D. S., Porter, M. A., Wymbs, N. F., Grafton, S. T., Bullmore, E., & Sporns, O. (2012). The economy of brain
Carlson, J. M., & Mucha, P. J. (2012). Robust detection of network organization. Nature Reviews Neuroscience,
dynamic community structure in networks. Chaos, 23, 13, 336–349.
013142. Bullmore, E. T., & Bassett, D. S. (2011). Brain graphs: Graphical
Bassett, D. S., Wymbs, N. F., Porter, M. A., Mucha, P. J., Carlson, models of the human brain connectome. Annual Reviews of
J. M., & Grafton, S. T. (2011). Dynamic reconfiguration of Clinical Psychology, 7, 113–140.
human brain networks during learning. Proceedings of the Butts, C. T. (2009). Revisiting the foundations of network
National Academy of Sciences, 108, 7641–7646. analysis. Science, 325, 414–416.
Bassett, D. S., Wymbs, N. F., Rombach, M. P., Porter, M. A., Buzsaki, G. (2011). Rhythms of the brain (1st ed.). Oxford
Mucha, P. J., & Grafton, S. T. (2013). Core–periphery University Press.
organisation of human brain dynamics. PLoS Computational Caeyenberghs, K., Leemans, A., Heitger, M. H., Leunissen, I.,
Biology, 9, e1003171. Dhollander, T., Sunaert, S., et al. (2012). Graph analysis of
Bassett, D. S., Yang, M., Wymbs, N. F., & Grafton, S. T. (2014). functional brain networks for cognitive control of action in
Learning-induced autonomy of sensorimotor systems. traumatic brain injury. Brain, 135, 1293–1307.
Retrieved from arxiv.org/abs/1403.6034. Caeyenberghs, K., Leemans, A., Leunissen, I., Gooijers, J.,
Behrens, T. E., Johansen-Berg, H., Woolrich, M. W., Smith, Michiels, K., Sunaert, S., et al. (2014). Altered structural
S. M., Wheeler-Kingshott, C. A., Boulby, P. A., et al. (2003). networks and executive deficits in traumatic brain injury
Non-invasive mapping of connections between human patients. Brain Structure and Function, 219, 193–209.
thalamus and cortex using diffusion imaging. Nature Calhoun, V. D., Miller, R., Pearlson, G., & Adali, T. (2014).
Neuroscience, 6, 750–757. The chronnectome: Time-varying connectivity networks
Biswal, B., Yetkin, F. Z., Haughton, V. M., & Hyde, J. S. (1995). as the next frontier in fMRI data discovery. Neuron, 84,
Functional connectivity in the motor cortex of resting human 262–274.
brain using echo-planar MRI. Magnetic Resonance in Calhoun, V. D., Pekar, J. J., McGinty, V. B., Adali, T., Watson,
Medicine, 34, 537–541. T. D., & Pearlson, G. D. (2002). Different activation dynamics
Blank, I. A., Kanwisher, N., & Fedorenko, E. (2014). A functional in multiple neural systems during simulated driving. Human
dissociation between language and multiple-demand systems Brain Mapping, 16, 158–167.
revealed in patterns of bold signal fluctuations. Journal of Camchong, J., MacDonald, A. W. R., Bell, C., Mueller, B. A., &
Neurophysiology, 112, 1105–1118. Lim, K. O. (2011). Altered functional and anatomical
Bollobás, B. (2001). Random graphs. Cambridge, UK: connectivity in schizophrenia. Schizophrenia Bulletin, 37,
Cambridge University Press. 640–650.
Borgatti, S. P., & Everett, M. G. (1999). Models of core/ Cao, H., Plichta, M. M., Schafer, A., Haddad, L., Grimm, O.,
periphery structures. Social Networks, 21, 375–395. Schneider, M., et al. (2014). Test-retest reliability of fMRI-
Borgers, C. S. E., & Kopell, N. (2008). Gamma oscillations based graph theoretical properties during working memory,
mediate stimulus competition and attentional selection in a emotion processing, and resting state. Neuroimage, 84, 888.
cortical network model. Proceedings of the National Chemla, S., & Chavane, F. (2010). A biophysical cortical column
Academy of Sciences, 105, 18023–18028. model to study the multi-component origin of the vsdi signal.
Braun, U., Plichta, M. M., Esslinger, C., Sauer, C., Haddad, L., Neuroimage, 53, 420–438.
Grimm, O., et al. (2012). Test-retest reliability of resting-state Chen, G., Ward, B. D., Xie, C., Li, W., Wu, Z., Jones, J. L., et al.
connectivity network characteristics using fMRI and graph (2011). Classification of Alzheimer disease, mild cognitive
theoretical measures. Neuroimage, 59, 1404–1412. impairment, and normal cognitive status with large-scale
Braver, T. S. (2012). The variable nature of cognitive control: network analysis based on resting-state functional MR
A dual mechanisms framework. Trends in Cognitive Sciences, imaging. Radiology, 259, 213–221.
16, 106–113. Ching, S., Brown, E. N., & Kramer, M. A. (2012). Distributed
Bressler, S. L., & Richter, C. G. (2015). Interareal oscillatory control in a mesoscale cortical network model: Implications
synchronization in top–down neocortical processing. for seizure suppression. Physical Review E, 86.
