Grand challenges for global brain sciences

Understanding the brain and curing its diseases are among the most exciting challenges of our time. Consequently, national, transnational, and private parties are investing billions of dollars (USD). To efficiently join forces, the Global Brain Workshop 2016 was hosted at Johns Hopkins University’s Kavli Neuroscience Discovery Institute on April 7–8. A second workshop, Open Data Ecosystem in Neuroscience took place July 25–26 in Washington D.C. to continue the discussion specifically about computational challenges and opportunities. A third conference, Coordinating Global Brain Projects, took place in New York City on September 19th in association with the United Nations General Assembly. So vast are both the challenges and the opportunities that global coordination is crucial.

To find ways of synergistically studying the brain, the kick-off workshop welcomed over 60 scientists, representing 12 different countries and a wide range of brain science subdisciplines. They were joined by 15 observers from various national and international funding organizations, including NIH, NSF, IARPA, the Kavli Foundation, and the Simons Foundation. Participants were engaged weeks before the conference and charged with coming up with ambitious projects that are both feasible and internationally inclusive, on par with the International Space Station (i.e., worthy of a global, decade-long effort). Over the course of 36 hours, scientists discussed, debated, and gathered feedback, ultimately proposing several “grand challenges for global brain sciences” that were refined by working groups. The workshop was covered in a media piece in Science April 15, 2016¹.

The group began with 60+ ideas, each forged independently by one of the scientific participants. Each participant proposed a unique challenge that was designed to meet the following desiderata:

Interestingly, a lot of the proposed ideas were similar to one another and others were complementary. This allowed the group to converge on three grand challenges for global brain sciences, each depending on a common universal resource. As each of these four projects gain momentum, we encourage readers to get in touch (details provided below).

Challenge 1: What makes our brains unique?

Both within and across species, brain structure is known to exhibit significant variability across many orders of magnitude in scale, including Anatomy, Biochemistry, Connectivity, Development, and gene Expression (ABCDE). It remains mysterious how and why the nervous system tightly regulates certain properties, while allowing others to vary. Understanding the design principles governing variability may hold the key to understanding intelligence and subjective experience, as well as the influence of variability on health and function.

This grand challenge is a global project to coordinate the construction of comprehensive multiscale maps of the ABCDE’s of multiple brains from multiple species using multiple cognitive and mental health disease models. Within a decade, we expect to have addressed this challenge in brains including, but not limited to, Drosophila, zebrafish, mouse, and marmoset, and to have developed tools to conduct massive neurocartographic analyses. Indeed, many existing datasets will play a crucial role in seeding this project, including data from the Human Brain Project, IARPA’s MICrONS project, and Z-Brain to name a few. The result will be a state-of-the-art “Virtual NeuroZoo” with fully annotated data and analytic tools for analysis and discovery. This virtual NeuroZoo can be utilized by neuroscientists and citizens alike, both as a reference and for educational materials. By incorporating disease models, we explicitly link this challenge with the third challenge. Global discussions around this project are now beginning via the tags “neurostorm” and “neurozoo” at the neuroinformatics discussion forum NeuroStars (https://neurostars.org/).

Challenge 2: How does the brain solve complex computational problems?

Brains remain the most computationally advanced machines for a large array of cognitive tasks - whether navigating hazardous terrain, translating languages, conducting surgery, or recognizing emotional states - despite the fact that modern computers can utilize millions of training samples, megawatts of power, and tons of hardware. While the ABCDEs establish the “wetware” upon which our brains can solve such computations, to understand the mechanisms we need to measure, manipulate, and model neural activity simultaneously across many spatiotemporal resolutions and scales - including wearables, embedded sensors, and actuators - while animals are exhibiting complex ecological behaviors in naturalistic environments.

This grand challenge is a global project to investigate a single naturalistic behavior that is ecologically relevant across phylogenies, such as foraging, and measure brain and body properties across spatial, temporal, and genetic scales. The challenge differs from previous efforts in three key ways. First, it requires studying animals in complex and naturalistic environments. Second, it requires coordinated attacks at many different scales by many different investigators while the animals are performing the same complex behaviors. We envision groups of 20–30 investigators all operating together on shared data and experimental design. Third, the richness of the mental repertoire of cognition suggests that deciphering its codes will require many parallel investigations to uncover different facets of brain function. These experiments in turn will produce multiscale models of neural systems with the potential to accomplish computational tasks that no current computer system can perform. Mechanistic studies, guided by theoretical models, will help to ask how perturbations of those systems lead to aberrant function, linking this challenge with the next one. Global discussions around this project are now beginning via the tags “neurostorm“ and “GlobalBrainLab” at NeuroStars.

