This week was our second Dangerous Ideas social, and we saved the stream of it on UStream. We’ve created a round table where we get together each week and talk about some cool tech stuff related or unrelated to what we are working on. You can tune in on Thursday’s at 4pm and watch live, or you can check out the recorded stream. We also take note of the links we create on our wiki.
Here at the Whole Brain Project, we are constantly looking for talented folks in the open source community who are interested in directing their creative talents towards the intersection of computer science and neuroscience. Makes for a great volunteer internship on a resume. Think you’ve got what it takes? Fill out our volunteer sign up form and find out! We’ll get back to you shortly.
Our Booth at the Society for Neuroscience annual meeting is set up and running. You can find us at booth number 4033, which is across from posters FFF. Come check us out!
A new and improved version of the Whole Brain Catalog was released Monday, May 10. The program was restructured so that version 0.7.7 offers faster start up times and more stability than ever before. The 3D library was also upgraded to the recently released Lightweight Java Game Library (LWJGL) 2.3 so that 3D visualization is now smoother and more secure. The Catalog’s virtual brain model was further enhanced by the addition of new 3D meshes of the eyes, optic tracts, optic chiasm, and skull.
Version 0.7.7 features exciting new capabilities and data that have been integrated into the Whole Brain Catalog. Additional simulated network models were added to expand the Catalog’s existing collection. An experimental version of simulation support was introduced in version 0.7.7, which allows users to load and run NeuroML simulations through the Whole Brain Catalog client. The experimental version of the Catalog’s volume rendering tool, which was featured in the previous release of version 0.7.6.1, was expanded so that it can now be run on Windows operating systems.
No computer can compete with the sheer computational speed and dynamic abilities of the human brain. Yet. A team of scientists from Michigan Technological University and the National Institute for Materials Science in Japan recently developed a new parallel processing chip has demonstrated brain-like computational power at speeds that could rival the fastest supercomputers.
The processor’s lightning-fast speed is the result of its organic structure. It is composed of organic molecules arranged so that they can simultaneously exchange information and work together toward one solution. This design is a major improvement from the serial processors that have been used in modern computers for the last 60 years. The parallel processor can not only compute at a much faster speed than serial processors, but can also perform more operations at the same time.
Many scientists agree that the brain is the fastest computer in existence because of the extensive connectivity and integration that occurs between neurons. The developers of the new organic processor argue that since it can integrate multiple signals like the brain can, it will be able to solve complex problems that today’s parallel processors are unable to compute. They have already used it to simulate two natural phenomena (heat diffusion and cancer proliferation), demonstrating its complex computational ability. The next step is to expand this technology into larger components for use in a lot of different common applications. We can’t even imagine the effects that parallel processing could have on our everyday lives. These “smart” chips could conceivably be added to almost every common technology to make them more human-like. The possibilities seem limitless.
HP recently released exciting new advancements that surfaced during the research and development of a novel circuit element called the memristor. Researchers from the HP labs first proved the existence of the memristor in April of 2008, at which time they began development of the physical prototype. Originally, the memristor generated public excitement because it opened up the possibility of more energy-efficient computers by eliminating the energy-consuming boot-up process. Unlike classic resistors and transistors, the memristor is about to maintain its electrical state even after the computer has been turned off, thereby maintaining memory of its behavior.
In an article published in Nature April 8, 2010, HP scientists reported that memristors may be more than energy-efficient processor elements. New arrangements and architectures involving memristors have demonstrated their logical capabilities, such as memory storage and Boolean processing. Moreover, the most recent prototypes are much faster than the transistors that are currently used. With its increased processing speed, enhanced capabilities, and conservative properties, the memristor is sure to appear in a slew of future electronic devices from laptops to hand-held videogame devices.
In addition, the public is abuzz from mention of the possibility that memristors may be able to perform some level of cognitive function. HP scientists claimed that memristor behavior mimics the biological behavior of neurons in the brain, and can therefore be used to construct an artificial device that can execute cognitive performance. Building a functional brain out of memoristors and other electrical elements is a long shot at best, but it is an exciting idea nonetheless.