Current Opinion in Neurobiology, 31, 62–66. Cisler, J. M., James, G. A., Tripathi, S., Mletzko, T., Heim, C., Hu,
Brodmann, K. (1909). Vegleichende lokalisationslehre der X. P., et al. (2012). Differential functional connectivity within
grosshirnde. Barth. an emotional regulation neural network among individuals
Brown, J. A., Terashima, K. H., Burggren, A. C., Ercoli, L. M., resilient and suceptible to the depressogenic effects of early
Miller, K. J., Small, G. W., et al. (2011). Brain network local life stress. Psychological Medicine, 43, 507.
interconnectivity loss in aging APOE-4 allele carriers. Cole, M. W., Bassett, D. S., Power, J. D., Braver, T. S., &
Proceedings of the National Academy of Sciences, 108, Petersen, S. E. (2014). Intrinsic and task-evoked network
20760–20765. architectures of the human brain. Neuron, 83, 238–251.
Buckner, R. L., & Krienen, F. M. (2013). The evolution of Cole, M. W., Laurent, P., & Stocco, A. (2013). Rapid instructed
distributed association networks in the human brain. task learning: A new window into the human brain’s unique
Trends in Cognitive Sciences, 17, 648–665. capacity for flexible cognitive control. Cognitve, Affective,
Buckner, R. L., Sepulcre, J., Talukdar, T., Krienen, F. M., Liu, H., and Behavioral Neuroscience, 13, 1–22.
Hedden, T., et al. (2009). Cortical hubs revealed by intrinsic Cole, M. W., Yarkoni, T., Repovs, G., Anticevic, A., & Braver,
functional connectivity: Mapping, assessment of stability, and T. S. (2012). Global connectivity of prefrontal cortex predicts
relation to Alzheimer’s disease. Journal of Neuroscience, cognitive control and intelligence. Journal of Neuroscience,
29, 1860–1873. 32, 8988–8999.
Bullmore, E., Barnes, A., Bassett, D. S., Fornito, A., Kitzbichler, Conaco, C., Bassett, D. S., Zhou, H., Arcila, M. L., Degnan, S. M.,
M., Meunier, D., et al. (2009). Generic aspects of complexity Degnan, B. M., et al. (2012). Functionalization of a

16 Journal of Cognitive Neuroscience Volume X, Number Y

 protosynaptic gene expression network. Proceedings of Fedorenko, E. (2014). The role of domain-general cognitive
the National Academy of Sciences, 109(Suppl. 1), control in language comprehension. Frontiers in Psychology,
10612–10618. 5, 1–17.
Corbetta, M., & Shulman, G. L. (2002). Control of goal-directed Fedorenko, E., & Thompson-Schill, S. L. (2014). Reworking
and stimulus-driven attention in the brain. Nature Reviews the language network. Trends in Cognitive Science, 18,
Neuroscience, 3, 201–215. 120–126.
Craddock, R. C., Jbabdi, S., Yan, C.-G., Vogelstein, J. T., Fekete, T., Wilf, M., Rubin, D., Edelman, S., Malach, R., &
Castellanos, F. X., Di Martino, A., et al. (2013). Imaging Mujica-Parodi, L. R. (2013). Combining classification with
human connectomes at the macro scale. Nature Methods, fMRI-derived complex network measures for potential
10, 524–539. neurodiagnostics. PLoS One, 8, e62867.
Crossley, N. A., Mechelli, A., Scott, J., Carletti, F., Fox, P. T., Fornito, A., Zalesky, A., Bassett, D., Meunier, D., Yucel, M.,
McGuire, P., et al. (2014). The hubs of the human Wood, S. J., et al. (submitted). Genetic influences on
connectome are generally implicated in the anatomy of economical properties of human functional cortical networks.
brain disorders. Brain, 137, 2382–2395. Fornito, A., Zalesky, A., Pantelis, C., & Bullmore, E. T. (2012).
David, O., Cosmelli, D., & Friston, K. J. (2004). Evaluation of Schizophrenia, neuroimaging and connectomics.
different measures of functional connectivity using a neural Neuroimage, 62, 2296–2314.
mass model. Neuroimage, 21, 659–673. Fortunato, S. (2010). Community detection in graphs. Physics
Dawson, D. A., Cha, K., Lewis, L. B., Mendola, J. D., & Shmuel, Reports, 486, 75–174.
A. (2013). Evaluation and calibration of functional network Friston, K. J. (1994). Functional and effective connectivity
modeling methods based on known anatomical connections. in neuroimaging: A synthesis. Human Brain Mapping, 2,
Neuroimage, 67, 331–343. 56–78.
de Reus, M. A., & van den Heuvel, M. P. (2013). The Gates, K. M., & Molenaar, P. C. (2012). Group search algorithm
parcellation-based connectome: Limitations and extensions. recovers effective connectivity maps for individuals in
Neuroimage, 80, 397–404. homogeneous and heterogeneous samples. Neuroimage,
Deco, G., Jirsa, V. K., & McIntosh, A. R. (2011). Emerging 63, 310–319.
concepts for the dynamical organization of resting-state Gerhart, J., & Kirschner, M. (2007). The theory of facilitated
activity in the brain. Nature Reviews Neuroscience, 12, variation. Proceedings of the National Academy of Sciences,
43–56. 104(Suppl. 1), 8582–8589.
