Situating the default-mode network along a principal

gradient of macroscale cortical organization

Daniel S. Marguliesa,1, Satrajit S. Ghoshb,c, Alexandros Goulasd, Marcel Falkiewicza, Julia M. Huntenburga,e,
Georg Langsf,g, Gleb Bezginh, Simon B. Eickhoffi,j, F. Xavier Castellanosk,l, Michael Petridesm, Elizabeth Jefferiesn,o,
and Jonathan Smallwoodn,o
a
Max Planck Research Group for Neuroanatomy & Connectivity, Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig 04103, Germany;
b
McGovern Institute for Brain Research, Massachusetts Institute of Technology, Cambridge, MA 02139; cDepartment of Otolaryngology, Harvard Medical
School, Cambridge, MA 02115; dDepartment of Computational Neuroscience, University Medical Center Hamburg-Eppendorf, Hamburg 20246, Germany;
e
Neurocomputation and Neuroimaging Unit, Department of Education and Psychology, Free University of Berlin, Berlin 14195, Germany; fDepartment of
Biomedical Imaging and Image-Guided Therapy, Computational Imaging Research Laboratory, Medical University of Vienna, Vienna A-1090, Austria;
g
Computer Science and Artificial Intelligence Laboratory, Massachusetts Institute of Technology, Cambridge, MA 02139; hMcConnell Brain Imaging Centre,
Montreal Neurological Institute, McGill University, Montreal, QC, Canada H3A 2B4; iInstitute for Neuroscience and Medicine, Research Center Jülich, Juelich
52428, Germany; jInstitute of Clinical Neuroscience and Medical Psychology, Heinrich Heine University, Duesseldorf 40225, Germany; kChild Study Center,
Department of Child and Adolescent Psychiatry, New York University Langone Medical Center, New York, NY 10016; lNathan Kline Institute for Psychiatric
Research, Orangeburg, NY 10962; mCognitive Neuroscience Unit, Montreal Neurological Institute, McGill University, Montreal, QC, Canada H3A 2B4; nDepartment
of Psychology, University of York, York YO10 5DD, United Kingdom; and oYork Neuroimaging Centre, University of York, York YO10 5DD, United Kingdom

Edited by Peter L. Strick, University of Pittsburgh, Pittsburgh, PA, and approved September 9, 2016 (received for review May 27, 2016)

Understanding how the structure of cognition arises from the in the middle temporal cortex and the most anterior temporal
topographical organization of the cortex is a primary goal in cortex that capture the meaning of what we see, hear, and do (12–
neuroscience. Previous work has described local functional gradi- 15). Similarly, in the prefrontal cortex, a rostral–caudal gradient has

NEUROSCIENCE
ents extending from perceptual and motor regions to cortical been proposed, whereby goals become increasingly abstract in an-
areas representing more abstract functions, but an overarching terior areas more distant from motor cortex, because they are
framework for the association between structure and function is increasingly removed from selection processes that operate on
still lacking. Here, we show that the principal gradient revealed by specific motor representations (5, 16–19). Much like the function–
the decomposition of connectivity data in humans and the macaque structure correspondence elucidated by topographic maps within
monkey is anchored by, at one end, regions serving primary sensory/ sensory and motor areas (20, 21), these processing gradients provide
motor functions and at the other end, transmodal regions that, in a systematic mapping between spatial position and a functional
humans, are known as the default-mode network (DMN). These spectrum of increasingly abstract representations (22).
DMN regions exhibit the greatest geodesic distance along the cortical Processing gradients have proven useful for understanding the
surface—and are precisely equidistant—from primary sensory/motor relation between specific regions and function in separate domains:
morphological landmarks. The principal gradient also provides an Mesulam (23) observed that the emergence of more abstract
organizing spatial framework for multiple large-scale networks and functional classes of cortex may follow a similar trajectory, hy-
characterizes a spectrum from unimodal to heteromodal activity in a pothesizing that abstract categories emerge from the convergence
functional metaanalysis. Together, these observations provide a of information across modalities (Fig. 1C). This notion has recently
characterization of the topographical organization of cortex and in- been extended by Buckner and Krienen (24), who proposed the
dicate that the role of the DMN in cognition might arise from its
“tethering hypothesis,” arguing that association cortex gains its
position at one extreme of a hierarchy, allowing it to process trans-
functional attributes through its increasing spatial distance from
modal information that is unrelated to immediate sensory input.
the constraints that determine the functional specialization of
topography | connectivity | cortical organization | default-mode network | Significance
gradients

We describe an overarching organization of large-scale connec-

A key assumption in neuroscience is that the topographical
structure of the cerebral cortex provides an organizing
principle that constrains its cognitive processes. Recent advances
tivity that situates the default-mode network at the opposite end
of a spectrum from primary sensory and motor regions. This to-
pography, based on the differentiation of connectivity patterns,
in the field of human connectomics have revealed multiple large-
is also embedded in the spatial distance along the cortical surface
scale networks (1–3), each characterized by distinct functional
between these respective systems. In addition, this connectivity
profiles (4). Some are related to basic primary functions, such as
gradient accounts for the respective positions of canonical net-
movement or perceiving sounds and images; some serve well-
works and captures a functional spectrum from perception and
documented, domain-general functions, such as attention or
action to more abstract cognitive functions. These results suggest
cognitive control (5–8); and some have functional characteristics
that the default-mode network consists of regions at the top of a
that remain less well-understood, such as the default-mode network
representational hierarchy that describe the current cognitive
(DMN) (9, 10). Although the topography of these distinct distrib-
landscape in the most abstract terms.
uted networks has been described using multiple methods (1–3), the
reason for their particular spatial relationship and how this con- Author contributions: D.S.M., M.P., E.J., and J.S. designed research; D.S.M. performed research;
strains their function remain unclear. D.S.M., S.S.G., M.F., J.M.H., G.L., G.B., and S.B.E. contributed new reagents/analytic tools; D.S.M.
Advances in mapping local processing streams have revealed analyzed data; and D.S.M., S.S.G., A.G., M.F., J.M.H., G.L., G.B., S.B.E., F.X.C., M.P., E.J., and J.S.
spatial gradients that support increasingly abstract levels of repre- wrote the paper.

