Introduction

An understanding of the neuronal mechanisms underlying visual perception has played a major role in shaping our understanding of how the brain extracts information concerning the world. In turn, insight into the development of this process is enormously conditioned by existing theories of brain function.

Over the last century, anatomical and physiological studies have provided a description of the point-to-point connectivity of the visual system where neighboring relations in the retina are conserved in their central projections. Hence, the primary visual cortex has been shown to contain a retinotopic map of the visual world in which each point is represented by neurons with specialized receptive fields that encode basic visual features. Initially, the primary visual cortex was thought to function as the cortical retina before relaying its activity to additional areas in the surrounding association cortex. It was in the association cortex that the important but highly mysterious business of seeing was thought to take place. Increasingly, the system came to be construed as being hierarchically organized, and successive levels were thought to subserve progressively higher functions. Essentially, the visual system was seen to be a highly passive system, whereby information in the retinal image was extracted in central structures, as reflected in the changing receptive field organization at successive levels (Hubel, 1995). The understanding of corticogenesis, which evolved in parallel, was that the early development of the cortex was dictated by the peripheral sense organ (Van der Loos and Woolsey, 1973), whereas later stages were dictated by the sensory experience of the animal (Blakemore and Cooper, 1970).

Recently, our understanding of the neuronal mechanisms underlying vision has shifted from a passive role of analysis of the retinal image to one of inference and the construction of constancy. Increasingly, the visual system is seen as a dynamic one where the individual stations are involved together in a computational process aimed at determining the probabilities of feature constellations in the visual environment (Scannell and Young, 2002; Young, 2000; Young, 2001; Zeki, 1993). In some ways, this modern synthesis has much to do with the Platonic understanding of the brain, in which the sensory impressions were compared to ideas of the world. In parallel to the release of central states from the dictatorship of the sensory periphery, developmental biologists are increasingly detecting the intrinsic constraints, largely of a molecular nature, which determine early neuronal development of higher levels of the visual system (Rakic, 1988).

One can argue, therefore, that our understanding of the development of the brain being environmentally driven or, alternatively, determined by internal constraints has largely been influenced by theories of brain function. The intrinsic and extrinsic control of cortical development has been epitomized in protomap and protocortex theories (O'Leary, 1989; Rakic, 1988). Although these theories have been considered by some to be antagonistic, it is becoming increasingly clear that normal development of the cortex involves a synthesis of both intrinsic and extrinsic control (Yuste and Sur, 1999).

Early studies of cortical connectivity tended to be dominated by a view that immature connections were imprecise and that mature patterns of connections were largely the consequence of pruning of exuberant, early-formed connections (Innocenti et al., 1977). More recently, it has become clear that, although some early-formed connections undergo pruning, others exhibit precise connectional features from the onset of their formation (Dehay et al., 1988a). Distinguishing between these two possibilities is important because different contributions of each might be expected to underlie the development of very different functions. In this chapter, the authors shall attempt to give an up-to-date account of where development of the connectivity of the visual cortex is thought to be driven by activity in the periphery and where connectivity is prespecified, presumably by molecular cues that lay down the basic structure of the system.

It is becoming increasingly evident that the physiology of the visual cortex can be interpreted usefully in terms of feedforward and feedback mechanisms underlying the hierarchical organization of the cortex (Felleman and Van Essen, 1991; Lamme and Roelfsema, 2000; Shao and Burkhalter, 1996). This approach provides a conceptual framework for interpreting the feedforward input from the thalamus to the cortex in terms of the recurrent excitation provided by the intracortical and intercortical processing (Douglas et al., 1995; Shao and Burkhalter, 1996). This approach turns out to be highly innovative because instead of viewing the role and development of geniculostriate and corticocortical connectivity as separate entities, it shows that these two sets of connections, in fact, share key features. Hence, the development of feedforward projections from the thalamus to the cortex exhibits common features with the development of feedforward corticocortical connections; moreover, there are fundamentally different developmental mechanisms operant in the feedback and feedforward pathway.