Traditional neurophysiology has gone a long way via the classic method for studying brain functions, where the responses of individual neurons to external stimuli are examined one at a time. In recent decades, new neurophysiological and neuroimaging techniques have been developed to study neural circuits and specific brain networks that support a specific function. Although these localized networks have advanced our knowledge about the details of how the cooperation and competition between neural processes take place, a crucial question in neuroscience still involves how the brain can support functions that transcend the functions of specialized networks.
Previous studies of functional interactions between large-scale brain networks have identified broad neural networks that operate in apparent competition and cooperation with one another. For example, some networks support internally oriented processing, and others mediate attention to external stimuli. Recently, it has been found that with challenging tasks, cooperation amongst internal networks increase, and the amount of increase co-varies with better performance in the specific task. Although this level of understanding is available, what is still missing is a mechanistic understanding of how cooperation and competition give rise to the emergent physiological functions in a self-organized sense.
The anatomical connectivity between different brain regions remains relatively stable, but the cooperative and competitive interactions between them are dynamic. With the emergence of the BRAIN initiative and efforts to develop high-density neural sensors, dense ensemble time-series
recordings will give us a better glimpse of how cooperative and competitive interactions through the whole brain process achieve specific functions. Realistically, we should also remember that the study on cooperative and competitive interactions between brain networks is generally challenging in part because of the challenges in understanding the dynamics of highly nonlinear elements interacting in complex networks. As such, we should place as much emphasis on new acquisition tools in neuroscience as we should in advancing our theoretical models that make predictions about how local competitive and cooperative subunits give rise to predictive, emergent functions. In parallel with this, there is an unmet need of evolving time-invariant statistical methods toward scalable, dynamic methods that can quantify dynamic interactions in a large stochastic network of systems over physiologically relevant timescales. These new dynamic statistical methods will have the potential to be applied to ensemble time series reflecting neurophysiological data as well as those of social behaviors.
Ideally, the advancement of sensors, modeling, and statistical approaches will help us understand the dynamic aspects of cooperative and competitive mechanisms within the brain as well as across social species. These approaches, if carefully developed in unison, have the potential to not only shed light on contributions to individual differences in behavior, but also on how the flexibility of normal brain functions is disrupted in neurological disorders.
Can the brain generate complex functions that transcend those of specialized networks by virtue of the patterns of cooperation and competition of overlapping functional networks?
How can we develop a quantitative method to dynamically track the evolution of cooperative and competitive interactions of segregated brain regions?
What are the similarities and differences in the application of the concepts of cooperation and competition to brain networks compared to whole organisms?
Cocchi L, Zalesky A, Fornito A, and Mattingley JB. Dynamic cooperation and competition between brain systems during cognitive control. Trends in Cognitive Sciences 2013;17(10):493-501.
Fornito A, Harrisona BJ, Zaleskya A, and Simonsd JS. Competitive and cooperative dynamics of large-scale brain functional networks supporting recollection. Proceedings of the National Academy of Sciences of the United States of America 2012;109(31):12788-12793.
Kim S, Quinn CJ, Kiyavash N, and Coleman TP. Dynamic and succinct statistical analysis of neuroscience data. Proceedings of the IEEE 2014;102(5):683-898.
IDR TEAM MEMBERS
Dennis E. Discher, University of Pennsylvania
Carla Koehler, Univeersity of California, Los Angeles
Ping-I Lin, Cincinnati Children’s Hospital Medical Center
Floria Mora-Kepfer Uy, University of Miami
Lilianne R. Mujica-Parodi, Stony Brook University
Naoki Saito, University of California, Davis
Nicholas C. Weiler, University of California, Santa Cruz
IDR TEAM SUMMARY—GROUP 9
Nicholas Weiler, NAKFI Science Writing Scholar University of California, Santa Cruz
IDR Team 9 was asked to explore how general principles of cooperation and competition can inform the scientific community’s understanding of brain networks.
The team examined examples of cooperative and competitive interactions across multiple scales of brain activity—from competitive principles governing the wiring of neural circuits to cooperative interactions between functionally specialized brain regions that ultimately let an organism perceive the world and produce adaptive behavior.