Dennis, E. L., Jahanshad, N., Toga, A. W., McMahon, K. L., Giessing, C., Thiel, C. M., Alexander-Bloch, A. F., Patel, A. X., &
de Zubicaray, G. I., Martin, N. G., et al. (2012). Test-retest Bullmore, E. T. (2013). Human brain functional network
reliability of graph theory measures of structural brain changes associated with enhanced and impaired attentional
connectivity. Medical Image Computing and Computer- task performance. Journal of Neuroscience, 33, 5903–5914.
Assisted Intervention, 15, 305–312. Ginestet, C. E., Nichols, T. E., Bullmore, E. T., & Simmons, A.
Deuker, L., Bullmore, E. T., Smith, M., Christensen, S., Nathan, (2011). Brain network analysis: Separating cost from topology
P. J., Rockstroh, B., et al. (2009). Reproducibility of graph using cost-integration. PLoS One, 6, e21570.
metrics of human brain functional networks. Neuroimage, Ginestet, C. E., & Simmons, A. (2011). Statistical parametric
47, 1460–1468. network analysis of functional connectivity dynamics during a
Diedrichsen, J., Balsters, J. H., Flavell, J., Cussans, E., & working memory task. Neuroimage, 55.
Ramnani, N. (2009). A probabalistic mr atlas of the human Grimm, S., Beck, J., Schuepbach, D., Hell, D., Boesiger, P.,
cerebellum. Neuroimage, 46, 39–46. Bermpohl, F., et al. (2008). Imbalance between left and right
Doron, K. W., Bassett, D. S., & Gazzaniga, M. S. (2012). Dynamic dorsolateral prefrontal cortex in major depression is linked to
network structure of interhemispheric coordination. negative emotional judgment: An fMRI study in severe major
Proceedings of the National Academy of Sciences, 109, depressive disorder. Biological Psychiatry, 63, 369–376.
18661–18668. Hagmann, P., Cammoun, L., Gigandet, X., Meuli, R., Honey,
Dosenbach, N. U., Fair, D. A., Miezin, F. M., Cohen, A. L., C. J., Wedeen, V. J., et al. (2008). Mapping the structural
Wenger, K. K., Dosenbach, R. A., et al. (2007). Distinct brain core of human cerebral cortex. PLoS Biology, 6, e159.
networks for adaptive and stable task control in humans. Hampson, M., Tokoglu, F., Sun, Z., Schafer, R. J., Skudlarski, P.,
Proceedings of the National Academy of Sciences, 104, Gore, J. C., et al. (2006). Connectivity-behavior analysis
11073–11078. reveals that functional connectivity between left BA39 and
Dosenbach, N. U., Nardos, B., Cohen, A. L., Fair, D. A., Power, Broca’s area varies with reading ability. Neuroimage, 31,
J. D., Church, J. A., et al. (2010). Prediction of individual 513–519.
brain maturity using fMRI. Science, 329, 1358–1361. He, H., Sui, J., Yu, Q., Turner, J. A., Ho, B. C., Sponheim, S. R.,
Eickhoff, S. B., Paus, T., Caspers, S., Grosbras, M. H., Evans, A., et al. (2012). Altered small-world brain networks in
Zilles, K., et al. (2007). Assignment of functional activations to schizophrenia patients during working memory performance.
probabalistic cytoarchitectonic areas revisited. Neuroimage, PLoS One, 7, e38195.
36, 511–521. He, Y., Chen, Z., Gong, G., & Evans, A. (2009). Neuronal
Ekman, M., Derrfuss, J., Tittgemeyer, M., & Fiebach, C. J. networks in Alzheimer’s disease. Neuroscientist, 15, 333–350.
(2012). Predicting errors from reconfiguration patterns in He, Y., Chen, Z. J., & Evans, A. C. (2007). Small-world
human brain networks. Proceedings of the National anatomical networks in the human brain revealed by cortical
Academy of Sciences, 109, 16714–16719. thickness from MRI. Cerebral Cortex, 17, 2407–2419.
Esslinger, C., Walter, H., Kirsch, P., Erk, S., Schnell, K., Arnold, Heinzle, J., Wenzel, M. A., & Haynes, J. D. (2012). Visuomotor
C., et al. (2009). Neural mechanisms of a genome-wide functional network topology predicts upcoming tasks.
supported psychosis variant. Science, 324, 605. Journal of Neuroscience, 32, 9960–9968.
Farbota, K. D., Bendlin, B. B., Alexander, A. L., Rowley, H. A., Heitger, M. H., Ronsse, R., Dhollander, T., Dupont, P.,
Dempsey, R. J., & Johnson, S. C. (2012). Longitudinal Caeyenberghs, K., & Swinnen, S. P. (2012). Motor learning-
diffusion tensor imaging and neuropsychological correlates induced changes in functional brain connectivity as revealed
in traumatic brain injury patients. Frontiers in Human by means of graph-theoretical network analysis. Neuroimage,
Neuroscience, 6. 61, 633.

Medaglia, Lynall, and Bassett 17

Herculano-Houzel, S. (2010). Coordinated scaling of cortical and Langley, P., Laird, J. E., & Rogers, S. (2009). Cognitive
cerebellar numbers of neurons. Frontiers in Neuroanatomy, architectures: Research issues and challenges. Cognitive
4, 1–8. Systems Research, 10, 141–160.
Hermundstad, A. M., Bassett, D. S., Brown, K. S., Aminoff, E. M., Latora, V., & Marchiori, M. (2001). Efficient behavior of small-
Clewett, D., Freeman, S., et al. (2013). Structural foundations world networks. Physical Review Letters, 87, 198701.
of resting-state and task-based functional connectivity in Latora, V., & Marchiori, M. (2003). Economic small-world
the human brain. Proceedings of the National Academy of behavior in weighted networks. The European Physics
Sciences, 110, 6169–6174. Journal B, 32, 249–263.
