The Journal of , January 7, 2015 • 35(1):1–3 • 1

Journal Club

Editor’s Note: These short, critical reviews of recent papers in the Journal, written exclusively by graduate students or postdoctoral fellows, are intended to summarize the important findings of the paper and provide additional insight and commentary. For more information on the format and purpose of the Journal Club, please see http://www.jneurosci.org/misc/ifa_features.shtml.

Considering the Impact of Large-Scale Network Interactions on Cognitive Control

X Fiona Kumfor,1,2,3 Nadene Dermody,1,4 and Muireann Irish1,2,4 1Neuroscience Research Australia, and 2ARC Centre of Excellence in Cognition and its Disorders, 3School of Medical Sciences, and 4School of Psychology, University of New South Wales, Sydney, NSW 2031, Australia Review of Jilka et al.

In everyday life, we often face situations is recruited in response to attention- provide important insights into the con- in which we must compete with an grabbing changes in the environment, and tribution of large-scale network dys- overriding/pre-potent response to inhibit it is anchored by the dorsal anterior cin- function on cognition and behavior. To our behavior and act in line with situa- gulate cortex and orbital frontoinsular this end, a recent study by Jilka et al. tional demands. For example, on most cortices, with robust connections with (2014) used behavioral and neuroimag- days, the drive home from work is a rela- subcortical and limbic structures (Zhou et ing methods to determine how cognitive tively automatic behavior that can be un- al., 2010). Conversely, the DMN is acti- control, and the underlying neural net- dertaken with limited conscious effort. vated when current situational demands works subtending this process, is altered But if a dog ran unexpectedly into traffic, are insufficient to capture our attention by TBI. you would need to quickly identify this (e.g., during monotonous tasks); it en- Previous work by the authors revealed change in the environment and stop your compasses a distributed set of regions that TBI-related damage to the white mat- ongoing habitual actions to prevent an including the medial and lateral tempo- ter tract connecting the right anterior accident. This ability to rapidly halt ac- ral cortices and inferior lateral parietal insula to the presupplementary motor tions that are already underway in re- cortices, centered on midline “hubs,” area/dorsal anterior cingulate cortex sponse to a change in the environment including the dorsomedial prefrontal (rAI-preSMA/dACC), located within the or internal state is dependent on cogni- and posterior cingulate cortices (Buck- network, is associated with a fail- tive control; the ability to “override or ner et al., 2008). Crucially, the commu- ure to deactivate the DMN (Bonnelle et augment reflexive behavior and habit- nication within, and the interaction al., 2012). In their recent study, Jilka et al. ual reactions to orchestrate behavior in between, large-scale networks is neces- (2014) sought to extend this research accord with our intentions” (Miller, sary for adaptive cognitive function. to determine whether damage to the 2000, p. 59). Cognitive control therefore provides rAI-preSMA/dACC tract impairs dy- The advent of resting-state functional an interesting conceptual model for namic interactions between the salience connectivity techniques investigating the potential interactions network and the DMN, manifesting in has led to the delineation of several large- between large-scale brain networks pro- disrupted cognitive control after TBI. scale functional brain networks that are posed to support higher-order cognitive The authors used two tasks to investi- differentially recruited contingent on sit- processes. gate the capacity for cognitive control in uational demands. Two key networks are Cognitive control is often compro- TBI relative to healthy control partici- the and the default mode mised after traumatic brain injury (TBI), pants: a stop signal task and a motor network (DMN). The salience network with individuals experiencing severe diffi- switching task. On the stop signal task, culties in inhibiting and regulating their participants were shown either left or

