diff --git a/programming/exercises/psths.m b/programming/exercises/psths.m
index 2decc04..f9f21ac 100644
--- a/programming/exercises/psths.m
+++ b/programming/exercises/psths.m
@@ -1,16 +1,15 @@
 load('ampullary.mat')
+load('electroreceptor_stimulus.mat')
 sample_rate = 20000; % Hz
-max_time = 0;
-for i = 1:size(times,2)
-    max_time = max([max_time, max(times{i})]);
-end
-%% create PSTH on the basis of the interspike intervals
+max_time = 10;
+
+%% create instantaneous firing rate on the basis of the interspike intervals
 t = times{1};
 firing_rate = [0 1./diff(t)];
 start = 1;
-resp = zeros(1, round(max_time * sample_rate));
+response = zeros(1, round(max_time * sample_rate));
 for i = 1:length(t)
-    resp(1,start:round(t(i) * sample_rate)) = firing_rate(i);
+    response(1,start:round(t(i) * sample_rate)) = firing_rate(i);
     start = round(t(i) * sample_rate);
 end
 fig = figure();
@@ -18,7 +17,7 @@ set(gcf, 'PaperUnits', 'centimeters');
 set(gcf, 'PaperSize', [11.7 9.0]);
 set(gcf, 'PaperPosition',[0.0 0.0 11.7 9.0]);
 set(gcf,'Color', 'white')
-plot((1/sample_rate:1/sample_rate:max_time), resp)
+plot((1/sample_rate:1/sample_rate:max_time), response)
 xlabel('time [s]')
 ylabel('firing rate [Hz]')
 ylim([0 300])
@@ -26,6 +25,7 @@ xlim([0 1])
 title('instanataneous firing rate')
 saveas(fig, 'isi.pdf','pdf')
 
+
 %% create PSTH using the binning method
 bin_width = 0.0125; % s
 edges = 0:bin_width:max_time;
@@ -39,22 +39,33 @@ for i = 1:size(times,2)
         firing_rate = firing_rate + (n / bin_width / size(times,2));
     end
 end
+
+response = zeros(1, round(max_time * sample_rate));
+start_index = 1;
+for i = 1:length(edges)
+    end_index = round(edges(i) * sample_rate);
+    response(start_index:end_index) = firing_rate(i);
+    start_index = end_index +1; 
+end
+
 fig = figure();
 set(gcf, 'PaperUnits', 'centimeters');
 set(gcf, 'PaperSize', [11.7 9.0]);
 set(gcf, 'PaperPosition',[0.0 0.0 11.7 9.0]);
 set(gcf,'Color', 'white')
-plot(time, firing_rate)
+plot(0:1/sample_rate:max_time-1/sample_rate, response)
 ylim([0 300])
 xlim([0 1])
 xlabel('time [s]')
 ylabel('firing rate [Hz]')
 title('binning method')
 saveas(fig, 'binning.pdf', 'pdf')
+
+
 %% create PSTH using the kernel-convolution method
 kernel_width = 0.0125; %s
 binary_spikes = zeros(size(times,2), round(max_time*sample_rate));
-resps = zeros(size(binary_spikes));
+responses = zeros(size(binary_spikes));
 window = hann(kernel_width*sample_rate,'symmetric');
 window = window/sum(window);
 
@@ -65,16 +76,25 @@ for i = 1:size(times,2)
       temp(1) = 1;
     end
     binary_spikes(i, temp) = 1;
-    resps(i,:) = conv(binary_spikes(i,:), window, 'same')*sample_rate;
+    responses(i,:) = conv(binary_spikes(i,:), window, 'same')*sample_rate;
 end
 fig = figure();
 set(gcf, 'PaperUnits', 'centimeters');
 set(gcf, 'PaperSize', [11.7 9.0]);
 set(gcf, 'PaperPosition',[0.0 0.0 11.7 9.0]);
 set(gcf,'Color', 'white')
-plot((1/sample_rate:1/sample_rate:max_time), mean(resps,1))
+
+subplot(2,1,1)
+plot((1/sample_rate:1/sample_rate:max_time), mean(responses,1))
 ylim([0 300])
 xlim([0 1])
 xlabel('time [s]')
 ylabel('firing rate [Hz]')
 title('convolution method')
+
+subplot(2,1,2)
+plot(stimulus_strong(:,1), stimulus_strong(:,2))
+ylim([-1 1])
+xlim([0 1])
+xlabel('time [s]')
+ylabel('stimulus intensity [arb. units]')
diff --git a/projects/project_vector_strength/export_data.py b/projects/project_vector_strength/export_data.py
index 2cf08bf..4984fdc 100644
--- a/projects/project_vector_strength/export_data.py
+++ b/projects/project_vector_strength/export_data.py
@@ -76,7 +76,7 @@ def load_spike_data(cell, max_time):
     if data[0][2][-1]/1000 < max_time:
         return None
     else:
-        return data[0][2][data[0][2] <= max_time*1000]
+        return data[0][2][data[0][2] <= max_time*1000]/1000
 
 
 if __name__ == '__main__':
@@ -103,12 +103,11 @@ if __name__ == '__main__':
             index += 1
             continue
         spikes = load_spike_data(cell, 20)
-       
         if spikes is None:
             print 'not enough baseline spikes'
             index += 1
             continue
-        eod_data = load_eod_data(cell, 20, 200000)
+        eod_data = load_eod_data(cell, 20, 20000)
         name = cell.split('/')[-1] + '_baseline.mat'
         sp.savemat(name, {'spike_times':spikes, 'eod':eod_data})
         found += 1
diff --git a/statistics/matlab/my_gradient_descent.m b/statistics/matlab/my_gradient_descent.m
index 264362b..4ca3e84 100644
--- a/statistics/matlab/my_gradient_descent.m
+++ b/statistics/matlab/my_gradient_descent.m
@@ -38,7 +38,7 @@ plot(x,y,'ok');
 %%
 
 t = linspace(-5,5,100);
-param0 = [0,0];
+param0 = [1,1];
 step = 0.01;
 param = param0;