Biomedical Engineering Reference
In-Depth Information
the x axis are bins in which spikes occurred for neuron 1, and
y axis are bins in which spikes for neuron y.
3. Over all the trials in the behavioral session, this Cartesian
matrix can be summed to generate a
joint-peristimulus time
matrix
. This matrix is referred to as the
raw matrix
.
4. It is important to correct for simple modulations in firing
rate; to make this correction, perform steps 1-3 for perievent
matrices in which trial orders are shuffled (preferred) or the
trials are shifted by one trial. The matrix from these shuffled
data is the
shift predicted matrix
.
5. Subtract the shift-predicted matrix from the raw matrix.
6. To compare between neurons, normalization is required. By
dividing by the product of standard deviations of the peristim-
ulus time histograms of each neuron, one can normalize the
JPST matrix to units of correlation.
7. The two diagonals provide interesting data. The 45-degree
diagonal provides a measure of time-varying correlation (also
referred to as a correlation time histogram, or CTH), while
the perpendicular diagonal (135 degrees) provides a measure
of cross-correlation.
This analysis is a standard technique and is readily performed
by existing software packages (NeuroExplorer; Nex Technologies,
Littleton, MA; also available from http://mulab.physiol.upenn.
edu/) and discussed in detail in the literature (
Fig. 7.5A and B
)
(27)
. Here, we provide a brief sample of code to calculate the raw
matrix:
trials = 50; bins = 100; % 50 trials, 100 bins
data1_binned = round(rand(trials, bins)); %sample
random data
data2_binned = round(rand(trials, bins)); %sample
random data
rawMatrix = zeros(bins, bins); %% preallocate matrices;
data1_hist = zeros(1, bins); data2_hist = data1_hist;
%% for loop to increment through trials
for i = 1:trials;
x = find(data1_binned(i,:)>0); % find spikes
y = find(data2_binned(i,:)>0); % find spikes
% find matches and sum the spike matrix
for j = 1:length(x); for k = 1:length(y);
rawMatrix(x(j),y(k)) = rawMatrix(x(j),y(k))...
+ data1_binned(i,x(j))+data2_binned(i,y(k))-1 ;
end; end;
% increments the PSTH
data1_hist(x) = data1_hist(x) + data1_binned(i,
x);
data2_hist(y) = data2_hist(y) + data2_binned(i, y);
end
imagesc(rawMatrix); axis xy;
