SpineML The Spiking Neural Mark-up Language

Models

All SpineML models are contained within the main SpineML repository alongside the schemas.

The examples below are all from the repository and are classified as either SpineCreator examples or Toolchain examples. SpineCreator examples have a project file, toolchain examples do not and are pure SpineML. Toolchain examples can however be imported into SpineCreator and are examples pass validation against the schema. To test validation you can use a tool such as xmllint.

SpineCreator Example Models

Gurney, Prescott and Redgrave Basal Ganglia model

This is a rate coded model of the Basal Ganglia. Details can be found in the following paper

A computational model of action selection in the basal ganglia. II. Analysis and simulation of behaviour K. Gurney, T. J. Prescott, P. Redgrave

Link to SpineML Model

Gurney, Prescott and Redgrave Basal Ganglia model, spiking implementation

This is an integrate and fire neuron spiking version of the rate coded model of the Basal Ganglia found above.

Link to SpineML Model

Striatal Microcircuit Model

This is a SpineML implementation of a Striatal microcircuit model. Details can be found in the paper:

Dopamine-modulated dynamic cell assemblies generated by the GABAergic striatal microcircuit Mark D. Humphries, , Ric Wood, Kevin Gurney

Link to SpineML Model

Angular Velocity Detector Unit model

This is a model of the angular velocity detection system in the insect from the following paper.

Cope A., Sabo C., Gurney K., Vasilaki E., and Marshall J. A. R. (2016), “A Model for an Angular Velocity-Tuned Motion Detector Accounting for Deviations in the Corridor-Centering Response of the Bee,” PLoS Computational Biology. doi:10.1371/journal.pcbi.1004887.s001

A SpineCreator viewable model is available below.

Link to SpineML Model

If you wish to execute the model and reporoduce the results from the paper then you should also download the complete repository and follow the detailed instructions.

Link to SpineML Model

SpineML toolchain examples

These example consist of pure SpineML with no SpineCreator project files or metadata. They can be imported into SpineCreator, but do not have 2D or 3D layout data.

Brette Benchmark Model

Components

Name Description XML PNG
LIF Leaky Integrate and Fire Neuron Body /public/images/Xml_icon.png Png icon.png
FixedWeight Fixed Weight Synaptic Update /public/images/Xml_icon.png Png icon.png
Curr_exp Exponentially Decaying Post-Synaptic Current /public/images/Xml_icon.png Png icon.png

Network

A network of two populations of Excitatory and Inhibitory neurons. Based upon the model described in; Romain Brette et al. “Simulation of networks of spiking neurons: A review of tools and strategies”, 2007.

High Level Network Layer Model

/public/images/Xml_icon.png

High Level Network Layer Model (split version with maximum population size 100)

/public/images/Xml_icon.png (1.9MB)

Experiment

Experiment file which runs the Brett benchmark for a period of 1 second recording all spike and voltage values.

Experiment Layer Model

/public/images/Xml_icon.png

Brette Benchmark Model (Using PyNN Neurons)

This is the same as above however the synapse model is integrated into the neuron body. The Post-Synapse component acts as a pass-through.

Components

Name Description XML PNG
IF_curr_exp Leaky Integrate and Fire Neuron Body with Exponentially Decaying Post-Synaptic Current /public/images/Xml_icon.png Png icon.png
PyNN_WeightUpdate A Fixed Weight Synaptic Update /public/images/Xml_icon.png Png icon.png
PyNN_PostSynapse A Pass-through Post-Synapse (Relays Impulses to the Post-Synaptic Neuron) /public/images/Xml_icon.png Png icon.png

Network

A network of two populations of Excitatory and Inhibitory neurons. Based upon the model described in; Romain Brette et al. “Simulation of networks of spiking neurons: A review of tools and strategies”, 2007.. Replicates the standard Brett Benchmark model however the PyNN_PostSynpases redirects any impulse events to the post synaptic neuron body which models the dynamics. PyNN uses separate synaptic currents for excitatory and inhibitory synapses and as such negative synaptic weights are not required.

High Level Network Layer Model

/public/images/Xml_icon.png

Experiment

Experiment file which runs the Brett benchmark for a period of 1 second recording all spike and voltage values.

Experiment Layer Model

/public/images/Xml_icon.png