SVM: The maximum of posterior probabilities greater than the posterior probability of the label

I use fitcecoc to classify data and I calculate the posterior probabilities. I expect that the posterior probability of the predicted class label will be the maximum of the probabilities, but sometimes it doesn't.

classifier = fitcecoc(x, y, 'FitPosterior',1);
[pred,~,~,posterior] = predict(classifier, test);

For example if the labels goes from 0 to 9:

predictedClassLabel = 1;
postProb = [0.0000    0.3943    0.5000    0.0000    0.1029    0.0000    0.0001    0.0001    0.0004    0.0022];

Is it normal? What is the reason of it?