sentation, often extending along adjacent cortical regions in a The authors declare no conflict of interest.
stepwise manner (11). In the visual domain, for example, the ventral This article is a PNAS Direct Submission.
occipitotemporal object stream transforms simple visual features, Freely available online through the PNAS open access option.
coded by neurons in primary visual cortex, into more complex visual 1
To whom correspondence should be addressed. Email: [email protected].
descriptions of objects in anterior inferior temporal cortical regions This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
and ultimately, contributes to multimodal semantic representations 1073/pnas.1608282113/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1608282113 PNAS Early Edition | 1 of 6

 A B

C D E

Fig. 1. The principal gradient of connectivity in both the (A) human and (B) macaque monkey cortices shows a spectrum between unimodal regions (dark blue) and
transmodal regions (sienna), which in the human cortex, peaks in regions corresponding to the DMN. The proximity of colors can be interpreted as greater similarity of
connectivity patterns. (C) The illustration of connectivity organization suggested by Mesulam (23) proposes a hierarchy of processing from distinct unimodal areas to
integrative transmodal areas. Labels Gradient 1 and Gradient 2, which were not included in the original figure, correspond to the results in D. Modified from ref. 23. (D) A
scatter plot of the first two connectivity embedding gradients. Gradient 1 extends between primary sensorimotor and transmodal regions (red). Gradient 2 separates
somatomotor and auditory cortex (green) from visual cortex (blue). Histograms depicting the distribution of values are presented on the respective axes. (E) Colors from the
scatter plot are presented on the cortical surface for anatomical orientation. A1, primary auditory; ag, angular gyrus; cing, anterior cingulate cortex; ifg, inferior frontal
gyrus; infs, intermediate frontal sulcus; L, limbic; M1, primary motor; mfg, middle frontal gyrus; mtc, middle temporal cortex; P, parietal; Pf, prefrontal; phf, para-
hippocampal formation; pmc, posteromedial cortex; ps, principal sulcus; S1, primary somatosensory; sfg, superior frontal gyrus; V1, primary visual; vmpfc, ventromedial
prefrontal cortex.

primary cortex. These viewpoints suggest that there may be structure of the cortex? (iii) Does the organization captured by the
macroscale gradients that integrate information across multiple principal gradient account for the spatial distribution of large-scale
domains into progressively more abstract representations, in which networks and the associated functions across the cortex? (iv) Do
local gradients within specific cortical systems could be situated these observations provide a framework for understanding the
and understood. functional role of the DMN in cognition?
One large-scale cortical system with function that remains
unclear is the DMN. Initially identified through its tendency to Results
deactivate during externally oriented tasks (25), the DMN has We began our analysis by characterizing the components describ-
since been shown to activate in tasks that depend on informa- ing the maximum variance in functional connectivity patterns—
tion retrieved from memory, such as remembering the past or
the extent to which nodes agree in the spatial distribution of cor-
thinking about the future, or considering the mental states of
relations—across the human cerebral cortex (Fig. 1 and Fig. S1).
others (reviews are in refs. 10 and 26). The DMN is also known
to play a role in states that are less related to ongoing environ- The functional connectivity matrix consisted of 91,282 cortical
mental events, such as daydreaming and mind wandering (27–30), and subcortical “grayordinates” with a resolution of 2 mm from the
and contributes to lapses in external processing (31). A consensus preprocessed dense connectome S900 release of the Human
view on the role of the DMN in human cognition is still lacking, Connectome Project (HCP) (39). These data were based on 1 h of
however, because of the increasing number of cognitive domains in resting-state fMRI data acquired from 820 healthy adult individ-
which it has been implicated. As well as playing an active role during uals. No further processing of the connectivity matrices beyond
states, such as autobiographical memory retrieval, social cognition, those already implemented by the HCP, which included minimal
and future thinking, the DMN has recently been shown to operate spatial smoothing of 2 mm FWHM (40), was conducted.
in concert with regions implicated in cognitive control during Rather than delineating discrete network parcellations, we
complex working memory tasks (32–36). This emerging evidence implemented a method that captures gradients in connectivity
illustrates that the DMN is not tied to a specific form of in- patterns over space—a cortical feature termed “connectopies”
formational content, leading to suggestions that it acts as a hub that (41). This method, known as diffusion embedding (42), allows
integrates representational information across the cortex (30, 37).
local and long distance connections to be projected into a
To understand the topographic organization of the cerebral
cortex at the macroscale (38), we explore how the principal vari- common space more effectively than approaches that use linear
ance in cortical connectivity relates to the topography of structure dimensionality reduction, such as principal component analysis
and function by addressing four key questions. (i) Is there a mac- (SI Materials and Methods). The resultant components, which we
roscale gradient of connectivity in the human brain that reflects the describe here as “gradients,” are unitless and identify the posi-
systematic integration across modalities in a hierarchical fashion? tion of nodes along the respective embedding axis that encodes
(ii) Does this macroscale organization relate to the geometric the dominant differences in nodes’ connectivity patterns.

