Models of activity-dependent neural development.

Abstract

What makes a useful model of neural development? One important contribution of modeling is to demonstrate that proposed biological mechanisms can be sufficient to account for experimental results. The Von der Malsburg model is a classic example. But such demonstrations alone do not provide tools to experimentally distinguish one mechanism from another. To draw such distinctions, the connection between measurable biological quantities and developmental outcomes must be established. Perhaps the most important task for the future of developmental modeling is to deepen the connection between theory and experiment. Experimentally, this requires detailed and difficult measurements or experimental perturbations of the correlations among inputs and the intracortical connectivity existing during development. Simultaneous measurement of the maps of spatial phase and orientation of mature simple cells will provide important information for the understanding of orientation column development. Theoretically, the number of open problems is enormous. How will inclusion of additional plasticity mechanisms, such as sprouting and retraction of synapses or plasticity of intracortical connections, alter the analytical understanding thus far achieved? What precisely determines the width of orientation columns in the model presented here? Can the relationship between ocular dominance and orientation columns be understood from developmental rules in a testable way? The existing framework may be extended to a three-dimensional cortex and to more complex models of intracortical connectivity. It may also be applied to other developmental phenomena including the development of lamination in the LGN (Shatz and Stryker, 1988; Hahm et al., 1991), the formation of visual maps in experimentally altered auditory cortex (Roe et al., 1990, 1992), and the mapping of visual and auditory maps in the optic tectum (Knudsen and Brainard, 1991; Brainard and Knudsen, 1993). For each system the goal is to develop testable predictions as to the patterns of activity and connectivity that could or could not lead to the results observed given a proposed mechanism of plasticity. Incorporation of deeper levels of biophysical realism will extend, deepen, and perhaps fundamentally alter the framework presented here. An important goal for the future will be to understand the computational and functional significance of developmental rules. Activity-dependent, competitive mechanisms of synaptic plasticity appear to play an important role in many processes of late neural development, where an initially rough connectivity pattern refines to a precise, mature pattern. A prominent example is the formation of ocular dominance columns in the visual cortex of many mammals. These processes may be modeled at several levels.(ABSTRACT TRUNCATED AT 400 WORDS)