Editorial Focus on "Invariant and heritable local cortical organization as revealed by fMRI"

Hugo Merchant, Ravi S. Menon
2018 Journal of Neurophysiology  
A laudable goal of open neuroscience is to make publicly available the massive and complex neural data collected with modern neuroscience techniques from substantial portions of the nervous system in humans and other mammal species. One such open dataset is the Human Connectome Project restingstate functional magnetic resonance imaging (fMRI) time series across hundreds of healthy young adults (Smith et al. 2013 ). This rich dataset provides the opportunity to determine the rules of cortical
more » ... ctional connectivity based on the correlated dynamics in the hemodynamic signal underlying fMRI. The article of Christova and Georgopoulos (Christova and Georgopoulos 2018) used this carefully acquired and curated dataset to study the effect of cortical distance on the neural interactions of six cortical areas of 854 subjects, including monozygotic and dizygotic twins, nontwin siblings, and nonrelated individuals. As a first step, the time series of the resting-state blood level oxygenation-dependent (BOLD) signal was prewhitened to avoid nonstationarities in the series themselves that could lead to erroneous associations in the voxel-by-voxel pairwise crosscorrelations (Langheim et al. 2006) . Thus a Box-Jenkins autoregressive integrative moving average (ARIMA) modeling analysis was performed, and the zero-lag pairwise cross-correlations (r 0 ) were computed on the ARIMA residuals (called innovations) across brain areas (Christova et al. 2011; Merchant et al. 2014) . The connections within the superior frontal, precentral, postcentral, superior parietal, inferior parietal, and lateral occipital areas showed a complex internal pattern of functional connectivity. The authors, however, focused their efforts on the change in r 0 as a function of distance (D). They found a strong and orderly decrease in r 0 with D, which was properly explained by the power law: r 0 ϭ kD Ϫb . The exponent b quantifies the rate by which r 0 declines with distance along the cortex. Notably, b was similar between hemispheres but showed a decrease in steepness in the anteroposterior axis, with the steepest b in superior frontal and the least steep b in lateral occipital. Although the exponent b was consistent among areas, it showed a large variation among individuals, suggesting that b is characteristic of an individual brain. In fact, Christova and Georgopoulos calculated the intraclass correlation coefficient of b in four group pairs and found evidence for the heritability of b. Specifically, they observed statistically significant positive intraclass correlations, in descending order, for monozygotic twins, dizygotic twins, and nontwin siblings but no intraclass correlation of b for nonrelated individuals. The power law of the zero-lag pairwise cross-correlations with distance indicates that cortical neural interactions are strong in close vicinity to a cortical area and become weaker with increasing distance. This connectivity distribution is characteristic of scale-free networks that are known to show a robust architecture in which information can be transferred and integrated with a high level of efficiency in a variety of networks such as the World Wide Web, and the brain is no exception (Barabási and Bonabeau 2003) . Indeed, anatomical studies have shown that neurons and brain regions that are spatially close have a larger probability of being connected, whereas connections between spatially remote neurons or brain regions are less likely (Averbeck and Seo 2008; Braitenberg and Schüz 1998; Hellwig 2000) . In addition, at the functional level it has been shown that cortical ensembles form synchronous activity patterns, called avalanches, whose size also follow the power law (Plenz and Thiagarajan 2007). Hence, the present results support the notion that the architecture of human cortical circuits strongly follows the power law. This network configuration should have profound implications in the efficiency of information processing within and between cortical areas. The decreasing gradient in the steepness of the power law in the anteroposterior axis of the cortex indicates that local interactions are larger in frontal areas than posterior areas, whereas in the latter the medium-length to distant connections are more preponderant. This variation in spatial tuning emphasizes the functional specialization of cortical areas. For example, the sizes of spatial receptive fields increase along posterioranterior visual hierarchy (Lennie 1998), the extent of local circuits processing voluntary movement information in motor cortex is large (Merchant and Georgopoulos 2018; Naselaris et al. 2005 Naselaris et al. , 2006 , and a posterior-anterior hierarchy exists for cognitive abstraction within prefrontal cortex (Badre and D'Esposito 2009). A recent study demonstrated that the timescale at which the spontaneous activity of neurons fluctuates is an intrinsic signature of the complexity of information processing in primate cortical microcircuits (Murray et al. 2014). In concordance with changes in b observed by Christova and
doi:10.1152/jn.00429.2018 pmid:29975168 fatcat:6poq6jo5kvb2bmehka7gxtyfla