Neuroscience  

Neuroscientists have used a complex digital brain simulation of a rat brain that combines functional connectivity analysis and network modeling to investigate synaptic mechanisms of memory formation and pattern completion in hippocampal CA3 network.

Neuroscientists have used advanced brain simulations to unravel the complex synaptic mechanisms of pattern completion in hippocampus of the brain. The researchers' findings suggest that the rules of synaptic connectivity between CA3 pyramidal cells in the hippocampus contribute to the remarkable efficiency of pattern completion.

The hippocampal CA3 region is known to play a key role in learning and memory. One of the most remarkable properties of the network is its ability to retrieve previously stored memories from incomplete or degraded versions, a phenomenon that is widely known as pattern completion.

Related articles

• Researchers Looking To Map The Brain Find It Even More Complex Than They Thought 
• Four Genes Appear To Affect Human Memory 
• Human Memory Implants May Soon Be Tested 

It is widely accepted that the synapses between CA3 pyramidal cells, the recurrent CA3–CA3 synapses, play a key role in pattern completion, but how this exactly works is still a mystery.

"The results provide a nice demonstration of how the Hopfield quote “build it, and you understand it” can be successfully applied to important questions in neuroscience."

Now, in  recent research published in the journal Science, neuroscientists Jose Guzman, Alois Schlögl, Michael Frotscher, and Peter Jonas have investigated these mechanisms by combining functional connectivity analysis and network modeling. Their findings suggest that the rules of synaptic connectivity between CA3 pyramidal cells contribute to the remarkable efficiency of pattern completion.

Previous theories of the hippocampal formation often depicted the region as a network of highly interconnected cells. The neuroscientists from the Institute of Science and Technology Austria (IST Austria) tested this hypothesis using a technique that allows monitoring the connection between electrical signals in up to eight neurons at the same time.

They made several highly surprising observations. First, they found that connectivity was sparse, with an average connection probability of approximately 1%. This massively challenges the dogma of a network of highly connected cells. Even more surprisingly, they discovered that connectivity in the network is not random, but exhibits connectivity motifs that occur much more frequently than expected for a random network. The structure of the hippocampal CA3 network may be somewhat reminiscent of a “small world” architecture as found in social networks. Finally, the authors revealed that synaptic connections between two cells are mediated by only one or two synaptic contacts. This is also remarkable because much higher numbers have been found for excitatory synaptic connections in the neocortex.

What did these 'rules' man for memory formation? To address this question, Jonas,who leads the cellular neuroscience group and his team built a model of the CA3 network that incorporates many of these new experimental observations. In contrast to many previous studies, the network was implemented in full size, so that all 330,000 CA3 neurons of the rat hippocampus were simulated.

They found that a full-size network model with realistic connectivity of 1% was indeed able to perform the network computation of pattern completion. Also, they discovered that the presence of connectivity motives increased, under certain conditions, the performance of the network. Finally, the design of synaptic connections based on one or two synaptic contacts also seems useful for pattern completion, apparently because it minimizes redundancy in the flow of information in the network.

For the researchers, both macro- and microconnectivity facilitate pattern completion in the CA3 cell network. “The results provide a nice demonstration of how the Hopfield quote “build it, and you understand it” can be successfully applied to important questions in neuroscience,” states Jonas.

SOURCE  IST Austria

By  33rd Square

Embed

Brain Simulation  

The restructured work of the European Human Brain Project was recently showcased to the public, with the research to date being made available to the wider scientific community. While the project aims of fully simulating a human brain have been scaled back, it continues to represent an important focus for neuroscience and brain-inspired computation. 

Understanding the human brain is one of the greatest challenges facing 21st century science. If we can rise to the challenge, we can gain profound insights into what makes us human, develop new treatments for brain disease and build revolutionary new computing technologies

The Human Brain Project (HBP) is a EU-funded initiative to create and operate a research infrastructure, to help advance neuroscience, medicine and computing. The 10-year Project began in 2013 and involves leading scientists at more than 100 universities and research centres across Europe, China, Japan the USA, including the Allen Brain Institute.