Higham, D. J., Rasajski, M., & Przulj, N. (2008). Fitting a Leunissen, I., Coxon, J. P., Geurts, M., Caeyenberghs, K.,
geometric graph to a protein-protein interaction network. Michiels, K., Sunaert, S., et al. (2013). Disturbed cortico-
Bioinformatics, 24, 1093–1099. subcortical interactions during motor task switching in
Holme, P., & Saramäki, J. (2012). Temporal networks. Physics traumatic brain injury. Human Brain Mapping, 34,
Reports, 519, 97–125. 1254–1271.
Illari, P. M., & Williamson, J. (2012). What is a mechanism? Liang, X., Zou, Q., He, Y., & Yang, Y. (2013). Coupling of
Thinking about mechanisms across the sciences. European functional connectivity and regional cerebral blood flow
Journal for Philosophy of Science, 2, 119–135. reveals a physiological basis for network hubs of the human
Jeong, B., Wible, C. G., Hashimoto, R., & Kubicki, M. (2009). brain. Proceedings of the National Academy of Sciences, 110,
Functional and anatomical connectivity abnormalities in left 1929–1934.
inferior frontal gyrus in schizophrenia. Human Brain Ma, S., Calhoun, V. D., Eichele, T., Du, W., & Adali, T. (2012).
Mapping, 30, 4138–4151. Modulations of functional connectivity in the healthy and
Kinnison, J., Padmala, S., Choi, J. M., & Pessoa, L. (2012). schizophrenia groups during task and rest. Neuroimage,
Network analysis reveals increased integration during 62, 1694–1704.
emotional and motivational processing. Journal of Madden, D. J., Bennett, I. J., & Song, A. W. (2009). Cerebral
Neuroscience, 32, 8361–8372. white matter integrity and cognitive aging: Contributions
Kinnunen, K. M., Greenwood, R., Powell, J. H., Leech, R., from diffusion tensor imaging. Neuropsychology Review,
Hawkins, P. C., Bonnelle, V., et al. (2011). White matter 19, 415–435.
damage and cognitive impairment after traumatic brain Mantzaris, A. V., Bassett, D. S., Wymbs, N. F., Estrada, E., Porter,
injury. Brain, 134, 440–463. M. A., Mucha, P. J., et al. (2013). Dynamic network centrality
Kirschner, M., & Gerhart, J. (1998). Evolvability. Proceedings of summarizes learning in the human brain. Journal of Complex
the National Academy of Sciences, 95, 8420–8427. Networks, 1, 83–92.
Kitzbichler, M. G., Henson, R. N., Smith, M. L., Nathan, P. J., & Maslov, S., & Sneppen, K. (2002). Specificity and stability in
Bullmore, E. T. (2011). Cognitive effort drives workspace topology of protein networks. Science, 296, 910–913.
configuration of human brain functional networks. Journal of McHaffie, J. G., Stanford, T. R., Stein, B. E., Coizet, V., &
Neuroscience, 31, 8259–8270. Redgrave, P. (2005). Subcortical loops through the basal
Klados, M. A., Kanatsouli, K., Antoniou, I., Babiloni, F., ganglia. Trends in Neurosciences, 28, 401–407.
Tsirka, V., Bamidis, P. D., et al. (2013). A graph theoretical McKeown, M. J., Makeig, S., Brown, G. G., Jung, T. P.,
approach to study the organization of the cortical Kindermann, S. S., Bell, A. J., et al. (1998). Analysis of
networks during different mathematical tasks. PLOS One, fMRI data by blind separation into independent spatial
8, e71800. components. Human Brain Mapping, 6, 160–188.
Koenigs, M., & Grafman, J. (2009). The functional Meunier, D., Fonlupt, P., Saive, A.-L., Plailly, J., Ravel, N., &
neuroanatomy of depression: Distinct roles for ventromedial Royet, J.-P. (2014). Modular structure of functional networks
and dorsolateral prefrontal cortex. Behavioural Brain in olfactory memory. Neuroimage, 95, 264–275.
Research, 201, 239–243. Meunier, D., Lambiotte, R., Fornito, A., Ersche, K. D., &
Kopell, N. J., Gritton, H. J., Whittington, M. A., & Kramer, M. A. Bullmore, E. T. (2009). Hierarchical modularity in human
(2014). Beyond the connectome: The dynome. Neuron, brain functional networks. Frontiers in Neuroinformatics,
83, 1319–1320. 3, 37.
Kose, F., Budczies, J., Holschneider, M., & Fiehn, O. (2007). Michael, A. M., Calhoun, V. D., Andreasen, N. C., & Baum, S. A.
Robust detection and verification of linear relationships to (2008). A method to classify schizophrenia using inter-task
generate metabolic networks using estimates of technical spatial correlations of functional brain images. Conference
errors. BMC Bioinformatics, 8. Proceedings—IEEE Engineering in Medicine and Biology
Koziol, L. F., & Budding, D. E. (2009). Subcortical structures Society, 5510–5513.
and cognition (1st ed.). Springer. Milenkovic, T., Filippis, I., Lappe, M., & Przulj, N. (2009).
