Synthesizing Symbolicism and Connectionism

Paul Thagard, Professor of Philosophy at UWaterloo and fellow member of the Centre for Theoretical Neuroscience, wrote a blog post calling the Semantic Pointer Architecture (SPA) a new synthesis in cognitive science, bringing together symbolicist and connectionist approaches. He writes:

As in connectionism, semantic pointers are patterns of firing in large neural populations, but Eliasmith has figured out how to make them also work like symbols in high-level reasoning. ... SPA provides a detailed, neurologically plausible, and mathematically rigorous account of how the dynamics of embodiment, embedding, and action work.

Read more at his blog post. Also check out more of Dr. Thagard's writing at Psychology Today!


Open Mind 2000 Page Book Released

A great collection of recent philosophy of mind/neuroscience has just been released at http://www.open-mind.net. It's equivalent to a 2000 page book, and Dr. Eliasmith is one of the many contributors. Check it out! Here's the start of the press release:

The MIND Group, run by the Mainz-based philosophy professor Thomas Metzinger, has chosen an unusual and innovative way to celebrate a special anniversary. Instead of organizing a one-off event, such as a conference, Professor Thomas Metzinger and Dr. Jennifer Windt are editing a collection of articles that document state-of-the-art research on the mind and the brain, consciousness, and the self. The collection will be freely available online at http://www.open-mind.net to anyone interested and will subsequently be published as a 2,000-page book. The project is supported by a local team of advanced undergraduate and graduate students at Johannes Gutenberg University Mainz (JGU). The contributions were written by 92 junior and senior members of the MIND Group, including internationally renowned researchers working in various areas of philosophy, psychology, and the neurosciences. The collection, which is being announced to the international press, commemorates the 20th meeting of the MIND Group and its more than 10 years of existence.


Nengo 2.0 released

The Nengo team is pleased to announce the release of Nengo 2.0, a Python library for building and simulating large-scale neural models. Nengo can create sophisticated neural simulations with sensible defaults in few lines of code, yet is extensible and flexible enough to use spiking and non-spiking neuron types in the same model, get input directly from hardware, drive robots, and simulate models on diverse computing resources.

This is the first release of Nengo that is implemented entirely in Python, and integrates well with tools like Matplotlib and IPython. Currently, it is designed to be used programmatically, but a browser-based graphical interface is under active development.

Features

Nengo has support for models using the following neuron types, which can be combined in the same model.

  • Rectified linear
  • Sigmoid
  • Leaky integrate-and-fire (spiking and non-spiking)
  • Adaptive leaky integrate-and-fire (spiking and non-spiking)
  • Izhikevich
  • Direct mode (in which mathematical functions are computed directly, rather than approximated from neural activity)

Nengo does not require online learning in the form of synaptic weight changes; however, learning rules do exist for situations when the objective function is not known ahead of time. Nengo has support for the following learning rules:

  • PES rule (minimizes an error signal)
  • BCM rule
  • Oja rule

See the documentation for more details and example usage.

Changes in Nengo 2.0

Nengo 2.0 is a completely new code base that implements all commonly used features of Nengo 1.4.

Links


Postdoc Position in Large-scale Brain Modeling

Centre for Theoretical Neuroscience, University of Waterloo

We are recruiting for a theoretical neuroscientist interested in building large-scale functional brain models. The successful candidate will become an integral part of the team that has built the world's largest functional brain model, Spaun. This work was published in Science, and received coverage around the world on CNN, BBC, CBC, and hundreds of other outlets. We are building the next generation of computational platforms to run such models. This position will focus on using supercomputers (e.g., IBM BlueGene) to run models written in the Nengo environment. Nengo is a general neural simulation environment that, in addition to Spaun, has been used to simulate adaptive arm controllers, state-of-the-art vision systems, natural language processors, reinforcement learners, and novel robotic controllers among many other projects. Postdocs will be encouraged to pursue their research program with these state-of-the-art tools.

Qualifications: PhD in theoretical neuroscience or related field; experience with MPI, BlueGene, or other HPC environments desirable; python experience an asset.

Start date: As soon as a suitable candidate is found

Submit: CV, statement of research interests, three references

Duration: 1 year renewable for a 2nd year

Lab: CNRG http://compneuro.uwaterloo.ca/

Contact: celiasmith@uwaterloo.ca (Chris Eliasmith)

Funding for the position will be provided by TalentEdge and a corporate sponsor.


Member of the College of the Royal Society

Chris Eliasmith has just been named an inaugural member of the Royal Society of Canada’s College of New Scholars. The inaugural cohort is 92 top faculty members from across Canada. Chris is a pioneering figure in the rapidly developing field of theoretical neuroscience. His most visible contribution has been in the development of Spaun, the world’s largest functional model of the brain. Spaun has garnered wide acclaim, both in the academy and in the popular press. Behind Spaun’s success is a deeply influential body of work implementing an innovative approach to the problem of how brains encode meaning.

College Member Citations 2014

Official press release - Sept. 16, 2014

Chris is a professor in the Philosophy department and Systems Design Engineering. He is also the director of the Centre for Theoretical Neuroscience and a Canada Research Chair in Theoretical Neuroscience.