The project is centred on six platforms including:

• The Neuroinformatics Platform: registration, search, analysis of neuroscience data.
• The Brain Simulation Platform: reconstruction and simulation of the brain.
• The High Performance Computing Platform: computing and storage facilities to run complex simulations and analyse large data sets.
• The Medical Informatics Platform: searching of real patient data to understand similarities and differences among brain diseases.
• The Neuromorphic Computing Platform: access to computer systems that emulate brain microcircuits and apply principles similar to the way the brain learns.
• The Neurorobotics Platform: testing of virtual models of the brain by connecting them to simulated robot bodies and environments.

The project also conducts related research and theoretical studies on brain structure and function, and looks at the ethical and societal implications such work.

After the project was started in 2013 with Henry Markram in charge of a whopping $1.3 billion Euro budget, there were some organizational issues. The project is now more focused on data tools and software that are not exclusively aimed at simulating the brain. It has also undergone a large scale review and restructuring.

Despite the difficulties, the Human Brain Project has announced the release of initial versions of its six Information and Communications Technology (ICT) platforms to users outside the Project. These platforms are designed to help the scientific community to accelerate progress in neuroscience, medicine, and computing.

The platforms released this week consist of prototype hardware, software tools, databases and programming interfaces, which will be refined and expanded in a collaborative approach with users.

The first version of the Brain Simulation Platform is being built for release at the end of the Ramp-Up Phase. This effort builds on previous work in the Blue Brain Project.

Related articles

• Mini Human Brain Grown In Petri Dish 
• Synaptic Transistors Point To Computer Hardware That Learns 
• Researchers Develop Neuromorphic Chip With Cognitive Abilities 

The brain simulation will provide researchers with an internet accessible Brain Simulation “Cockpit”, allowing them to perform experiments in silico.  This will involve investigating the relationships between different levels of biological organisation in the healthy and the diseased brain and preparing the way for the re-implementation of neuronal circuits in neuromorphic hardware.

The neuromorphic computing platform intends to release two different neuromorphic computing prototypes as part of the ramp-up. The two systems being constructed are key computing resources for the HBP. One cluster is features 128 Intel i7-2600 cores with a total Linpack benchmark performance of 3.2 Teraflops for computation.

The cluster is being optimised for efficient communication with the other systems and includes 32 I7-nodes dedicated to operating the highspeed links to individual wafers for closed-loop virtual robotics experiments. Users will access the Platform via a single  point of access web portal shared among  all the platforms.

The Neurorobotics Platform will allow researchers to design and run simple experiments in cognitive neuroscience using simulated robots linked to simplified versions of the project's brain models.

Karlheinz Meier, Co-leader of the Neuromorphic Platform, said, “The HBP invites scientists everywhere to work with our prototype Platforms and give us their feedback. This will help us improve their functionality and ease of use, and hence their value to society.”

The platforms are designed to help researchers to advance faster and more efficiently, by sharing data and results, and exploiting advanced capabilities. The platforms should enable closer collaboration between scientists to create more detailed models and simulations of the brain.

Since the US launched its own BRAIN Initiative, the restructuring of the European initiative has also been repositioned to avoid research overlap and hopefully will accelerate this important area of research.

The videos below provide further detail on some of the platform project work undertaken for the HBP so far.

SOURCE  Human Brain Project

By 33rd Square

Embed

 Mind Uploading  

Researchers from the OpenWorm project have successfully mapped all the connections between the a roundworm's 302 neurons simulated them in software that controls a small robot.

For those interested in the Singularity, it has long been understood that there are no physical laws limiting us from taking our connectome, or mind file and duplicating it in another substrate, such as a robot. This technical challenge has just had a proof of concept made by uploading the brain of a roundworm into a robot.

Researchers from the OpenWorm project have successfully mapped all the connections between Caenorhabditis elegans’ 302 neurons and managed to simulate them in software.

