Spiking-timing dependent plasticity (STDP) is a biologically observed rule of synaptic strengthening and weakening. This model implements an abstract, feed-forward STDP network which learns from binocular input, and converges to neuronal units which are strikingly similar to those observed in the primate visual cortex.
However, for this to occur, it is crucial that the input dataset is acquired using a geometry as similar to the primate visual system as possible. Departures from this geometry lead to results which qualitatively describe measurements made under anormal rearing conditions.
Reference : Chauhan, T., Masquelier, T., Montlibert, A., & Cottereau, B. (2018). Emergence of binocular disparity selectivity through Hebbian learning. The Journal of Neuroscience, 38(44), 9563–9578. [doi]