Features of spatial and functional segregation and integration of the primate connectome revealed by trade-off between wiring cost and efficiency

Abstract

The primate connectome, possessing a characteristic global topology and specific regional connectivity profiles, is well organized to support both segregated and integrated brain function. However, the organization mechanisms shaping the characteristic connectivity and its relationship to functional requirements remain unclear. The primate brain connectome is shaped by metabolic economy as well as functional values. Here, we explored the influence of two competing factors and additional advanced functional requirements on the primate connectome employing an optimal trade-off model between neural wiring cost and the representative functional requirement of processing efficiency. Moreover, we compared this model with a generative model combining spatial distance and topological similarity, with the objective of statistically reproducing multiple topological features of the network. The primate connectome indeed displays a cost-efficiency trade-off and that up to 67% of the connections were recovered by optimal combination of the two basic factors of wiring economy and processing efficiency, clearly higher than the proportion of connections (56%) explained by the generative model. While not explicitly aimed for, the trade-off model captured several key topological features of the real connectome as the generative model, yet better explained the connectivity of most regions. The majority of the remaining 33% of connections unexplained by the best trade-off model were long-distance links, which are concentrated on few cortical areas, termed long-distance connectors (LDCs). The LDCs are mainly non-hubs, but form a densely connected group overlapping on spatially segregated functional modalities. LDCs are crucial for both functional segregation and integration across different scales. These organization features revealed by the optimization analysis provide evidence that the demands of advanced functional segregation and integration among spatially distributed regions may play a significant role in shaping the cortical connectome, in addition to the basic cost-efficiency trade-off. These findings also shed light on inherent vulnerabilities of brain networks in diseases. Author summaryThe intricate primate structural connectome, as the network substrate for distributed and integrated brain function, is shaped by fundamental physical factors and functional requirements. We addressed a trade-off between two competing basic factors of wiring cost and processing efficiency as well as additional requirements of functional segregation and integration on regional structural connectivity profiles by applying cost-efficiency trade-off model to reconstruct the macaque cortical network. We also compared this model with a generative model combining spatial distance and topological similarity. The trade-off model balancing the two basic factors recovered the main topological features that were explicitly specified as the objective functions in the generative model. Moreover, 67% of all connections and most regional connectiv...

Spatial Embedding Imposes Constraints on Neuronal Network Architectures

“…One particularly influential study considered 13 generative models that all incorporated a wiring probability that increased with distance [111]. Consistent with other work, the authors found that the model that only included the wiring minimization constraint was unable to recreate long distance connections of individual connectomes in humans [8,11,111]. Successive generative models were then added that attempted to recreate certain aspects of topology in addition to these geometric constraints [111].…”

“…The processes and influences that guide the formation of structural connections in neural systems are disparate and varied [7,8]. Evidence from genetics suggests that neurons with similar functions as operationalized by similar gene expression tend to have more similar connection profiles than neurons with less similar functions [7,9,10].…”

“…Collectively, these studies demonstrate the influence of parsimonious wiring rules on complex network topology. Future work could be directed to better understand the aspects of connectome topology that remain unexplained and thus may arise from more subtle rules [8].…”

Specificity and robustness of long-distance connections in weighted, interareal connectomes

“…More recently and using a different macaque network dataset, ref. 41 suggested that long-distance connections, those unanticipated by a model that penalized the formation of costly connections, form a dense cluster and may also perform some segregative functions.…”

Genetic influences on hub connectivity of the human connectome

“…We also note that our model fits (Figs 4B-E) are in the same range as those reported by Betzel et al (33), indicating that our discrepant results are not due to differences in model accuracy. Finally, while our analysis mimics prior comprehensive evaluations of generative network models based on cost-value trade-offs (32,33), other formulations and approaches are also possible (116)(117)(118) and we cannot completely rule out a role for stochastic processes in shaping hub connectivity. In particular, more abstract models suggest that stochastic wiring, acting in concert with developmental changes in brain geometry and heterogeneous timing of connection formation across regions, can indeed generate networks with brain-like properties (34,71,119).…”