Scientists published the first map of the worm's synaptic connections, or connectome, in 1986 and a refined draft in 2006.

The brain of the roundworm has just 1000 cells, of which only 302 are neurons with 7,000 connections or synapses.

In comparison, the human brain has an estimated 86 billion neurons, and over 10¹⁴ synapses.

"We believe brain research must accelerate, we are taking matters into our own hands. If we cannot build a computer model of a worm, the most studied organism in all of biology, we don’t stand a chance to understand something as complex as the human brain."

Experiments like this help accomplish the eventual uploading of more complex organisms, leading towards the post-human goal of uploading a human consciousness. More importantly, this research is helping us better understand how brains work as a system and could lead to more effective treatment, or maybe even a cure, for diseases such as Alzheimer’s.

The researchers do note that this is just a first step. The brain simulation still isn’t 100% exact, as the researchers had to simplify the process that triggers an artificial neuron to fire.

Related articles Researchers Develop Protein to Better Track Brain Activity  Simulating Roundworms May Help Us Understand of the Basic Principles of the Brain  David Dalrymple Update On Project NEMALOAD

“Because we believe brain research must accelerate, we are taking matters into our own hands. If we cannot build a computer model of a worm, the most studied organism in all of biology, we don’t stand a chance to understand something as complex as the human brain. We must crawl before we can walk!” OpenWorm explains on their website.

On the robot, sensors replace the biological inputs that the animal's brain would receive. Motor neuron simulators in the software, running on an on-board Raspberry Pi computer, then drive the robot’s motors as if they were right and left groups of muscles.

So, without being explicitly programmed to do so, the robot moved back and forth and avoided objects using only a simulated brain. The robot behaves much like a real worm would, given similar sensory inputs. Activating the front sensor stops forward movement and, touching the front and rear sensors makes the robot move forward and back.

You might even say, the robot 'thinks' it is a worm.

The OpenWorm robot seems to show that a stimulated digital brain might behave like a biological brain does.  Conceivably, if we eventually map a human brain with similar resolution and supply it with stimulation in a virtual or physical environment, we would produce a human copy in robotic substrate.




SOURCE  iProgrammer

By 33rd Square

Embed

 Neuroscience  

In a recent special seminar at the Sagol School of Neuroscience at Tel Aviv University, Dr Henry Markram, head of the HBP discussed the project and the principles they are using to uncover the secrets of the brain.

The European Human Brain Project (HBP) is a billion dollar mega-project with dedicated respectively to Neuroinformatics, Brain Simulation, High Performance Computing, Medical Informatics, Neuromorphic Computing and Neurorobotics components.  In fact, the project recently expanded with new participants.

"If you can't measure it, you are going to have to predict it."

In a recent special seminar at the Sagol School of Neuroscience at Tel Aviv University, Dr Henry Markram, head of the HBP discussed the project and the principles they are using to uncover the secrets of the brain.

For Markram, understanding the human brain can be broken down into the following problems:

1. Translating the knowledge of the brain into a benefit for society.  Markram discusses how despite our growing knowledge of the brain, the development of medicines for the brain is slowing, and pharmaceutical companies are abandoning the development of neuroscience due to the costs.

2. It is impossible to experimentally map the brain.  Markram postulates that the 55 different types of neurons in the brain leading to trillions of connections.  According to his calculations, it would take over 2000 man-years to map a piece of the brain the size of a pinhead at a cost of two billion dollars.   For Markram, "You are never going to get enough data to model the brain."  His solution to this is called predictive neuroscience.  "If you can't measure it, you are going to have to predict it," he says.

Related articles Ken Hayworth In Depth on Uploading and Brain Preservation  Human Brain Project and Graphene Win Billion Euro FET Flagship Funding  The Human Brain Project Goals

Neuroscience has to apply the methods of physics to simulate the brain according to Markram.  In the case of the connectome, the Human Brain Project is taking the predictive route, in contrast to the American BRAIN Initiative's measurement plan.

