Sensorimotor cortex reinforcement learning of 2-joint virtual arm reaching (Neymotin et al. 2013)

 Download zip file   Auto-launch 
Help downloading and running models
Accession:150245
"... We developed a model of sensory and motor neocortex consisting of 704 spiking model-neurons. Sensory and motor populations included excitatory cells and two types of interneurons. Neurons were interconnected with AMPA/NMDA, and GABAA synapses. We trained our model using spike-timing-dependent reinforcement learning to control a 2-joint virtual arm to reach to a fixed target. ... "
Reference:
1 . Neymotin SA, Chadderdon GL, Kerr CC, Francis JT, Lytton WW (2013) Reinforcement learning of 2-joint virtual arm reaching in a computer model of sensorimotor cortex Neural Computation 25(12):3263-93 [PubMed]
Model Information (Click on a link to find other models with that property)
Model Type: Realistic Network;
Brain Region(s)/Organism:
Cell Type(s): Neocortex V1 pyramidal corticothalamic L6 cell; Neocortex U1 pyramidal intratelencephalic L2-5 cell; Neocortex V1 interneuron basket PV cell; Neocortex fast spiking (FS) interneuron; Neocortex spiking regular (RS) neuron; Neocortex spiking low threshold (LTS) neuron;
Channel(s):
Gap Junctions:
Receptor(s): GabaA; AMPA; NMDA;
Gene(s):
Transmitter(s): Gaba; Glutamate;
Simulation Environment: NEURON;
Model Concept(s): Synaptic Plasticity; Learning; Reinforcement Learning; STDP; Reward-modulated STDP; Sensory processing;
Implementer(s): Neymotin, Sam [samn at neurosim.downstate.edu]; Chadderdon, George [gchadder3 at gmail.com];
Search NeuronDB for information about:  Neocortex V1 pyramidal corticothalamic L6 cell; Neocortex V1 interneuron basket PV cell; Neocortex U1 pyramidal intratelencephalic L2-5 cell; GabaA; AMPA; NMDA; Gaba; Glutamate;
/
a2dmodeldb
readme.html
drspk.mod *
infot.mod *
intf6_.mod *
misc.mod *
nstim.mod *
stats.mod *
vecst.mod *
arm.hoc
basestdp.hoc
col.hoc
colors.hoc *
declist.hoc *
decmat.hoc *
decnqs.hoc *
decvec.hoc *
default.hoc *
drline.hoc *
filtutils.hoc *
geom.hoc
grvec.hoc *
hinton.hoc *
infot.hoc *
init.hoc
labels.hoc *
misc.h *
mosinit.hoc
network.hoc
nload.hoc
nqs.hoc *
nqsnet.hoc *
nrnoc.hoc *
params.hoc
python.hoc
pywrap.hoc *
run.hoc
samutils.hoc *
screenshot.png
sense.hoc *
setup.hoc *
simctrl.hoc *
stats.hoc *
stim.hoc
syncode.hoc *
trainedplast.nqs
units.hoc *
xgetargs.hoc *
                            
// $Id: hinton.hoc,v 1.46 2009/10/23 21:58:22 billl Exp $

// load_file("hinton.hoc")
load_file("colors.hoc")

//* decl
declare("nilsl",nil,"ghint",nil,"cmat",new Vector(),"sepvec",new Vector(),"sepvec1",new Vector())
declare("sepy",0,"yoff",0,"xsep",0,"ysep",0)
nilsl = new SectionList()
wdt=1
rot=1 // rotate so that the trace appears from lower L to upper R
scly=1
// func mod101 () { return($1%101) }
// cmat.resize(500*100)
// cmat.indgen()
// cmat.apply("mod101")
// plot2d(ghint,-90,50)
// draw2d(ghint,cmat,500)

printf("Hinton plot typical sequence: plot2d(ghint) wdt=0.1 draw2d(ghint,cmat,sqrt(cmat.size))\n")

//* plot2d(g[,min,max,ratio]); plot2d(g[,vec,ratio]); plot2d(g,ratio)
proc plot2d () { local ratio,width
  if (!isassigned($o1)) {
    $o1 = new PlotShape(nilsl,0)
    flush_list.append($o1)
  }
  c1($o1,63)
  ratio=1 width=500
  if (numarg()>=2) {
    if (argtype(2)==1) {
      $o1.scale($o2.min,$o2.max)
      if (numarg()>=3) ratio=$3
    } else {
      if (numarg()==2) ratio=$2
      if (numarg()>=3) $o1.scale($2,$3)
      if (numarg()>=4) ratio=$4
    }
  } else $o1.scale(0,100)
  $o1.view(0,100,0,100,100,50,width,ratio*500)
  $o1.erase_all
}

// drawppm(vec,file,[cols]) draw bitmap directly into file
proc drawppm () { local a localobj v1,v2,v3
  a=allocvecs(v1,v2,v3)
  if (numarg()>=3) cols=$3
  c1m(clm) // clm holds an nqs of color values
  v2.resize(clm.size(1))
  v2.rgb(clm.v,clm.v[1],clm.v[2],1)
  v1.copy($o1)
  v1.scale(0,63)
  v1.rnd() // indices into v2
  v3.resize(v1.size)
  v3.index(v2,v1)
  if (scly>1) {
    v1.transpose(v3,cols)
    v2.resample(v1,scly)
    v3.resize(v2.size)
    v3.transpose(v2,rows*scly)
  }
  tmpfile.wopen($s2)
  v3.ppmwr(tmpfile,cols)
  dealloc(a)
}