2 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1608282113 Margulies et al.

The Principal Gradient in Humans and Macaque Monkeys. The prin- Additional gradients describing progressively less connectivity vari-
cipal gradient (Fig. 1A), which accounts for the greatest variance ance are available in Fig. S1.
in connectivity in the human brain (Fig. S2), is anchored at one
end by the primary and unimodal visual, somatosensory/motor, DMN Peaks of the Principal Gradient Are Equidistant from Primary
and auditory regions. At the other end are regions including the Areas. Having characterized the topography of a principal gra-
angular gyrus, rostral anterior cingulate, posteromedial cortex, dient in connectivity, we next investigated whether it is related to
middle temporal gyrus, and middle and superior frontal gyri— the intrinsic geometry of the cortex. To do so, we examined
regions that, in humans, are collectively described as the DMN. whether regions at the extreme of the DMN end occupy spatial
Regions situated between the two extreme ends of the principal locations that are maximally distant along the cortical surface
gradient include the inferior frontal sulcus, the intraparietal from unimodal regions. We selected seven peak cortical nodes
sulcus, and the inferior temporal sulcus, constituting hetero- across the DMN clusters of the principal gradient and calculated
modal integration and higher-order cognitive regions. the minimum geodesic distance from all other nodes to any of
The initial proposal of Mesulam (23) was motivated by tract- these “seed” nodes (additional description of methods is in SI
tracing studies conducted in the macaque monkey. To determine Materials and Methods).
whether our method would generalize to these forms of data, we Fig. 2 shows that cortical distance reproduces many features of
performed the same embedding analysis on a publicly available the spatial embedding of the principal gradient. Four of the peak
database of tract-tracing studies conducted in the macaque monkey. DMN nodes are equidistant from the central sulcus, which is the
The principal gradient of the macaque monkey cerebral cortex is topographical landmark of primary somatosensory/motor cortex.
presented in Fig. 1B and similar to the human functional connec- Likewise, we observe a similar correspondence with the calcarine
tivity-based results, anchored at one end by visual and somatosen- sulcus, marking the location of primary visual cortex. More gener-
sory/motor regions and at the other end by higher-order transmodal ally, distance clearly increases with lower principal gradient values,
regions in the temporal lobe and the medial and lateral prefrontal with an especially rapid transition in the connectivity gradient be-
cortexes. The cross-species correspondence of the principal gradient tween 25 and 40 mm and plateaus at the extremes (Fig. 2B). This
relationship is, nevertheless, captured by a linear fit (R2 = 0.55). It is

NEUROSCIENCE
suggests that this axis of connectivity variation is phylogenetically
noteworthy that unimodal regions are at least 40 mm from the
conserved and may represent a primary dimension of cortical ex-
DMN peaks. In similar analyses of macaque monkey cortical dis-
pansion (43).
tance (Fig. S3), we observed a comparable distance threshold for
The topography of the principal gradient in both the human and
unimodal regions. In sum, this analysis shows that the principal
macaque monkey is consistent with the claim that cortical con-
connectivity gradient reflects macrostructural features of cortical
nectivity is organized along a dimension spanning primary/unimodal
organization: the nodes corresponding to one extreme end of the
and transmodal regions—a hypothesis that is summarized sche-
gradient—core regions of the DMN—are maximally distant from
matically along the Gradient 1 dimension in Fig. 1C. However, for
regions that directly govern perception and action.
this spectrum to indicate hierarchical integration across distinct
modalities, the following connectivity component should distinguish The Principal Gradient Captures the Spatial Layout of Large-Scale
between primary modalities as indicated by the dimension Gradient Networks. We next examined the extent to which the principal
2 in Fig. 1C. gradient captures the macroscale layout of intrinsic functional
Consistent with the hypothesis by Mesulam (23) (Fig. 1C), the connectivity networks. Despite the high reproducibility of large-
component accounting for the second-most variance in connectivity scale resting-state networks (1, 44–46), there is no clear over-
in the human brain differentiates regions solely within the unimodal arching spatial schema to explain the transition of one network
end of the principal gradient (Fig. 1D). One end of the spectrum is to another. We examined the widely used seven-network par-
characterized by regions of the occipital cortex implicated in pro- cellation by Yeo et al. (2) with respect to the position of each
cessing visual input, whereas the opposite end includes the so- network along the principal gradient (Fig. 3A). [Results using the
matosensory and motor regions surrounding the central sulcus as 17-network parcellation from ref. 2 are presented in Fig. S4.]
well as the auditory regions of the temporal perisylvian region (Fig. Fig. 3 shows that networks are not randomly distributed along
1E). The convergence described by the first two connectivity gra- this dimension: instead, as shown in the box plots in Fig. 3B,
dients across sensory/motor modalities and toward a singular set of cortical nodes from the same network tend to cluster at similar
nodes within transmodal cortex is consistent with the claim that the positions. Importantly, the DMN identified in this parcellation
principal gradient is organized along a dimension that integrates (Fig. 3, red) occupies one extreme position along the principal
unimodal regions in a hierarchical manner (Fig. 1C). Moreover, the gradient and is maximally separated from visual (Fig. 3, purple)
principal gradient, anchored at one end by the DMN, contains and motor (Fig. 3, blue) networks, which are at the other extreme.
within it several local processing gradients that have already been One exception is the limbic network (Fig. 3, beige), which includes
described within the temporal and frontal lobes (12–15, 17–19). an extensive range of values. However, the spatial distribution of

