A model for the formation of ocular dominance stripes

The paper describes a model of competition that explains the formation of the ocular dominance stripes found in layer IVc of cat and monkey visual cortex. The main proposal is that synapses exert effects on the growth of other synapses, and that these effects extend over distances of at least 600 μm and vary in magnitude and sign within this distance. Interactions between like type synapses are assumed to be stimulating for distances up to about 200 μm, and inhibitory for distances of 200-600 μm. The reverse is true of interactions between synapses of opposite eye type, where the effects are inhibitory for distances up to about 200 μm and stimulating for longer ones. The interactions are assumed to be circularly symmetric. Growth of, for example, right eye synapses at one point will therefore (a) encourage local growth of right eye synapses and inhibit local growth of left eye synapses and (b) encourage growth of left eye synapses and inhibit growth of right eye synapses in an annular ring surrounding the point of initial increase. At the start of development, right and left eye synapses are assumed to be intermixed randomly within layer IVc. Computer simulations show that a wide variety of conditions incorporating these assumptions will lead to the formation of stripe patterns. These reproduce many of the morphological features of monkey ocular dominance stripes, including Y- and Н-type branches and terminations, the tendency for stripes to run at right angles into the boundaries of the pattern, and to narrow at branch points. The model can explain the effects of monocular deprivation on stripe morphology if it is assumed that the effectiveness of deprived eye synapses in determining rates of growth locally is reduced. The existence of a critical period for these effects can be explained if it is assumed that lateral growth of terminals does not occur, and that some factor such as a limited availability of postsynaptic sites decreases the rate of growth of synapses as their density approaches a maximum. The model can be generalized to account for pattern formation in other systems, such as zebra or mackerel skin, where similar striped patterns occur. In this context, the simplest model based on diffusion that will produce a pattern of stripes requires that one cell type should secrete two substances, one of which stimulates growth or differentiation of its parent cell type and has a low rate of diffusion or is rapidly inactivated, and another that inhibits growth or differentiation and either has a higher rate of diffusion or is less rapidly inactivated. A preliminary account of some of these results has appeared elsewhere (Swindale 1979).