One of the key findings of Markram's team that is helping develop their algorithms for brain simulation is that neurons do not connect to other neurons specifically, but to other regions of the brain.  For the areas they do not yet understand, they are creating 'black boxes' for areas of further study.

3. We don't know how to translate our knowledge from animals into humans.  In the Human Brain Project they are working to build up a data ladder of how gene expression operates between mice and humans.  They will use this to infer and predict what is going on in the human brain.  This is necessary, as the neuroscience work is primarily done on animals, and human cells cannot be explored directly (yet).

4. Obtain an integrative, multi-scale view of the brain under any condition.  This is where the essential brain simulation of the HBP takes prominence.  In order to run a simulation of a human brain, there is a research initiative to develop the computer technology necessary to run such simulations.

In a full (or partial) brain simulation scientists will be able to test environmental factors, chemical interactions and other potential areas on a brain.

5. Symptom-based, subjective classification of brain disease.  Now, it is unknown how inter-related and interconnected brain diseases are.  Markram suggests that diseases like Alzheimer's and autism may be related.  Through a better understanding of the brain, and increased research collaboration such an understanding may be made.

6. Brain to technology.  If we can develop technologies based on an understanding of the brain, there is a great potential for business.  Markram admits that this was a key factor in obtaining funding for the HBP from the European Commision.  Neuromorphic chips development is one area the project is focused on.

7.  Take Society on a Voyage of the Brain.  The team will be constructing a building, a museum of sorts, based on the research progress at the HBP.  The building will be located in Lausanne, Switzerland, but will be partnered with institutions world-wide.

During the question section of the presentation (at ~52:00 mark), the question of consciousness is re-engaged.  Markram, when answering if his project will uncover if consciousness is an emergent property or not, answers that he believes the HBP will learn the neural correlates of behavior and decision-making. Essentially Markram predicts that the project will find the neural states associated with particular behaviors.

"My personal view is that consciousness is a critical mass of causal interactions," says Markram. "What happens when there is a critical mass is another question."


SOURCE  Biological Sciences, NeuroScience

By 33rd Square

Embed

 Neuroscience  

For the first time, neuroscientists have systematically mapped the white matter "scaffold" of the human brain, the critical communications network that supports brain function.

For the first time, neuroscientists have systematically identified the white matter "scaffold" of the human brain, the critical communications network that supports brain function.

Their work, published in the open-source journal Frontiers in Human Neuroscience, has major implications for understanding brain injury and disease. By detailing the connections that have the greatest influence over all other connections, the researchers offer not only a landmark first map of core white matter pathways, but also show which connections may be most vulnerable to damage.

"We coined the term white matter 'scaffold' because this network defines the information architecture which supports brain function," said senior author John Darrell Van Horn of the USC Institute for Neuroimaging and Informatics and the Laboratory of Neuro Imaging at USC.

"While all connections in the brain have their importance, there are particular links which are the major players," Van Horn said.

Graphical representation of human brain connectivity scaffold.
Image Source -  USC Institute for Neuroimaging and Informatics

Using MRI data from a large sample of 110 individuals, lead author Andrei Irimia, also of the USC Institute for Neuroimaging and Informatics, and Van Horn systematically simulated the effects of damaging each white matter pathway.

Related articles Breakthrough Technique Gives Unprecedented View Inside The Brain  Brain Surgeons Find The Neurological Basis of Human Speech  Researchers Demonstrate Matrix-Style Learning

They found that the most important areas of white and gray matter don't always overlap. Gray matter is the outermost portion of the brain containing the neurons where information is processed and stored. Past research has identified the areas of gray matter that are disproportionately affected by injury.

But the current study shows that the most vulnerable white matter pathways – the core "scaffolding" – are not necessarily just the connections among the most vulnerable areas of gray matter, helping explain why seemingly small brain injuries may have such devastating effects.

