The notion of distance is a key factor for the organization of the map because it allows to compare the input signal and the "prototype" signal .However, the euclidian distance is not adapted to compare two pictures. In the following examples, we compare an image  with two different images. The results illustrate the necessity to find another criteria to compute the similarity of two patterns.

 

In order to solve this problem we use our own distance which results are shown in the following picture.

The inputs of the cortical map

Unlike the Kohonen map, each neuron of the cortical map (in blue) receives a different input . A neuron input comes from a restricted area of the retina (in green). To preserve topological property inherent to self organizing map, the inputs of two neighboring neurons are kept similar by a partial recovering.

The emergence of neurons

The neurons with high activity (i.e. neurons which "prototype" signal is similar to input signal) are the most relevant during the learning phase. An competitive and iterative algorithm is used in order to extract neurons that are locally relevant. We define local interactions (typically Mexican hat function) that a neuron creates in its neighborhood.

In the following picture, a high activity neuron is represented by a white square, a low activity neuron by a black one.

t=0 the activity is linked to the similarity between the input pattern and the prototype pattern of a neuron.	t=3 two neurons with hign activity emerge.	t=6 t=9 due to local interactions, neurons of the neighborhood benefits from the activity of the two local winners.