What Roles Do Individual Neurons Play in Higher Visual Integration?

Central integration of visual information in mammals and other vertebrates involves both a hierarchical organization and a parallel organization. That is, several parallel pathways each consist of a hierarchy of projections to higher visual centers through multiple neurons and synapses. Three classes of visual pathways project from the retina to the lateral geniculate nucleus (LGN), each consisting of different kinds of matched retinal ganglion cells and LGN cells that differ in size and in response properties. In primates these classes have been termed magnocellular (large cells), parvocellular (smaller, medium-sized cells), and koniocellular (still smaller cells). Box Extension 14.3 describes how these parallel paths convey different sorts of visual information, and asks how individual neurons and ensembles of neurons process visual information in the brain.

Parallel visual pathways are probably a general feature of mammalian visual systems, but have been most studied in cats and primates. The parallel pathways in cats are termed Y-, X-, and W-type, probably corresponding to the magno-, parvo, and koniocellular classes in primates. Neurons of the three pathways carry somewhat different visual information. The medium-sized parvocellular cells are more responsive to the fine detail of stationary visual patterns. The large magnocellular cells are more rapidly conducting and more responsive to stimulus changes and movements. The small koniocellular cells have complex and varied receptive-field properties that do not concern us here. The separation of magnocellular, parvocellular, and koniocellular pathways is preserved in the lateral geniculate nucleus, and each of the three classes of pathways projects to a different depth in the primary visual cortex. Studies indicate that cortical simple cells receive primarily parvocellular input, whereas magnocellular input predominates for complex cells. Central visual pathways thus have elements of both parallel and hierarchical organization, the details of which are complicated and continue to elicit controversy.

The ways in which CNS neurons function in visual perception are also not clear. Simple and complex cells can be envisioned as contrast detectors, responsive to line segments and contrast edges at particular orientations. We may see the world as a sort of line drawing composed of the activities of many such contrast detectors. There are several theories about how higher visual integration areas process this information; one theory, called the feature detector model, suggests that higher integrating neurons are “tuned” to respond to more restricted features of the visual world, such as faces, letters, and so forth. Critics argue that ensembles of neurons, rather than individual neurons, are likely to make such discriminations, and so individual neurons are unlikely to have such strict requirements. Despite these criticisms, single-neuron recordings in several mammalian species reveal neurons (in inferior temporal cortex) that respond specifically to faces, and in some humans, to faces of specific individuals such as Jennifer Aniston or Halle Berry! Thus there is some evidence that individual neurons can act as feature detectors, but there remains a gulf between our understanding of the physiology of visual neurons and our questions about mechanisms of visual perception.

References

Kreiman, G. 2007. Single unit approaches to human vision and memory. Curr. Opin. Neurobiol. 17: 471–475.

Quironga, R. Q., L. Reddy, G. Kreiman, C. Koch, and I. Fried. 2005. Invariant visual representation by single neurons in the human brain. Nature 435:1102–1107.

Schiller, P. 2010. Parallel information processing channels created in the retina. Proc. Natl. Acad. Sci. U.S.A. 107: 17087–17094.