A B

Fig. 2. (A) The minimum geodesic distance (in millimeters) from each point on the cortical surface to seven seed nodes located in the positive peaks of the principal
gradient. Morphological landmarks of primary areas denoted by white dotted lines, such as the central sulcus (cs; somatosensory/motor), calcarine sulcus (cals; visual),
and transverse temporal gyrus (tt; auditory), are equidistant from the surrounding DMN peaks (illustrated by arrows). Gray lines mark the calculated equidistant line.
(B) The contour scatter plot shows the negative relationship between geodesic distance from seven positive peak locations and the principal gradient (R2 = 0.55).

Margulies et al. PNAS Early Edition | 3 of 6

A B data from studies in the macaque monkey found a corresponding
gradient, providing initial evidence that this axis of connectivity
variation may be phylogenetically conserved. The observation that
the principal gradient corresponds to the intrinsic geometry of the
cortex—regions in the DMN have the greatest geodesic distance
along the cortical surface from primary sensory/motor areas—
C D further indicates this axis may provide a crucial blueprint for cor-
tical organization (Fig. 2). We also found that large-scale networks
are arranged along this axis, with the same transitions between
consistently adjacent networks occurring throughout the cortex
(Fig. 3). Finally, a task-based metaanalysis characterizing the func-
tional attributes of this gradient showed a spectrum of increasing
abstraction that follows the transition from unimodal cortex to the
extreme end of the gradient in the DMN (Fig. 4).
The location of the DMN at one extreme end of the principal
gradient provides an organizing principle for understanding its role
in cognition. First, these findings provide anatomical support for
Fig. 3. (A) The principal gradient values from each of seven networks (2) are why the DMN has been associated with processes that are un-
presented as (B) box plots ordered by the mean value. (C) Illustrative cutouts related to immediate stimulus input, such as daydreaming or mind
taken from A to show the repeated patterns of network spatial adjacency wandering (27, 28, 30). The DMN is at a maximal distance from
captured by the principal gradient. Arrows in A indicate the corresponding systems involved in perception and action in both functional con-
orientation of the cutouts. (D) A schematic of the spatial relationships of ca- nectivity and anatomical space, indicating that the neural activity in
nonical resting-state networks (2) applying the schema suggested in ref. 23 these regions is likely to be comparably insulated from direct en-
presented in Fig. 1C. dmn, default-mode network; dorsal attn, dorsal attention
vironmental input (49, 50). Second, the location of the DMN as
network; sal, salience network; somato/mot, somatosensory/motor network.
equidistant from all sensory/motor systems is aligned with its broad
range of functions that requires integration between multiple
this network may be accounted for by low signal to noise within sensory systems, including episodic (51) and semantic memory (52–
the original data used for parcellation (2), and it may, thus, not 54), social cognition (55, 56), goal-directed working memory tasks
accurately reflect the connectivity of its constituent regions. (26, 32, 33, 35), and reward-guided decision making (57, 58). The
This analysis, therefore, shows that the principal gradient of two cardinal features of the DMN related to abstraction––stimulus
connectivity provides a framework for the spatial ordering of independence and content heterogeneity––can be accounted for by
large-scale networks. In addition, the principal gradient captures its position at the end of a topographical hierarchy that is equi-
similar, repeating transitions between these networks, which occur distant from unimodal systems, thus acting as a hub of integration
across cortical lobes (Fig. 3C). We represent this consistent ar- across multiple sensory modalities (37) (Fig. 3D).
rangement as a schematic illustration in Fig. 3D. Notably, outlier The principal gradient illustrates a broader topographic or-
gradient values for each network are located predominantly at ganization of large-scale connectivity (38) that accounts for the
their boundaries (Fig. S5), suggesting that, in some cases, the
principal gradient describes gradual connectivity transitions that
are obscured by discrete network parcellation.

Distribution of Functions Along the Principal Gradient. Our final

analysis explored whether the regions located at the DMN ex-
treme of the gradient serve functions that are abstracted from
perception and action. We conducted a metaanalysis using the
NeuroSynth database (47) [Figs. S6 and S7 show corresponding
analysis using the BrainMap database (48)] and examined the
association between a list of topic terms with regions of interest
created from five-percentile bins of the principal gradient. Topic
terms were sorted by their weighted average position along the
gradient, revealing a systematic shift in function. Fig. 4 shows
that the unimodal end is characterized by terms depicting acting
and perceiving, such as “motor,” “visual perception,” “multi-
sensory processing,” and “auditory processing,” whereas the end
characterized by the DMN emphasizes terms such as “social
cognition,” “verbal semantics,” and “autobiographical memory”—
tasks that rely on complex representations abstracted away from
specific sensory and motor processes. Between the extremes, we
observe domain-general functions, such as “cued attention,” “in-
hibition,” and “working memory,” in regions corresponding to the
dorsal attention and salience networks above (Fig. 3D).
Discussion Fig. 4. NeuroSynth metaanalysis of regions of interest along the principal
gradient using 24 topic terms. Terms are ordered by the weighted mean of their
Our analysis characterized a principal gradient of cortical orga-
location along the gradient. Sensory processing terms are located at the top
nization in the human connectome, which is anchored at one end followed by domain-general cognitive functions and then, higher-order abstract
by systems implicated in perceiving and acting, and at the other cognitive and memory-related processes. Similar results using the BrainMap
end by transmodal association regions, corresponding in humans database are available in SI Materials and Methods. autobiographical mem.,
to the DMN (Fig. 1). A comparative analysis using tract-tracing autobiographical memory; multisensory proc., multisensory processing.

