Starting from unoriented connections, neurons in the RF-LISSOM model develop oriented receptive fields and patterned lateral connections cooperatively and simultaneously. This input-driven self-organization represents the salient features of the training examples: the afferent connections develop feature detectors to distinguish between the inputs seen, and the inhibitory connections represent the long-range activity correlations between feature-selective cells. During visual processing, this information is used to eliminate redundant information, and enhance the selectivity of cortical cells. The self-organized map forms an accurate representation of the orientation and position of the input lines, suitable for use at higher levels and for further processing. If these self-organizing processes remain active even in the adult animal, they will continually act to adapt cortical response properties to match the visual environment. As discussed in the next chapter, this can result in phenomena such as tilt illusions and aftereffects.