Model of memory linking through memory allocation (Kastellakis et al. 2016)

 Download zip file 
Help downloading and running models
Accession:206249
Here, we present a simplified, biophysically inspired network model that incorporates multiple plasticity processes and explains linking of information at three different levels: (a) learning of a single associative memory (b) rescuing of a weak memory when paired with a strong one and (c) linking of multiple memories across time. By dissecting synaptic from intrinsic plasticity and neuron-wide from dendritically restricted protein capture, the model reveals a simple, unifying principle: Linked memories share synaptic clusters within the dendrites of overlapping populations of neurons
Reference:
1 . Kastellakis G, Silva AJ, Poirazi P (2016) Linking Memories across Time via Neuronal and Dendritic Overlaps in Model Neurons with Active Dendrites. Cell Rep 17:1491-1504 [PubMed]
Citations  Citation Browser
Model Information (Click on a link to find other models with that property)
Model Type: Realistic Network;
Brain Region(s)/Organism:
Cell Type(s): Abstract integrate-and-fire leaky neuron with dendritic subunits;
Channel(s):
Gap Junctions:
Receptor(s):
Gene(s):
Transmitter(s):
Simulation Environment: C or C++ program; C or C++ program (web link to model);
Model Concept(s): Active Dendrites;
Implementer(s): Kastellakis, George [gkastel at gmail.com];
/
stdmodel
distributionPlot
exportfig
figs
mtrand
README
allgraphs.m
allrun.m
an_brtest.m
an_stats.m
anmulti.py
ansims.py
barwitherr.m *
btagstats.m *
CImg.h *
constructs.cpp
constructs.h
defaults.m
dir2.m *
getspikedata.m *
getsynstate.m *
getsynstate2.m *
graphs.m *
hist_percents.m *
hist_with_errs.m *
interact.m *
intexp_constructs.cpp
job_sims.sh
kurtos.m *
lamodel.cpp
LICENSE *
make_graphs.m *
Makefile *
matlab.mat *
mtest.py
mtrand.cpp *
mtrand.h *
multi.py
multistats.m *
nextplot.m *
pairstrong.m *
repeated.m *
rotateXLabels.m *
run_1.sh
run_2strong.sh
run_2weak.sh
run_3.sh
run_all.sh
run_brov.sh
run_brtest.sh
run_btag.sh
run_dir.sh
run_ep.sh
run_gp.sh
run_gp2.sh
run_mult.sh
run_Nsparse.sh
run_pairstrong.sh
run_rep.sh
run_sims.sh
run_sparse.sh
run_sparseS2.sh
runloc.sh
runmany.sh
S2sparse.m *
savefig.m *
scratch.m *
sensitivity.m *
stats.m *
stats.py *
stderr.m *
strong2.m *
strongstrong.m *
submit_lamodel.sh *
three.m *
trevrolls.m *
vis.py *
weastrong.m *
wxglmodel *
wxglmodel.cpp *
wxglmodel.h *
wxmodel.cpp *
wxmodel.h *
                            
defaults
nruns=10;

totfiring = zeros(nruns, 4);
totactive = zeros(nruns, 4);

coract = zeros(nruns, 9);

%CUTOFF=5; % Hz

npatterns=1;
ndays  = 4;

brweights = zeros(ndays, npyrs*nbranches, nruns);
nrnweights = zeros(ndays, npyrs, nruns);
branch_syns = zeros(ndays, npyrs*nbranches, nruns);
brsynratio= zeros(ndays, nruns);



for run=1:nruns
        for ncase=1:ndays
            
            sfn=sprintf('./data/%s_%d_%d/spikesperpattern.dat', CONDITION, ncase, run-1);
            spk = load( sfn);
            spk = spk(1, 1:npyrs)/(stimduration/1000);
            