4 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1608282113 Margulies et al.

spatial arrangement of local processing streams throughout the Materials and Methods
cerebral cortex. Gradients in both the temporal and prefrontal The principal gradient was derived from human (39) and macaque (67, 68)
cortexes are apparent in Fig. 1, showing that these hierarchies connectivity matrices using diffusion embedding (42)—a nonlinear di-
are not isolated local phenomena; they emerge as elements of a mensionality reduction technique (Fig. 1). Geodesic distance along the cor-
spectrum that begins within input–output systems and ends with tical surface from peak nodes of the transmodal end of the principal
the DMN. Notably, our results are consistent with a recent gradient, presented in Fig. 2, was calculated using an exact distance algo-
modification of the rostral–caudal processing gradient described rithm (69, 70). For the comparison with canonical large-scale networks (Fig.
3), the principal gradient values were extracted from each of seven networks
within lateral frontal cortex (59, 60). Rather than the more
from ref. 2. Finally, binarized masks at five-percentile increments of the
rostral areas located farther along in the processing hierarchy principal gradient were used as regions of interest in a NeuroSynth (47)
(18, 19), two distinct hierarchical gradients of temporal- and metaanalysis (Fig. 4). Additional information regarding methods is available
feature-related abstraction converge in middle lateral prefrontal in SI Materials and Methods as well as Figs. S1–S7. All software used in this
cortex (60). The consistency between the principal gradient and study is openly available at https://neuroanatomyandconnectivity.github.io/
this revised lateral prefrontal hierarchy suggests that it may pro- gradient_analysis/.
vide a source for future studies investigating the detailed topog- All MRI data used in this study were publicly available and anonymized.
raphy of local processing streams. Participant recruitment procedures and informed consent forms, including
In addition to incorporating local processing streams within a consent to share deidentified data, were previously approved by the
global framework, the principal gradient situates discrete large- Washington University Institutional Review Board as part of the HCP (39).
scale connectivity networks along a continuous spectrum. With
ACKNOWLEDGMENTS. Data were provided by the HCP, and the Washington
recent advances in multimodal cortical parcellation (61), this University, University of Minnesota, and Oxford University Consortium (Princi-
approach provides a complementary means to describe the ge- pal Investigators David Van Essen and Kamil Ugurbil; Grant 1U54MH091657)
stalt of the cortical mosaic. Future studies are needed to better funded by 16 NIH Institutes and Centers that support the NIH Blueprint
characterize the types of transitions between different patterns of for Neuroscience Research, and the McDonnell Center for Systems Neuro-
science at Washington University. S.S.G. was partially supported by NIH
large-scale connectivity and identify where processing occurs in a Grants 1R01EB020740-01A1, 1P41EB019936-01A1, 3R01MH092380-04S2, and
stepwise (11) or “gradiential” manner (22).

NEUROSCIENCE
1U01MH108168-01. M.P. was supported by Canadian Institutes of Health Research
It is now widely accepted that the DMN is important, because it Foundation Grant FDN-143212. E.J. was supported by Biotechnology and Biolog-
permits cognitive processing that is independent of the here and ical Sciences Research Council (BB/J006963/1) and European Research Council
Grant 283530-SEMBIND. J.S. was supported by European Research Council Grant
now. This capacity is adaptive, because it permits flexibility: more WANDERINGMINDS–646927 and a grant from the John Templeton Foundation
abstract representations of a stimulus enable the generation of “Prospective Psychology Stage 2: A Research Competition” (to Martin Seligman).
alternative behaviors, allowing original and creative thoughts to G.L. is supported by NIH National Institute of Biomedical Imaging and Bioen-
emerge (62). Along those lines, a “positive–negative” axis of brain– gineering Neuroimaging Analysis Center P41EB015902; NIH National Institute
of Neurological Disorders and Stroke R01NS086905; and Austrian Science Fund
behavior covariation describes a similar connectivity spectrum, [Fonds zur Förderung wissenschaftlicher Forschung (FWF)] I2714-B31, KLI 544-
distinguishing the DMN from sensory/motor regions (63). Beyond B27, and Jubiläumsfonds der Österreichischen Nationalbank 15356. S.B.E. is
supporting states of creativity and planning (64, 65), the DMN has supported by the National Institute of Mental Health (R01-MH074457), the
also been implicated in almost all psychiatric conditions (66), in- Helmholtz Portfolio Theme “Supercomputing and Modeling for the Human
Brain,” and the European Union Seventh Framework Programme (FP7/2007-
dicating that there may be costs as well as benefits from the ca- 2013) under Grant Agreement 604102. The opinions expressed in this publica-
pacity to apprehend the world as it might be rather than seeing it tion are those of the authors and do not necessarily reflect the views of the
as it is right now. John Templeton Foundation.

