Methods for classifying shapes of receptive fields

In visual cortex, it has been shown that cells that are ideal local edge detectors seem to coexist with other types, such as cells with non-oriented receptive fields and cells with narrower spatial frequency tuning [2]. This finding challenges the traditional notion that simple cells are just local edge detectors [3]. In signal processing it is common to use a mixture of dictionaries to provide coding efficiency [4]. Each dictionary is made up of variations of a single shape primitive that is repeated under transformations, such as in different positions and sizes. Recently it was demonstrated in a model of visual cortex that different classes of experimentally found receptive fields (oriented and nonoriented receptive fields) can be learned from natural images in an unsupervised fashion [1].