Across Columns Horizontal

As noted earlier, a strong historical bias can be found in favor ofthe description of modular circuits respecting the laminar organization and organized along the depth (vertical) dimension ofthe gray matter. Similarly, at the functional level, a strong bias can be noted in the elucidation of feedforward sequential streams in processing. An analogy can, however, be made with computers and the problem of minimization of the wire length (Peters and Kaiserman-Abramof 1970; Chklovskii et al. 2002; Mitchison 1991; Mead 1989). To avoid a complete connectivity pattern and reduce the physical size of the global system, a hardware configuration often used is to stack-over interface bus-cards, each dedicated to input only or output only. In that respect, the vertical dimension does not carry processing, and only the lateral dimension is used to wire the computing architecture. This analogy suggests that, rather than looking for vertically organized columns, one should concentrate on the pattern of horizontal connectivity (Figure 19.3e, f) to characterize the functional specialization of the cortical network under consideration.

Excitatory horizontal connectivity from visual to prefrontal cortex exhibits a patchy layout: the interpatch spacing is roughly double that of the patch diameter. This patchy architecture has been studied extensively in primary visual cortex and shows a strong correlation between anatomy and function. Cells that are connected through long excitatory links tend to belong to columns with the same functional preference. Some authors, however, have moderated the impact of the principle "those alike tend to wire together." This schema neglects the spatial organization of axons and dendritic structure of the target cell. The matching of the composite size of the axonal terminal distribution from the presynaptic cells with that of the dendritic spread of the target cell could be the result of an optimization process maximizing the diversity of inputs collected by a given neuron (Malach 1992).

Several arguments can be listed to support the fit between the function and the anatomy of the horizontal network. During development in strabismic cats, anomalous horizontal connectivity links are formed between distant cortical territories corresponding to the same eye-dominance (Schmidt et al. 1997; Trachtenberg and Stryker 2001). Another example can be observed with a sensory substitution protocol imposing a rewiring of the input to the auditory thalamus at an early stage of development. If visual input is provided to auditory thalamus at that time, the auditory cortex develops a visual competence and an orientation preference map. The horizontal connectivity anatomy in the rewired A1 cortex resembles that of a control area V1, in terms of anteroposterior/ mediolateral biases, and not to that of a normal auditory cortex (Sharma et al. 2000). To progress in this direction, work must be undertaken to characterize the factors, linked with activity and the sensory code, that determine the number of patches, the extent of their distribution, their input distribution, and their output distribution, both during normal and abnormal development.

0 0

Post a comment