Kramer, M. A., Eden, U. T., Kolaczyk, E. D., Zepeda, R., Optimized null model for protein structure networks. PLoS
Eskandar, E. N., & Cash, S. S. (2010). Coalescence and One, 4, e5967.
fragmentation of cortical networks during focal seizures. Miller, E. K., & Cohen, J. D. (2001). An integrative theory of
Journal of Neuroscience, 30, 10076–10085. prefrontal cortex function. Annual Review of Neuroscience,
Kramer, M. A., Eden, U. T., Lepage, K. Q., Kolaczyk, E. D., 24, 167–202.
Bianchi, M. T., & Cash, S. S. (2011). Emergence of persistent Mishkin, M., & Ungerleider, L. G. (2014). Contribution of
networks in long-term intracranial EEG recordings. Journal striate inputs to the visuospatial functions of parieto-
of Neuroscience, 31, 15757–15767. preoccipital cortex in monkeys. Journal of Gerontology
Kramer, M. A., Kolaczyk, E. D., & Kirsch, H. E. (2008). Emergent Series A, 6, 57–77.
network topology at seizure onset in humans. Epilepsy Miyake, A., Friedman, N. P., Emerson, M. J., Witzki, A. H., &
Research, 79, 173–186. Howerter, A. (2000). The unity and diversity of executive
Lancichinetti, A., & Fortunato, S. (2012). Consensus clustering functions and their contributions to complex “frontal lobe”
in complex networks. Scientific Reports, 2. tasks: A latent variable analysis. Cognitive Psychology, 41,
Langer, N., Pedroni, A., & Jäncke, L. (2013). The problem of 49–100.
thresholding in small-world network analysis. PLoS One, 8, Mokhtari, F., & Hossein-Zadeh, G. A. (2013). Decoding brain
e53199. states using backward edge elimination and graph kernels in

18 Journal of Cognitive Neuroscience Volume X, Number Y

 fMRI connectivity networks. Journal of Neuroscience Porter, M. A., Onnela, J.-P., & Mucha, P. J. (2009). Communities
Methods, 212, 259–268. in networks. North American Mathematical Society, 56,
Moretti, L., Cristofori, I., Weaver, S. M., Chau, A., Portelli, J. N., 1082–1097, 1164–1166.
& Grafman, J. (2012). Cognitive decline in older adults with a Power, J. D., Cohen, A. L., Nelson, S. M., Wig, G. S., Barnes,
history of traumatic brain injury. Lancet Neurology, 11, K. A., Church, J. A., et al. (2011). Functional network
1103–1112. organization of the human brain. Neuron, 72, 665–678.
Moussa, M. N., Vechlekar, C. D., Burdette, J. H., Steen, M. R., Power, J. D., & Petersen, S. E. (2013). Control-related systems in
Hugenschmidt, C. E., & Laurienti, P. J. (2011). Changes in the human brain. Current Opinion in Neurobiology, 23,
cognitive state alter human functional brain networks. 223–228.
Frontiers in Human Neuroscience, 5, 83. Qin, J., Wei, M., Liu, H., Yan, R., Luo, G., Yao, Z., et al. (2013).
Mucha, P. J., Richardson, T., Macon, K., Porter, M. A., & Onnela, Large scale brain functional networks support sentence
J.-P. (2010). Community structure in time-dependent, comprehension: Evidence from both explicit and implicit
multiscale, and multiplex networks. Science, 328, 876–878. language tasks. PLOS One, 8, e80214.
Muller, A. M., & Meyer, M. (2014). Language in the brain at rest: Qin, J., Wei, M., Liu, H., Yan, R., Luo, G., Yao, Z., et al. (2014).
New insights from resting state data and graph theoretical Abnormal brain anatomical topological organization of the
analysis. Frontiers in Human Neuroscience, 8, 228. cognitive-emotional and the frontoparietal circuitry in major
Nagano-Saito, A., Martinu, K., & Monchi, O. (2014). Function depressive disorder. Magnetic Resonance in Medicine, 72,
of basal ganglia in bridging cognitive and motor modules 1397–1407.
to perform an action. Frontiers in Human Neuroscience, Raichle, M. E. (2011). The restless brain. Brain Connectivity, 1,
8, 187. 3–12.
Nelson, S. M., Cohen, A. L., Power, J. D., Wig, G. S., Miezin, Raj, A., Kuceyeski, A., & Weiner, M. (2012). A network diffusion
F. M., Wheeler, M. E., et al. (2010). A parcellation scheme for model of disease progression in dementia. Neuron, 73,
human left lateral parietal cortex. Neuron, 67, 156–170. 1204–1215.
Newman, M. E. J. (2010). Networks: An introduction. Oxford Reijmer, Y. D., Leemans, M. K. L. J., Brundel, A., & Biessels, G. J.
University Press. (2013). Disruption of the cerebral white matter network is
Newman, M. E. J., & Girvan, M. (2004). Finding and evaluating related to slowing of information processing speed in
community structure in networks. Physical Review E, 69, patients with type 2 diabetes. Diabetes, 62, 2112–2115.
026113. Richiardi, J., Achard, S., Bunke, H., & Van De Ville, D. (2013).
Niogi, S. N., Mukherjee, P., Ghajar, J., Johnson, C., Kolster, R. A., Machine learning with brain graphs: Predictive modeling
Sarkar, R., et al. (2008). Extent of microstructural white approaches for functional imaging in systems neuroscience.
matter injury in postconcussive syndrome correlates with Signal Processing Magazine, IEEE, 30, 58–70.
impaired cognitive reaction time: A 3T diffusion tensor Richiardi, J., Eryilmaz, H., Schwartz, S., Vuilleumier, P., &
imaging study of mild traumatic brain injury. American Van De Ville, D. (2011). Decoding brain states from fMRI
Journal of Neuroradiology, 29, 967–973. connectivity graphs. Neuroimage, 56, 616–626.
