Using Spatio-Temporal Correlations to Learn 1 Topographic Maps for Invariant Object Recognition 2

16 The retinal image of visual objects can vary drastically with changes of viewing angle. 17 Nevertheless, our visual system is capable of recognizing objects fairly invariant of viewing 18 angle. Under natural viewing conditions, different views of the same object tend to occur in 19 temporal proximity, thereby generating temporal correlations in the sequence of retinal 20 images. Such spatial and temporal stimulus correlations can be exploited for learning 21 invariant representations. We propose a biologically plausible mechanism that implements 22 this learning strategy using the principle of self-organizing maps. We developed a network of 23 spiking neurons that uses spatio-temporal correlations in the inputs to map different views of 24 objects onto a topographic representation. After learning, different views of the same object 25 are represented in a connected neighborhood of neurons. Model neurons of a higher 26 processing area which receive unspecific input from a local neighborhood in the map show 27 view-invariant selectivities for visual objects. The findings suggest a functional relevance of 28 cortical topographic maps. 29