I was recently interviewed by Scope, a publication established in 2005 to showcase the work undertaken by the students in the MIT Graduate Program in Science Writing. The interview was about a research project led by Chi-Sang Poon, whose MIT group has designed a chip emulating in detail the dynamics of brain synapses, the junctions between neurons. Read the rest of this entry »
-
Silicon synapses
| December 8, 2011Comments: Leave a commentAlso posted in Brain-interfaces -
The name of the Outstein
| September 29, 2011
On Friday, September 16, 2011 Boston University chartered a vibrant new center to house research in Computational Neuroscience and Neural Technology (CompNet). In addition to some new areas of emphasis CompNet will support many aspects of the research mission of the former Department of Cognitive and Neural Systems (CNS). The closing of the CNS Department affords an opportunity to reflect on an epoch through the lens of the Outstein symbol that came to be its de facto logo. On the left, the Outstein logo. Read the rest of this entry »Comments: Leave a commentAlso posted in Computing -
Study Computational Neuroscience at Boston University
| September 21, 2011
The Computational Neuroscience PhD specialization of Boston University’s Graduate Program for Neuroscience provides students with a uniquely specialized curriculum that supplements core neuroscience coursework with advanced training in a wide array of computational methods for studying the nervous system and developing neuroscience-related technologies. Topics of study include: neural network modeling, neural dynamics, sensory, motor, and cognitive modeling, statistical modeling, sensory and motor prosthesis, brain-machine interfaces, neuroinformatics, neuromorphic engineering, and robotics. Coursework is chosen from the wide array of computational and neuroscience courses offered by the many departments and programs of the main Boston University campus and the BU School of Medicine. Students pursue their research interests in laboratories across the University and have the opportunity to combine hands on experimental research with highly sophisticated computational analysis.Comments: 1 Comment -
Computing in the Neocortex
| July 29, 2011
More than anything, the neocortex makes us human, so it has been said. Humans are better than any other living things at reading blog posts, scheduling daily activities, and filling out tax forms, among other things mundane and not. Much progress has been made localizing certain functions to certain areas of the brain, in the neocortex in particular. Other questions remain unanswered. These include how function arises from form: how do the individual neurons cooperate together to process and combine information? What is the role of each of the six neocortical layers in information processing? What impact does network connectivity have on the shape of dynamics? How do neuronal oscillations and rhythms help process information? How are different aspects of cognition coordinated? These questions are often difficult or impossible to answer from in-vivo measurements, not only because it is currently impossible to measure the state of all neurons in the brain, but also because knowledge of the state of each neuron would create an insurmountably large dataset that would be difficult to interpret. Read the rest of this entry » -
Hierarchical Temporal Memory
| May 10, 2011
Numenta is a neuromorphic technology company, founded in 2005 by Dileep George (then a graduate student at Stanford), and Jeff Hawkins (founder of Palm Computing). Numenta’s Hierarchical Temporal Memory (HTM) is a hierarchical Bayesian model based on the theory of cortical function outlined in Hawkins's book, On Intelligence. It has been used in commercial applications such as a video monitoring application, and web analytics. This post introduces Neurdon readers to HTM. Read the rest of this entry » -
Spiking Neural Networks in Python (Part 1)
| March 6, 2011
Previously I demonstrated how to represent and simulate models of individual neurons using Python. In this next series of articles I'm going to show how Python can be used to simulate connected networks of spiking neurons. Modeling and evaluating these networks can give us an insight into how populations of neurons in the brain interact to give rise to various observed phenomena. Lateral and feedback recurrent networks are of particular interest for computational simulation as attempting to analytically describe these systems quickly becomes intractable. First I'm going to start off with some of the most basic recurrent networks. Read the rest of this entry »Comments: 3 CommentsAlso posted in Computing -
Neural Modeling with Python (Part 5)
| February 22, 2011
We've seen how Python can be an effective tool for simply and efficiently implementing and simulating four different neural models using Python: leaky Integrate-and-fire, Hodgkin-Huxley, Izhikevich, and active compartments. However, other than some convenience methods and simple tricks, these examples could have been done in MATLAB without too much pain for an experienced user. So where does Python really stand out? In my first post I argued that a major advantage of Python is it's power and flexibility as a general programming language. In this case, its extensibility has allowed it to become the interface of choice for many powerful neural simulator tools. Read the rest of this entry »Comments: Leave a commentAlso posted in Computing -
Neural Modeling with Python (Part 4)
| February 11, 2011
So far I've presented three different models for simulating neural spiking dynamics. A key assumption that was made but never stated in each of these examples was that the neuron we were modeling had no defined morphology. In other words, we were looking at models that assumed the neuron was a dimensionless sphere or point. These point neurons can be very effective for studying the behavior of large-scale spiking neural networks (e.g. Izhikevich), but are impractical if you want to investigate how anatomical features of a neuron contribute to signal propagation. For this we return to the Hodgkin-Huxley model and cable theory. Read the rest of this entry »Comments: Leave a commentAlso posted in Computing -
Neural Modeling with Python (Part 3)
| February 2, 2011
So far we've looked at how to simulate a simple LIF model neuron and a complex Hodgkin-Huxley model neuron. The LIF neuron is computationally simple but physiologically implausible, while Hodgkin-Huxley gives us a very good representation of actual neural dynamics but is parameter-heavy and computationally expensive. An intriguing compromise between the two exists -- one that can generate a wide variety of observed neural spiking behavior while doing so with limited computational demand. It is called the quadratic integrate-and-fire model neuron, or simply Izhikevich neuron. Read the rest of this entry » -
Neural Modeling with Python (Part 2)
| January 26, 2011
In my last post, I demonstrated how to simulate and plot a simple leaky integrate-and-fire (LIF) neuron using Python. The LIF neuron provides a simple representation of a spiking neuron, but lacks biological plausibility especially when it comes to the actual spike generation. A neural model that does have a solid foundation in physiology is that originally proposed by Alan Hodgkin and Andrew Huxley in 1952. Read the rest of this entry »Comments: Leave a commentAlso posted in Computing