Olaf Sporn’s previous monograph  tries to build bridges between neuroscientists and network scientists by emphasizing the importance of networks in the study of complex biological systems such as the human brain. This new book builds on that framework and, inspired by the National Institutes of Health (NIH) Human Connectome Project, tries to identify key scientific goals in the study of the brain from the perspective of networks. The author candidly assesses the state of the art in this field, which is still in its infancy; analyzes the empirical techniques that can be used to deepen our understanding of the human brain; bluntly exposes their severe limitations; and suggests promising avenues for future progress.
The book starts with two introductory chapters on the role of structure in the study of the brain. Complexity as found in the human brain resists reductionist explanations and requires the theoretical framework provided by network models. These models, when applied to the brain, lead to “connectomics,” a term coined by the author, in 2005, by analogy to genomics. When laying out the rationale behind this emerging field, the author draws a parallel not only between the connectome and the genome, but also between the study of networks in the brain and systems biology. He even resorts to architecture, where form often follows function, when he stresses the importance of the structure of the human brain as the foundation for its rich dynamics.
The third chapter focuses on the most significant challenges that lie ahead in the study of the networks of the brain. First of all, the sheer size of the networks makes them unassailable using current technology, since the human brain is estimated to contain billions of neurons and a potential quadrillion synaptic connections. Furthermore, no technique currently exists that can take a snapshot of a complete human brain at the subcellular level. A canonical human brain cannot be built by combining parts from different individuals, since the brain exhibits a rich variability among different individuals at the neuronal level. Moreover, an additional layer of complexity is attributable to plasticity, the continuous changes in neural pathways and synapses experienced throughout the life of a living organism.
Once the stage is set for the study of connectomics and its limitations, three chapters address the different perspectives offered by the available empirical techniques that can be used to study the brain. These perspectives are often complementary in the quest for understanding the behavior of the human brain:
- No noninvasive techniques can be used to study the brain at the micro level. Electron and light microscopy can be used to analyze local regions of the brain with the help of computer vision algorithms that extract features from images taken from different slices of a given sample.
- At the macro level, tract tracing and magnetic resonance techniques can be used in vivo and offer a broad-brush glimpse into the human brain.
- To analyze brain dynamics, noninvasive neuroimaging technology is often used. Indirect (and necessarily incomplete) measurements of the brain activity can be obtained with techniques based on hemodynamic signals, such as the blood-oxygen-level-dependent (BOLD) contrast employed by functional magnetic resonance imaging (fMRI), or electromagnetic signals, such as those obtained by electroencephalography (EEG) and magnetoencephalography (MEG). These can be used to infer structural and dynamical properties of the brain, and help us discover characteristic patterns in brain activity.
Existing techniques, however, suffer from an important limitation: the true network of the brain cannot be directly measured in living humans.
The final chapters cover recent discoveries related to the network architecture of the brain and computer models that are being developed to test hypotheses in “computational connectomics.” Brain networks are clearly nonrandom, and they show a neat modular, and apparently hierarchical, architecture that neuroscientists are now beginning to unravel, often with the help of unsupervised machine learning techniques (community detection algorithms, in social network jargon). The networks of the brain are also subject to spatial and energy efficiency constraints, which must be taken into account when trying to decipher their structure.
From a complementary point of view, different computer models have been proposed to analyze some facets of the brain, from Jirsa’s virtual brain simulation (http://thevirtualbrain.org/) and Markram’s Blue Brain Project to create a synthetic mammalian brain at the molecular level, to a supercomputer simulation designed by IBM Research that models a cortex at the scale of a cat cortex, which is not the same as simulating an actual cat cortex. As Sporns soberly acknowledges, “so far, neither model has made significant contact with empirical research on human brain anatomy or physiology” (p. 167). Building a truly functional model of the brain requires the discovery of many currently unknown details of brain internals.
In any case, networks provide a suitable theoretical framework for delving into the human brain. Sporns surveys the current state of the art and explains that much is yet to be done in the fields of connectomics or network neuroscience. In the future, the connectome is poised to be the “ome” for neuroscience, in the same way that the genome provides the key foundation for the molecular biology of the cell. Mapping the structure of the human brain--producing its wiring diagram in effect--would be equivalent to sequencing the genome: a key milestone in the (hypothetical) future of neuroscience. It would not be enough to wholly understand the human brain, just as sequencing the genome has not been enough to understand all the complex biochemical mechanisms within a living cell, but it would certainly open up fascinating new avenues of research.