1. Damoiseaux JS, et al. (2006) Consistent resting-state networks across healthy subjects. 17. Petrides M (2005) From Monkey Brain to Human Brain. A Fyssen Foundation
Proc Natl Acad Sci USA 103(37):13848–13853. Symposium, eds Dehaene S, Duhamel J-R, Hauser MD, Rizzolatti G (MIT Press, Cam-
2. Yeo BTT, et al. (2011) The organization of the human cerebral cortex estimated by bridge, MA), pp 293–314.
intrinsic functional connectivity. J Neurophysiol 106(3):1125–1165. 18. Badre D (2008) Cognitive control, hierarchy, and the rostro-caudal organization of the
3. Power JD, et al. (2011) Functional network organization of the human brain. Neuron frontal lobes. Trends Cogn Sci 12(5):193–200.
72(4):665–678. 19. Badre D, D’Esposito M (2009) Is the rostro-caudal axis of the frontal lobe hierarchical?
4. Smith SM, et al. (2009) Correspondence of the brain’s functional architecture during Nat Rev Neurosci 10(9):659–669.
activation and rest. Proc Natl Acad Sci USA 106(31):13040–13045. 20. Kaas JH (1997) Topographic maps are fundamental to sensory processing. Brain Res
5. Petrides M (2005) Lateral prefrontal cortex: Architectonic and functional organiza- Bull 44(2):107–112.
tion. Philos Trans R Soc Lond B Biol Sci 360(1456):781–795. 21. Kaas JH (1987) The organization of neocortex in mammals: Implications for theories
6. Duncan J (2010) The multiple-demand (md) system of the primate brain: Mental of brain function. Annu Rev Psychol 38:129–151.
programs for intelligent behaviour. Trends Cogn Sci 14(4):172–179. 22. Goldberg E (1989) Gradiental approach to neocortical functional organization. J Clin
7. Cole MW, Yarkoni T, Repovs G, Anticevic A, Braver TS (2012) Global connectivity of pre- Exp Neuropsychol 11(4):489–517.
frontal cortex predicts cognitive control and intelligence. J Neurosci 32(26):8988–8999. 23. Mesulam MM (1998) From sensation to cognition. Brain 121(Pt 6):1013–1052.
8. Petrides M (2015) Lateral and dorsomedial prefrontal cortex and the control of 24. Buckner RL, Krienen FM (2013) The evolution of distributed association networks in
cognition. Brain Mapping: An Encyclopedic Reference, Volume 2: Anatomy and the human brain. Trends Cogn Sci 17(12):648–665.
Physiology, Systems, ed Toga AW (Academic, London), pp 417–422. 25. Shulman GL, et al. (1997) Common blood flow changes across visual tasks. II. De-
9. Greicius MD, Krasnow B, Reiss AL, Menon V (2003) Functional connectivity in the creases in cerebral cortex. J Cogn Neurosci 9(5):648–663.
resting brain: A network analysis of the default mode hypothesis. Proc Natl Acad Sci 26. Spreng RN, Grady CL (2010) Patterns of brain activity supporting autobiographical
USA 100(1):253–258. memory, prospection, and theory of mind, and their relationship to the default mode
10. Raichle ME (2015) The brain’s default mode network. Annu Rev Neurosci 38:433–447. network. J Cogn Neurosci 22(6):1112–1123.
11. Sepulcre J, Sabuncu MR, Yeo TB, Liu H, Johnson KA (2012) Stepwise connectivity of 27. Mason MF, et al. (2007) Wandering minds: The default network and stimulus-independent
the modal cortex reveals the multimodal organization of the human brain. J Neurosci thought. Science 315(5810):393–395.
32(31):10649–10661. 28. Christoff K, Gordon AM, Smallwood J, Smith R, Schooler JW (2009) Experience sam-
12. Mishkin M, Ungerleider LG (1982) Contribution of striate inputs to the visuospatial pling during fMRI reveals default network and executive system contributions to
functions of parieto-preoccipital cortex in monkeys. Behav Brain Res 6(1):57–77. mind wandering. Proc Natl Acad Sci USA 106(21):8719–8724.
13. Goodale MA, Milner AD (1992) Separate visual pathways for perception and action. 29. Stawarczyk D, Majerus S, Maquet P, D’Argembeau A (2011) Neural correlates of
Trends Neurosci 15(1):20–25. ongoing conscious experience: Both task-unrelatedness and stimulus-independence
14. Patterson K, Nestor PJ, Rogers TT (2007) Where do you know what you know? the rep- are related to default network activity. PLoS One 6(2):e16997.
resentation of semantic knowledge in the human brain. Nat Rev Neurosci 8(12):976–987. 30. Smallwood J, et al. (2016) Representing representation: Integration between the
15. Visser M, Jefferies E, Embleton KV, Lambon Ralph MA (2012) Both the middle tem- temporal lobe and the posterior cingulate influences the content and form of
poral gyrus and the ventral anterior temporal area are crucial for multimodal se- spontaneous thought. PLoS One 11(4):e0152272.
mantic processing: Distortion-corrected fMRI evidence for a double gradient of 31. Weissman DH, Roberts KC, Visscher KM, Woldorff MG (2006) The neural bases of
information convergence in the temporal lobes. J Cogn Neurosci 24(8):1766–1778. momentary lapses in attention. Nat Neurosci 9(7):971–978.
16. Koechlin E, Ody C, Kouneiher F (2003) The architecture of cognitive control in the 32. Vatansever D, Menon DK, Manktelow AE, Sahakian BJ, Stamatakis EA (2015) Default
human prefrontal cortex. Science 302(5648):1181–1185. mode dynamics for global functional integration. J Neurosci 35(46):15254–15262.