//** draw2d(g,mat,cols) draw matrix
proc he () { ghint.erase_all() }
// may want to have additional sepvecy in future -- for now only using with squares 
proc draw2d () { local i,j,k,ny,nx,xs,ys
  if (numarg()>=3) nx=$3 else nx=cols
  if (sepvec.size>0){ revec(sepvec1,0)
    if (xsep!=ysep) printf("WARNING sepvec1 won't work\n")
    j=0    
    for i=0,sepvec.size-2 sepvec1.append(sepvec.x[i]+(j+=xsep))
  }
  xs=ys=0
  ny=$o2.size/nx
  if (ny!=int(ny)) {printf("draw2dERR %g x %g ??\n",nx,ny) return}
  if (!rot) {
    for i=0,nx-1 { ys=0
      if (sepvec.size>0) {if (sepvec.contains(i)) xs+=xsep} else xs+=xsep
      for j=0,ny-1  { k=ny-1-j  // k is reverse of j
                   // ptr            xloc         yloc                               xsz    ysz
       if (sepvec.size>0) { if (sepvec.contains(j)) ys+=ysep } else ys+=ysep
       $o1.hinton(&$o2.x[j*nx+i],xs+(i+.5)*wdt,-ys+yoff+(k+.5)*wdt*scly+sepy*(k+.5), wdt, wdt*scly) 
      }
    }
  } else for i=0,nx-1 { xs=0
    if (sepvec.size>0) {if (sepvec.contains(i)) ys+=ysep} else ys+=ysep
    for j=0,ny-1 {
      if (sepvec.size>0) { if (sepvec.contains(j)) xs+=xsep } else xs+=xsep
      $o1.hinton(&$o2.x[j*nx+i],xs+(j+.5)*wdt,ys+yoff+(i+.5)*wdt*scly+sepy*(k+.5), wdt, wdt*scly) 
    }
  }
  $o1.size(0, i*wdt, yoff, yoff+j*wdt*scly)
  $o1.exec_menu("Shape Plot")
  $o1.exec_menu("Zoom in/out")
  $o1.flush
}

//** drawhalf(g,mat,cols) draws the lower half of correlation matrix
proc drawhalf () { local i,j,k,ny,nx,n localobj v1,g
  g=$o1 v1=$o2
  if (numarg()>=3) cols=nx=$3 else nx=cols
  ny=nx
  if ((cols-1)!=v1.size*2/cols) {printf("drawhalfERR %g,%g ??\n",nx,v1.size*2/nx+1) return}
  n=0
  for i=0,nx-2 for j=i+1,ny-1 {
    g.hinton(&v1.x[n], (j+.5)*wdt, yoff+(i+.5)*wdt*scly+sepy*(k+.5), wdt, wdt*scly) 
    n+=1
  }
  g.size(0, i*wdt, yoff, yoff+j*wdt*scly)
  g.exec_menu("Shape Plot")
  g.exec_menu("Zoom in/out")
  g.flush
}


//** hgg(color_vec, yloc_vec, xloc[, xwidth, flag]) flag means horizontal
// like gg() but draw vec vs vec as a hinton plot
proc hgg () { local xyloc,flag,max,ii,a,wdt,wflag
  if (numarg()==0) {
    print "hgg(color_vec, yloc_vec, xloc[, xwidth (vec or num), flag]) flag means horizontal"
    return }
  a=allocvecs(1)
  xyloc=$3
  flag=wflag=0
  wdt=$o2.max/12 // to put 12 in a row next to each other and get square
  if (numarg()>=4) if (argtype(4)==1) wflag=1 else wdt=$4
  if (numarg()>=5) flag=$5
  mso[a].deriv($o2,1,1) mso[a].append(mso[a].mean) mso[a].add(0.01*mso[a].mean)
  for ii=0,$o1.size-1 {
    if (wflag) wdt=$o4.x[ii]+1e-3 // should be non-zero width
    if (flag){
      ghint.hinton(&$o1.x[ii],$o2.x[ii]+mso[a].x[ii]/2, xyloc, mso[a].x[ii], wdt)
    } else  {         // color   xloc           yloc                  xsz       ysz
      ghint.hinton(&$o1.x[ii], xyloc, $o2.x[ii]+mso[a].x[ii]/2, wdt, mso[a].x[ii]) }
      // print $o1.x[ii], xyloc, $o2.x[ii]+mso[a].x[ii]/2, wdt, mso[a].x[ii]
  }
  dealloc(a)
}

proc showghint () {
  ghint.exec_menu("Shape Plot")
  ghint.exec_menu("View = plot")
  ghint.fastflush
  doEvents()
}

//** anim2d(g,mat,cols) draws slowly
proc anim2d () { local i,j,ny,nx
  $o1.erase_all ny=$o2.size/nx
  if (numarg()>=3) nx=$3 else nx=cols
  wdt=.01
  for i=0,nx-1 {
    for j=0,ny-1 $o1.hinton(&$o2.x[j*nx+i],(i+.5)*wdt,(j+.5)*wdt,wdt)
    $o1.flush
    doEvents()
  }
}

Loading data, please wait...