A novel view on the visual cortex

Almost forty years ago, David H. Hubel and Torsten N. Wiesel won the Nobel Prize in Physiology or Medicine for their seminal work on the neuronal mechanisms of vision. One of their groundbreaking findings was that they identified orientation selective neurons in the primary visual cortex (V1) as an essential building block of visual perception.

In the visual cortex of cats, the researchers found neurons that prefer a specific orientation of a light bar over all the others. They concluded that these neurons could act as edge detectors. This property arises in the cortex for the first time; neurons in the retina and in the lateral geniculate nucleus (the relay station between the retina and V1), still respond equally to all orientations.

As we meanwhile know from applying modern imaging methods, orientation selective neurons are arranged on the cortical surface in an orderly fashion: They form what we call an orientation map. Neurons of similar selectivities are close to each other, with transitions between different orientations being generally smooth. However, these maps also exhibit specific singularities: spots where cells with all orientations lie next to each other (so-called pinwheel centers) and fractures (so-called non-linear zones). Many modeling studies, taking different theoretical perspectives, have tried to identify the mechanisms how these orientation selective neurons come into being, and how orientation maps with the observed properties arise from this.

In a new publication in the journal Biological Cybernetics, Sadra Sadeh and Stefan Rotter from the Bernstein Center Freiburg and the cluster of excellence BrainLinks-BrainTools have suggested a new model that uses biologically realistic connectivity, accounting for both statistics and geometry of neuronal projections from the lateral geniculate nucleus to the cortex. In their model, each cortical neuron is contacted by many neurons of the geniculate nucleus, which renders it sensitive to light stimuli in a large region of the visual field. In addition, connections that link the geniculate nucleus with the cortex are arranged on the cortical surface in a non-homogeneous fashion. Combining these facts, which were actually found in experiments, into a theoretical model produces orientation maps that fit data from neurobiological experiments better than previous models.

The model of the Freiburg scientists also provides an explanation for other properties of these orientation maps, in particular their inter-relations with the map of retinotopy (how the visual field is mapped onto the cortical surface) as well as with the map of ocular dominance (how cortical neurons differentially respond to both eyes). The next step is to quantify the properties of these maps, and to further test the predictions of the model.

 

Original publication:

Sadra Sadeh and Stefan Rotter (2013) Statistics and geometry of orientation selectivity in primary visual cortex. Biological Cybernetics, e-Publication ahead of print

Image caption:

(A) Assuming clustered projection patterns from thalamus to cortex (black circles) in the model, neurons exhibit orientation selectivity as a result of elongated receptive fields (shown for the sample neuron located on the white cross). Moreover, orientation selectivity is arranged in an orderly fashion, similar to the maps reported in cat and monkey striate cortex (B).