Owen, J. P., Ziv, E., Bukshpun, P., Pojman, N., Wakahiro, M., Richiardi, J., Gschwind, M., Simioni, S., Annoni, J. M., Greco, B.,
Berman, J. I., et al. (2013). Test-retest reliability of Hagmann, P., et al. (2012). Classifying minimally disabled
computational network measurements derived from the multiple sclerosis patients from resting state functional
structural connectome of the human brain. Brain connectivity. Neuroimage, 62, 2021–2033.
Connectivity, 3, 160–176. Ritter, P., Schirner, M., McIntosh, A. R., & Jirsa, V. K. (2013).
Palacios, E. M., Sala-Llonch, R., Junque, C., Roig, T., Tormos, The virtual brain integrates computational modeling and
J. M., Bargallo, N., et al. (2012). White matter integrity related multimodal neuroimaging. Brain Connectivity, 3, 121–145.
to functional working memory networks in traumatic brain Roberts, R. E., Anderson, E. J., & Husain, M. (2013). White
injury. Neurology, 78, 852–860. matter microstructure and cognitive function. Neuroscientist,
Palva, J. M., Monto, S., Kulashekhar, S., & Palva, S. (2010). 19, 8–15.
Neural synchrony reveals working memory networks and Rombach, M. P., Porter, M. A., Fowler, J. H., & Mucha, P. J.
predicts individual memory capacity. Proceedings of the (2012). Core–periphery structure in networks. arXiv,
National Academy of Sciences, 107, 7580–7585. 1202.2684.
Palva, S., Monto, S., & Palva, J. M. (2010). Graph properties of Ronan, L., Voets, N. L., Hough, M., Mackay, C., Roberts, N.,
synchronized cortical networks during visual working Suckling, J., et al. (2012). Consistency and interpretation of
memory maintenance. Neuroimage, 49, 3257–3268. changes in millimeter-scale cortical intrinsic curvature across
Pandit, A. S., Expert, P., Lambiotte, R., Bonnelle, V., Leech, R., three independent datasets in schizophrenia. Neuroimage,
Turkheimer, F. E., et al. (2013). Traumatic brain injury 63, 611–621.
impairs small-world topology. Neurology, 80, 1826–1833. Rybarsch, M., & Bornholdt, S. (2012). Binary threshold
Parks, E. L., & Madden, D. J. (2013). Brain connectivity and networks as a natural null model for biological networks.
visual attention. Brain Connectivity, 3, 317–338. Physical Review E, 86, 026114.
Passingham, R. E., Rowe, J. B., & Sakai, K. (2013). Has brain Sanabria-Diaz, G., Melie-García, L., Iturria-Medina, Y., Alemán-
imaging discovered anything new about how the brain Gómez, Y., Hernández-González, G., Valdés-Urrutia, L., et al.
works? Neuroimage, 66, 142–150. (2010). Surface area and cortical thickness descriptors reveal
Plenz, D., Niebur, E., & Schuster, H. G. (2014). Criticality in different attributes of the structural human brain networks.
neural systems. Wiley. Neuroimage, 50, 1497–1510.
Poldrack, R. A. (2014). Is “efficiency” a useful concept in Sasson, E., Doniger, G. M., Pasternak, O., Tarrasch, R., & Assaf,
cognitive neuroscience? Developmental Cognitive Y. (2013). White matter correlates of cognitive domains in
Neuroscience, 11, 12–17. normal aging with diffusion tensor imaging. Frontiers in
Pomarol-Clotet, E., Canales-Rodriguez, E. J., Salvador, R., Sarro, Neuroscience, 7.
S., Gomar, J. J., & Vila, F. (2010). Medial prefrontal cortex Schapiro, A. C., Rogers, T. T., Cordova, N. I., Turk-Browne,
pathology in schizophrenia as revealed by convergent N. B., & Botvinick, M. M. (2013). Neural representations of
findings from multimodal imaging. Molecular Psychiatry, events arise from temporal community structure. Nature
15, 823–830. Neuroscience, 16, 486–492.

Medaglia, Lynall, and Bassett 19

Schiff, S. (2012). Neural control engineering. Cambridge, MA: Sporns, O. (2014). Contributions and challenges for network
MIT Press. models in cognitive neuroscience. Nature Neuroscience, 17,
Schwarz, A. J., & McGonigle, J. (2011). Negative edges and 652–660.
soft thresholding in complex network analysis of resting Sporns, O., Tononi, G., & Kötter, R. (2005). The human
state functional connectivity data. Neuroimage, 55, connectome: A structural description of the human brain.
1132–1146. PLoS Computational Biology, 1, e42.
Seeley, W. W., Menon, V., Schatzberg, A., Keller, J., & Glover, Steffens, D. C., Taylor, W. D., Denny, K. L., Bergman, S. R., &
G. H. (2007). Dissociable intrinsic connectivity networks for Wang, L. (2011). Structural integrity of the uncinate fasciculus
salience processing and executive control. Journal of and resting state functional connectivity of the ventral
Neuroscience, 27, 2349–2356. prefrontal cortex in late life depression. PLoS One, 6, e22697.
