p = [-10 -5 0 5 10];

t = [0 0 1 1 1];

y = sim(net,p)

e = t-y

perf = mse(e)

In this above example, 'p' is input vector, 't' is the target vector, 'net' is the network model variable, 'y' is the neural network output and 'e' is difference between original target and neural network outputs.

mse(error_vector) - it returns the MSE value of neural network output.