Received Oct. 9, 2014; revised Nov. 12, 2014; accepted Nov. 14, 2014. behavior to meet internal goals. Diffuse right arrows, and were asked to press the M.I. is supported by an ARC Discovery Early Career Researcher Award axonal injury occurs after TBI, and the corresponding left or right key. On a mi- (DE130100463). integrity of pathways con- nority of trials (20%), however, partici- Correspondence should be addressed to Dr. Muireann Irish, Neuro- necting network hubs is disrupted. pants were shown a “stop signal” (red dot) science Research Australia, Sydney, NSW 2031, Australia. E-mail: [email protected]. Thus, large-scale network function is and were required to inhibit their re- DOI:10.1523/JNEUROSCI.4213-14.2015 impaired (Sharp et al., 2014). Studying sponse. In contrast, on the motor switch- Copyright © 2015 the authors 0270-6474/15/350001-03$15.00/0 individuals with TBI can therefore ing task, participants learned to respond 2 • J. Neurosci., January 7, 2015 • 35(1):1–3 Kumfor et al. • Journal Club to blue targets with their left hand and red flexible switching of attention to - reasoning, are associated with abnormal targets with their right hand. Crucially, on generated, internal states (Burgess et al., modulation of the DMN by the salience a minority of trials (20%), participants 2007). A further region of interest in this network (Chiong et al., 2013). This find- were instructed to switch their response. regard is the posterior cingulate cortex, ing, together with Jilka et al.’s (2014) re- The authors used psychophysiological one of the putative midline hubs of the sults, has important implications for interaction analysis (PPI) of functional DMN, as this region has been proposed other cognitive and behavioral features of MRI data collected as subjects performed to interact with frontoparietal atten- this syndrome. For example, socially dis- these two tasks to investigate functional tional networks to regulate between in- inhibited behavior is a prominent feature connectivity between salience network ternal and external forms of cognition of . Given these regions of interest (i.e., right anterior (Leech et al., 2012). new insights into the putative role of the insula, dorsal anterior cingulate cortex) We suggest that exploration of the salience network, it is reasonable to pro- and the DMN, as well as a region of in- functional coupling between the salience pose that in social settings the salience terest outside the salience network (i.e., network and other functional brain net- network must first identify the presence of right inferior frontal gyrus) and the works via the insula might further our un- relevant social cues, following which acti- DMN. derstanding of cognitive control from a vation of the central-executive network is The results revealed that in controls, network perspective. While Jilka et al. coordinated to guide decision-making successful performance on both the inhi- (2014) interpreted their findings in terms regarding appropriate behavior. In this bition and switching tasks was associated of the functional coupling between the sa- way, degeneration of the insula node of with stronger correlations (increased lience network and the DMN, the right the salience network may prevent the functional connectivity) between the right frontoinsular node of the salience network flexible switching between the central- anterior insula node of the salience net- has also been implicated in switching be- executive network and the DMN in work and the DMN. Importantly, deficits tween the DMN and the central-executive response to contextual/environmental in inhibition and switching seen behav- network (Sridharan et al., 2008). This demands, potentially giving rise to a iorally in TBI were associated with de- large-scale functional brain network en- number of hallmark behavioral and creased functional connectivity between compasses the dorsolateral prefrontal cognitive abnormalities observed in this the right anterior insula node and the cortex and the posterior parietal cortex syndrome. While our interpretation at DMN compared with controls. More- and is important for higher-level goal- this point is purely speculative, exami- over, in TBI, greater damage to the directed behavior, decision-making, nation of the characteristic impairments rAI-preSMA/dACC tract within the sa- and . The insula is in frontotemporal dementia, from this lience network was associated with weaker uniquely positioned to serve as an inter- network perspective, via targeted tasks functional connectivity between the face between the salience network, assessing cognitive control, will provide right anterior insula and the DMN dur- DMN, and central-executive network, important insights into the role of large- ing inhibition and switching. Together, with reciprocal connections between scale neural network interactions in these results indicate that the right ante- sensory, motor, limbic, and association supporting cognition and behavior. rior insula represents a crucial node brain regions, enabling integration of In conclusion, Jilka et al.’s (2014) study within the salience network, which po- information from across the brain. How localizes the right anterior insula as a key tentially enables dynamic interactions the findings by Jilka et al. (2014) fit node within the salience network for between the salience and default mode within this broader framework of large- modulating large-scale functional brain networks. scale network interactions will be an im- network activity to support cognitive The study by Jilka et al. (2014) pro- portant area for future research. control. We propose that future work vides an elegant demonstration of the dy- A second important consideration is adopting this convergent approach will namic interactions between functional how such network interactions may be af- prove particularly useful for under- brain networks in supporting complex fected in other clinical populations. Simi- standing the origins of a host of dys- cognitive acts essential for adaptive func- lar to TBI, neurodegenerative disorders functional behaviors typically observed tioning. The exposition of specific sites are increasingly being viewed as “network in neurodegenerative and psychiatric within the salience network (i.e., the right phenomena” and thus offer a unique op- populations. anterior insula) that potentially support portunity to study the impact of large- interactions between the salience network scale functional network disruption on References and the DMN is noteworthy and corrob- complex behaviors. The behavioral vari- Bonnelle V, Ham TE, Leech R, Kinnunen KM, orates previous work implicating the right ant of frontotemporal dementia repre- Mehta MA, Greenwood RJ, Sharp DJ (2012) anterior insula and anterior cingulate cor- sents a syndrome of immense interest in Salience network integrity predicts default tex across a range of higher-order pro- this context, given that the earliest sites of mode network function after traumatic brain cesses including cognitive control and pathology reside in the frontoinsular cor- injury. Proc Natl Acad Sci U S A 109:4690– 4695. CrossRef Medline performance monitoring. It has been tices, reflecting progressive degeneration Buckner RL, Andrews-Hanna JR, Schacter DL suggested that these regions form a highly of the salience network (Zhou et al., (2008) The brain’s default network. Ann N Y interconnected core system for task- 2010). Clinically, these patients display Acad Sci 1124:1–38. CrossRef Medline dependent control of goal-directed behav- disinhibited and inappropriate behaviors, Burgess PW, Dumontheil I, Gilbert SJ (2007) ior and sensory processing (Dosenbach et pointing toward impaired cognitive con- The gateway hypothesis of rostral prefrontal al., 2007). One remaining issue, however, trol, which is generally taken to reflect cortex (area 10) function. Trends Cogn Sci 11: concerns how the current findings relate to prefrontal cortical degeneration. Recent 290–298. CrossRef Medline Chiong W, Wilson SM, D’Esposito M, Kayser influential theories which view the rostral work, however, has revealed that impair- AS, Grossman SN, Poorzand P, Seeley WW, as a gateway between ments in higher-level cognitive abilities in Miller BL, Rankin KP (2013) The salience monitoring of external states and the frontotemporal dementia, such as moral network causally influences default mode Kumfor et al. • Journal Club J. Neurosci., January 7, 2015 • 35(1):1–3 •3

