When we get input in the first column, we then compare it to the neuron in the next column and the next column after that.

We do this until we get to the output layer and find the best match. This system is similar to that of the human brain.