Learning spatial transformations through STDP (Davison, Frégnac 2006)

 Download zip file   Auto-launch 
Help downloading and running models
Accession:64261
A common problem in tasks involving the integration of spatial information from multiple senses, or in sensorimotor coordination, is that different modalities represent space in different frames of reference. Coordinate transformations between different reference frames are therefore required. One way to achieve this relies on the encoding of spatial information using population codes. The set of network responses to stimuli in different locations (tuning curves) constitute a basis set of functions which can be combined linearly through weighted synaptic connections in order to approximate non-linear transformations of the input variables. The question then arises how the appropriate synaptic connectivity is obtained. This model shows that a network of spiking neurons can learn the coordinate transformation from one frame of reference to another, with connectivity that develops continuously in an unsupervised manner, based only on the correlations available in the environment, and with a biologically-realistic plasticity mechanism (spike timing-dependent plasticity).
Reference:
1 . Davison AP, Fr├ęgnac Y (2006) Learning cross-modal spatial transformations through spike timing-dependent plasticity. J Neurosci 26:5604-15 [PubMed]
Model Information (Click on a link to find other models with that property)
Model Type: Realistic Network;
Brain Region(s)/Organism: Generic;
Cell Type(s):
Channel(s):
Gap Junctions:
Receptor(s): GabaA; AMPA;
Gene(s):
Transmitter(s):
Simulation Environment: NEURON;
Model Concept(s): Synaptic Plasticity; Long-term Synaptic Plasticity; Unsupervised Learning; STDP;
Implementer(s): Davison, Andrew [Andrew.Davison at iaf.cnrs-gif.fr];
Search NeuronDB for information about:  GabaA; AMPA;
// Template for a network cell based on the IntFire4 class
// for use with the NetLayer and LayerConn classes.

// SYNTAX:
// if4nc = new IntFire4nc(paramvec)
// 
// paramvec is a 4-element vector containing the values of taum, taue, taui1
// and taui2.

// syn
//   reference to the actual IntFire4. Called syn as the IntFire4 artificial
//   cell acts as both cell and post-synaptic mechanism, i.e. it receives the
//   network events directly.

// spiketimes
//   Vector into which the times of spike in the IntFire4 are recorded.

// set_background(weight)
// set_background(weight,rate)
// set_background(weight,rate,start)
// set_background(weight,rate,start,noise)
// set_background(weight,rate,start,noise,number)
//  All except the first argument optional. Sets parameters of the built-in
//  NetStim. The NetStim is created the first time this procedure is called.
//  rate is in spikes/second, i.e. netstim.interval = 1000.0/rate

// memb_init() 
// memb_init(m_init)
// With no arguments, sets the initial value of m to the stored value
// (default 0). With one argument, changes the stored value and sets
// the initial value to this.

// source
// For external objects to query the source of spike events. For IntFire4
// it is just a reference to syn.

// Andrew Davison, UNIC, CNRS, Aug 2003


begintemplate IntFire4nc
  public syn, spiketimes, set_background
  public memb_init, source, record
  public bg, bgconn // for debugging
  
  objref syn, spiketimes, bg, bgconn
  objref rec, source, nil, fih, this
  
  proc init() {
    syn = new IntFire4(0.5)
    syn.taum = $o1.x[0]
    syn.taue = $o1.x[1]
    syn.taui1 = $o1.x[2]
    syn.taui2 = $o1.x[3]
    source = syn
    spiketimes = new Vector()
    //rec = new NetCon(syn,nil)
    //rec.record(spiketimes)
    fih = new FInitializeHandler("memb_init()",this)
    m_init = 0
  }
  
  proc record() {
    if ($1 == 1) {
      rec = new NetCon(syn,nil)
      rec.record(spiketimes)
    } else {
      objref rec
    }
  }
  
  proc memb_init() {
    if (numarg() == 1) {
      m_init = $1
    }
    syn.m = m_init
  }
  
  proc set_background() { local na
    if (!object_id(bg)) { // if the background stimulus does not exist, create it
      bg = new NetStim(0.5)
      bgconn = new NetCon(bg,syn,1,0,0.0)
    }
    bgconn.weight = $1
    na = numarg()
    if (na > 1) {
      if ($2 > 1e-9) {
	bg.interval = 1000.0/$2
      } else {
	bg.interval = 1e12
      }
      if (na > 2) {
	bg.start = $3
	if (na > 3) {
	  bg.noise = $4
	  if (na > 4) {
	    bg.number = $5
	  }
	}
      }
    }      
  }
  
endtemplate IntFire4nc

Loading data, please wait...