Abstract
Simulations of neural circuits are bounded in scale and speed by available computing resources, and particularly by the differences in parallelism and communication patterns between the brain and high-performance computers. SpiNNaker is a computer architecture designed to address this problem by emulating the structure and function of neural tissue, using very many low-power processors and an interprocessor communication mechanism inspired by axonal arbors. Here we demonstrate that thousand-processor SpiNNaker prototypes can simulate models of the rodent barrel system comprising 50,000 neurons and 50 million synapses. We use the PyNN library to specify models, and the intrinsic features of Python to control experimental procedures and analysis. The models reproduce known thalamocortical response transformations, exhibit known, balanced dynamics of excitation and inhibition, and show a spatiotemporal spread of activity though the superficial cortical layers. These demonstrations are a significant step toward tractable simulations of entire cortical areas on the million-processor SpiNNaker machines in development.
Highlights
The rodent somatosensory cortex is principally concerned with processing information from the whiskers, and is organized (Petersen et al, 2009)
We firstly ran parameter-sweeping simulations to verify that the model satisfies the relationship between excitatory and inhibitory balance described analytically by Brunel; secondly, we reproduced the thalamocortical response transformations observed and simulated, respectively, by Simons and Carvell (1989) and Kyriazi and Simons (1993); and we simulated a chain of five www.frontiersin.org barrel columns in parallel to show the scale of models feasible on SpiNNaker
Simulating neural circuits is a promising approach to improving our understanding of brain function
Summary
The rodent somatosensory cortex is principally concerned with processing information from the whiskers, and is organized (Petersen et al, 2009). In common with the other sensory cortices, the barrel cortex is radially organized into granular (layer 4), supragranular (layers 1–3) and infragranular layers (layers 5, 6). A barrel column is defined as a cylinder through all cortical layers, with a cross-sectional area equal to that of the granular-layer barrel. Axonal projections from the ventral posteromedial nucleus of the thalamus, which convey sensory signals from the whiskers, primarily innervate the granular layer. Signals flow within a barrel column from granular to supragranular layers and in turn to infragranular layers (Lefort et al, 2009). This relatively clear and well understood organization makes the barrel cortex a good candidate for investigations of cortical microcircuitry
Sign-in/Register to view highlights & summary 
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call