ABSTRACT
Almost all neuroscientists consider that understanding brain structure will aid understanding of brain function. This assumption is reproduced in studies at every level of the nervous system, and of every presumed functional subsystem. At the level of the whole cortex, for example, many researchers assume that information processing is closely determined by the inputs, the internal connectivity and computations, and the outputs of the network of areas and nuclei that make up the brain. Defining the connectivity of brain structures has
all of biology. However, this success has brought with it a problem that is very similar to that faced by some other biological disciplines: the quantity and complexity of connection data, and their dispersion through an extensive and idiosyncratic literature make it very difficult indeed to derive reliable conclusions about the information they collectively bear about the organisation of the system. An example of the scale of this problem is provided by noting that more than 14,000 individual reports of connections between different gross structures of the rat brain have been made in the last 20 years (Burns and Young, 2000). Data so numerous and complex provide excellent opportunities for the derivation of false conclusions if examined only informally, simply through the ease by which inconvenient data can be overlooked. Similarly, for the macaque visual system, V1 is known to be connected to more than 50 other structures (e.g. Young et al., 1995). More than 300 ipsilateral cortico-cortical connections have been described between at least 30 differentiable visual processing regions (e.g. Felleman and Van Essen, 1991). These connections, together with the connections that visual areas make with other cortical regions, constitute a cortical network defined by almost a thousand gross connections (e.g. Young, 1993, 1995). Furthermore, a plethora of callosal and other commissural connections link the two hemispheres; and the cortical visual systems stand upon a thalamo-cortical network of almost equal complexity (e.g. Scannell et al., 1999).