network activity during moral reasoning. with the . J Neurosci 34: Sridharan D, Levitin DJ, Menon V (2008) A criti- Brain 136:1929–1941. CrossRef Medline 10798–10807. CrossRef Medline cal role for the right fronto- Dosenbach NU, Fair DA, Miezin FM, Cohen AL, Leech R, Braga R, Sharp DJ (2012) Echoes of the in switching between central-executive and Wenger KK, Dosenbach RA, Fox MD, Snyder brain within the posterior cingulate cortex. default-mode networks. Proc Natl Acad Sci AZ, Vincent JL, Raichle ME, Schlaggar BL, Pe- J Neurosci 32:215–222. CrossRef Medline U S A 105:12569–12574. CrossRef Medline tersen SE (2007) Distinct brain networks for Miller EK (2000) The prefontral cortex and cog- Zhou J, Greicius MD, Gennatas ED, Growdon adaptive and stable task control in humans. nitive control. Nat Rev Neurosci 1:59–65. ME, Jang JY, Rabinovici GD, Kramer JH, Proc Natl Acad Sci U S A 104:11073–11078. CrossRef Medline Weiner M, Miller BL, Seeley WW (2010) Di- CrossRef Medline Sharp DJ, Scott G, Leech R (2014) Network vergent network connectivity changes in be- Jilka SR, Scott G, Ham T, Pickering A, Bonnelle V, dysfunction after traumatic brain injury. havioural variant frontotemporal dementia Braga RM, Leech R, Sharp DJ (2014) Dam- Nat Rev Neurol 10:156–166. CrossRef and Alzheimer’s disease. Brain 133:1352– age to the salience network and interactions Medline 1367. CrossRef Medline