Margulies et al. PNAS Early Edition | 5 of 6

33. Konishi M, McLaren DG, Engen H, Smallwood J (2015) Shaped by the past: The default 64. Spreng RN, Stevens WD, Chamberlain JP, Gilmore AW, Schacter DL (2010) Default
mode network supports cognition that is independent of immediate perceptual in- network activity, coupled with the frontoparietal control network, supports goal-
put. PLoS One 10(6):e0132209. directed cognition. Neuroimage 53(1):303–317.
34. Spreng RN, et al. (2014) Goal-congruent default network activity facilitates cognitive 65. Beaty RE, et al. (2014) Creativity and the default network: A functional connectivity
control. J Neurosci 34(42):14108–14114. analysis of the creative brain at rest. Neuropsychologia 64:92–98.
35. Crittenden BM, Mitchell DJ, Duncan J (2015) Recruitment of the default mode net- 66. Broyd SJ, et al. (2009) Default-mode brain dysfunction in mental disorders: A sys-
work during a demanding act of executive control. eLife 4:e06481. tematic review. Neurosci Biobehav Rev 33(3):279–296.
36. Krieger-Redwood K, et al. (2016) Down but not out in posterior cingulate cortex: 67. Stephan KE, et al. (2001) Advanced database methodology for the Collation of
Deactivation yet functional coupling with prefrontal cortex during demanding se- Connectivity data on the Macaque brain (CoCoMac). Philos Trans R Soc Lond B Biol Sci
mantic cognition. Neuroimage 141:366–377. 356(1412):1159–1186.
37. van den Heuvel MP, Sporns O (2013) Network hubs in the human brain. Trends Cogn 68. Bakker R, Wachtler T, Diesmann M (2012) CoCoMac 2.0 and the future of tract-tracing
Sci 17(12):683–696. databases. Front Neuroinform 6:30.
38. Jbabdi S, Sotiropoulos SN, Behrens TE (2013) The topographic connectome. Curr Opin 69. Mitchell JSB, Mount DM, Papadimitriou CH (1987) The discrete geodesic problem.
Neurobiol 23(2):207–215. SIAM J Comput 16(4):647–668.
39. Van Essen DC, et al. (2013) The WU-Minn human connectome project: An overview. 70. O’Rourke J (1999) Computational geometry column 35. Int J Comp Geom Appl 9(4-5):
Neuroimage 80:62–79. 513–515.
40. Glasser MF, et al. (2013) The minimal preprocessing pipelines for the human con- 71. Marcus DS, et al. (2011) Informatics and data mining tools and strategies for the
nectome project. Neuroimage 80:105–124. human connectome project. Front Neuroinform 5:4.
41. Haak KV, Marquand AF, Beckmann CF (2016) Connectopic mapping with resting-state 72. Fischl B (2012) FreeSurfer. Neuroimage 62(2):774–781.
fMRI. arXiv:1602.07100. 73. Jenkinson M, Bannister P, Brady M, Smith S (2002) Improved optimization for the
42. Coifman RR, et al. (2005) Geometric diffusions as a tool for harmonic analysis and robust and accurate linear registration and motion correction of brain images.
structure definition of data: Diffusion maps. Proc Natl Acad Sci USA 102(21):7426–7431. Neuroimage 17(2):825–841.
43. Hill J, et al. (2010) Similar patterns of cortical expansion during human development 74. Jenkinson M, Beckmann CF, Behrens TEJ, Woolrich MW, Smith SM (2012) FSL.
and evolution. Proc Natl Acad Sci USA 107(29):13135–13140. Neuroimage 62(2):782–790.
44. Biswal BB, et al. (2010) Toward discovery science of human brain function. Proc Natl 75. Van Essen DC, Glasser MF, Dierker DL, Harwell J, Coalson T (2012) Parcellations and
hemispheric asymmetries of human cerebral cortex analyzed on surface-based at-
Acad Sci USA 107(10):4734–4739.
lases. Cereb Cortex 22(10):2241–2262.
45. Wang D, et al. (2015) Parcellating cortical functional networks in individuals. Nat
76. Van Essen DC (2004) Surface-based approaches to spatial localization and registration
Neurosci 18(12):1853–1860.
in primate cerebral cortex. Neuroimage 23(Suppl 1):S97–S107.
46. Gordon EM, Laumann TO, Adeyemo B, Petersen SE (October 13, 2015) Individual
77. Von Bonin G, Bailey P (1947) The Neocortex of Macaca Mulatta (University of Illinois
variability of the system-level organization of the human brain. Cereb Cortex.
Press, Champaign, IL).
47. Yarkoni T, Poldrack RA, Nichols TE, Van Essen DC, Wager TD (2011) Large-scale auto-
78. Bezgin G, Vakorin VA, van Opstal AJ, McIntosh AR, Bakker R (2012) Hundreds of brain
mated synthesis of human functional neuroimaging data. Nat Methods 8(8):665–670.
maps in one atlas: Registering coordinate-independent primate neuro-anatomical
48. Fox PT, Lancaster JL (2002) Opinion: Mapping context and content: The BrainMap
data to a standard brain. Neuroimage 62(1):67–76.
model. Nat Rev Neurosci 3(4):319–321.
79. Lafon S, Lee AB (2006) Diffusion maps and coarse-graining: A unified framework for