Sepulcre, J. (2012). Stepwise connectivity of the modal cortex Stevens, A. A., Tappon, S. C., Garg, A., & Fair, D. A. (2012).
reveals the multimodal organization of the human brain. Functional brain network modularity captures inter- and
Journal of Neuroscience, 32, 10649–10661. intra-individual variation in working memory capacity.
Sepulcre, J. (2014). Integration of visual and motor functional PLOS One, 7, e30468.
streams in the human brain. Neuroscience Letters, 567, Stoodley, C. J., & Schmahmann, J. D. (2009). Functional
68–73. topography in the human cerebellum: A meta-analysis of
Shamay-Tsoory, S. G., Aharon-Peretz, J., & Perry, D. (2008). neuroimaging studies. Neuroimage, 44, 489–501.
Two systems for empathy: A double dissociation between Sugranyes, G., Kyriakopoulos, M., Dima, D., O’Muircheartaigh,
emotional and cognitive empathy in inferior frontal gyrus J., Corrigall, R., Pendelbury, G., et al. (2012). Multimodal
versus ventromedial prefrontal lesions. Brain, 201, 617–627. analyses identify linked functional and white matter
Shen, H., Wang, L., Liu, Y., & Hu, D. (2010). Discriminative abnormalities within the working memory network in
analysis of resting-state functional connectivity patterns of schizophrenia. Schizophrenia Research, 138, 136–142.
schizophrenia using low dimensional embedding of fMRI. Supekar, K., Menon, V., Rubin, D., Musen, M., & Greicius, M. D.
Neuroimage, 49, 3110–3121. (2008). Network analysis of intrinsic functional brain
Shenoy, K. V., Kaufman, M. T., Sahani, M., & Churchland, M. M. connectivity in Alzheimer’s disease. PLoS Computational
(2011). A dynamical systems view of motor preparation: Biology, 4, e1000100.
Implications for neural prosthetic system design. Progress in Szczepanski, S. M., & Knight, R. T. (2014). Insights into human
Brain Research, 192, 33–58. behavior from lesions to the prefrontal cortex. Neuron, 83,
Shirer, W. R., Ryali, S., Rykhlevskaia, E., Menon, V., & Greicius, 1–17.
M. D. (2012). Decoding subject-driven cognitive states with Taleb, N. (2007). Black swans and the domains of statistics.
whole-brain connectivity patterns. Cerebral Cortex, 22, The American Statistician, 61, 198–200.
158–165. Telesford, Q. K., Morgan, A. R., Hayasaka, S., Simpson, S. L.,
Siebenhuhner, F., Weiss, S. A., Coppola, R., Weinberger, D. R., Barret, W., Kraft, R. A., et al. (2010). Reproducibility of graph
& Bassett, D. S. (2013). Intra- and inter-frequency brain metrics in fMRI networks. Frontiers in Neuroinformatics, 4.
network structure in health and schizophrenia. PLoS One, Thompson-Schill, S. L., Swick, D., Farah, M. J., D’Esposito, M.,
8, e72351. Kan, I. P., & Knight, R. T. (1998). Verb generation in patients
Silver, J. M., McAllister, T. W., & Arciniegas, D. B. (2009). with focal frontal lesions: A neuropsychological test of
Depression and cognitive complaints following mild neuroimaging findings. Proceedings of the National
traumatic brain injury. American Journal of Psychiatry, Academy of Sciences, 95, 15855–15860.
166, 653–661. Thorne, T., & Stumpf, M. P. (2007). Generating confidence
Simpson, S. L., Hayasaka, S., & Laurienti, P. J. (2011). intervals on biological networks. BMC Bioinformatics, 8.
Exponential random graph modeling for complex brain Turk-Browne, N. B. (2013). Functional interactions as big data
networks. PLoS One, 6, e20039. in the human brain. Science, 342, 580–584.
Simpson, S. L., Moussa, M. N., & Laurienti, P. J. (2012). An Vaishnavi, S. N., Vlassenko, A. G., Rundle, M. M., Snyder, A. Z.,
exponential random graph modeling approach to creating Mintun, M. A., & Raichle, M. E. (2010). Regional aerobic
group-based representative whole-brain connectivity glycolysis in the human brain. Proceedings of the National
networks. Neuroimage, 60, 1117–1126. Academy of Sciences, 107, 17757–17762.
Smith, S. M., Fox, P. T., Miller, K. L., Glahn, D. C., Fox, Valente, T. W., Coronges, K., Lakon, C., & Costenbader, E.
P. M., Mackay, C. E., et al. (2009). Correspondence of the (2008). How correlated are network centrality measures?
brain’s functional architecture during activation and rest. Connect (Tor), 28, 16–26.
Proceedings of the National Academy of Sciences, 106, van den Heuvel, M. P., Mandl, R. C., Stam, C. J., Kahn, R. S., &
13040–13045. Hulshoff Pol, H. E. (2010). Aberrant frontal and temporal
Smith, S. M., Miller, K. L., Salimi-Khorshidi, G., Webster, M., complex network structure in schizophrenia: A graph
Beckmann, C. F., Nichols, T. E., et al. (2011). Network theoretical analysis. Journal of Neuroscience, 30,
modelling methods for fMRI. Neuroimage, 54, 875–891. 15915–15926.
