"Our visual system can effortlessly recognize hundreds of thousands of objects in spite of tremendous variation in their appearance, resulting, for instance, from changes in object position and pose. Achieving such an invariant representation of the visual world is an extremely challenging computational problem that even the most advanced artificial vision systems are not fully able to solve. This is why understanding the neuronal mechanisms underlying object vision is one of the major challenges of systems neuroscience and a crucial step towards developing artificial vision systems and visual prostheses.Little is known yet about how the brain develops and maintains invariant object representations. The leading theory is that visual neurons exploit the spatiotemporal continuity of visual experience (i.e., the natural tendency of different object views to occur nearby in time) to learn to produce similar responses for temporally contiguous stimuli, so as to factorize object identity from other variables (such as position, size, etc.). This Unsupervised Temporal Learning (UTL) strategy has been instantiated in a number of computational frameworks, but its empirical investigation has received little attention. My proposal will use the visual system of the rat to address key questions about the nature of UTL and other learning theories, such as their impact on recognition behavior and object representations at both single-neuron and population level, and their role during early postnatal development. This will be achieved through a highly multidisciplinary approach, including high-throughput behavioral testing, in vivo neuronal recordings, immediate-early gene labeling, controlled-rearing in virtual visual environments, and computational modeling. This will lead to ground-breaking insights into the learning principles that sculpt the cortical representations of visual objects through unsupervised exposure to the spatiotemporal statistics of visual experience."