Functional Models of Receptive Fields

Receptive fields are crucial elements in the brain’s processing of information. They filter out meaningful components from the stream of incoming data and define higher order features. They allow for the storage and further processing of such features both in individual sensory modalities and in multimodal integration. On the output side, motor fields similarly support the generation of patterns of coordinated movement. In this chapter, we study specific examples from the visual system that are paradigmatic for receptive fields also in other parts of the brain. Center-surround processing in the retina and lateral geniculate nucleus (LGN) enhance intensity and color contrast and remove irrelevant information from the images. The standard model here is based on Gaussians and convolution. Neurons in the visual cortex are characterized by their selectivity for edge orientation that is modeled by Gabor functions, convolutions, and point nonlinearities. As an example of a more derived feature, we study the detection of visual motion; it is based on spatiotemporal orientation selective receptive fields with a squaring nonlinearity.