Song, S., Liu, Y., Zhou, Y., Wang, K., Yu, C., & Jiang, T. (2009). van den Heuvel, M. P., & Sporns, O. (2011). Rich-club
Conference Proceedings—IEEE Engineering in Medicine organization of the human connectome. Journal of
and Biology Society, 2009, 2212–2215. Neuroscience, 31, 15775–15786.
Sporns, O. (2006). Small-world connectivity, motif composition, van den Heuvel, M. P., & Sporns, O. (2013). Network hubs
and complexity of fractal neuronal connections. Biosystems, in the human brain. Trends in Cognitive Sciences, 17,
85, 55–64. 683–696.
Sporns, O. (2010). Networks of the brain. Cambridge, MA: van Wijk, B. C., Stam, C. J., & Daffertshofer, A. (2010).
MIT Press. Comparing brain networks of different size and connectivity
Sporns, O. (2011). The human connectome: A complex density using graph theory. PLoS One, 5, e13701.
network. Annals of the New York Academy of Sciences, 1224, Wang, J., Wang, L., Zang, Y., Yang, H., Tang, H., Gong, Q., et al.
109–125. (2009). Parcellation-dependent small-world brain functional
Sporns, O. (2012). From simple graphs to the connectome: networks: A resting-state fMRI study. Human Brain
Networks in neuroimaging. Neuroimage, 62, 881–886. Mapping, 30, 1511–1523.

20 Journal of Cognitive Neuroscience Volume X, Number Y

Wang, J. H., Zuo, X. N., Gohel, S., Milham, M. P., Biswal, B. B., & sensorimotor putamen and frontoparietal cortex during
He, Y. (2011). Graph theoretical analysis of functional brain motor chunking in humans. Neuron, 74, 936–946.
networks: Test-retest evaluation on short- and long-term Xu, L., Zhang, H., Hui, M., Long, Z., Jin, Z., Liu, Y., et al. (2014).
resting-state functional MRI data. PLoS One, 6, e21976. Dynamic reconfiguration of human brain betwork during
Wang, L., Li, Y., Metzak, P., He, Y., & Woodward, T. S. (2010). learning. Neuroscience, 261, 7641–7646.
Age-related changes in topological patterns of large-scale Ystad, M., Hodneland, E., Adolfsdottir, S., Haasz, J., Lundervold,
brain functional networks during memory encoding and A. J., Eichele, T., et al. (2011). Cortico-striatal connectivity and
recognition. Neuroimage, 50, 862–872. cognition in normal aging: A combined dti and resting state
Warren, D. E., Power, J. D., Bruss, J., Denburg, N. L., Waldron, fMRI study. Neuroimage, 55, 24–31.
E. J., Sun, H., et al. (2014). Network measures predict Yu, Q., Sui, J., Liu, J., Plis, S. M., Kiehl, K. A., Pearlson, G., et al.
neuropsychological outcome after brain injury. Proceedings (2013). Disrupted correlation between low frequency power
of the National Academy of Sciences, 111, 14247–14252. and connectivity strength of resting state brain networks in
Watanabe, T., Hirose, S., Wada, H., Imai, Y., Machida, T., schizophrenia. Schizophrenia Research, 143, 165–171.
Shirouzu, I., et al. (2013). A pairwise maximum entropy Zalesky, A., Fornito, A., & Bullmore, E. (2012). On the use of
model accurately describes resting-state human brain correlation as a measure of network connectivity.
networks. Nature Communications, 4, 1370. Neuroimage, 60, 2096–2106.
Watts, D. J., & Strogatz, S. H. (1998). Collective dynamics of Zalesky, A., Fornito, A., Harding, I. H., Cocchi, L., Yucel, M.,
‘small-world’ networks. Nature, 393, 440–442. Pantelis, C., et al. (2010). Whole-brain anatomical networks:
Wen, W., Zhu, W., He, Y., Kochan, N. A., Reppermund, S., Does the choice of nodes matter? Neuroimage, 50,
Slavin, M. J., et al. (2011). Discrete neuroanatomical networks 970–983.
are associated with specific cognitive abilities in old age. Zhang, Y., Fan, L., Zhang, Y., Wang, J., Zhu, M., Zhang, Y., et al.
Journal of Neuroscience, 31, 1204–1212. (2014). Connectivity-based parcellation of the human
Woodward, J. (2011). Scientific explanation. In E. N. Zalta (Ed.), posteromedial cortex. Cerebral Cortex, 24, 719–727.
The Stanford encyclopedia of philosophy. Zhou, J., Gennatas, E. D., Kramer, J. H., Miller, B. L., & Seeley,
Wu, T. C., Wilde, E. A., Bigler, E. D., Yallampalli, R., McCauley, W. W. (2012). Predicting regional neurodegeneration from
S. R., Troyanskaya, M., et al. (2010). Evaluating the the healthy brain functional connectome. Neuron, 73,
relationship between memory functioning and cingulum 1216–1227.
bundles in acute mild traumatic brain injury using diffusion Zuo, X. N., Ehmke, R., Mennes, M., Imperati, D., Castellanos,
tensor imaging. Journal of Neurotrauma, 27, 303–307. F. X., Sporns, O., et al. (2012). Network centrality in the
Wymbs, N. F., Bassett, D. S., Mucha, P. J., Porter, M. A., & human functional connectome. Cerebral Cortex, 22,
Grafton, S. T. (2012). Differential recruitment of the 1862–1875.

Medaglia, Lynall, and Bassett 21