"Sometimes people experience a head injury which seems severe but from which they are able to recover. On the other hand, some people have a seemingly small injury which has very serious clinical effects," says Van Horn, associate professor of neurology at the Keck School of Medicine of USC. "This research helps us to better address clinical challenges such as traumatic brain injury and to determine what makes certain white matter pathways particularly vulnerable and important."

The researchers compare their brain imaging analysis to models used for understanding social networks. To get a sense of how the brain works, Irimia and Van Horn did not focus only on the most prominent gray matter nodes – which are akin to the individuals within a social network. Nor did they merely look at how connected those nodes are.

Image Source -  USC Institute for Neuroimaging and Informatics

Rather, they also examined the strength of these white matter connections, i.e. which connections seemed to be particularly sensitive or to cause the greatest repercussions across the network when removed. Those connections which created the greatest changes form the network "scaffold."

"Just as when you remove the internet connection to your computer you won't get your email anymore, there are white matter pathways which result in large scale communication failures in the brain when damaged," Van Horn said.

When white matter pathways are damaged, brain areas served by those connections may wither or have their functions taken over by other brain regions, the researchers explain. Irimia and Van Horn's research on core white matter connections is part of a worldwide scientific effort to map the 100 billion neurons and 1,000 trillion connections in the living human brain, led by the Human Connectome Project and the Laboratory of Neuro Imaging at USC.

Irimia notes that, "these new findings on the brain's network scaffold help inform clinicians about the neurological impacts of brain diseases such as multiple sclerosis, Alzheimer's disease, as well as major brain injury. Sports organizations, the military and the US government have considerable interest in understanding brain disorders, and our work contributes to that of other scientists in this exciting era for brain research."


SOURCE  University of Southern California

By 33rd Square

Subscribe to 33rd Square

 Human Brain Project  

The multi-billion Euro Human Brain Project, co-funded by the European Union, plans to use supercomputers to model the human brain and then use the research to simulate drugs and treatments for diseases, create learning artificial intelligence and much more.

Members of the Human Brain Project officially kicked off their decade-long global project this week. The most ambitious neuroscience project in the world, with multiple sub-projects, the aim is to find a deeper and more meaningful understanding of  how the human brain operates.

The Human Brain Project (HBP) comprises 135 research institutions throughout Europe and is coordinated through the Ecole polytechnique fédérale de Lausanne (EFPL). At the project's launch,  neuroscientists, doctors, computer scientists, and roboticists will begin to refine the project in across the research platforms including neuroinformatics, brain simulation, high-performance computing, medical informatics, neuromorphic computing and neurorobotics, each composed of technological tools and methods to ensure the project’s objectives will be met. So far 13 sub-projects have been established.

Related articles US Set To Enter The Brain Race With The Brain Activity Map Project  Human Brain Project and Graphene Win Billion Euro FET Flagship Funding  BRAIN Project Infographic Released

The researchers will set up and test the platforms over the next 30 months and in 2016, these platforms should be ready for testing by the Human Brain Project scientists and researchers from around the world.

As the extended video above demonstrates, the HBP researchers will have to manage enormous amounts of data. The mission of the neuroinformatics platform will be to extract the maximum amount of information possible from these sources and integrate it into a cartography that encompasses all the brain’s organizational levels, from the individual cell all the way up to the entire brain.

Neurorobotics, this research platform will focus on integrating neural network simulations into robots (including highly accurate virtual ones), who will benefit from new aptitudes such as learning abilities or resiliency.

Another important component will be to create neuro-inspired technologies. Neuromorphic chips that can imitate how networks of neurons function and take learn will be developed and expanded.

All of the sub-elements will feed to the main project, so coordinating everything is a huge management task —in terms of the people, resources and the massive amounts of data that will be collected and analyzed.  

The Human Brain Project hopes the results of the ten year effort will be information and knowledge that can be transferred into the development of new medical and information technologies.

The U.S. National Institutes of Health in May also announced an attempt to map the brain, the BRAIN Initiative.