The team identified a fundamental disconnect between new research demonstrating brain-wide changes in network connectivity in people with neurological disorders such as autism and epilepsy and more basic research into mechanistic explanations of these disorders at the level of neurons and synaptic connections.
The team chose to narrow the focus of their mission to develop a research strategy to bridge this gap in scientific approaches to neurological disease. They chose to focus on epileptic seizures as a relatively simple disordered network state around which to develop a broad strategy for examining the connection between specific competitive and cooperative brain mechanisms and the healthy and diseased network dynamics they produce.
The Problem: The Black Box
The human brain contains approximately 86 billion neurons. Each neuron contacts thousands to hundreds of thousands of its fellows in intricate and overlapping networks. Our consciousness is the flux of electrical activity through these webs of connectivity.
Modern psychiatry frequently treats brain diseases as whole-organ problems, the team noted. Many epilepsy drugs work by wholesale manipulation of the chemical signals used by diverse networks across the brain. Taking a different approach, neurosurgeons try to eliminate epilepsy by cutting out chunks of brain tissue suspected of sparking seizures.
However, in recent years, new technology has made possible a more fine-grained mapping of neural connections and the flow of activity between regional brain networks. Large-scale research efforts—such as the Human Connectome Project and the Obama Administration’s Brain Initiative—have arisen to extend these technological developments.
The team determined that the time seemed ripe for a new approach that could connect diseased activity patterns at the whole-brain level to the small-scale mechanisms that cause them. In particular, they emphasized the need to determine how potential mechanisms affect competitive and cooperative interactions at the under-researched middle scale of brain networks.
Network Connectivity and Psychiatric Disease
Human neuroimaging experiments suggest that at a broad scale, the brain is organized into a so-called “small-world” architecture, in which tightly interconnected local networks are linked to one another by rarer long-distance connections. This network organization enables highly efficient information transfer and may also keep network activity in a stable, balanced state. Departures from “small-worldness” may be a key factor in the development of psychiatric and neurological disease.
The team focused in particular on epilepsy because seizures—excessive
network activity linked to out-of-control positive feedback loops—are a relatively straightforward network phenomenon. Previous research has proposed a number of potential genetic and molecular mechanisms that may generate seizures in epileptic brains.
Neuroimaging experiments show that patients with epilepsy have many more long-distance connections than control subjects. This expanded connectivity in patients with epilepsy may contribute to producing seizures that can spread rapidly through the brain. Such changes in connectivity may be caused by synaptic plasticity—the ability of neurons to dynamically shift and reform their connections with one another. But the mechanisms that cause these changes are not yet understood.
The team set itself the goal of laying out a new strategy to bridge these scales: to investigate the mechanistic basis (at the microscale) for observations of altered functional connectivity in epilepsy (at the macroscale). The team proposed examining mid-scale network mechanisms of cooperation and conflict in the context of three potential classes of mechanism—resource competition, plasticity regulation, and circuit repair—which may form the basis for modulation of synaptic plasticity and maladaptive consequences for network activity.
The Strategy: Top Down
The team outlined a strategy of beginning with the ultimate emergent consequence of altered brain networks—in this case epileptic seizures—and drilling down in scale to identify the mechanisms behind them. Their logic was that establishing a clear large-scale network phenomenon in humans should allow subsequent experimental investigations of possible causes for that phenomenon to stay focused and directed.
For each potential constraint they identified, the team outlined a two-pronged experimental approach to examine connections between defects at the macrolevel (whole brain) and microlevel (neurons and synapses).
Competition for Metabolic Resources
The brain is the body’s primary consumer of energy, by far. The brain accounts for 2% of the body’s mass, but 20% of its energy consumption, over twice that of any other organ. Synapse production and maintenance requires energy as a resource. Finite resources require optimization, balancing gains and costs of synaptic density.
Understanding how energy metabolism affects the onset of seizures in brain networks might shed new light on why the ketogenic diet, which generally stabilizes the body’s rate of energy use, is as effective as medication in preventing seizures in many people.