Scientists at the Massachusetts Institute of Technology are currently developing a dynamic computer program that will automatically trace neuronal projections to produce an all-inclusive, high-resolution connectome. Today, many researchers continue to trace axonal processes by hand through a series of brain slice images to determine where the connections occur. This process is meticulous and slow, and could drag out the process of connectivity mapping over centuries. However, the new tool that researchers are working on at MIT should computerize and significantly speed up the tracing process so that the creation of a comprehensive connectome may be achieved within a more reasonable time period.
Viren Jain and postdoctoral associate Srinivas Turaga of the Seung Lab at MIT are using automated machine learning to develop their highly intuitive computer program that will automatically track neural connections through segmented brain images. Jain and Turaga feed the computer a series of segmentations through which connections have already been traced as examples. This method allows the computer to come up with its own algorithm and to correct the algorithm as necessary after exposure to new data. Therefore, when the computer is given a new set of segmentations, it will be able to trace a connection without further human input.
Computerized mapping of neuronal connections is a major step in the area of connectivity mapping, but more advancements in processing speed and accuracy need to be made before a full connectome can be realized. For more information about Jain’s self-teaching program, check out “Mapping the Brain” on MIT News.
The Whole Brain Catalog participated in the San Diego Science Festival’s second annual Expo Day on Saturday, March 27, at Petco Park. The Science Festival is geared towards families and young students to inspire the next generation of scientists, mathematicians, and engineers. Expo Day, the final day of the Science Festival, is a free event that features booths and interactive activities from leading science, technology, engineering, and mathematics organizations throughout San Diego. Check out this article from the UCSD’s campus-wide newsletter for more information about the Science Festival and Expo Day.
The National Biomedical Computation Resource (NBCR), the Community Cyberinfrastructure for Advanced Marine Microbial Ecology Research and Analysis (CAMERA) and the Neuroscience Information Framework (NIF) joined the Whole Brain Catalog to represent the Center for Research in Biological Systems at Expo Day. The beautiful plasma screen continuously running Drew Berry’s animation of neural structures and events was a big hit, drawing and maintaining a crowd throughout most of the day. Sarah Maynard and Chris Aprea interacted with festival-goers of all ages, explaining aspects of neuroscience and the basic concepts of the Whole Brain Catalog. Many of the children were even able to interact with the computer on brain maps (http://brainmaps.org/), zooming in and out of the brain images. Throughout the day, many students showed interest in joining the lab for summer internships with CRBS. Other CRBS researches that participated in Expo Day included Wilfred Li, Teri Simas, Guy Perkins, and Shulei Shu.
We hope our participation in Expo Day succeeded in broadening our reputation within the community and inspiring young minds to explore future career paths in science and technology. We certainly had a lot of fun!
Mapping neural connectivity is a major topic of interest and research in neuroscience. A connectivity map, or “connectome,” would be an invaluable tool for neuroscientists by providing a physiological backbone on which new theories and ideas could be based. Moreover, it could serve as the basis for countless technological advancements in the fields of computer science and neuroinformatics. However, this task is considerably more complicated than it may sound. Each of the brain’s 100 billion neurons forms anywhere from 1,000 to 10,000 synapses with other neurons. Mapping this enormous volume of connections is a truly daunting task. In fact, it has been estimated that generating a high resolution connectome could take longer than the average lifetime using the technology that exists today. Therefore, new tactics must be employed to make significant progress in connectivity mapping.
Giorgio Ascoli, founding Director of the Center for Neural Informatics, Structures, & Plasticity at George Mason University’s Krasnow Institute for Advanced Study, has taken an alternative approach in his work on neural connectivity. In “The Coming of Age of the Hippocampome,” Ascoli describes his statistical method of mapping circuits at the neuron level within a specific brain region or structure. Creating such a high-resolution map is still a lofty goal, but limiting the neuron types and applying statistical algorithms significantly simplifies the issue. Ascoli is currently working on the circuitry map of the rodent hippocampus, which he has termed the “hippocampome.”
The possibility of a hippocampome is an exciting development in neural connectivity. Such a map could help to define and further understand neuron function and clarify how general circuitry works in the brain. However, the hippocampus is one of the most widely studied brain structures, and Ascoli’s method may not be applied as successfully to mapping other, less documented areas of the brain.