            %pop = spk(spk>=CUTOFF);
            pop = spk(spk>=CUTOFF);
            totfiring(run, ncase) = mean(spk, 2);
            totactive(run, ncase) = sum(spk>CUTOFF,2)/npyrs;



            ff = sprintf('./data/%s_%d_%d/synstate.dat', CONDITION, ncase, run-1);
            ss = load(ff);

            for i=1:size(ss,1)
                bid=ss(i,2);
                nid=ss(i,3);
                srcid=ss(i,5);
                bstrength = ss(i,6);
                w=ss(i,7);
                if (srcid ==0 && bid <= npyrs*nbranches)
                    brweights( ncase, bid+1, run) = brweights(ncase, bid+1, run) + w;
                    brstrengths(ncase, bid+1)=bstrength;
                    nrnweights( ncase, nid+1,run) = nrnweights(ncase, nid+1,run) + w;
                end
                if (srcid ==0 && bid <= npyrs*nbranches &&  w > 0.7)
                    branch_syns(ncase, bid+1, run) = branch_syns(ncase, bid+1, run)+1;
                end
            end
            
            brsynratio(ncase,run) = sum(branch_syns(ncase,:, run)>3)/(nbranches*npyrs);
        end
end

close all
mf = mean(totfiring,1);
sf = std(totfiring,0,1)/sqrt(nruns);

mact = mean(totactive,1);
sact = std(totactive,0,1)/sqrt(nruns);

barwitherr(100.* sact(1,:,1), 100.* mact(1,:,1));
title('% coding neurons')
%ylabel('% Active neurons')
xlabel('Day')
ylim([0,80]);

export_fig(sprintf('./figs/%s_pops.pdf',CONDITION), '-transparent')

figure
barwitherr(sf(1,:,1), mf(1,:,1));

title('Average firing rate [Hz]')
%ylabel('Avg firing rate [Hz]')
%xlabel('Number of trainings')
%ylim([0,70]);
export_fig(sprintf('./figs/%s_rates.pdf',CONDITION), '-transparent')

figure
hs=hist(branch_syns(1, :), [0:8]);

bar(hs(:,2:end)/sum(hs(:)))
title('1st day')
ylim([0,0.2]);
export_fig(sprintf('./figs/%s_brsyns_1day.pdf',CONDITION), '-transparent')

figure
hs=hist(branch_syns(4, :), [0:8]);
bar(hs(:,2:end)/sum(hs(:)))
ylim([0,0.2]);
title('4th day')
export_fig(sprintf('./figs/%s_brsyns_4day.pdf',CONDITION), '-transparent')


figure
barwitherr(100.0*std(brsynratio,0,2)/sqrt(nruns), 100.0*mean(brsynratio,2))
title('Branches with >2 potentiated synapses')
ylabel('Percentage')
xlabel('Day')
ylim([0,16]);
export_fig(sprintf('./figs/%s_brsyns.pdf',CONDITION), '-transparent')


figure
aa = mean(branch_syns(:,:)');
ss = std(branch_syns(:,:)');
errorbar(ss, aa, 'o')
title('Potentiated synapses per branch');
ylabel('Number of synapses')
xlabel('Day')
ylim([0,6]);
export_fig(sprintf('./figs/%s_syn_per_branch.pdf',CONDITION), '-transparent')



figure
tweights = (squeeze(sum(nrnweights,2)));

barwitherr(std(tweights,0,2), mean(tweights,2))

%barwitherr(100.0*std(brsynratio,0,2)/sqrt(nruns), 100.0*mean(brsynratio,2))
title('Total Syn Weights')
ylabel('Total Syn Weight')
xlabel('Day')
ylim([0,12000]);
export_fig(sprintf('./figs/%s_tweights.pdf',CONDITION), '-transparent')


CONDITION
b1 = branch_syns(1,:);
b1 = b1(b1>0);
b4 = branch_syns(4,:);
b4 = b4(b4>0);
mean(b1)
std(b1)
mean(b4)
std(b4)
[h,p] = ttest2(b1, b4)