Macrolevel approaches: The team proposed measuring energy consumption in real time in patients under different levels of glycemic load. Specifically, they proposed using magnetic resonance spectroscopy and positron emission tomography to measure local dynamics of creatine and glucose levels respectively in the brain.
Microlevel approaches: The team proposed measuring the distribution and activity of mitochondria within neurons and around synapses, as well as local metabolite profiles measured under different dietary parameters.
Regulation of Excitation and Inhibition
Brain networks strive to maintain a balance between excitatory signaling using the neurotransmitter glutamate and inhibitory signaling using the neurotransmitter GABA. Inhibition is crucially important in neural networks—without it, a highly interconnected excitatory network is subject to positive feedback loops that produce runaway excitation, also known as a seizure. A reduced ratio of GABA synapses to glutamate synapses has been hypothesized to play a role in epilepsy. An open question is why this imbalance occurs in the first place.
Macrolevel approaches: The team proposed taking advantage of recent advances in ultra-high-field (7 Tesla) magnetic resonance spectroscopy to measure local levels of glutamate and GABA in the human brain in real time. The team proposed that drugs could be used to disrupt the balance of excitation and inhibition to directly test the consequences for brain-wide network activity.
Microlevel approaches: The team proposed probing the mechanisms behind the changes in excitatory/inhibitory balance seen in human neuroimaging using drugs, optogenetic control of neuronal activity, and measurement of synaptic density in mice.
Trauma Repair and Plasticity
Epilepsy is a relatively common consequence of brain injury, though seizures may not develop for years after the original brain damage. One reason is that the brain must balance the need for plasticity to repair damaged
circuits with the danger that too much plasticity may disrupt previously stable network relationships.
One way the brain limits plasticity is to impose structural barriers on synaptic formation. In particular, extracellular matrix structures called perineuronal nets (PNNs) grow around neurons in later stages of brain development to restrict further alterations to neural circuits. Previous research has linked the formation of PNNs to the closure of critical periods for learning and stabilization of network function.
Following injury, resident microglia or blood-derived white cells may locally remove structural constraints on plasticity using soluble factors such as matrix-degrading enzyme matrix metalloprotease (MMP9). Neurons themselves may also secrete MMP9 following injury. Reinstating plasticity enables the brain to repair damage and networks to reestablish themselves. However, heightened plasticity also means that injured brain circuits become susceptible to excessive synaptic sprouting, an overcompensation that may impact network function and potentially lead to seizures.
Another potential consequence of injury is the formation of glial scars, an additional structural barrier that may regionally isolate neural circuits. By analogy to cardiac scars that form following heart injury, these barriers may disrupt local circuits by forcing them to reroute activity around the barriers, creating new stress points that may be related to seizure activity.
Macrolevel approaches: The team proposed developing models of network connectivity that include effects of structural barriers and rewiring. These models can be tested to explore the consequences of these factors for the percolation of activity through networks following injury. In addition, existing drugs can inhibit proteases such as MMP9 and allow retrospective analyses of patients to determine the optimal degree of structural plasticity that balances circuit repair and stability.
Microlevel approaches: The team proposed live imaging of the distribution of matrix and glial scar cells after brain damage in mice through thinned skulls, combined with ion current visualization to track neural activity.
Societal Impact: Meeting in the Middle
The team proposed that these experiments would produce multiscale data, which could be used to computationally model how different mechanistic factors influence neural network function. Such results have the potential to reveal how the ongoing equilibria neural circuits must
maintain—managing resource competition, balancing excitation and inhibition, and regulating plasticity and repair—and interact with specific genetic and physical risk factors to produce neurological disorders such as epilepsy. Understanding these chronic influences on the function of neural circuits may also enable researchers to target relevant mechanisms to prevent age-based cognitive decline.
In addition, from a purely scientific perspective, the team argued, this research strategy would allow neuroscientists to finally bridge the profound gap between human neuroimaging and the mechanisms underlying brain function.