The club-sandwich mystery

'THE whole organization of the human lateral geniculate nucleus (LGN) cries aloud that something is being segregated. The question before us here is simply: what?" It was with this question that the late Gordon Walls opened his engaging monograph on the LGN, the knee-shaped nucleus of the mammalian brain that lies in the visual pathway between the retina and the striate cortex'. A paper by Schiller, Logothetis and Charles on page 68 of this issue2 offers an answer to Walls's question. Medical students are still told that the LGN is a mere relay station. But to the neurobiologist the LGN is provocative in its exquisite organization. In Old World primates and in man, the lateral genicu-late consists of six layers, each eye providing the inputs to three layers (see figure). The two layers that lie on the inside of the geniculate 'knee' consist of large cells and hence are known as the magnocellular laminae'; their main reti-nal inputs are from the axons of the large cells that Perry and Cowey3 called Pa cells. The remaining four layers consist of small cells and so are known as the 'parvocellular laminae'; their main retina1 inputs are from the class of cells known as P(3 cells. If we number the six laminae from the inside of the knee outwards, then layers 1, 4 and 6 of each geniculate are drawn from the contralateral eye and layers 2, 3 and 5 from the ipsilateral. Each lamina forms a map of the contra-lateral half of the visual field, and, what is most remarkable, the six maps are in precise alignment4. Gordon Walls' likened the lateral geniculate to a club sandwich: the toothpick piercing the sandwich corresponds to a single direction in visual space. Why is the lateral geniculate stratified? How is the task of analysing the visual world distributed among the laminae (and their cortical projections)? Why does the developing visual system go to such trouble to align the six maps in the LGN? And why indeed does the LGN exist? These must be interdependent questions. But it is the second of them that has been most explicitly addressed, and three types of evidence have been offered. First, electrophysiological recordings have been made from the primate LGN while various stimuli were presented to the eye. When the stimulus varies only slowly in space (that is, it is of low spatial frequency), parvocellular units typically exhibit …