The hierarchical processing of the ventral visual pathways was revealed in Hubel and Weisel’s seminal 1962 experiments with cats’ eyes. They linked the retina to the cortex and exposed the orientation selectivity of neurons in the V1 region of the visual cortex. Since Hubel and Weisel’s study, a lot of research has been conducted investigating the theory that the ventral visual stream hierarchically processes increasingly complex feature representations – a process that peaks in feature complexity at the inferior temporal cortex.

According to Yamins et al., the inferior temporal cortex  can support “invariant object categorization” for a broad spectrum of tasks. This hierarchical processing is analogous to the series of stacked layers that form modern neural networks. In the case of convolutional neural networks (CNNs), feature maps are created through convolutions and sets of filters, which are activated by different features of varying levels of complexity. Yamins et al., suggest that the later layers in a CNN selectively process specific objects while becoming more tolerant to natural variability and noise in the image – the deeper CNN networks “correspond intuitively to architecturally specialized sub-regions like those observed in the ventral visual stream.”