Challenge 3: How can we augment clinical decision-making to prevent disease and restore brain function?

Psychiatric and neurological illnesses levy enormous burdens upon humanity: impairment, suffering, financial costs, and loss of productivity. Despite a growing awareness of the challenges, clinicians consistently battle the lack of objective tests to guide clinical decision-making (e.g., diagnosis, selection of treatments, prognosis). Compounding these limitations are societal stigmas regarding mental illness that increase the suffering of patients and their families. The ABCDEs of neurobiological variability, when coupled with multiscale mechanistic models of cognition, will provide new approaches to neurobiologically informed clinical decision making.

This grand challenge is a global project to transform clinical decision-making via incorporating neural mechanisms of dysfunction. This will require collecting, organizing and analyzing human and non-human anatomical and functional data. These data (such as ADNI and ADHD-200), and the tools developed to explore and discover novel treatment therapies, will be the foundation upon which the next decades of experiments and clinical decisions will be based. The distributed and multimodal nature of these datasets further motivate the need for an all-purpose computational platform, upon which models of disease can be developed, deployed, tested, and refined.

A universal resource

All three of the grand challenges for global brain sciences represent severe methodological challenges, both technological and computational. The technological developments required for each of the challenges are non-overlapping. In contrast, regardless of the nature of the scientific questions or data modalities involved, each project will require computational capabilities including collecting, storing, exploring, analyzing, modeling, and discovering data. Although neuroscience has developed a large number of computational tools to deal with existing datasets (for example, resources in http://www.nitrc.org/), the datasets proposed here bring with them a whole suite of new challenges, including scale and complexity.

This resource would be a comprehensive computational platform, deployed in the cloud, that will provide web services for all the current “pain points” in daily neuroscience practice associated with big data. This resource will realize a new era of brain sciences, one in which the bottlenecks to discovery transition away from data collection and processing to data enriching exploring, and modeling. While science has always benefitted from standing on the shoulders of giants, this will enable science to stand on the shoulders of everyone. Today, essentially every practicing neuroscientist’s productivity is limited due to computational resources, access to data or algorithms, or struggling with determining which data and algorithms are best suited to answer the most pressing questions of our generation. This resource will create a future where those limitations will feel as archaic as fitting the data with paper and pencil feels today. For further details, see an article written by the Neuro Cloud Consortium called called “To the Cloud! A Grassroots Proposal to Accelerate Brain Science Discovery”².

Societal considerations

Each nation affords different opportunities and restrictions, owing to ethical, policy, and cultural considerations. Because these grand challenges are inherently inclusive, manifesting them will require understanding and mitigating issues that arise in cross-cultural endeavors. Indeed, addressing the vast diversity of partnerships in such an endeavor is a challenge in itself. We therefore recommend the following. First, form a ‘cultural sensitivity committee’ to consider and investigate potentially sensitive issues. Second, bolstered by their research, establish cross-cultural collaboration education materials, including written guidelines and videos, which will be recommended to all participating scientists. Third, to deepen the understanding of transnational collaborations, develop trainee exchange programs in which participating trainees will spend six months to a year working and training in a foreign country. This will also facilitate cross-cultural knowledge dissemination and fertilization. Fourth, require frequent assessments to ensure maintenance of cultural sensitivities. These assessments will feedback into the educational material and be used to modify the exchange programs.

Next steps

Crucial to the success of this endeavor is a sequence of actionable steps that the community can follow. Because we are not proposing any additional funding, realizing the eventual goals of these grand challenges will rely on marshalling existing funds. Due to the incoming leadership changes, both on national and transnational levels, quick action is of the essence. Therefore, we have taken the following steps:

We have created a webpage, http://neurox.io, containing a bibliography of reports that resulted from this conference, as well as a list of all scientific participants and observers who attended the original brainstorming meeting in April that led to this document. We will also be monitoring comments on NeuroStars (https://neurostars.org/), a community forum for neuroscience and neuroinformatics related queries, with the tag “neurostorm” for further discussion. Finally, we held an outpost at the NeuroData booth #4126 at the 2016 Society for Neuroscience conference (https://www.sfn.org/annual-meeting/neuroscience-2016) to discuss these issues further. We were encouraged by visitors who felt inspired by this idea to join the discussion, engage.