Use of additional input, which is called bias, of neuron improves properties of the neuron. It allows moving the threshold of activation function. Use of bias increases calculations, because the additional weight have to be determined. That was presented for one and two-dimensional neurons. In the event of neurons with more inputs the situation is similar, however drawing geometrical interpretation of activation function and results of normalizatios would be impossible. For example let's consider 4-input neuron. Normalization of input vectors would be equal to move these points to the sphere in 4 dimensions, which is not possible to illustration. The only way to present this problem would be a 3D animation.