SOURCE  Human Brain Project

By 33rd Square

Subscribe to 33rd Square

 
Artificial Intelligence

Recently Nikola Danaylov of Singularity 1 on 1 interviewed Chris Eliasmith and discussed a variety of topics including the story behind his desire to create the breakthrough brain simulation, SPAUN.  Eliasmith is working on the Semantic Pointer Architecture as a new model of the brain and is using SPAUN to explore the idea, and will also soon be releasing a book, How To Build A Brain, that will outline the concept.

In November last year, the system SPAUN was announced as major breakthrough in brain simulation. Created by Chris Eliasmith the director of the Centre for Theoretical Neuroscience at University of Waterloo SPAUN,which stands for Semantic Pointer Architecture Unified Network, is a computer model that can recognize numbers, remember them, figure out numeric sequences, and even write them down with a robotic arm.

Recently Nikola Danaylov of Singularity 1 on 1 interviewed Eliasmith and discussed a variety of topics such as: the story behind his desire to create a whole brain simulation and the hardware requirements to run it; whether SPAUN has thoughts and feelings and how would we know if it did; the ethical issues behind creating a brain-in-a-vat artificial intelligence; the relationship between philosophy and engineering; his upcoming book How to Build a Brain; Eliasmith’s thoughts on Deep Blue, Watson, Blue Brain, SyNAPSE and Ray Kurzweil‘s How to Create a Mind: The Secret of Human Thought Revealed; and his take on the technological singularity.

Eliasmith's work on SPAUN is pointing to the need for embodiment to generate an artificial brain.  "Ultimately, I suspect you are going to have a difficult time of actually doing this well if you do not put it into a body.  And so creating a brain-in-a-vat might just really end in the thought experiment that it is.  You might not have a successful  adaptive, effective brain if you do not put it in a body." Eliasmith adds, "So, robotics is definitely something that is important for us in trying to make sure these models are sufficiently robust and sort of...animal like, to explain the kinds of behavior that we are interested in."

When comparing SPAUN to other systems like Watson, Eliasmith suggests that the SPAUN approach is different in that it exhibits a variety of behaviors, and more importantly it cannot switch between these tasks. This essentially means that Watson is not as brain-like as SPAUN. Watson and Deep Blue are highly specialized, and the SPAUN model is not.  As a path to Artificial General Intelligence, Eliasmith's approach is therefore potentially very important.
The basis of SPAUN is Eliasmith's Semantic Pointer Architecture hypothesis. Briefly, the semantic pointer hypothesis states:

Higher-level cognitive functions in biological systems are made possible by semantic pointers. Semantic pointers are neural representations that carry partial semantic content and are composable into the representational structures necessary to support complex cognition.

The term 'semantic pointer' was chosen because the representations in the architecture are like 'pointers' in computer science (insofar as they can be 'dereferenced' to access large amounts of information which they do not directly carry). However, they are 'semantic' (unlike pointers in computer science) because these representations capture relations in a semantic vector space in virtue of their distances to one another, as typically envisaged by connectionists.

SPAUN can download at nengo.ca.

According to  Eliasmith “We Have Not Yet Learned What The Brain Has To Teach Us!”

Eliasmith is also head of the Computational Neuroscience Research Group (CNRG) at the Centre for Theoretical Neuroscience at Waterloo. This group is developing and applying a general framework for modeling the function of complex neural systems (the Neural Engineering Framework or NEF). The NEF is grounded in the principles of signal processing, control theory, statistical inference, and good engineering design. It provides a rational and robust strategy for simulating and evaluating the function of a wide variety of specific biological neural circuits. Members of the group applied the NEF to projects characterizing sensory processing, motor control, and cognitive function.

Some members of the CNRG have recently begun developing applications of related principles to problems in machine intelligence. Specifically, they are constructing novel methods for automatic text understanding that can be used to support classification and clustering.




SOURCE  Singularity Weblog

By 33rd Square

Subscribe to 33rd Square