49. Kiebel SJ, Daunizeau J, Friston KJ (2008) A hierarchy of time-scales and the brain. PLoS
dimensionality reduction, graph partitioning, and data set parameterization. IEEE
Comput Biol 4(11):e1000209.
Trans Pattern Anal Mach Intell 28(9):1393–1403.
50. Friston K (2013) Life as we know it. J R Soc Interface 10(86):20130475.
80. Ye AQ, et al. (2015) The intrinsic geometry of the human brain connectome. Brain
51. Schacter DL, Addis DR (2007) The cognitive neuroscience of constructive memory:
Inform 2(4):197–210.
Remembering the past and imagining the future. Philos Trans R Soc Lond B Biol Sci
81. Atasoy S, Donnelly I, Pearson J (2016) Human brain networks function in connectome-
362(1481):773–786.
specific harmonic waves. Nat Commun 7:10340.
52. Binder JR, Desai RH, Graves WW, Conant LL (2009) Where is the semantic system? a
82. Von Luxburg U (2007) A tutorial on spectral clustering. Stat Comput 17(4):395–416.
critical review and meta-analysis of 120 functional neuroimaging studies. Cereb
83. Tenenbaum JB, De Silva V, Langford JC (2000) A global geometric framework for
Cortex 19(12):2767–2796. nonlinear dimensionality reduction. Science 290(5500):2319–2323.
53. Jefferies E (2013) The neural basis of semantic cognition: Converging evidence from 84. Langs G, Golland P, Tie Y, Rigolo L, Golby AJ (2010) Functional geometry alignment
neuropsychology, neuroimaging and TMS. Cortex 49(3):611–625. and localization of brain areas. Adv Neural Inf Process Syst 1:1225–1233.
54. Constantinescu AO, O’Reilly JX, Behrens TEJ (2016) Organizing conceptual knowledge 85. Langs G, Tie Y, Rigolo L, Golby AJ, Golland P (2010) Localization of language areas in
in humans with a gridlike code. Science 352(6292):1464–1468. brain tumor patients by functional geometry alignment. Proceedings of the MICCAI
55. Amodio DM, Frith CD (2006) Meeting of minds: The medial frontal cortex and social Workshop on Computational Imaging Biomarkers for Tumors, eds Shen D, et al.
cognition. Nat Rev Neurosci 7(4):268–277. (Beijing, China). Available at www.spl.harvard.edu/publications/item/view/2112. Ac-
56. Amft M, et al. (2015) Definition and characterization of an extended social-affective cessed September 9, 2016.
default network. Brain Struct Funct 220(2):1031–1049. 86. Langs G, Menze BH, Lashkari D, Golland P (2011) Detecting stable distributed patterns
57. Fellows LK (2011) Orbitofrontal contributions to value-based decision making: Evi- of brain activation using gini contrast. Neuroimage 56(2):497–507.
dence from humans with frontal lobe damage. Ann N Y Acad Sci 1239:51–58. 87. Langs G, et al. (2011) Learning an atlas of a cognitive process in its functional ge-
58. Chau BKH, Kolling N, Hunt LT, Walton ME, Rushworth MFS (2014) A neural mecha- ometry. Inf Process Med Imaging 22:135–146.
nism underlying failure of optimal choice with multiple alternatives. Nat Neurosci 88. Langs G, et al. (2014) Decoupling function and anatomy in atlases of functional con-
17(3):463–470. nectivity patterns: Language mapping in tumor patients. Neuroimage 103:462–475.
59. Goulas A, Uylings HBM, Stiers P (2014) Mapping the hierarchical layout of the 89. Langs G, Golland P, Ghosh SS (2015) Predicting activation across individuals with resting-
structural network of the macaque prefrontal cortex. Cereb Cortex 24(5):1178–1194. state functional connectivity based multi-atlas label fusion. Medical Image Computing
60. Nee DE, D’Esposito M (March 21, 2016) The hierarchical organization of the lateral and Computer-Assisted Intervention–MICCAI 2015, Lecture Notes in Computer Science,
prefrontal cortex. eLife, 10.7554/eLife.12112. eds Navab N, Hornegger J, Wells WM, Frangi AF (Springer, Berlin), Vol 9350, pp 313–320.
61. Glasser MF, et al. (2016) A multi-modal parcellation of human cerebral cortex. Nature 90. Fox PT, Lancaster JL, Laird AR, Eickhoff SB (2014) Meta-analysis in human neuroimaging:
536(7615):171–178. Computational modeling of large-scale databases. Annu Rev Neurosci 37:409–434.
62. Haggard P (2008) Human volition: Towards a neuroscience of will. Nat Rev Neurosci 91. Laird AR, et al. (2011) Behavioral interpretations of intrinsic connectivity networks.
9(12):934–946. J Cogn Neurosci 23(12):4022–4037.
63. Smith SM, et al. (2015) A positive-negative mode of population covariation links brain 92. Rottschy C, et al. (2013) Differentiated parietal connectivity of frontal regions for
connectivity, demographics and behavior. Nat Neurosci 18(11):1565–1567. “what” and “where” memory. Brain Struct Funct 218(6):1551–1567.

6 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1608282113 Margulies et al.