Spreading associative neural network recognizes the shape and position of an object simultaneously

Paying attention to the spatial recognition system in the brain (i.e. parietal cortex), we proposed a spreading associative neural network (SAN net) which had double spreading neural layers. We investigated whether the SAN net could represent multiple spatial frames of reference and multiple object shape information simultaneously. The SAN net not only recognized the shape of the object (e.g. human faces and Arabic numerals) in the two-dimensional space irrespective of its position, but also recognized its position irrespective of its shape in the input pattern at the same time. The nonlinear transformation using spatial double spreadings in the SAN net is crucial for the simultaneous recognition of both the shape and the spatial position of an object. The results showed that not only multiple spatial frames of reference but also multiple object shape information could be represented simultaneously by the SAN net.